linking microbial community structure and function …
TRANSCRIPT
LINKING MICROBIAL COMMUNITY STRUCTURE AND FUNCTION WITH TROPICAL FOREST RECOVERY
By
A. Peyton Smith
A Dissertation in partial fulfillment of
the requirements for the degree of
Doctor of Philosophy
(Department of Soil Science)
at the
UNIVERSITY OF WISCONSIN-MADISON
2013
Date of final oral examination: 08/27/13 The dissertation is approved by the following members of the Final Oral Committee: Erika Marín-Spiotta, Assistant Professor, Geography Teri C. Balser, Professor, Soil and Water Science (University of Florida-Gainesville)
Phillip W. Barak, Professor, Soil Science Christopher J. Kucharik, Associate Professor, Agronomy
Matthew D. Ruark, Assistant Professor, Soil Science
© Copyright by A. Peyton Smith 2013
All Rights Reserved
i
TABLE OF CONTENTS ABSTRACT………………………………………………………………………..………ii
ACKNOWLEDGEMENTS…………………………………………………………..………iv
INTRODUCTION, OVERVIEW AND BACKGROUND:...………………………………………………………………….………1
Tables and Figures………...….…………………………….……….……..18 References………………………………………………….……….……...20
CHAPTER 1: Seasonal and successional changes in soil microbial community structure during
reforestation of a tropical post-agricultural landscape…..…...…………….27 Tables and Figures…………………………………………………...…….51 References……………………………………………………..…….……..63
CHAPTER 2: Microbial community composition rapidly responds to changes in aboveground
succession…………………………………………….………………..…..73 Tables and Figures…………………………………………………...….....85 References……………………………………………………………….....89
CHAPTER 3: Linking microbial ecology and soil organic matter aggregate stabilization with tropical land cover change…………………..……………………………..96 Tables and Figures……………………………………………….…….…120 References………………………………………………………....……...133
CHAPTER 4: Shifts in the functional capacity of soil microbial communities with tropical forest
regeneration on abandoned pastures……………………………………...141 Tables and Figures…………………………………………………….….163 References…………………………………………………..………….....175
CONCLUSION: Microbial succession, recovery, structure-function links………………181
ii
ABSTRACT
Soil microorganisms regulate fundamental biochemical processes in plant litter decomposition
and soil organic matter (SOM) transformations and are thus, important drivers for ecosystem
processes and biogeochemical cycles. Most studies largely ignore the role of microbial
communities in mediating the response of soil carbon (C) to land-use change, even though
microbes are central players in soil C dynamics. In order to predict how land cover change
affects belowground carbon storage, an understanding of how forest floor and soil microbial
communities respond to changes in vegetation, and the consequences for SOM formation and
stabilization, is fundamental. Using a well-replicated, long-term successional chronosequence
where data on aboveground plant communities and SOM dynamics have already been
established, I am investigating the effects of natural post-agricultural forest regeneration on
microbial communities and belowground C cycling in the subtropical wet forest life zone of
southeastern Puerto Rico. My primary objectives include to: (1) characterize microbial
community composition and activity during 90 years of forest recovery on former pastures, (2)
investigate links between microbial community structure, function and soil organic carbon
(SOC), and (3) identify direct links between microbial community composition and microbial
functional gene diversity.
Results show that land cover type, or forest successional stage, predicts microbial
community structure in this landscape. At the same time, microbial community structure and
activity varied by season and year stressing the importance of a multiple, temporal, sampling
strategy when investigating microbial community dynamics. Within a year following woody
biomass encroachment of an abandoned pasture, I detected a shift in the soil microbial
iii
community from a pasture-associated community to an early secondary forest community in one
of the replicate pasture sites. This data supports a direct link between aboveground and
belowground biotic community structures and highlights the importance of long-term repeated
sampling of microbial communities in dynamic ecosystems. Successional control over microbial
community composition with forest recovery has potential implications for nutrient cycling with
changes in vegetation cover.
This study also revealed the importance of mineral interactions in defining the
relationship between soil aggregates, microbial communities and SOM storage in highly
weathered tropical soils. The fungal –to-bacterial ratio decreased with diminishing aggregate
size, while the ratio between gram-positive and gram-negative bacteria increased in the silt and
clay fraction. Differences in microbial composition among soil aggregates may influence SOC
mineralization and storage within soil aggregates as these microbial functional groups use
different sources of SOC. The fungal-to-bacterial ratio was also important in shaping microbial
functional gene diversity of genes involved in C, N and P cycling, linking microbial composition
to functional potential in SOM transformations. Further, links between microbial community
composition, function and overall ecosystem function persist even during shifts in microbial
structure and function with forest regeneration. Microbial communities appear to recover in
structure and function to near original, or primary forest conditions following 40-70 years of
secondary forest regeneration.
As more regions in the tropics experience post-agricultural reforestation, understanding
patterns in belowground community structure and function can improve predictions of the fate of
ecosystem C with an increase in forest cover.
iv
ACKNOWLEDGEMENTS
First and foremost I would like to thank Erika Marín-Spiotta and the Biogeo Lab at the
University of Wisconsin-Madison. Without her continued support and insight, this research
would not be where it is today. In addition, members of her lab, the Biogeo Lab in the Dept. of
Geography, have graciously adopted me and have provided countless hours of help in not just
running analyses, but also in preparing me for conferences. I have had the good fortune of
working with and mentoring several undergraduates and laboratory assistants that have spent
countless hours homogenizing, grinding, weighing, etc. my many soil samples.
I would also like to thank Teri Balser for her continued advising despite her move to
Gator land and the world of Deanship. Teri has also provided much intellectual and financial
support for this research throughout my time at the University of Wisconsin. I also want to thank
her for her support in my teaching and learning pursuits and the opportunity to spread the gospel
of The N Game
1
INTRODUCTION: OVERVIEW AND BACKGROUND
Shifts in microbial community structure and function with tropical land use change
1. Introduction
Land use and land cover change affects the global carbon (C) cycle, biodiversity and ecosystem
processes, goods and services. Just in regards to C emissions alone, land use change released an
estimated, annual mean of 156Pg C yr-1 from 1850-2000, with 63% of those emissions coming
from tropical land use change (Houghton 2003). Soils are the largest non-sedimentary terrestrial
reservoir of C, holding two to three times more C in soil organic matter (SOM) than in the
atmosphere or aboveground vegetation (Houghton 2007). Thus changes in soil C from tropical
land use change may play a major role altering global C fluxes.
Land use and land cover change have also been shown to alter microbial communities
with consequences on soil C (Nusslein and Tiedje 1999, Waldrop et al. 2000, Burke et al. 2003,
Balser et al. 2006, Bissset et al. 2011, Potthast et al. 2012). As soil microorganisms control
multiple input and loss pathways of C and N from soils, the effects of land use and land cover
change on microbial communities may drive changes in soil C fluxes to the atmosphere. Thus it
is important to understand how soil microbial communities respond to land use change and their
potential effects on soil C retention and loss.
Microorganisms are central players in regulating important ecological processes such as
decomposition, nutrient cycling, and SOM transformations and stabilization. Changes in overall
biomass or the abundance of key microbial groups can affect pathways and rates of
biogeochemical processes (Doherty and Gutknecht 2012, Potthast et al. 2012). Biogeochemical
consequences of microbial shifts may be due to preferential use or selective preservation of
2
different substrates by distinct microbial communities (Moorhead and Sinsabaugh 2006,
Paterson et al. 2008, Strickland et al. 2009). For example, microbes in forests may be better
adapted to degrading recalcitrant C compounds than communities in grasslands (Cleveland et al.
2003). Poll et al. (2006) suggested that fungi preferentially degrade aboveground plant litter-C,
while bacteria are more likely able to assimilate soil-based C substrates. Bacterial groups may
use different SOM fractions: gram negative bacteria have been shown to associate preferentially
with young and labile C (Potthast et al 2010, Drissner et al. 2007) and gram positive with older
soil C pools (Kramer and Gleixner 2008). Overall, effects of land conversion and management
on microbes is expected to influence C stabilization in soils (Kandeler et al. 1996, Waldrop and
Firestone 2004, Potthast et al. 2012). Understanding the links between microbial community
structure, function and ecosystem processes with land use and land cover change is therefore
important in our attempts to understand and predict the responses of soil C.
Most land change research has centered on deforestation and the conversion of grasslands
to agriculture and pasture, which have been the dominant trends in land cover changes globally
(Ramankutty and Foley 1999, Pongratz et al. 2008). In many parts of the world, however, forest
cover is expanding, primarily as a result of agricultural abandonment and secondary succession,
but also due to woody encroachment on grasslands with fire suppression, establishments of
timber plantations, or promotion of reforestation projects for carbon sequestration (Brown and
Lugo 1990, Aide et al. 2000, Caspersen et al. 2000, Aide and Grau 2004). As most research has
focused on changes in aboveground biomass and species changes during forest growth, we are
still unable to predict how belowground C and nutrient pools will respond to reforestation
(Marín-Spiotta and Sharma 2013). Even less is known about how the microbial community
responds to forest recovery, and the consequences of this on soil C and N dynamics (da C Jesus
3
et al. 2009, Hafich et al 2012). As such, my dissertation research investigates the role of
microbial composition and activity in SOM dynamics during secondary forest succession on
former pastures.
My dissertation research addresses several main questions: (1) what is the effect of
tropical forest regeneration on soil microbial community biomass, composition and functional
activity? (2) which environmental variables help explain patterns in microbial community
dynamics with land cover change ? and (3) how does microbial community composition link to
function in SOM cycling with forest regeneration? I am addressing these questions using a well-
studied long-term forest successional chronosequence on former pastures in the wet subtropical
forest life zone of Puerto Rico. The sites include three replicates each of: active pastures,
secondary forests (now 20, 30, 40, 70, and 90 years old), and old-growth (primary) forests that
have never been converted to pastures (>100 years old). Sites are all located at similar elevation
on well-drained, silty clay, loamy Oxisols (Los Guineos series) within five km of each other, and
receive similar rainfall amounts allowing us to investigate the direct effects of a change in
aboveground vegetation on soil microbial community dynamics.
2. Dissertation Overview
My thesis is divided into four data chapters. The first two chapters focus on the short and long-
term effects of forest regeneration on soil microbial community biomass, composition and
activity. Chapter 1 is a multi-season, multi-year study that characterizes microbial community
composition and enzymatic activity during 90 years of forest recovery on former pastures,
investigates links between microbial community structure, function and soil organic carbon
(SOC), and identifies environmental variables that may help explain patterns in microbial
4
community structure and function along the entire forest regeneration chronosequence. Chapter 2
explores whether soil microbial succession follows or precedes aboveground succession by
tracking and comparing microbial community structure with forest encroachment of an
abandoned pasture with other pasture and secondary forest sites.
The final two chapters investigate potential links between microbial community
composition and functional diversity with SOM structure and cycling with forest regeneration.
Chapter 3 links microbial community composition with the distribution of soil organic matter
(SOM) among soil aggregate fractions to answer the questions: (1) are different microbial groups
associated with different SOM pools, (2) how do these relationships differ with changes in
vegetation during tropical forest succession? Soil C, nitrogen (N) and microbial composition via
phospholipid fatty acid analysis (PLFA) were measured in separated soil aggregate fractions
from active pastures, early secondary forests (40 years old), late secondary forests (90 years old)
and remnant primary forests (> 100 years old). Chapter 4 connects microbial composition and
functional diversity using GeoChip, a high-throughput, gene-based metagenomic functional gene
microarray, with tropical forest recovery. GeoChip contains probes that specifically target genes
coding for enzymes involved in C, N and phosphorus (P) cycling. Overall, this research
investigates links between above and belowground succession, connects microbial community
structure with SOM structure and relates microbial community composition with function in
SOM function along a tropical land cover change chronosequence. A brief review of literature
addressing land use and land cover change effects on microbial community dynamics is
described below.
3. Background
5
3.1 Microbial community structure with tropical land use change
Land use change can directly affect soil physical, chemical and biological properties including
microbial community structure, defined here as biomass, composition and diversity. Out of the
few studies that investigate the effects of tropical land use change on soil microbial communities,
land conversions are mostly focused on shifts between forest and agricultural systems, including
pastures (section 3.1.1, Borneman and Triplett 1997, Nusslein and Tiedje, 1999, Cleveland et al.
2003, Bossio et al. 2005, Templer et al. 2005, Chaer et al 2009, da C Jesus et al. 2009,
Montecchia et al 2011, Ormeño-Orillo et al, 2012, Potthast et al. 2012), forest and plantations
(section 3.1.2, Waldrop et al. 2000, Dinesh et al. 2004, Bossio et al. 2005, Templer et al 2005)
and between primary and secondary forests (section 3.1.3, Templer et al. 2005, da C Jesus et al.
2009, Sandoval-Perez et al. 2009, Ormeño-Orillo et al. 2012).
3.1.1 Forests versus agriculture
Some of the greatest differences in microbial community structure between land use and land
cover types occur during land conversion from forest to pasture or agriculture. In volcanic soils
in Hawaii, forest conversion to pasture resulted in a 49% change in microbial community
composition (Nusslein and Tiedje, 1999). Using 16S rRNA profiling, microbial communities
from young (40 year old) and old (100 year old) sugar cane cropping systems were more related
to each other than to adjacent montane and premontane forest communities (Montecchia et al.
2011). In the Amazon highlands, differences between forest and agricultural soils were also
greater than within different forest or agricultural types, with Firmicutes dominating the bacterial
community composition of forest soils and Bacteroidetes making up the majority of bacteria in
crop and pasture soils (da C Jesus et al. 2009). Using phospholipid fatty acid analysis (PLFA),
6
Chaer et al. (2009) reported greater abundance of actinobacteria and fungi in agricultural soils
and an increase of arbuscular mycorrhizal fungi (AMF) and gram-negative bacteria in forest
soils. Burke et al. (2003) also reported greater actinobacterial and fungal abundance in
agricultural soils compared to forest soils using PLFA. Microbial biomass doubled from pasture
to forest in a Costa Rican Oxisol (Cleveland et al. 2003). Bornemann and Triplett (1997)
measured greater species diversity in forest sites compared to pasture sites.
However, there are also studies that report greater biomass or microbial diversity in
pastures compared to forests. For example, while microbial composition changed between
forests and pastures, bacterial diversity was greater in the pasture communities in the Brazilian
Amazon (da C Jesus et al. 2009). Pothastt et al (2012) also measured greater microbial biomass,
activity and fungal abundance in pasture soils compared to forest soils. Contrary to Burke et al.
(2003) and Chaer et al. (2009), Bossio et al. (2005) reported greater actinobacteria and fungi via
PLFA biomarkers in the forested soils compared to agricultural soils. Thus, while microbial
community structure consistently shifts with forest conversion to agriculture and pastures, the
effect on composition, biomass and diversity is not universal across studies. Variations in how
microbial community structure responds to forest conversion is most likely due to differences in
soil physical and chemical structure associated with land use change that drives microbial
community dynamics (see section 4.1 below for a detailed discussion on different ecological
drivers of microbial community dynamics).
3.1.2 Tree plantations versus native forests
Similar to shifts in forest conversion to cropping systems or pastures, microbial communities
differ between native forests and plantation forests but the effects on microbial structure also
7
differs among studies. Waldrop et al. (2000) measured greater microbial biomass in native
forests compared to pineapple plantations of differing ages in Tahiti. Similar to Waldrop et al.
(2000), microbial biomass C and N was greater in forest soils compared to plantation soils in
South India and the Domincan republic (Dinesh et al. 2004, Templer et al. 2005, respectively).
Microbial PLFA structure differed more between forest and plantations than within different
plantation types with greater gram-positive bacteria in the forest soils and greater actinobacteria
and fungi in the plantation soils (Waldrop et al. 2000). In contrast, ergosterol (an indicator of
fungi) was greater in all forest sites (evergreen, semi-evergreen and broadleaf deciduous) versus
adjacent coconut, arecanut and rubber plantation sites (Dinesh et al. 20004). Microbial
community structure via PLFA and Denaturing Gradient Gel Electrophoresis (DGGE) in soils
from western Kenya appeared to be similar between native forests, woodlots and tea plantations
using principal components analysis (Bossio et al. 2005).
3.1.3 Secondary forests versus primary forests
Microbial community structure between secondary forests and primary forests are often more
similar than between other land cover types. Bacterial community diversity and richness (via
Terminal Restriction Fragment Length Polymorphism, T-RFLP, and 16S rRNA) were more
similar between secondary forests (both young and old) and primary forests than between forest
and agricultural soils in the Brazilian Amazon (de C Jesus et al 2009). Microbial biomass C and
N did not change between secondary forests < 5 years old), secondary forests aged 5-7 years and
old primary forest sites in a Caribbean wet forest zone (Templer et al. 2005). Further, microbial
biomass C and N increased with age since agricultural abandonment in both abandoned pastures,
mixed regenerating forest sites (mixed garden with regenerated trees) and secondary forest sites
8
suggesting that microbial communities recover structure with time (Templer et al. 2005).
Ormeño-Orillo et al. (2012) also showed microbial recovery with time; Bradyrhizobium diversity
increased back to primary forest levels in secondary forest sites. In Costa Rica, secondary forests
regenerated on pasture grasslands recovered soil physical and microbial properties than adjacent
managed grasslands (Hafich et al. 2012). In my dissertation research, microbial community
composition via PLFA in older secondary forest soils (70 and 90 years old) was similar to
primary forest soils and different from young secondary forest soils (20, 30 and 40 years old)
(see Chapter 1). In addition, microbial functional diversity in SOM cycling differed between
early secondary forest soils and late secondary forest soils (see Chapter 4). This suggests that
microbial communities have the potential to recover with time. This is explored below in the
discussion (section 4.3).
3.2 Microbial community activity with tropical land use change
Microbial activity can be measured in a variety of ways: respiration, potential nitrification or
denitrification rates, extracellular enzyme activities, etc. These activities are often used as
indicators of microbial function as they directly relate to fundamental soil processes such as
decomposition, nutrient cycling and SOM transformations. Extracellular enzymes, which are the
main microbial agents of decomposition in soils, are often measured to detect the effects of
vegetation and land management shifts on microbial community function (Burns and Dick,
2002). Extracellular enzymes are involved in fundamental soil biogeochemical cycling
processes, OM decomposition and formation, the availability of essential plant nutrients, and
greenhouse gas production. Important enzymes involved in OM decomposition and nutrient
cycling catalyze the breakdown of dominant plant litter compounds, such as cellulose,
9
hemicellulose, and lignin, and control the release of plant- and microbe-available nutrients from
organic forms (Sinsabaugh et al. 2002). Cellulases (glucosidases), hemicellulases (xylanases),
lignin degrading enzymes (phenol oxidases), nitrogen hydrolyzing enzymes (ureases, acetyl-
glucosiminidases) and phosphatases are all commonly measured to assess microbial activity in
SOM and nutrient cycling (Table 1). Extracellular enzymatic activity is regulated by
environmental and biochemical factors: such as temperature, moisture soil pH, and substrate
availability (Tabatabai 1994, Tate 2002). Mineralogy and soil structure also affects enzymatic
activity as enzymes themselves can become isolated from their substrates via sorption to mineral
surfaces or occlusion within soil aggregates and micropores (Burns 1982, Tate 2002,
Quiquampoix et al. 2002).
Enzyme activities vary across different land uses (Table 2). While many studies report
significant effects of land use or land cover change on enzyme activities, there are few consistent
patterns in enzyme values associated with land use types. For example, pastures can have higher
(Vallejo et al. 2010) or lower (Sandoval-Perez et al. 2009, Vallejo et al. 2010) phosphatase
activity compared to forests. Additionally, there can be no difference between pastures and
forests (Acosta-Martinez et al. 2007, Smith et al. In prep, see Chapter 1). This is similar to
results for β-glucosidase activities; pastures can have higher activities (Acosta-Martinez et al.
2007, Sotomayor et al. 2009), lower activities (Vallejo et al. 2010) or no change in activities
(Smith et al. In prep, see Chapter 1) compared to forests. Despite various and often conflicting
results for enzyme activities with land use change, agricultural soils have lower enzyme activities
relative to natural systems (forest or grassland) or pasture systems (Acosta-Martinez et al. 2007,
Sotomayor et al. 2009).
10
Lack of clear trends in enzyme activities between land uses can be a result of a variety of
methodological and ecological factors. For example, enzyme assay conditions are not
standardized across laboratories (German et al. 2011). Therefore, substrate concentrations,
incubation times and temperatures, assay pH conditions, fluorescent compounds used, soil mass,
etc. can vary among studies, which has the potential to alter values measured (German et al.
2011, Burns et al. 2013 and many others). Ecologically, there are different biotic and abiotic
drivers regulating microbial enzyme activities and soil enzyme turnover such as soil moisture,
pH, soil structure and soil type, C, N, etc (Burns and Dick 2002). Shifts in these properties within
land uses in a study can mask the effects of land use change. For example, different soil types
between replicate sites for agriculture, plantation and forests made it difficult to tease out land
use change effects in a study in Western Kenya (Bossio et al. 2005). A more detailed explanation
of drivers of microbial community composition and activity is included below (see Section 4.1).
Despite the high variability in enzyme activities across studies of tropical land use
change, there are distinct differences in microbial enzyme activities between temperate and
tropical ecosystems (Sinsabaugh et al. 2008, Waring et al. 2013). Ratios of β-glucosidase and
NAGase activities to phosphatase activities are lower in tropical soils compared to temperate
soils indicating a greater microbial demand for P (Waring et al. 2013). In my dissertation
research, soil phosphatase activities across all sites and collection dates was 12 to 500 times
higher (averaging approximately 6000 – 12000 µmolhr-1g-1 soil) than all other enzymes
measured and was also one of the highest activities measured (approximately 1800 – 26,000
µmol hr-1g-1) in litter samples (Smith et al. In prep. see Chapter 1). This is consistent with
common thought that older, more weathered soils (such as those found in the tropics) are
depleted in P (Walker and Syers, 1976). Phosphorus can also be limited by sorption to variable-
11
charged clays (Sollins et al. 1988) and high amounts of precipitation and leaching (Santiago et al.
2005) associated with tropical soils. Low P availability may increase microbial investment in
producing P acquiring enzymes, such as phosphatase (Sinsabaugh and Follstad Shah, 2012).
4. Discussion
4.1 Variations in microbial community dynamics with tropical land use change due to
different ecological drivers of community structure and function
Differences in the effects of tropical land use change on microbial community properties among
studies was largely due to differences in the mechanisms controlling community structure and
function. On a broad scale, mechanisms that drive microbial community structure are similar to
drivers of macro-community structure: physiological limitations, competition and dispersal
processes (Paul and Clark 1989, Morris and Blackwood 2007). Physiological limitations refer to
the specific range of environmental conditions such as pH, salinity, oxygen, temperature and
moisture in which different populations can grow and reproduce. Thus, the soil physical and
chemical structure plays a strong role in regulating microbial community structure (Morris and
Blackwood 2007). For example, soil properties such as pH, base saturation and Al3+ explained
31% of the variations in bacterial community structure between agricultural (pasture and
cropping systems), secondary forests (both young and old) and primary forests in the Brazilian
Amazon region (da C Jesus et al. 2009). Microbial structure is commonly correlated with soil pH
in the tropical studies reported here (da C Jesus et al. 2009, Sandoval-Perez et al. 2009, Potthast
et al. 2012), as well as in global soil analyses (Fierer and Jackson 2006, Lauber et al. 2009,
Rousk et al. 2010). Soil moisture was also commonly correlated with microbial community
dynamics during tropical land use change (Bossio et al. 2005, Ushio et al. 2009, Eaton et al.
12
2011). However, in the majority of studies examined, soil C and N were strongly correlated with
both microbial biomass and enzyme activities (Waldrop et al. 2000, Groffman et al. 2001, Bossio
et al. 2005, Templer et al. 2005, Ushio et al. 2009, Eaton et al. 2011, Potthast et al. 2012, Santos
et al. 2012). Overall, the effect of soil type on microbial community structure and activity most
always outweighs the effects of land use change (Burke et al. 2003, Bossio et al. 2005, Acosta-
Martinez et al. 2007).
While the majority of literature focuses on the relationship between microbial community
dynamics and soil C, N, the effects of soil P concentrations are often overlooked. Yet, P is one of
the primary limiting nutrients in tropical ecosystems on highly-weathered soils (Reed et al.
2011). Further, studies that include both chemical and microbiological-associated measurements
of P (P fractions such as Porg or Pinorg or activity of P-acquiring enzymes, etc.) show that P is an
important component in microbial processes (Cleveland et al. 2002, Eaton et al. 2011, Liu et al.
2012, Waring et al. 2013) and overall ecosystem function (Vitousek 1984, Cleveland et al. 2011)
with tropical land use change. Low P availability constrains processes such as decomposition and
microbial utilization of labile C in highly weathered tropical soils (Cleveland et al. 2002). Lui et
al. (2012) suggest that the effects of low P availability on microbial biomass and composition are
more pronounced in N-rich environments. In a meta-analysis of soil extracellular enzyme
activities in the tropics, Waring et al. (2013) suggest that P limitations may alter C cycling
processes by reducing microbial investments into producing C-based decomposition enzymes
due to necessary up-regulation of P acquiring enzymes.
Resource quality is also an important mechanism driving microbial community structure
and function (Lavelle and Spain, 2001). Land use change alters aboveground vegetation, which
alters inputs (both quality and quantity) to the soil and microbial community. Differences in the
13
chemistry of leaf and root litter can influence the composition and activity of the microbial
community (Wardle and Lavelle 1997, Zak et al. 2003, Carney and Matson 2006, Potthast et al.
2010, Talbot and Treseder 2012, Ushio et al. 2012). Higher microbial activity and biomass in
tropical pasture soils compared to forest soils was attributed to the higher nutrient and substrate
availability associated with grassland litter (Potthast et al. 2012). In my dissertation research, the
majority of soil properties is held relatively constant across land use and land cover types
allowing me to directly attribute changes in microbial composition and activity to changes in
aboveground vegetation. In Chapter 2, I show how microbial community composition shifts
within a year of forest development on abandoned pastures. I attribute this to a change in plant
inputs from grassland litter to woody leaf and root inputs (see Chapter 2) which have known
different chemistries (Marin-Spiotta et al. 2008).
Further. soil heterogeneity plays a part in the high variability in microbial community
structure and activity with tropical land use change (Templer et al. 2005, González-Cotréz et al.
2011, Smith et al. In prep. See Chapter 1). Spatial variability in microbial community structure
and function has long been recognized in soils (Ettema and Wardle 2002). Yet, highly weathered
clay soils under diverse tropical forest vegetation can be especially heterogeneous spatially
(Carvalheiro and Nepstad 1996, Decaens and Rossi 2001, Townsend et al. 2008). Soil C, nutrient
concentrations and redox conditions can vary at the micro-scale (Pett-Ridge and Firestone 2005,
Teh and Silver 2006, Templer et al. 2008, DeAngelis et al. 2010). High diversity in microsite
conditions, among and within soil aggregates can help explain the high variability in observed
extracellular enzyme activity (Schimel et al. 2005).
14
4.2 Despite variability, agricultural systems reduce microbial diversity and function
In the majority of studies investigating tropical land use or land cover change on microbial
community dynamics, microbial community structure (as biomass, composition or diversity) and
function (respiration, enzyme activities or nutrient mineralization) were significantly reduced in
the most conventional or intensively managed agricultural systems compared to forested sites
(Bossio et al. 2005, Chaer et al. 2009, Montecchia et al. 2011, Ormeño-Orillo et al. 2012, Santos
et al. 2012). For studies that solely compared plantation systems to other forest systems and not
conventional cropping systems, the land use that was most disturbed or intensively managed had
significantly less biomass, respiration and nutrient cycling rates compared with other forests
(Templer et al 2005). Lower microbial diversity, biomass and activity in agricultural soils are
often attributed to fewer plant residue inputs, poorer quality, or less labile, plant inputs and
nutrients (Liu et al. 2006, Montecchia et al. 2011). Lower microbial diversity and functional
capacity in agriculture or highly disturbed soils can result in reduced microbial stability, such as
lower resistance and resilience (or recovery) from future disturbance (Chaer et al. 2009). This
may alter important ecosystem processes such as nutrient cycling or soil C stabilization (Chaer et
al. 2009). Consistent with this theory, conventional agricultural practices have been shown to
decrease soil C stocks in both temperate and tropical ecosystems (Six et al. 2002).
The detrimental influence of conversion to cropping systems from forest cover on
microbial community structure and function was not seen, however, in conversion of forests to
pasture. Pasture systems often had a higher rate of microbial respiration, enzyme activities and
greater biomass, diversity compared to forest systems (de C Jesus et al. 2009, Potthast et al.
2012). Pastures, unlike cropping systems, provide more root and litter inputs into the soil that
15
microbial communities readily use (Six et al. 2002). Additionally, pasture litter is often more
labile than forest litter (Baldock et al. 1992, 1997). Vallejo et al. (2010) showed that a 12-year-
old silvopastoral system (forested pasture) had similar microbial function as an adjacent primary
forest. Studies investigating the effect of other tropical agroforesty systems, especially cropping
systems with tree cover, on microbial community dynamics are limited (Vallejo et al. 2010).
4.3 Microbial communities recover with time
When tropical agricultural systems or pastures are abandoned or allowed to regenerate into
forests, the microbial community can recover many attributes of its structure and function to
nearly original forest community conditions (Templer et al. 2005, da C Jesus et al. 2009,
Sandoval-Perez et al. 2009). However, microbial community recovery time varied between
studies. Within 5-7 years of subtropical broadleaf forest regeneration on agricultural crops,
Templer et al. (2005) reported that soil properties, microbial biomass and microbial activity
(mineralization, nitrification, and respiration) recovered to similar values to that of the
undisturbed old forest site in the Dominican Republic. Mixed gardens that had been abandoned
for less than 5 years also showed signs of recovery, but had not yet reached similar levels to that
of the older secondary forest soils (5-7 years since abandonment) (Templer et al. 2005). In
regenerated highland wet forests in Western Amazon, bacterial community composition shifted
to the composition of the bacterial communities located in the primary forests within 5-30 years
of forest regeneration (da C Jesus et al. 2009). Sandoval-Perez et al. (2009) show that in a
tropical dry forests system in Mexico, both soil and microbiological properties more closely
resemble primary forest soils after 26 years of forest regeneration. In my dissertation research,
the soil microbial community recovers structure (see Chapter 1) and function (see Chapter 4) to
16
that of the remnant primary forest between 40-70 years of secondary forest regeneration in a
mid-elevation wet forest in southeastern Puerto Rico.
Despite various recovery times for microbial communities in each of the studies,
microbial communities respond to changes in land use and land cover quite rapidly (Templer et
al. 2005, see Chapter 2). Within a year of forest conversion to agriculture, microbial biomass C
and N, and microbial respiration and denitrification were significantly reduced (Templer et al.
2005). Further, the microbial community shows signs of recovery within less than 5 years of
agricultural abandonment and near full recovery within 5-7 years (Templer et al. 2005). This is
more rapid than the recovery of aboveground forest structure or soil organic carbon (SOC)
reported for tropical forest regeneration (Brown and Lugo 1990, Hughes et al. 1999, Rhoades et
al. 2000, Marín-Spiotta et al. 2007). For example, SOC did not recover to original forest
conditions following secondary forest regeneration for approximately 73 years in a humid
tropical forest zone in Mexico (Hughes et al. 1999). On the other hand, Rhoades et al. (2000)
report that soil C stocks recovered within 20 years in regenerated forests in Ecuador. Secondary
forest tree composition recovered to a similar, but not the same, composition, physoca; structure
and stem density of primary forest trees after 60 years of forest regeneration of abandoned
pastures in the chronosequence used in my dissertation research (Marín-Spiotta et al. 2007).
Thus, it appears that the microbial community not only recovers to pre-agricultural conditions
over time, but that they begin to recover more rapidly than other belowground and aboveground
properties and processes.
17
5. Conclusion
Tropical land changes alters microbial community structure and function with direct
consequences on biogeochemical cycling and nutrient pools. However, the magnitude and
direction of change (i.e. increases or decreases in biomass, diversity) are not consistent between
studies investigating similar land use conversions. This is most likely due to differences in soil
physical and chemical properties, management, ecosystem ages and soil types with land use
change across studies. Shifts in microbial community structure and function are driven by
various ecological properties that were not consistent among studies. This makes generalizations
about how land use change impacts microbial structure, function and overall ecosystem function
difficult. Further, there are few studies that investigate the impact of reforestation or forest
recover on microbial community dynamics despite the increasing trend of forest regeneration in
Latin America and the Caribbean. My dissertation research aims to address this gap in our
understanding of microbial community recovery in post-agricultural forests.
18
6. Tables and Figures
Table 1. Common extracellular enzymes measured and their function in SOM cycling and
decomposition.
SOM component Enzyme Function
β-glucosidase Involved in cellulose decomposition and produces bioavailable glucose.
Cellobiohydrolase Catalyzes the hydrolysis of cellulose, producing cellobiose (which is easily degraded in to glucose).
α-glucosidase Starch-degrading enzyme that releases glucose.
C cycling
xylosidase Involved in hemi-cellulose degradation. Catalyzes the hydrolysis of xylan, producing xylose.
N cycling N- acetylglucosaminidase (NAGase)
Involved in the degradation of chitin, a main component of fungal cell walls and insect exoselectons. Produces mineralizable N.
P cycling Phosphatase Involved in P mineralization. Transforms organic P into phosphate (a plant-available form of P).
* Modified from German et al. 2011
19
Table 1. Summary of results for studies investigating the effects of tropical land use change on extracellular enzyme activity.
Study Land Use Location Soil Type β-glucosidase NAGase Phosphatase
Acosta-Martinez et al. 2007 Agriculture, forest and
pasture
Puerto Rico Inceptisol,
Ultisol, Oxisol
pasture > forest =
agriculture
pasture > forest =
agriculture
forest = pasture >
agriculture Sotomayor et al. 2009 Pasture, agriculture and 2
plantation forests
(Leucaena and Eucalyptus)
Puerto Rico Vertisol plantation forests
> pasture >
agriculture
Eucalyptus >
Leucaena
plantation=
pasture >
agriculture
Sandoval-Perez et al. 2009 Pasture, primary forest,
secondary forest
Mexico Entisol Primary = secondary
forest > pasture Waldrop et al. 2000 Pineapple plantations
(young and old), adjacent
native forest
Tahiti Mollisol young plantation
≥ forest ≥ old
plantation
young plantation >
forest = old plantation Sjogersten et al. 2011** Palm swamp, mixed forest
swamp, Anacardaceae
forest swamp, sawgrass
Panama Histosol sawgrass >
Anacardaceae
forest swamp >
mixed forest
swamp = palm
swamp
sawgrass >
Anacardaceae
forest swamp >
mixed forest
swamp = palm
swamp
sawgrass > mixed
forest swamp >
Anacardaceae forest
swamp = palm
swamp
Ushio et al. 2010 Broadleaf forest trees,
conifer trees
Malaysia NA Conifers =
Broadleaf trees
Conifers>Broadleaf
trees Vallejo et al. 2010* Pasture, native forest,
silvopastoral
chronosequence
Colombia Mollisol primary forest =
12 yr silvopasture
> pasture = 8 yr
silvopasture = 3yr
silvopasture
12 yr silvopasture > 8
yr silvopasture > 3 yr
silvopasture = pasture
= primary forest
Bossio et al. 2005* Primary forest, tea
plantation, agriculture
(continuous maize)
Kenya NA (mixed types) tea plantation =
primary forest >
agriculture
tea plantation ≥
primary ≥
agriculture
agriculture > tea
plantation ≥ primary
forest Dinesh et al 2004** Forests (evergreen, semi-
evergreen and moist
deciduous), Plantations
(coconut, arecanut and
rubber)
India Inceptisol forests >
plantations
forests > plantations
*Enzyme values normalized to soil C. ** assayed for phophomonoesterase, not acid phosphatase
Enzyme activities are significantly different using mean comparisons (as reported in each study) when separated by ">". When separated by "≥" land uses shared one letter out
of two represented in the study, and "=" signifies that land uses were not significantly different (i.e. shared the same letter).
20
7. References
Acosta-Martínez, V., L. Cruz, et al. (2007). "Enzyme activities as affected by soil properties and land use in a tropical watershed." Applied Soil Ecology 35(1): 35-45. Aide, T. and H. Grau (2004). "Globalization, migration, and Latin American ecosystems." Science 305: 1915–1916. Aide, T. M., M. L. Clark, et al. (2012). "Deforestation and Reforestation of Latin America and the Caribbean (2001–2010)." Biotropica 0(0): 1-10. Aide, T. M., J. K. Zimmerman, et al. (2000). "Forest regeneration in a chronosequence of tropical abandoned pastures: implications for restoration ecology." Restoration Ecology 8: 328-338. Baldock, J., J. M. Oades, et al. (1997). "Assessing the extent of decomposition of natural organic materials using solid-state 13C NMR spectroscopy." Australian Journal of Soil Research 35: 1061-1083. Baldock, J. A., J. M. Oades, et al. (1992). "Aspects of the chemical structure of soil organic materials as revealed by solid-state 13C NMR spectroscopy." Biogeochemistry 16: 1-42. Balser, T. C., K. D. McMahon, et al. (2006). "Bridging the gap between micro - and macro-scale perspectives on the role of microbial communities in global change ecology." Plant and Soil 289(1-2): 59-70. Bissett, A., A. E. Richardson, et al. (2011). "Long-term land use effects on soil microbial community structure and function." Applied Soil Ecology 51: 66-78. Borneman, J. and E. W. Triplett (1997). "Molecular microbial diversity in soils from eastern Amazonia: evidence for unusual microorganisms and microbial population shifts associated with deforestation." Applied and Environmental Microbiology 63(7 ): 2647-2653. Bossio, D. A., M. S. Girvan, et al. (2005). "Soil Microbial Community Response to Land Use Change in an Agricultural Landscape of Western Kenya." Microbial Ecology 49(1): 50-62. Brown, S. and A. E. Lugo (1990). "Tropical secondary forests " Journal of Tropical Ecology 6(1): 1-32. Burke, R. A., M. Molina, et al. (2003). "Stable Carbon Isotope Ratio and Composition of Microbial Fatty Acids in Tropical Soils." Journal of Environmental Quality 32: 198-206. Burns, R. G. (1982). "Enzyme activity I nsoil: Location and a possible role in microbial ecology.” Soil Biology and Biochemistry 14: 423-427.
21
Burns, R. G., J. L. DeForest, et al. (2013). "Soil enzymes in a changing environment: Current knowledge and future directions." Soil Biology and Biochemistry 58: 216-234. Burns, R. G. and R. P. Dick (2002). Enzymes in the environment: activity, ecology, and applications., Marcel Dekker, Inc. Carney, K. M. and P. A. Matson (2006). "The Influence of Tropical Plant Diversity and Composition on Soil Microbial Communities." Microbial Ecology 52(2): 226-238. Carvalheiro (de Oliveira ), K. and D. C. Nepstad (1996). "Deep soil heterogeneity and fine root distribution in forests and pastures of eastern Amazonia." Plant and Soil 182 (2): 279-285. Caspersen, J. P., S. W. Pacala, et al. (2000). "Contributions of land-use history to carbon accumulation in U.S. forests." Science 290: 1148-1151. Chaer, G., M. Fernandes, et al. (2009). "Comparative Resistance and Resilience of Soil Microbial Communities and Enzyme Activities in Adjacent Native Forest and Agricultural Soils." Microbial Ecology 58(2): 414-424. Cleveland, C. C., A. R. Townsend, et al. (2002). "Phosphorus Limitation of Microbial Processes in Moist Tropical Forests: Evidence from Short-term Laboratory Incubations and Field Studies." Ecosystems 5(7): 0680-0691. Cleveland, C. C., A. R. Townsend, et al. (2003). "Soil Microbial Dynamics and Biogeochemistry in Tropical Forests and Pastures, Southwestern Costa Rica." Ecological Applications 13(2): 314-326. da C Jesus, E., T. L. Marsh, et al. (2009). "Changes in land use alter the structure of bacterial communities in Western Amazon soils." The ISME Journal 3(9): 1004-1011. DeAngelis, K. M., W. L. Silver, et al. (2010). "Microbial communities acclimate to recurring changes in soil redox potential status." Environmental Microbiology 12(12): 3137-3149. Decaëns, T. and J. P. Rossi (2001). "Spatio–temporal structure of earthworm community and soil heterogeneity in a tropical pasture." Ecography 24(6): 671-682. Dinesh, R., S. G. Chaudhuri, et al. (2004). "Soil biochemical and microbial indices in wet tropical forests: effects of deforestation and cultivation." Journal of Plant Nutrition and Soil Science 167(1): 24-32. Docherty, K. M. and J. L. Gutknecht (2012). "The role of environmental microorganisms in ecosystem responses to global change: current state of research and future outlooks." Biogeochemistry 109(1): 1-6.
22
Drissner, D., H. Blum, et al. (2007). "Nine years of enriched CO2 changes the function and structural diversity of soil microorganisms in a grassland." European Journal of Soil Science 58(1): 260-269. Eaton, W. (2010). "Microbial community indicators of soil development in tropcail secondary forests CR." Ecological Restoration 28(3): 236-238. Fierer, N. and R. B. Jackson (2006). "The diversity and biogeography of soil bacterial communities." Proceedings of the National Academy of Sciences of the United States of America 103: 626-631. German, D. P., M. N. Weintraub, et al. (2011). "Optimization of hydrolytic and oxidative enzyme methods for ecosystem studies." Soil Biology and Biochemistry 43(7): 1387-1397. González-Cortés, J. C., M. Vega-Fraga, et al. (2012). "Arbuscular mycorrhizal fungal (AMF) communities and land use change: the conversion of temperate forests to avocado plantations and maize fields in central Mexico." Fungal Ecology 5(1): 16-23. Groffman, P. M., W. H. McDowellb, et al. (2001). "Soil microbial biomass and activity in tropical riparian forests." Soil Biology and Biochemistry 33: 1339-1348. Hafich, K., E. J. Perkins, et al. (2012). "Implications of land management on soil microbial communities and nutrient cycle dynamics in the lowland tropical forest of northern Costa Rica." Tropical Ecology 53(2): 215-224. Houghton, R. A. (2003). "Revised estimates of the annual net flux of carbon to the atmosphere from changes in land use and land management 1850–2000." Tellus 55B: 378-390. Houghton, R. A. (2007). "Balancing the global carbon budget." Annual Review of Earth and Planetary Science 35: 313-347. Hughes, R. F., J. B. Kauffman, et al. (1999). "Biomass, Carbon, and Nutrient Dynamics of Secondary Forests in a Humid Tropical Region of México." Ecology 80(6): 1892-1907. Kandeler, F., C. Kampichler, et al. (1996). "Influence of heavy metals on the functional diversity of soil microbial communities." Biology and Fertility of Soils 23: 299-306. Kramer, C. and G. Gleixner (2006). "Variable use of plant- and soil-derived carbon by microorganisms in agricultural soils." Soil Biology and Biochemistry 38(11): 3267-3278. Lauber, C. L., M. Hamady, et al. (2009). "Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial community structure at the continental scale. ." Applied and Environmental Microbiology 75: 5111. Lavelle, P. and A. V. Spain (2001). Soil Ecology. Dordrecht, Netherlands, Kluwer Academic Publishers.
23
Liu, L., P. Gundersen, et al. (2012). "Effects of phosphorus addition on soil microbial biomass and community composition in three forest types in tropical China." Soil Biology and Biochemistry 44(1): 31-38. Marín-Spiotta, E., R. Ostertage, et al. (2007). "Long-term patterns in tropical reforestation: Plant community composition and aboveground biomass accumulation.” Ecological Applications 17(3): 828-839. Marín-Spiotta, E., S. Sharma, et al. (2013). "Carbon storage in successional and plantation forest soils: a tropical analysis." Global Ecology and Biogeography 22(1): 105-117. Marín-Spiotta, E., C. W. Swanston, et al. (2008). "Chemical and mineral control of soil carbon turnover in abandoned tropical pastures." Geoderma 143(1-2): 49-62. Montecchia, M. S., O. S. Correa, et al. (2011). "Multivariate approach to characterizing soil microbial communities in pristine and agricultural sites in Northwest Argentina." Applied Soil Ecology 47(3): 176-183. Moorhead, D. L. and R. L. Sinsabaugh (2006). "A theoretical model of litter decay and microbial interaction." Ecological Monographs 76: 151-174. Morris, S. J. and C. B. Blackwood (2007). The ecology of soil organisms, Oxford: Elsevier. Nusslein, K. and J. M. Tiedje (1999). "Soil bacterial community shift with change from forest to pasture vegetation in a tropical soil." Applied Environmental Microbiology 65: 3622–3626. Paterson, E., G. Osler, et al. (2008). "Labile and recalcitrant plant fractions are utilised by distinct microbial communities in soil: Independent of the presence of roots and mycorrhizal fungi." Soil Biology and Biochemistry 40(5): 1103-1113. Paul, A. and F. E. Clark (1989). Soil Microbiology and Biochemistry. San Diego, Academic Press. Pett-Ridge, J. and M. K. Firestone (2005). "Redox fluctuation structures microbial communities in a wet tropical soil." Applied and environmental microbiology 71(11): 6998-7007. Poll, C., J. Ingwersenm, et al. (2006). "Mechanisms of solute transport affect small-scale abundance and function of soil microorganisms in the detritusphere." European Journal of Soil Science 57: 583-595. Pongratz, J., C. Reick, et al. (2008). "A reconstruction of global agricultural areas and land cover for the last millennium." Global Biogeochemical Cycles 22(3).
24
Potthast, K., U. Hamer, et al. (2010). "Impact of litter quality on mineralization processes in managed and abandoned pasture soils in Southern Ecuador." Soil Biology and Biochemistry 42(1): 56-64. Potthast, K., U. Hamer, et al. (2011). "Land-use change in a tropical mountain rainforest region of southern Ecuador affects soil microorganisms and nutrient cycling." Biogeochemistry 111(1-3): 151-167. Quiquampoix, H., S. Servagent-Noinville, et al. (2002). Enzyme adsorption on soil mineral surfaces and consequences for the catalytic activity Enzymes in the environment R. P. D. Richard G. Burns. New York, NY, Marcel Dekker New York 285-306. Ramankutty, N. and J. A. Foley (1999). "Estimating historical changes in global land cover: Croplands from 1700 to 1992." Global biogeochemical cycles 13(4): 997-1027. Reed, S. C., A. R. Townsend, et al. (2011). Phosphorus Cycling in Tropical Forests Growing on Highly Weathered Soils. Phosphorus in Action E. K. E.K. Bunemann. Berlin Heidelberg Springer-Verlag Soil Biology 26: 339-369. Rhoades, C. C., G. E. Eckert, et al. (2000). "Soil carbon differences among forest, agriculture, and secondary vegetation in lower montane Ecuador." Ecological Applications 10(2): 497-505. Rousk, J., P. C. Brookes, et al. (2010). "Investigating the mechanisms for the opposing pH relationships of fungal and bacterial growth in soil." Soil Biology and Biochemistry 42(6): 926-934. Sandoval-Pârez, A. L., M. E. Gavito, et al. (2009). "Carbon, nitrogen, phosphorus and enzymatic activity under different land uses in a tropical, dry ecosystem." Soil Use and Management 25(4): 419-426. Santiago, L. S., E. A. Schuur, et al. (2005 ). "Nutrient cycling and plant-soil feedbacks along a precipitation gradient in lowland Panama." Journal of Tropical Ecology 21(4): 461-470. Schimel, J. P., J. Bennett, et al. (2005). Microbial community composition and soil nitrogen cycling: is there really a connection? . Biological diversity and function in soils. M. B. U. R.D. Bardgett, D.W. Hopkins. Cambridge, UK, Cambridge University Press: 171-188. Sinsabaugh, R. L. and J. J. Follstad Shah (2012). "Ecoenzymatic stoichiometry and ecological theory." Annual Review of Ecology, Evolution, and Systematics 43: 313-343. Sinsabaugh, R. L., C. L. Lauber, et al. (2008). "Stoichiometry of soil enzyme activity at global scale." Ecology Letters 11(11): 1252-1264. Six, J., C. Feller, et al. (2002). "Soil organic matter, biota and aggregation in temperate and tropical soils - Effects of no-tillage." Agronomie 22(7-8): 755-775.
25
Sollins, P. (1998). "Factors influencing species composition in tropical lowland rain forest: does soil matter?" Ecology 79(1): 23-30. Sotomayor-Ramírez, D., Y. Espinoza, et al. (2009). "Land use effects on microbial biomass C, β-glucosidase and β-glucosaminidase activities, and availability, storage, and age of organic C in soil." Biology and Fertility of Soils 45(5): 487-497. Strickland, M., C. Lauber, et al. (2009). "Testing the Functional Significance of Microbial Community Composition." Ecology 90: 441-451. Tabatabai, M. (1994). Soil Enzymes. Methods of Soil Analysis: Microbial and Biochemical Properties. R. R. Weaver. Madison, WI, SSSA. 2: 775-833. Talbot, J. M. and K. K. Treseder (2012). "Interactions among lignin, cellulose, and nitrogen drive litter chemistry–decay relationships." Ecology 93(2): 345-354. Tate, R. L. (2002). Microbiology and Enzymology of Carbon and Nitrogen Cycling Enzymes in the Environment. R. P. D. Richard G. Burns, CRC Press. Teh, Y. A. and W. L. Silver (2006). "Effects of soil structure destruction on methane production and carbon partitioning between methanogenic pathways in tropical rain forest soils." Journal of Geophysical Research 111(G1). Templer, P. H., P. M. Groffman, et al. (2005). "Land use change and soil nutrient transformations in the Los Haitises region of the Dominican Republic." Soil Biology and Biochemistry 37(2): 215-225. Templer, P. H., W. L. Silver, et al. (2008). "Plant and microbial controls on nitrogen retention and loss in a humid tropical forest.” Ecology 89(11): 3030-3040. Townsend, A., G. Asner, et al. (2008). "The biogeochemical heterogeneity of tropical forests." Trends in Ecology & Evolution 23(8): 424-431. Ushio, M., T. C. Balser, et al. (2012). "Effects of condensed tannins in conifer leaves on the composition and activity of the soil microbial community in a tropical montane forest." Plant and Soil 365(1-2): 157-170. Ushio, M., T. Miki, et al. (2009). "Phenolic control of plant nitrogen acquisition through the inhibition of soil microbial decomposition processes: a plant-microbe competition model." Microbes and environments 24(2): 180-187. Vallejo, V. E., F. Roldan, et al. (2010). "Soil enzymatic activities and microbial biomass in an integrated agroforestry chronosequence compared to monoculture and a native forest of Colombia." Biology and Fertility of Soils 46(6): 577-587.
26
Waldrop, M. P., T. C. Balser, et al. (2000). "Linking microbial community composition to function in a tropical soil." Soil Biology and Biochemistry 32: 1837-1846. Waldrop, M. P. and M. K. Firestone (2004). "Altered utilization patterns of young and old soil C by microorganisms caused by temperature shifts and N additions." Biogeochemistry 67(2): 235-248. Walker, T. W. and J. K. Syers (1976). "The fate of phosphorus during pedogenesis." Geoderma 15(1): 1-19. Wardle, D. A. and P. Lavelle (1997). Linkages between soil biota, plant litter quality and decomposition. Driven by nature: Plant litter quality and decomposition. G. Cadisch and K. E. Giller. Wallingford, UK, CAB Intl: 107-124. Waring, B. G., S. R. Weintraub, et al. (2013). "Ecoenzymatic stoichiometry of microbial nutrient acquisition in tropical soils." Biogeochemistry. Zak, D. R., W. E. Holmes, et al. (2003). "Plant diversity, soil mirobial communities and ecosystem funtion: Are there any links?" Ecology 84(8): 2042-2050.
27
CHAPTER 1:
Seasonal and successional changes in soil microbial community structure during
reforestation of a tropical post-agricultural landscape
Abstract
Soil microorganisms regulate fundamental biochemical processes in plant litter decomposition
and soil organic matter (SOM) transformations. In order to predict how land cover change affects
belowground carbon storage, an understanding of how forest floor and soil microbial
communities respond to changes in vegetation, and the consequences for SOM formation and
stabilization, is fundamental. We measured microbial community composition and activity
across a long-term chronosequence of secondary forests regrowing on abandoned pastures in the
wet subtropical forest life zone of Puerto Rico. Here we report intra- and interannual data on
bulk soil and forest floor microbial community composition, via phospholipid fatty acid analysis,
PLFA, and microbial activity via extracellular enzyme activity, from replicate pastures,
secondary forests aged 20, 30, 40, 70, and 90-years, and primary forests. All our sites were
located on the same soil series with minimal differences in soil properties, allowing us to
examine the direct effects of a change in plant cover with forest regrowth on abandoned pastures
and during almost a century of forest succession. Despite intra- and inter-annual variability, there
was a persistent strong effect of land cover type and forest successional stage, or age, on overall
microbial community PLFA structure. Microbial communities differed between pastures, early
secondary forests and old secondary and primary forests, following successional shifts in tree
species composition. While successional patterns held across seasons, the importance of different
28
microbial groups driving these patterns differed seasonally. Extracellular enzyme activity did not
differ with forest succession, but varied by year and season. Few to no significant relationships
existed between microbial community parameters and soil pH, moisture, and carbon and nitrogen
concentrations or stocks. Our data show that land cover type, or forest successional stage, is a
better predictor of microbial community structure in this landscape. At the same time, microbial
community structure and activity varied by season and year stressing the importance of a
multiple, temporal, sampling strategy when investigating microbial community dynamics.
Successional control over microbial community composition with forest recovery has potential
implications for nutrient cycling with changes in vegetation cover. As more areas in the tropics
experience post-agricultural reforestation, understanding patterns in belowground community
structure and function can improve predictions of the fate of ecosystem carbon with an increase
in forest cover.
Keywords: Tropics, Land-Use Change, Soil, Litter, Microbial Communities, PLFA, Extracellular
Enzymes, and Forest Succession.
29
1. Introduction
Land conversion to crops and pasturelands are important drivers of carbon (C) feedbacks
between terrestrial ecosystems and the atmosphere, and the primary cause of habitat loss
affecting biodiversity in tropical and subtropical regions (Houghton 1995; Hoekstra, Boucher et
al. 2004; Houghton 2005). While deforestation has been the most studied land use transition in
the tropics, the opposite trend – forest regeneration or reforestation now characterizes many sites
on formerly cultivated or cleared lands and successional forests have become a dominant cover
type in the tropics (Grau et al. 2004; Wright 2005; Meiyappan and Jain 2012). This is especially
true in tropical Latin America and the Caribbean (Aide et al. 2012). In Puerto Rico, forest cover
increased from 13% in the 1940s to ca. 42% in the 1990s due to widespread agricultural
abandonment following a transition in regional sociopolitical and economic policies (Weaver
and Birdsey 1990; Helmer et al. 2002; Grau et al. 2003). In the Sierra de Cayey region of Puerto
Rico, the region specific to this study, forest cover increased from less than 20% to 62% in 60
years (Pascarella et al. 2000). Despite the broad geographic expansion of secondary forests
across the tropics, the effects of forest regeneration on C and biodiversity are underrepresented in
the literature, and large uncertainties surround the fate of C and species in forests growing on
disturbed land.
Secondary forests (defined here as forests regrowing on formerly deforested lands) have
great potential as C sinks through both their ability to store C in growing aboveground biomass,
and to store C as soil organic matter (SOM) belowground (Prentice 2001; Silver et al. 2004;
Marín-Spiotta et al. 2009). Yet the extent to which reforested soils act as a carbon sink or what
drives the response of soil C to changes in vegetation is still unclear (Hughes et al. 1999; Don et
al. 2011; Marín-Spiotta et al. 2013). While variables such as soil type, tree species composition
30
and time since agricultural abandonment can influence the fate of soil C with reforestation,
climatic variables like mean annual temperature (Marín-Spiotta et al. 2013) and precipitation
(Don et al. 2011) have been shown to be the most important predictors of soil carbon in tropical
post-agricultural forests.
Most studies largely ignore the role of microbial communities in mediating the response
of soil C to land-use change, even though microbes are central players in soil C dynamics. Shifts
in plant species composition during post-agricultural succession in the tropics have been well
documented (Silver et al. 2000; Lugo and Helmer 2004; Marín-Spiotta et al. 2007), but few have
followed changes in microbial community composition during reforestation (but see Hedlund
2002; Zhang et al. 2005; Macdonald et al. 2009). Changes in microbial community structure,
physiology and function have been shown to alter ecosystem processes, such as CO2 production,
plant litter decomposition, SOM transformations and nutrient cycling (Wardle and Putten 2002;
Schimel et al. 2007; Allison and Martiny 2008; McGuire and Treseder 2010). However, how
shifts in microbial community structure affect important biogeochemical processes and
ecosystem function is poorly understood, especially in tropical ecosystems. Our understanding of
microbial community dynamics in these highly species-diverse ecosystems can be further
complicated by the spatial and temporal heterogeneity in resource availability, process rates and
microbial activity.
Microbial communities are sensitive to changes in climate that occur with season and
time. Microbial biomass, composition and activity can respond to shifts in temperature and
precipitation with consequences for ecosystem function on a variety of timescales (Bardgett et al.
2005; Kardol et al. 2006; Treseder et al. 2011). Treseder et al. (2011) places a high importance
on understanding and incorporating temporal variations in microbial communities when
31
modeling ecosystem dynamics as microbial responses to seasonal changes in climate can have
positive and negative global change feedback potentials over short and long timescales. In a
multi-year experimental study on the effects of increased temperature, CO2, and N-deposition on
microbial community structure and function, inter-annual variability had a greater effect on
community dynamics than did any of the treatments (Gutknecht et al. 2012). In other studies
describing temporal or inter-annual variability in microbial community composition and activity,
most of the microbial community shifts are directly linked to changes in aboveground vegetation
(Grayston et al. 2001; Bardgett et al. 2005; Kardol et al. 2006). Plant effects on soil microbes can
occur through changes in leaf litter and root dynamics which alter the quality and quantity of
energy and nutrient inputs available to the soil microbial community (Wardle 2004).
Reforestation can provide a model system for revealing interdependence between plant and soil
communities by examining temporal changes in composition and ecosystem processes during
ecological succession.
We used a well-replicated, long-term successional chronosequence to evaluate the effects
of natural post-agricultural forest regeneration on microbial communities and belowground C
cycling in the subtropical wet forest life zone of southeastern Puerto Rico. Our objectives were to
(1) characterize microbial community composition and activity during 90-years of forest
recovery on former pastures, (2) investigate links between microbial community structure,
function and soil organic carbon (SOC) and (3) identify environmental variables that may help
explain patterns in microbial community structure and function. Previous work at these same
sites showed that while bulk SOC did not change with forest regeneration (Marín-Spiotta et al.
2009), the distribution and turnover of SOC in physical fractions varied among pastures,
32
secondary forests and primary forests (Marín-Spiotta et al. 2008). Here we explore the possibility
that microbial community dynamics may explain these patterns.
2. Materials and Methods
2.1 Field Site Description
This study takes advantage of previously established chronosequence plots (Marín-Spiotta et al.
2007) consisting of active pasture, secondary forests growing on pastures abandoned 20, 30, 40,
70 and 90 years ago, and primary forest sites that have not been under pasture or agricultural use.
All sites are located on private land, 580-700 m above seal level and within approximately five
km of each other in the Sierra de Cayey in southeastern Puerto Rico (18°01´ N, 66°05´ W). Mean
air temperature between 1971-2000 was 21.5ºC (Southeast Regional Climate Center
http://sercc.com/vclimateinfo/historical/historical_pr.html) with little seasonal variation (Daly et
al. 2003). Mean annual precipitation during the years we sampled (2010-2012) from nearby
Jajome Alto climate station (requested from the Caribbean Atmospheric Research Center,
http://atmos.uprm.edu/) was approximately 2184 mm, with monthly mean precipitation varying
from approximately 310 mm in the wet season (May-October) and 54 mm during the dry season
(November-April) (Figure 1). Across all sites, soils are characterized as very-fine, kaolinitic,
isothermic Humic Hapludox in the Los Guineos soil series using US soil taxonomy (Soil Survey
Staff, 2008). Vegetation at each site is described in detail in Marín-Spiotta et al. (2007). Forest
tree species composition differs among early successional secondary forests, late successional
secondary forests and primary forests.
2.2 Sampling and Experimental Design
33
We sampled three replicate sites for each land cover type (pasture, secondary forest 20, 30, 40,
70 and 90-years old, primary forest) with the exception of the 40 year-old secondary forests, as
one of the three replicate sites was recently lost due to residential development. Within each site,
we collected three replicate samples from a ~1 m2 area randomly distributed in aspect and
distance from the center of the plot (previously established and marked). Both forest floor and
soil were collected in the forest sites, whereas just soil was collected at the pasture sites.
Sampling occurred biannually (during both the wet and dry season) from July 2010 to July 2012
to account for any potential effects of season on microbial community profiles.
At each forest site, a portion of the forest floor litter was collected by hand in 20 cm2 area
(estimated) each sampling subplot prior to soil sampling. At all sites, soil was collected
immediately underneath the forest floor using a 4 mm diameter soil core up to 20 cm depth.
Several soil cores (5-8) were collected and combined, making one composite soil sample per
subplot for a total of three replicate samples per site, or 9 replicate samples per land cover type
(three subplots within a site, three replicate sites per land cover type). Following collection, soil
was stored in coolers and shipped to the University of Wisconsin-Madison. Subsamples were
processed separately for enzyme analysis, microbial PLFA and other physical and chemical
analyses. Subsamples of soil and litter for PLFA analysis were frozen within 24-36 hours of
shipping and then freeze-dried for later analysis. Subsamples intended for enzyme analysis were
stored at 4°C and processed within 1-5 days from arrival of samples in the lab.
2.3 Physical and Chemical Soil and Forest Floor Properties
Soil moisture content, pH, total C and nitrogen (N) concentrations (%) were determined for both
soil and forest floor sampled for all collection dates for soil moisture and pH and during January
34
and August of 2011 for C and N. Field moisture content was determined gravimetrically on
freshly sampled, field moist soils and forest floor. Briefly, 10 g of soil and 5 g of forest floor was
oven dried at 105°C or 60°C (for soil and litter, respectively) for 48 hours or until no change in
mass was observed. Water (%) by mass was calculated as [(wet mass – dry mass)/dry mass]*100
(Klute 1986). Soil and forest floor pH was measured on dried and ground samples using a
Sartorius PP-20 professional pH reader in a 1:1 (by volume) 1 M KCl slurry (Sparks, 1996).
Total C and N concentrations were determined on ground, air-dried soil and oven-dried
litter (60°C) using a Flash 2000 NC Analyzer (Thermo Scientific, Wilmington, Delaware) at
University of Wisconsin-Madison. Soil samples were ball-milled using a SPEX Sample Prep
mixer/Mill (Metuchen, New Jersey) and litter samples were ground through 40 and 60 mesh
screens on a Thomas Wiley Mini-Mill (Swedesboro, New Jersey). All samples were run in
duplicate with replicate error < 10% using aspartic acid as calibration and internal standards. As
this soil contains no inorganic C, total % C can be interpreted as organic carbon concentration
(% C). Forest floor and soil C-to-N ratios were calculated as molar ratios (Cleveland and Liptzin
2007).
2.4 Microbial Community Composition
Microbial community composition was measured using a hybrid phospholipid fatty acid (PLFA)
and fatty acid methyl ester (FAME) analysis protocol ( Smithwick, Turner et al. 2005). Briefly,
PLFAs were extracted from freeze-dried and homogenized soil (3 g) or litter (0.25 g) using a
specific ratio (1:2:0.9) of chloroform, methanol, and a phosphorus-buffer. After isolating and
concentrating the extracted PLFAs, they were then saponified, methylated, transferred to an
organic phase and then washed with a basic NaOH solution. Stock standards (9:0, 19:0) of
35
known concentrations (7.08 µg/ml for 19:0 and 9 µg/ml for 9:0) were then added to each sample.
Samples were run on a Hewlett-Packard 6890 Gas Chromatograph equipped with a flame
ionization detector and an Ultra 2 capillary column (Agilent Technologies Inc., Santa Clara, CA,
USA). Peaks were identified using bacterial fatty acid standards and peak identification software
(MIDI Inc, Newark, DE, USA). Peak areas are converted to µmol PLFA g soil -1 (absolute
abundance) using internal standard peaks (9:0, 19:0). Microbial biomass is calculated as the sum
of all peaks (as µmol PLFA g soil -1) identified less than 20.5 C atoms long (Vestal and White
1989; Zelles 1999). Phospholipid fatty acids identified and used as indicator species or in
biomass, guilds and ratios are detailed in Table 1. Due to large differences in soil and litter
microbial communities, indicator species were often used to represent microbial community
composition in soils while microbial guilds better represented microbial community composition
in litter communities.
2.5 Microbial Community Functional Activity
Microbial community functional activity was measured as extracellular enzyme potential
activity on soils and forest floor samples collected from January 2011 through July 2012. Key
enzymes involved in nutrient cycling processes, such as β-glucosidase, α- glucosidase,
cellobiohydrolase, xylosidase (involved in decomposition of cellulose and hemi-cellulose
compounds), N-acetylglucosaminidase (catalyzes the decomposition of chitin and nitrogen
polymers stored in SOM) and acid phosphatase (used for microbial phosphorous acquisition)
were measured using a modified fluorescent-linked substrate (4-methylumbelliferone, MUB)
microplate protocol (Tate 1994; Sinsabaugh et al. 1999; German et al. 2011) optimized for in situ
temperature and pH conditions (German et al. 2011). The homogenate was prepared using 1-2 g
36
fresh soil or 0.5 g fresh litter in 100 ml sodium acetate buffer (pH 5). Enzyme substrates
(200µM) were dispensed in 50 µl aliquots into 200 µl soil homogenate and incubated for one
hour (β-glucosidase, N-acetylglucosaminidase, phosphatase) or three hours (α-glucosidase,
cellobiohydrolase, xylosidase) at room temperature (which is similar to in situ soil temperatures).
Following incubation, 10 µl of 1M NaOH was added to stop the reaction and then the plates
were read (approx. 4 minutes following the NaOH addition) on a Beckman-Coulter DTX880
fluorescent microplate reader (Backman-Coulter, Fullerton, CA, USA). Potential enzyme
activities are measured as µmol enzyme hr-1 g-1 using the following equation (modified from
German et al. 2011):
TOTAL Activity (µmol hr-1 g -1) = (net F/ε) x (hr-1 incubation) x (assay volume x homogenate
volume-1) x (total buffer volume x dry sample wt-1)
where: Net Fluorescence (F) = [(average substrate value - homogenate control) - (substrate
control- plate blank)] and ε = (slope of MUB in presence of homogenate/assay volume).
2.6 Statistical Analysis
Statistical analysis was performed using JMP Pro Version 10 (SAS Inst. Inc., Cary, NC, USA).
Analysis of Variance (ANOVA) and mean comparisons of all pairs using Tukey-Kramer HSD
was performed on soil chemical and microbiological properties by land cover type and sampling
date. For those analyses, data was log transformed for normality when necessary. Principal
Component Analyses was performed on the arcsine-square root transformed, relative abundance
of PLFA biomarkers (Ramette 2007). Only indicator species (see Table 1) were used in principal
37
component analyses of soil communities, whereas all PLFA biomarkers (< 20.5 C chain length)
were analyzed for litter communities.
3. Results
3.1 Soil Microbial Community Biomass, Composition and Activity
Total soil microbial biomass, measured as the sum of all PLFAs, and the mass of individual
PLFA biomarkers differed by land cover and forest successional stage (Figure 2a-c). Pastures
and early secondary forests (20, 30 and 40 years following regeneration) had greater total
biomass than the late secondary (70 and 90 years following regeneration) (Figure 2a). While soil
bacterial biomass did not differ with forest regeneration, overall fungal biomass (sum of 16:1w5c,
18:1w9c, and 18:2w6,9c) was greater in the pastures and early secondary forests (Figure 2b). As
a result, the fungal-to-bacteria ratio decreased in the late secondary and primary forests (Figure
2c). This result was consistent when the indicator for arbuscular mycorrhizae (16:1w5c) was
removed from the calculations of fungal biomass.
Soil PLFA data revealed differences in microbial community structure by forest
successional stage. Pastures and early secondary forests differed from the older forests along the
first and second principal component axes (Figure 3). The successional patterns in community
structure and biomass were consistent when analyzing all data together (data collected biannually
from July 2010 until July 2012) or individually by season and collecting date.
The influence of forest regeneration on microbial community composition varied
depending on whether the absolute or relative abundance of PLFA biomarkers were considered
(Figures 4a-f). Absolute abundance is the total amount of lipid extracted per gram of soil, and is
a measure of biomass, while the relative abundance of a lipid in the sample (the amount of lipid
38
extracted per total lipid) might be considered a better measure of community change. In this case,
the relative abundance of the PLFA indicator for gram-positive bacteria was greatest (p < 0.05)
in the older forests, but the absolute abundance did not change with forest regeneration (Figure
4a). The opposite is true for the PLFA biomarker indicating methanotrophic bacteria (Figure 4f);
the absolute abundance was significantly higher in the pastures and early secondary forests, but
there was no difference in the relative abundance of this biomarker.
Soil microbial community structure also varied temporally, showing strong intra- and
interannual variability. Averaged across all land cover types, collection date had a significant
effect on soil microbial community structure along principal component one (PC1) and principal
component two (PC2) (p < 0.0001) in a principal components analysis (Figure 5). While rainfall
varies seasonally in this region (Figure 1), there was no consistent effect of season (wet versus
dry season) on soil microbial community structure from year to year. Community structure from
samples collected in August 2011 (wet season) and January 2012 (dry season) differed
significantly from that in soils from July 2010 (wet season), January 2011 (dry season) and July
2012 (wet season) along the first principal component (Figure 4). PLFA biomarkers indicating
gram-positive bacteria (15:0iso), anaerobic, gram-negative bacteria (19:0cyclo) and
actinobacteria (16:0 10methyl) described a greater proportion of the variation (91.8%, 88.0% and
83.5%, respectively) along PC1. Along PC2, soil microbial community structure in August 2011
(wet season) differed from January 2012 (dry season), but not from the other wet seasons (July
2010 and July 2012). The PLFA biomarkers for fungi (18:1w9c) and gram-negative bacteria
(16:1w7c) described the majority of variation (80.2% and 69.9%, respectively) along the second
principal component.
39
When analyzing each collection date individually, there were seasonal differences in the
PLFA biomarker having the greatest correlation with the first principal component. With the
exception of July 2012, the biomarker for anaerobic, gram-negative bacteria (19:0cyclo)
explained the greatest amount of variability (> 90%) during the wet seasons (July 2010, August
2011), whereas the biomarker for gram-positive bacteria (15:0iso) was responsible for the
highest variation (> 90 %) during the dry seasons (January 2011, January 2012).
High variability in potential activity of extracellular enzymes in soils within and among
site replicates masked any potential differences across the forest regeneration chronosequence (in
supplemental data). However, enzyme activities did show strong inter- and intra-annual
variability (Table 2). Across all sites and collection dates mean phosphatase activity was 12 to
500 times higher (averaging approximately 6000 – 12000 µmolhr-1g-1 soil) than all other
enzymes measured. N-acetylglucosaminidase (NAGase), beta-glucosidase and xylosidase
averaged between 300 and 1700 µmol hr-1g-1 soil, while alpha-glucosidase and cellobiohydrolase
had the lowest activities (approximately 20 – 300 µmol hr-1g-1 soil). Specific enzyme activity
normalized by soil organic C concentrations or total microbial biomass did not follow any
apparent pattern with forest regeneration, despite some significant differences with land cover
and forest age (data not shown).
3.2 Forest Floor Microbial Community Biomass, Composition and Activity
Forest successional age had a significant effect on total microbial biomass (p < 0.0047), total
bacteria (p < 0.0044), and the fungal-to-bacterial ratio (p < 0.0001) in the forest floor (litter)
averaged across all collection dates. However, while the microbial biomass, total bacteria and the
fungal-to-bacterial ratio of the litter communities varied with significant differences across the
40
chronosequence, there was no clear trend with forest succession (Figure 6 a-c). This was also
true when analyzing each collection date individually; there were few significant effects of forest
age or successional stage on biomass and the majority of abundant PLFA biomarkers.
Microbial community structure of the forest floor (p < 0.0001 for both PC1 and PC2)
varied with forest age when averaged across all collection dates (Figure 7), although the patterns
differed from those measured in the soil. While there were differences in forest floor microbial
communities among the forest ages, there was no consistent trend with successional stage: early,
late, or primary forest, as in the soils. The 70-yr secondary forests were significantly different
from all other forests along PC1 (p < 0.0001). Forest floor microbial community structure also
varied with collection date, but not by season (wet versus dry). For example, the samples from
July 2010 (wet season) showed the lowest variation in microbial structure among forest ages (i.e.
smallest variation between PC1 scores), while those from July 2012 (another wet season) showed
the greatest variability. When comparing microbial community biomass and guilds (Figure 7 a-d),
collection date had a stronger effect (p < 0.0001) on abundance than did forest succession (p-
values were higher and sometimes not even significant, data not shown). Despite this, there were
few consistent patterns in microbial composition with season or with year. In general, total
biomass and the abundance of select microbial guilds in the forest floor were greatest in August
2011 and January 2012 (wet season and dry season, respectively), and lowest in the wet season
of July 2010 (Figure 8 a-e).
Extracellular enzyme activity of the forest floor community varied by collection date
depending on the enzyme measured. Phosphatase, NAGase, cellobiohydrolase and xylosidase all
varied by date collected, but did not follow the same patterns in variation. For example,
phosphatase activity was higher in the wet seasons versus the dry seasons, while NAGase
41
activity was higher in both seasons in 2011 than in both seasons in 2012 (Figure 9). While
phosphatase was one of the greatest enzyme activities measured (approximately 1800 – 26,000
µmol hr-1g-1), it did not show the same striking magnitude of difference in the forest floor
samples compared to the soil (Table 2). Alpha-glucosidase and xylosidase, enzymes involved in
cellulose and hemi-cellulose decomposition, ranged between 500 – 1600 µmol hr-1g-1. Individual
enzyme activity (shown in supplemental data) and specific enzyme activity normalized by soil
organic C concentrations or total microbial biomass did not follow any apparent pattern with
forest regeneration, despite some significant differences with forest age (data not shown).
Microbial community structure and composition between the forest floor and soil
communities differed strongly. Microbial biomass and the absolute abundance of important
guilds, such as Gm+ bacteria, fungi and actinobacteria, were 2 to 5 times greater in the forest
floor compared to the soil (Figure 10 a-d). However, the relative contribution of gram-positive
bacteria and actinobacteria to overall biomass was greater in soils than in the forest floor. Fungi
had both greater absolute and relative abundance in the forest floor compared to the soils.
3.3. Environmental Controls on Microbial Parameters
There were few to no significant relationships among soil microbial community composition,
extracellular enzyme activity and edaphic properties (C, N, moisture, pH). Many of these
variables did not differ by land cover type of forest age. For example, soil % N and % C did not
change with land cover change, forest age or season (data not shown) with a mean % N of 0.26
(± 0.009) and % C of 3.36 (±0.112). The C:N ratio of the soil varied across land cover types, but
showed no pattern with forest succession (mean C:N 15.2 ± 0.172). The C:N ratio of the forest
floor litter varied by forest age with the 70-yr secondary forests having the lowest values: 26.0 ±
42
1.27 and 27.6 ± 0.91 in the dry and wet season, 2011 (data not shown). Soil moisture at the time
of collection varied with forest regeneration stage and season. In general, gravimetric soil
moisture content was higher in the wet seasons (0.55 ± 0.02) than in the dry seasons (0.64 ±
0.02) across the years and highest in the pastures and primary forests in both the wet and dry
seasons. Moisture content of the forest floor varied seasonally, with higher values in the wet
seasons (2.17 ± 0.07). Soil and forest floor pH changes across the forest regeneration
chronosequence ranging from 3.60 to 4.39 (± 0.06) in soil and 4.44 to 5.42 (± 0.07) for litter, but
not with year or season. Forest floor chemical properties such as pH, field moisture, carbon and
nitrogen concentrations could not explain the patterns in microbial community composition. Soil
microbial extracellular activity (total, specific or individual) also showed few relationships with
edaphic properties (% C, % N, moisture, pH) or with microbial parameters (total biomass,
individual PLFA biomarkers (both absolute abundance and relative percent). Despite a lack of
significant relationships between enzyme activities and microbial and soil properties, the activity
of individual enzymes were highly correlated. In particular, NAGase activity is positively
correlated with Beta-glucosidase (r2 = 0.85), alpha glucosidase (r2 = 0.71) and phosphatase (r2 =
0.69).
4. Discussion 4.1 Microbial Communities follow Successional Trends with Reforestation
Belowground microbial community structure showed successional patterns with reforestation of
abandoned pastures, despite large inter- and intra-annual variability in the microbial lipid data
and guilds responsible for differences between pastures and forests and with forest age. The
clustering of soil microbial community structure into early and late successional groups parallels
forest tree species composition, where the early secondary forest tree communities significantly
43
differed from the late secondary, and primary forests (Marín-Spiotta et al. 2007). Most studies on
microbial community composition or activity during land-use and land-cover change attribute
microbial responses to changes in soil type or properties (Bossio et al. 2005; Jia et al. 2005;
Acosta-Martínez et al. 2007) rather than to changes in aboveground plant communities (Zak et al.
2003). Our data did not reveal successional trends in soil moisture, pH, texture, C or N stocks or
the soil C-to-N ratio across the secondary forest chronosequence or any significant relationships
among these soil properties and microbial parameters. An earlier study on secondary forest
succession in the Loess Plateau, China, attributed a rapid increase in microbial biomass with a
parallel increase in soil organic C and total N during the first two decades of forest succession
despite sites sharing a similar soil type (Jia et al. 2005). All our sites were located on the same
soil series with minimal differences in soil properties, thus allowing us to examine the direct
effects of a change in plant cover with forest regrowth on abandoned pastures and during almost
a century of forest succession.
Shifts in vegetation can affect energy and nutrient inputs into the soil ecosystem and
thereby influence microbial composition and activity. Differences in the chemistry and quantity
of leaf and root litter and root exudates can influence the composition and activity of the
microbial community (Zak et al. 2003; Carney and Matson 2006; de Graaff et al. 2010; Potthast
et al. 2010; Talbot and Treseder 2012). The importance of different drivers for changes in
microbial communities (plant diversity, changes in litter input quantity and quality, or species
interactions) is debated in the literature (Waldrop et al. 2000; Cleveland et al. 2003; Paterson
2003; Bossio et al. 2005; Hooper et al. 2005; Paterson et al. 2008, Ushio et al. 2010). For
example, Carney and Matson (2006) showed that increased plant diversity and specific species
were linked to greater microbial diversity. Zak et al. (2003) attributed changes in microbial
44
composition to shifts in litterfall production rather than to plant species diversity per se. While
previous work at our sites showed no difference in C chemistry of annual leaf litterfall among the
forest ages (Ostertag et al. 2008), a preliminary test of forest floor chemistry using 13C-nuclear
magnetic resonance (NMR) spectroscopy showed potential chemical differences between the
early secondary forest litter and primary forest litter of non-composited litter (unpublished). The
primary forest differed from the 20yr old secondary forest in two main classes of carbon
compounds: methyl and alkyl-C compounds (such as lipids and plant waxes), and aromatic-C
(such as lignin and tannins) (Baldock et al. 1997). While this may be attributed to differences in
decompositional stages, it could also indicate differences in litter chemistry between the young
secondary and primary forest.
4.2 Microbial Communities show Strong Inter-annual and Intra-annual Differences
Temporal variation, both intra- and inter-annual, also played an important role in shaping soil
and forest floor microbial community structure and function. Different PLFA biomarkers
accounted for seasonal differences in community composition. In the wet seasons, the indicator
for anaerobic, gram-negative bacteria explained most of the variability in the successional
patterns. Shifts in precipitation and soil moisture have been shown to regulate microbial
community composition and function (Fierer, Schimel et al. 2003; Evans and Wallenstein 2011;
Bouskill, Lim et al. 2012). In other rainforests on similar soil types but higher rainfall (MAP
3500 mm) in Puerto Rico, soil oxygen concentrations show very large spatial and temporal
fluctuations, driving changes in bacterial populations (Pett-Ridge and Firestone 2005; Bouskill et
al. 2012). Our soils, while not as wet, show visible evidence of redoximorphic properties at the
microsite scale, likely facilitated by the high clay content, well-developed aggregate structure,
45
and high iron content. The higher correlation coefficient of the PLFA indicator for anaerobic,
gram-negative bacteria during wet seasons indicates temporal and quite possibly spatial
variations in anaerobic soil environments. Gram-positive bacteria, which were a strong indicator
of community differences among our sites during the dry seasons, have been shown to be most
resistant to moisture stress (Kaur et al. 2005; Schimel et al. 2007; Bouskill et al. 2012). This may
be due to a thick cell wall that better protects them from changes in moisture and osmotic
potential (Paul and Clark 1996; Madigan 2009).
While intra- and inter-annual variation in microbial community composition and activity
has been widely documented in the literature (Wardle 1998; Bardgett et al. 2005), its importance
and implications on ecosystem functioning and response to global change has only been recently
recognized (Treseder et al. 2011; Gutknecht et al. 2012). In a study on temperate grassland
succession, Kardol et al (2006) showed that temporal variation in plant-microbe interactions has
a strong influence on aboveground plant community dynamics and succession. This could be due
to seasonal and temporal changes in nutrient availability as mediated by the microbial
community (Bardgett et al. 2005). In a multi-year experimental grassland study, microbial
community composition varied more from year to year than with increased nitrogen,
precipitation, CO2 or temperature possibly masking the direct responses to the treatment
(Gutknecht et al. 2012). At the same time, the authors speculated that temporal variations in
community composition may have a direct influence on the magnitude and direction of responses
to the climate change treatments. Treseder et al. (2011) argue that temporal variations in
microbial community composition or activity could either accelerate or mitigate climate change
effects such as microbial respiration in response to increasing temperatures; theoretically
increased temperature results in increased respiration and thus, atmospheric CO2 input and yet in
46
experimental studies high respiration rates often declines with time. Changes in microbial
composition, biomass and activity, physiological acclimation or adaptation are all possible
mechanisms affecting warming-induced respiration responses (summarized in Talbot and
Treseder 2012). As we begin to understand the influence of time and season on microbial
structure and function, we can better understand and predict how this variability will affect the
scale and extent of microbial responses to global change.
4.3 Microbial Activity Reflects Soil Microsite Heterogeneity
While microbial community composition showed strong differences by forest successional class,
season and year, any potential effects of these variables on extracellular enzymes were masked
by the high variability in potential activity within and among sites. Enzyme activity was also not
correlated with measured soil and microbiological properties. Despite the fact that many studies
report a positive, linear relationship between enzyme activity and microbial biomass (Bossio et al.
2005; Acosta-Martínez et al. 2007), PLFA biomass at our sites was decoupled from activity
measurements for all six enzymes considered here. This may be due to operational limitations in
the definition and protocols for measuring extracellular enzyme activity. Methodologically,
extracellular enzyme activity is a measure of potential activity versus realized or in situ activity
(Burns 1978; Tate 1994; Sinsabaugh et al. 1999; Burns and Dick 2002). Enzyme values derived
from laboratory methods reflect sample and assay conditions which are often set at optimal
enzyme reaction conditions (DeForest 2009; German et al. 2011). Further, enzyme assays
measure the pool of all available enzymes in a sample; those that were actively being produced at
time of measurement in response to a substrate and residual enzymes that may have become
stabilized in the soil on mineral surfaces (Burns 1982; Quiquampoix et al. 2002; Tate 2002;
47
Allison 2006). As PLFA biomass is a measure of the active soil community (Tunlid et al. 1985,
Frostegard et al. 2011) and enzyme assays measure potential enzyme pools in the soil, a non-
significant relationship between the two is not entirely surprising in highly-weathered clay
Oxisols, which are characterized by variably charged clay minerals and high concentrations of
iron and aluminum oxides and hydroxides, which have been shown to stabilize enzymes
(Quiquampoix et al. 2002; Zimmerman and Ahn 2011). In a review of enzyme activities specific
to tropical soils, Waring et al. (2013) found no significant relationship between microbial
biomass carbon and β-glucosidase, N-acetylglucosaminidase and phosphatase activities.
Highly weathered clay soils under diverse tropical forest vegetation can be highly
spatially heterogeneous (Carvalheiro and Nepstad 1996; Decaens and Rossi 2001; Townsend et
al. 2008), with high variability in soil C, nutrient concentrations and redox at the micro-scale
(Pett-Ridge and Firestone 2005; Teh and Silver 2006; Templer et al. 2008; DeAngelis et al.
2010). High diversity in microsite conditions, among and within soil aggregates can help explain
the high variability in observed extracellular enzyme activity (Schimel et al. 2005). Microbial
structure and function in the soil matrix can vary spatially (Ettema and Wardle 2002; Balser et al.
2006) and temporally with changes in redox conditions and biogeochemical process rates (Pett-
Ridge and Firestone 2005; DeAngelis et al. 2010). The fluorometric, micro-plate assay for
measuring potential enzyme activity are often performed on a small quantity (~1g) of fresh soil
(Sinsabaugh et al. 1999). The ability to effectively homogenize a sample for enzyme analysis can
be limited in highly heterogeneous and clay-rich soils. Thus, the variability in results we
obtained for extracellular enzyme activities are more representative of soil microsite conditions
versus overall site or land cover conditions.
48
The high activity of phosphatase relative to other enzymes measured in all land
uses is consistent with studies in both temperate (Saviozzi et al. 2001; Waldrop and Firestone
2006; Trasar-Cepeda et al. 2008) and tropical ecosystems (Caldwell et al. 1999; Waldrop et al.
2000; Cleveland et al. 2002; Cleveland et al. 2003; Bossio et al. 2005; Acosta-Martínez et al.
2007; Sjögersten et al. 2010). Phosphatase activity is especially important in tropical soils that
are often phosphorus limited (Vitousek and Sanford 1986) as phosphatase catalyzes the
hydrolysis of ester-phosphate bonds, releasing bioavailable phosphorus. Soil phosphorus
limitation has been linked to reductions in primary productivity (Davidson et al. 2004; Cleveland
et al. 2011), SOM decomposition (Wieder et al. 2009) and changes in microbial processes
(Cleveland et al. 2002; Ilstedt and Singh 2005) and enzyme stoichiometry (Waring et al. 2013).
Phosphatase activity with changes in land cover is not so consistent, with some studies reporting
higher values in forests (Saviozzi et al. 2001; Sicardi et al. 2004; Waldrop and Firestone 2006;
Trasar-Cepeda et al. 2008), pastures (Chen et al. 2003) or no difference between forest and
pastures (Acosta-Martínez et al. 2007).
4.4. PLFA as an Appropriate Tool for Identifying Shifts in Soil Microbial Communities
PLFA is a relatively fast and affordable technique that allowed us for high replication of samples
and sites and for analysis of multiple time points. Our data shows that PLFA was useful for
identifying temporal and successional shifts in microbial community composition. In this study,
PLFA proved an effective tool for determining temporal changes in microbial community
biomass and composition for several reasons. PLFA characterizes active microbial community,
thus providing a snap-shot of the in situ composition (Tunlid et al. 1985) which allowed us to
detect differences in microbial communities between sample types (forest floor and soils) and
49
among land cover types. While PLFA doesn’t provide taxonomic information at a species level,
it provides gross functional group information such as gram-positive, gram-negative,
methanotrophic, anaerobic, fungi, etc. that has been fairly accurate for soils (Frostegård and
Bååth 1996; Zelles 1999; Balser et al. 2005; Frostegård et al. 2011). The method has been
successfully used in the past to best show both short- and long-term shifts due to environmental
changes (Ruess and Chamberlain 2010; Frostegård et al. 2011; Wixon and Balser 2013).
The ability of PLFA to detect differences in microbial communities differed for plant
forest floor and soil samples. We found that individual PLFA biomarkers commonly reported in
the literature for the identification of broad microbial groups, such as gram-positive or gram-
negative bacteria, in a diversity of soil types (Zelles 1997; Zelles 1999; Ruess and Chamberlain
2010) were not accurate indicators for plant litter. In fact, many of these indicator PLFAs were
not even present in the forest floor samples. Plant waxes and other plant compounds can interfere
with PLFA detection and identification, leading to misrepresented biomasses of select
biomarkers (Zelles 1996; Zelles 1997; Joergensen and Wichern 2008) For example in a study
that measured PLFAs of isolated microbial and plant species, Zelles (1997) showed that linoleic
acid, the biomarker generally indicating fungi (18:2w6,9c), is also identified in sterilized plant
material, which should not contain any active microbial biomass. This could confound accurate
measurements of fungal abundance in plant litter samples. However, the greater fungal biomass
we detected in forest floor samples relative to soil samples may not just be due to a
misinterpretation of PLFA biomarkers in litter samples. It is well documented that fungal
biomass is higher in decomposing litter samples (Findlay et al. 2002; Hättenschwiler et al. 2005).
One way we chose to minimize these differences in plant-identified biomarkers was to represent
microbial communities through the use of guilds, or multiple biomarkers representing key
50
microbial groups, such as gram-positive bacteria (Balser and Firestone 2005). Guilds were better
able to represent changes in the abundance of key microbial groups with season and succession
by combining the biomass of multiple indicators versus just one. The use of PLFA revealed large
differences in microbial communities in the soil, which were consistent with patterns in
aboveground tree species composition (described in Marín-Spiotta et al. 2007). The application
of PLFA to a variety of environmental samples will improve our interpretation of fatty acid
biomarkers.
5. Conclusions
Our data revealed strong seasonal and inter-annual differences in microbial community
composition with tropical forest succession. Despite the high temporal variability in community
structure, land cover and forest age were more important factors explaining microbial
community composition than soil properties along a 90-year old successional chronosequence on
abandoned pastures. We showed both short- (20 years) and long-term (90 years) microbial
community dynamics with forest recovery. Our multi-seasonal and multi-annual sampling
scheme allowed us to detect an overarching imprint of the aboveground forest community on soil
microbes. This study provides insight into successional dynamics of belowground communities
during forest regrowth on abandoned pastures, which may have implications for soil
biogeochemical cycling and ecosystem function in post-agricultural tropical forests.
51
-6. Tables and Figures Table 1. Microbial PLFA biomarkers and metrics usedCommunity
Category Community Metric PLFA Biomarker References1
Biomass sum named and unnamed PLFAs Fungal Biomass
16:1 w7c, 18:1 w9c, 18:2 w6,9c Tunlid et al. 1985, Zelles 1999
Bacterial Biomass
15:0iso, 15:0anteiso, 16:1w7c, 17:0anteiso, 17:0iso,17:0cyclo, 18:1w5c, 18:1w7c, 19:0cyclo
Zelles 1997, 1999
F:B ratio fungal biomass/ sum bacterial biomass
Bardgett et al. 1996, Kaur et al. 2005, Santruckova et al. 2003
Gram-positive bacteria
15:0iso Kaur et al. 2005, Zelles 1997, 1999,
Actinobacteria 16:0 10methyl Ratledge and Wilkinson 1988 Gram-negative bacteria
16:1 w7c Ratledge and Wilkinson 1988, Zelles 1999
Arbuscular Mycorrhizal Fungi
16:1 w5c Olsson et al. 1995, Olsson 1999
Saprotrophic Fungi 18:1 w9c Bardgett et al. 1996, Frostegard et al. 2011
Methanotrophic bacteria
18:1 w7c Sundh et al. 2000
Saprotrophic Fungi 18:2 w6,9c Frostegard and Baath 1996, Joergensen and Wichern 2008, Kaiser et al. 2010
Indicator Species*
Anaerobic, gram-negative bacteria
19:0cyclo Vestal and White 1989
Gram-positive bacteria
14:0iso, 15:0anteiso, 15:0iso, 16:0iso, 16:0anteiso, 17:0iso, 17:0anteiso
Actinobacteria 16:0 10methyl, 17:0 10methyl, 18:0 10methyl, 19:0 10methyl
Gram-negative bacteria
16:1w7c, 16:1w9c, 17:1w8c, 18:1w5c
Total Bacteria 16:0 10methyl, 17:0 10methyl, 18:0 10methyl, 19:0 10methyl, 14:0iso, 15:0anteiso, 15:0iso, 16:0iso, 16:0anteiso, 17:0iso, 17:0anteiso, 16:1w7c, 16:1w9c, 17:1w8c, 18:1w5c
Total Fungi 16:1w5c, 18:1w9c, 18:2w6,9c
Community Guilds**
F:B Guild ratio Total Fungi / Total Bacterial Guild
Balser and Firestone 2005, Mentzer et al. 2006, Waldrop and Firestone 2006, Williamson and Wardle 2007
Soil: Indicator Species Forest Floor: all named and unnamed PLFAs
Community Structure
Principal Component Analysis of PLFAs combined: all named and
unnamed PLFAs
Mentzer et al. 2006, Ushio et al. 2008, Chaer et al. 2009, Frostegard et al. 2011
1Not meant to be an exhaustive list of all publications supporting use of specific PLFA metrics and/or biomarkers. *Used solely for soil microbial community analysis and not for microbial community composition or structure of forest floor community. **Used for community composition and structure of forest floor analysis.
52 Table 2. Mean soil extracellular enzyme activities by collection date and land cover type. Standard error is a propagated error (SE) calculated as the square root of the sum of errors of sample reps and sites.
Year Season Land Cover Enzyme Activities (µmol g-1 hr-1) Beta NAG Phos Alpha CBH Xylo
2011 dry Pasture 1238.5 ± 526.83 782.25 ± 275.54 6294.14 ± 3112.76 58.24 ± 21.32 209.89 ± 84.18 813.19 ± 239.89 20 yr secondary forest 2446.73 ± 1112.17 2420.99 ± 2175.15 12849.31 ± 6524.72 100.9 ± 42.61 414.37 ± 191.1 1920.56 ± 721.78 30 yr secondary forest 2753.31 ± 2180.62 5945.53 ± 5563.18 16304.09 ± 12530.52 107.63 ± 53.64 552.89 ± 373.01 1532.44 ± 1054.51 40 yr secondary forest 1652.2 ± 265.66 1675.91 ± 570.93 9518.66 ± 4254.83 59.9 ± 18.03 293.71 ± 104.02 942.74 ± 348.91 70 yr secondary forest 1205.91 ± 577.24 863.06 ± 449.91 12715.93 ± 7232.34 56.3 ± 19.06 236.61 ± 112.11 945.3 ± 388.89 90 yr secondary forest 1350.57 ± 587.54 903.77 ± 405.77 8932.31 ± 3796.92 56.22 ± 42.88 294.84 ± 101.31 905.28 ± 125.05 Primary Forest 1105.67 ± 440.23 1267.18 ± 510.47 11976.52 ± 9145.54 56.41 ± 18.17 119.27 ± 43.91 428.55 ± 143.58
wet Pasture 449.29 ± 302.28 469.8 ± 285.47 3871.11 ± 2058.35 49.47 ± 18.79 173.67 ± 146.7 579.08 ± 345.77 20 yr secondary forest 443.32 ± 183.09 1132.59 ± 1098.93 7204.21 ± 2438.04 54.97 ± 19.19 276.37 ± 209.75 750.31 ± 253.51 30 yr secondary forest 501.01 ± 184.5 1220.88 ± 586.15 7244.44 ± 1421.03 47.71 ± 17.76 248.93 ± 108.59 588.92 ± 238.33 40 yr secondary forest 452.49 ± 134.44 590.76 ± 178.45 4629.39 ± 1307.63 42.31 ± 19.46 185.87 ± 97.86 522.67 ± 145.32 70 yr secondary forest 397.85 ± 139.68 554.24 ± 276.85 6889.63 ± 3417.13 58.48 ± 34.82 203.95 ± 159.96 450.93 ± 249.22 90 yr secondary forest 453.36 ± 185.38 470.53 ± 214.39 6929.52 ± 1399.4 57.91 ± 26.6 242.37 ± 195.59 685.51 ± 418.98 Primary Forest 191.66 ± 112.38 380.8 ± 187.78 6245.54 ± 2036.68 31.68 ± 10.69 63.99 ± 27.71 309.42 ± 169.9
2012 dry Pasture 887.23 ± 396.57 745.31 ± 360.77 7929.18 ± 3087 69.42 ± 40.92 183.02 ± 48.24 637.85 ± 155.27 20 yr secondary forest 552.14 ± 257.08 598.95 ± 275.31 6586.99 ± 4209.87 29.14 ± 7.22 102.88 ± 50.55 408.17 ± 107.74 30 yr secondary forest 589.21 ± 260.27 1410.82 ± 1027.43 8930.54 ± 2967.31 23.94 ± 9.22 98.15 ± 72.05 226.11 ± 144.08 40 yr secondary forest 1128.04 ± 575.83 1116.61 ± 467.34 9056.57 ± 4460.52 28.12 ± 14.71 122.28 ± 78.58 371.81 ± 221.35 70 yr secondary forest 457.63 ± 248.07 515.89 ± 258.95 9605.22 ± 4526.07 29.2 ± 13.31 88.07 ± 49.96 286.73 ± 159.17 90 yr secondary forest 597.03 ± 228.9 669.3 ± 226.48 9743.27 ± 4045 34.41 ± 9.34 145.64 ± 59.44 511.88 ± 107.44 Primary Forest 646.84 ± 389.63 1155.52 ± 1046.63 15828.62 ± 8659.57 39.29 ± 16.45 64.88 ± 32.01 262.24 ± 112.2
wet Pasture 376.15 ± 98.22 271.53 ± 80.03 3925.34 ± 1033.33 25.05 ± 8.56 92.12 ± 33.93 521.08 ± 139.13 20 yr secondary forest 295.25 ± 88.09 225.19 ± 112.62 3304.72 ± 847.08 15.06 ± 4.27 57.72 ± 16.86 380.24 ± 72.92 30 yr secondary forest 336.36 ± 253.62 464.67 ± 277.07 4652.9 ± 1392.6 14.77 ± 7.1 74.04 ± 23.29 265.75 ± 109.02 40 yr secondary forest 460.57 ± 103.02 342.72 ± 94.39 4369.88 ± 1039.68 18.31 ± 4.65 123.68 ± 39.03 434.92 ± 173.62 70 yr secondary forest 341.23 ± 99.62 320.16 ± 156.27 5818.55 ± 1363.71 24.27 ± 8.84 84.36 ± 28.05 403.5 ± 89.87 90 yr secondary forest 403.22 ± 85.76 341.77 ± 97.35 6401.72 ± 886.03 22.81 ± 7.92 112.4 ± 31.14 463.5 ± 156.45 Primary Forest 368.15 ± 172.44 430.64 ± 203.47 6718.68 ± 2715.09 22.76 ± 5.77 65.86 ± 38.66 244.06 ± 66.08
!
53
Figure 1. Daily precipitation from July 2010 to July 2012 at Jajome Alto Climate Station, Puerto Rico.
54
Figure 1. (a) Total PLFA biomass, (b) fungal biomass, and (c) fungal to bacterial (F:B) ratio across forest cover types. Means for each forest cover type (pasture, 20, 30, 40, 70 and 90-yr secondary forests and 100+ yr primary forests) are averaged across all collection dates. Error bars represent one standard error from the mean. Analysis of variance performed on log transformed data show a significant effect of forest cover type on total biomass, fungal biomass and F:B (p<0.0001 for all). Letters not shared by forest cover types indicate significant differences.
55
Figure 2. Principal component analysis of soil microbial community PLFA structure by forest cover. Indicator species of PLFA biomarkers used in analysis. Mean forest cover types (points) are averaged across all seasons. Error bars represent one standard error from the mean.
Pasture 20yr secondary
30yr secondary
40yr secondary
70yr secondary
90yr secondary
Primary forest
-‐2
-‐1.5
-‐1
-‐0.5
0
0.5
1
1.5
-‐1.5 -‐1 -‐0.5 0 0.5 1 1.5 2 PC1 36.4% (p<0.0001)
PC2 23.8% (p<0.0001)
56
Figure 3. Absolute (nmol g-1) and relative abundance (%) of indicator biomarkers for (a) gram-positive bacteria, 15:0iso, (b) actinobacteria, 16:0 10 methyl, (c) gram-negative bacteria, 16:1w7c, (d) anaerobic, gram-negative bacteria, 19:0 cyclo, (e) arbuscular mycorrhizal fungi, 16:1w5c, and (f) methanotrophic bacteria, 18:1w7c. Means for each forest cover type (pasture, 20, 30, 40, 70 and 90yr secondary forests and 100+ yr primary forests) are averaged across all collection dates. Error bars represent one standard error from the mean. Y-axis corresponds to absolute abundance, while relative abundance is expressed as a percent.
57
Figure 4. Principal component analysis of soil microbial community PLFA structure by collection date (season and year). Indicator species of PLFA biomarkers used in analysis. Collection dates are averaged across all seasons. Error bars represent one standard error from the mean.
Wet Season 2010 (July)
Dry Season 2011 (January)
Wet Season 2011 (August)
Dry Season 2012 (January)
Wet Season 2012 (July)
-‐1.5
-‐1
-‐0.5
0
0.5
1
-‐1.5 -‐1 -‐0.5 0 0.5 1 1.5
PC2 23.8% (p<0.0001)
PC1 36.4% (p<0.0001)
58
Figure 6. Microbial community composition of the forest floor litter by forest age. Means for each forest age are averaged across all collection dates. Error bars represent one standard error from the mean. Analysis of variance performed on log transformed data showed a significant effect of forest age on (a) total biomass (p<0.0047), (b) total bacteria (p<0.0044), (c) fungal-to-bacterial ratio (p<0.0001). Letters not shared indicate significant differences.
59
Figure 7. Principal component analysis of forest floor litter microbial community PLFA structure by forest cover, season and year. All PLFA biomarkers were used in the analysis. Error bars represent one standard error from the mean.
60
Figure 8. Microbial composition of forest floor across collection dates. Means for each collection date are averaged across all forest cover types. Error bars represent one standard error from the mean. Analysis of variance performed on log transformed data shows a significant effect of collection date on (a) total biomass, (b) fungal biomass, and (d) gram-negative bacteria abundance, (p<0.0001 for all). Letters not shared indicate significant differences.
61
Figure 9. Extracellular enzyme activities from the forest floor by season for 2011 and 2012. Mean activities are averaged across all forest ages. Error bars represent one standard error from the mean
0
5000
10000
15000
20000
25000
30000
Dry Season 2011 Wet Season 2011 Dry Season 2012 Wet Season 2012
Alpha-‐Glucosidase Xylosidase Cellobiohydrolase NAGase Beta-‐glucosidase Phosphatase
62
Figure 10. Microbial community composition of soil versus forest floor litter; (a) microbial biomass (p<0.0001), (b) gram-positive bacteria absolute and relative abundance (p<0.0001), (c) fungal biomass, both absolute and relative abundance (p<0.0001), and (d) actinobacteria absolute and relative abundance (p<0.0001). Means are averaged across all forest cover types and collection dates.
.
63
7. References
Acosta-Martínez, V., L. Cruz, et al. (2007). "Enzyme activities as affected by soil properties and land use in a tropical watershed." Applied Soil Ecology 35(1): 35-45.
Aide, T. M., M. L. Clark, et al. (2012). "Deforestation and Reforestation of Latin America and the Caribbean (2001–2010)." Biotropica 0(0): 1-10.
Allison, S. D. (2006). "Soil minerals and humic acids alter enzyme stability: implications for ecosystem processes." Biogeochemistry 81(3): 361-373.
Allison, S. D. and J. B. H. Martiny (2008). "Colloquium Paper: Resistance, resilience, and redundancy in microbial communities." Proceedings of the National Academy of Sciences 105(Supplement 1): 11512-11519.
Baldock, J., J. M. Oades, et al. (1997). "Assessing the extent of decomposition of natural organic materials using solid-state 13C NMR spectroscopy." Australian Journal of Soil Research 35: 1061-1083.
Balser, T. C. and M. K. Firestone (2005). "Linking microbial community composition and soil processes in a California annual grassland and mixed-conifer forest." Biogeochemistry 73(2): 395-415.
Balser, T. C., K. D. McMahon, et al. (2006). "Bridging the gap between micro- and macro-scale perspectives on the role of microbial communities in global change ecology." Plant and Soil 289(1-2): 59-70.
Balser, T. C., K. K. Treseder, et al. (2005). "Using lipid analysis and hyphal length to quantify AM and saprotrophic fungal abundance along a soil chronosequence." Soil Biology and Biochemistry 37(3): 601-604.
Bardgett, R., W. Bowman, et al. (2005). "A temporal approach to linking aboveground and belowground ecology." Trends in Ecology & Evolution 20(11): 634-641.
Bossio, D. A., M. S. Girvan, et al. (2005). "Soil Microbial Community Response to Land Use Change in an Agricultural Landscape of Western Kenya." Microbial Ecology 49(1): 50-62.
Bouskill, N. J., H. C. Lim, et al. (2012). "Pre-exposure to drought increases the resistance of tropical forest soil bacterial communities to extended drought." The ISME Journal 7(2): 384-394.
64
Burns, R. G. (1978). Soil Enzymes. London, Acadmin Press.
Burns, R. G. (1982). "Enzyme activity in soil: Location and a possible role in microbial ecology.” Soil Biology and Biochemistry 14: 423-427.
Burns, R. G. and R. P. Dick (2002). Enzymes in the environment: activity, ecology, and applications., Marcel Dekker, Inc.
Caldwell, B. A., R. P. Gri�ths, et al. (1999). "Soil enzyme response to vegetation disturbance in two lowland Costa Rican soils." Soil Biology and Biochemistry 31: 1603-1608.
Carney, K. M. and P. A. Matson (2006). "The Influence of Tropical Plant Diversity and Composition on Soil Microbial Communities." Microbial Ecology 52(2): 226-238.
Carvalheiro, K. d. O. and D. C. Nepstad (1996). " Deep soil heterogeneity and fine root distribution in forests and pastures of eastern Amazonia " Plant and Soil 182(2 ): 279-285.
Chen, C. R., L. M. Condron, et al. (2003). "Seasonal changes in soil phosphorus and associated microbial properties under adjacent grassland and forest in New Zealand." Forest Ecology and Management 177(1-3): 539-557.
Cleveland, C. C., A. R. Townsend, et al. (2002). "Phosphorus Limitation of Microbial Processes in Moist Tropical Forests: Evidence from Short-term Laboratory Incubations and Field Studies." Ecosystems 5(7): 0680-0691.
Cleveland, C. C., A. R. Townsend, et al. (2003). "Soil Microbial Dynamics and Biogeochemistry in Tropical Forests and Pastures, Southwestern Costa Rica." Ecological Applications 13(2): 314-326.
Cleveland, C. C., A. R. Townsend, et al. (2011). "Relationships among net primary productivity, nutrients and climate in tropical rain forest: a pan-tropical analysis." Ecology Letters 14(9): 939-947.
Davidson, E. A., C. J. R. d. Carvalho, et al. (2004). "Nitrogen and Phosphorus Limitation of Biomass Growth in a Tropical Secondary Forest." Ecological Applications 14(4): S150-S163.
65
de Graaff, M.-A., A. T. Classen, et al. (2010). "Labile soil carbon inputs mediate the soil microbial community composition and plant residue decomposition rates." New Phytologist 188(4): 1055-1064.
DeAngelis, K. M., W. L. Silver, et al. (2010). "Microbial communities acclimate to recurring changes in soil redox potential status." Environmental Microbiology 12(12): 3137-3149.
Decaens, T. and J. P. Rossi (2001). "Spatio-temporal structure of earthworm community and soil heterogeneity in a tropical pasture." Ecography 24: 671-682.
DeForest, J. L. (2009). "The influence of time, storage temperature, and substrate age on potential soil enzyme activity in acidic forest soils using MUB-linked substrates and l-DOPA." Soil Biology and Biochemistry 41(6): 1180-1186.
Don, A., J. Schumacher, et al. (2011). "Impact of tropical land-use change on soil organic carbon stocks - a meta-analysis." Global change biology 17(4): 1658-1670.
Ettema, C. H. and D. A.Wardle (2002). "Spatial soil ecology." Trends in Ecology & Evolution 17(4): 177-183.
Evans, S. E. and M. D. Wallenstein (2011). "Soil microbial community response to drying and rewetting stress: does historical precipitation regime matter?" Biogeochemistry 109(1-3): 101-116.
Fierer, N., J. P. Schimel, et al. (2003). "Influence of Drying-Rewetting Frequency on Soil Bacterial Community Structure." Microbial Ecology 45(1): 63-71.
Findlay, S. E. G., S. Dye, et al. (2002). "Microbial growth and nitrogen retention in litter of Phragmites australis compared to Typha angustifolia.” Wetlands 22(3): 616-625.
Frostegård, A. and E. Bååth (1996). "The use of phospholipid fatty acid analysis to estimate bacterial and fungal biomass in soil " Biology and Fertility of Soils 22(1-2 ): 59-65.
Frostegård, Å., A. Tunlid, et al. (2011). "Use and misuse of PLFA measurements in soils." Soil Biology and Biochemistry 43(8): 1621-1625.
German, D. P., M. N. Weintraub, et al. (2011). "Optimization of hydrolytic and oxidative enzyme methods for ecosystem studies." Soil Biology and Biochemistry 43(7): 1387-1397.
66
Grau, H. R., T. M. Aide, et al. (2004). "Trends and scenarios of the carbon budget in postagricultural Puerto Rico (1936–2060)." Global Change Biology 10: 1163-1179.
Grau, H. R., T. M. Aide, et al. (2003). "The Ecological Consequences of Socioeconomic and Land-Use Changes in Postagriculture Puerto Rico." Bioscience 53(12): 1159-1168.
Grayston, S. J., G. S. Griffith, et al. (2001). "Accounting for variability in soil microbial communities of temperate upland grassland ecosystems." Soil Biology and Biochemistry 33: 533-551.
Gutknecht, J. L. M., C. B. Field, et al. (2012). "Microbial communities and their responses to simulated global change fluctuate greatly over multiple years." Global change biology 18(7): 2256-2269.
Hättenschwiler, S., A. V. Tiunov, et al. (2005). "Biodiversity and Litter Decomposition in Terrestrial Ecosystems." Annual Review of Ecology, Evolution, and Systematics 36(1): 191-218.
Hedlund, K. (2002). "Soil microbial community structure in relation to vegetation management on former agricultural land." Soil Biology and Biochemistry 34: 1299-1307.
Helmer, E. H., O. Ramos, et al. (2002). "Mapping the Forest Type and Land Cover of Puerto Rico, a Component of the Caribbean Biodiversity Hotspot." Caribbean Journal of Science 38(3-4): 165-183.
Hoekstra, J. M., T. M. Boucher, et al. (2004). "Confronting a biome crisis: global disparities of habitat loss and protection." Ecology Letters 8(1): 23-29.
Hooper, D. U., F. S. Chapin, et al. (2005). "Effects of biodiversity on ecosystem functioning: A consensus of curent knowledge.” Ecological Monograph 75(1): 3-35.
Houghton, R. A. (1995). "Land-use change and the carbon cycle." Global Change Biology 1: 275-287.
Houghton, R. A. (2005). "Aboveground Forest Biomass and the Global Carbon Balance." Global Change Biology 11(6): 945-958.
67
Hughes, R. F., J. B. Kauffman, et al. (1999). "Biomass, Carbon, and Nutrient Dynamics of Secondary Forests in a Humid Tropical Region of México." Ecology 80(6): 1892-1907.
Ilstedt, U. and S. Singh (2005). "Nitrogen and phosphorus limitations of microbial respiration in a tropical phosphorus-fixing acrisol (ultisol) compared with organic compost." Soil Biology and Biochemistry 37(7): 1407-1410.
Jia, G.-m., J. Cao, et al. (2005). "Microbial biomass and nutrients in soil at the different stages of secondary forest succession in Ziwulin, northwest China." Forest Ecology and Management 217(1): 117-125.
Joergensen, R. and F. Wichern (2008). "Quantitative assessment of the fungal contribution to microbial tissue in soil." Soil Biology and Biochemistry 40(12): 2977-2991.
Kardol, P., T. Martijn Bezemer, et al. (2006). "Temporal variation in plant-soil feedback controls succession." Ecology Letters 9(9): 1080-1088.
Kaur, A., A. Chaudhary, et al. (2005). "Phospholipid fatty acid – A bioindicator of environment monitoring and assessment in soil ecosystem." Current Science 89(7): 1103-1112.
Klute, A. (1986). Methods of soil analysis. Part 1. Physical and mineralogical methods. Madison, WI, American Society of Agronomy
Lugo, A. E. and E. Helmer (2004). "Emerging forests on abandoned land: Puerto Rico’s new forests." Forest Ecology and Management 190(2-3): 145-161.
Macdonald, C. A., N. Thomas, et al. (2009). "Physiological, biochemical and molecular responses of the soil microbial community after afforestation of pastures with Pinus radiata." Soil Biology and Biochemistry 41(8): 1642-1651.
Madigan, M. T. (2009). Brock biology of microorganisms. San Francisco, CA, Pearson/ Banjamin Cummings.
Marin-Spiotta, E., R. Ostertag, et al. (2007). “Long-term patterns in tropical reforestation: plant community composition and aboveground biomass accumulation” Ecological Applications 17(3): 828-839.
68
Marín-Spiotta, E., S. Sharma, et al. (2013). "Carbon storage in successional and plantation forest soils: a tropical analysis." Global Ecology and Biogeography 22(1): 105-117.
Marin-Spiotta, E., W. L. Silver, et al. (2009). "Soil organic matter dynamics during 80 years of reforestation of tropical pastures." Global change biology 15(6): 1584-1597.
Marín-Spiotta, E., C. W. Swanston, et al. (2008). "Chemical and mineral control of soil carbon turnover in abandoned tropical pastures." Geoderma 143(1-2): 49-62.
McGuire, K. L. and K. K. Treseder (2010). "Microbial communities and their relevance for ecosystem models: Decomposition as a case study." Soil Biology and Biochemistry 42(4): 529-535.
Meiyappan, P. and A. K. Jain (2012). "Three distinct global estimates of historical land-cover change and land-use conversions for over 200 years." Frontiers of Earth Science 6(2): 122-139.
Ostertag, R., E. Marín-Spiotta, et al. (2008). "Litterfall and Decomposition in Relation to Soil Carbon Pools Along a Secondary Forest Chronosequence in Puerto Rico." Ecosystems 11(5): 701-714.
Pascarella, J. B., T. M. Aide, et al. (2000). "Land-Use History and Forest Regeneration in the Cayey Mountains, Puerto Rico." Ecosystems 3(3): 217-228.
Paterson, E. (2003). "Importance of rhizodeposition in the coupling of plant and microbial productivity." European Journal of Soil Science 54: 741-750.
Paterson, E., G. Osler, et al. (2008). "Labile and recalcitrant plant fractions are utilised by distinct microbial communities in soil: Independent of the presence of roots and mycorrhizal fungi." Soil Biology and Biochemistry 40(5): 1103-1113.
Paul, E. A. and F. E. Clark (1996). Soil Microbiology and Biochemistry. San Deigo, CA, Academin Press.
Pett-Ridge, J. and M. K. Firestone (2005). "Redox Fluctuation Structures Microbial Communities in a Wet Tropical Soil." Applied and Environmental Microbiology 71(11): 6998-7007.
69
Potthast, K., U. Hamer, et al. (2010). "Impact of litter quality on mineralization processes in managed and abandoned pasture soils in Southern Ecuador." Soil Biology and Biochemistry 42(1): 56-64.
Prentice (2001). The Carbon Cycles and Atmospheric Carbon Dioxide Contributions of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. J. T. Houghton. Cambridge, UK, Cambridge University Press.
Quiquampoix, H., S. Servagent-Noinville, et al. (2002). Enzyme adsorption on soil mineral surfaces and consequences for the catalytic activity Enzymes in the environment R. P. D. Richard G. Burns. New York, NY, Marcel Dekker New York 285-306.
Ramette, A. (2007). "Multivariate analyses in microbial ecology." FEMS Microbiology Ecology 62(2): 142-160.
Ruess, L. and P. M. Chamberlain (2010). "The fat that matters: Soil food web analysis using fatty acids and their carbon stable isotope signature." Soil Biology and Biochemistry 42(11): 1898-1910.
Saviozzi, A., R. Levi-Minzi, et al. (2001). "A comparison of soil quality in adjacent cultivated, forest and native grassland soils." Plant and Soil 223: 251-259.
Schimel, J. P., T. C. Balser, et al. (2007). "Microbial Stress-Response Physiology and Its Implications for Ecosystem Function." Ecology 88(6): 1386-1394.
Schimel, J. P., J. Bennett, et al. (2005). Microbial community composition and soil nitrogen cycling: is there really a connection? . Biological diversity and function in soils. M. B. U. R.D. Bardgett, D.W. Hopkins. Cambridge, UK, Cambridge University Press: 171-188.
Sicardi, M., F. Garcı́a-Préchac, et al. (2004). "Soil microbial indicators sensitive to land use conversion from pastures to commercial Eucalyptus grandis (Hill ex Maiden) plantations in Uruguay." Applied Soil Ecology 27(2): 125-133.
Silver, W. L., L. M. Kueppers, et al. (2004). "Carbon sequestration and plant community dynamics following reforestation of tropical pastures.” Ecological Applications 14(4): 1115-1127.
Silver, W. L., R. Ostertag, et al. (2000). "The Potential for Carbon Sequestration Through Reforestation of Abandoned Tropical Agricultural and Pasture Lands." Restoration Ecology 8(4).
70
Sinsabaugh, R. L., M. J. Klug, et al. (1999). Characterizing soil microbial communities Standard Soil Methods for Long-Term Ecological Research 318.
Sjögersten, S., A. W. Cheesman, et al. (2010). "Biogeochemical processes along a nutrient gradient in a tropical ombrotrophic peatland." Biogeochemistry 104(1-3): 147-163.
Smithwick, E. A. H., M. G. Turner, et al. (2005). "Variation in NH4+ mineralization and microbial communities with stand age in lodgepole pine (Pinus contorta) forests, Yellowstone National Park (USA)." Soil Biology and Biochemistry 37(8): 1546-1559.
Talbot, J. M. and K. K. Treseder (2012). "Interactions among lignin, cellulose, and nitrogen drive litter chemistry–decay relationships." Ecology 93(2): 345-354.
Tate, R. L. (1994). Soil Microbiology, Wiley and Sons, Incorporation.
Tate, R. L. (2002). MIcrobiology and Enzymology of Carbon and Nitrogen Cycling Enzymes in the Environment. R. P. D. Richard G. Burns, CRC Press.
Teh, Y. A. and W. L. Silver (2006). "Effects of soil structure destruction on methane production and carbon partitioning between methanogenic pathways in tropical rain forest soils." Journal of Geophysical Research 111(G1).
Templer, P. H., W. L. Silver, et al. (2008). "Plant and microbial controls on nitrogen retention and loss in a humid tropical forest.” Ecology 89(11): 3030-3040.
Townsend, A., G. Asner, et al. (2008). "The biogeochemical heterogeneity of tropical forests." Trends in Ecology & Evolution 23(8): 424-431.
Trasar-Cepeda, C., M. C. Leirós, et al. (2008). "Hydrolytic enzyme activities in agricultural and forest soils. Some implications for their use as indicators of soil quality." Soil Biology and Biochemistry 40(9): 2146-2155.
Treseder, K. K., T. C. Balser, et al. (2011). "Integrating microbial ecology into ecosystem models: challenges and priorities." Biogeochemistry 109(1-3): 7-18.
71
Tunlid, A., B. H. Baird, et al. (1985). "Determination of phospholipid ester-linked fatty acids and poly P-hydroxybutyrate for the estimation of bacterial biomass and activity in the rhizosphere of the rape plant Brassica napus (L.)." Canadian Journal of Microbiology 31: 1113-1119.
Vestal, J. R. and D. C. White (1989). "Lipid Analysis in Microbial Ecology." Bioscience 39(8): 535-541.
Vitousek, P. M. and R. L. Sanford (1986). "Nutrient Cycling in Moist Tropical Forest." Annual Review of Ecology and Systematics 17: 137-167.
Waldrop, M. P., T. C. Balser, et al. (2000). "Linking microbial community composition to function in a tropical soil." Soil Biology and Biochemistry 32: 1837-1846.
Waldrop, M. P. and M. K. Firestone (2006). "Seasonal Dynamics of Microbial Community Composition and Function in Oak Canopy and Open Grassland Soils." Microbial Ecology 52(3): 470-479.
Wardle, D. A. (1998). " Controls of temporal variability of the soil microbial biomass: A global-scale synthesis." Soil Biology and Biochemistry 30(13): 1627-1637.
Wardle, D. A. (2004). "Ecological Linkages Between Aboveground and Belowground Biota." Science 304(5677): 1629-1633.
Wardle, D. A. and W. H. v. d. Putten (2002). Chapter 14 Biodiversity, ecosystem functioning and above-ground-below-ground linkages. Biodiversity and ecosystem function: synthesis and perspectives. S. N. Michel Loreau, Paable Inchausti. Oxford, UK, Oxford University Press: 155-168.
Waring, B. G., S. R. Weintraub, et al. (2013). "Ecoenzymatic stoichiometry of microbial nutrient acquisition in tropical soils." Biogeochemistry.
Weaver, P. and R. A. Birdsey (1990). "Growth of Secondary Forest in Puerto Rico Between 1980 and 1985." Turrialba 40(1): 12-22.
Wieder, W. R., C. C. Cleveland, et al. (2009). "Controls over leaf litter decomposition in wet tropical forests." Ecology 90(12): 3333-3341.
72
Wixon, D. L. and T. C. Balser (2013). "Toward conceptual clarity: PLFA in warmed soils." Soil Biology and Biochemistry 57: 769-774.
Wright, S. J. (2005). "Tropical forests in a changing environment." Trends in Ecology & Evolution 20(10): 553-560.
Zak, D. R., W. E. Holmes, et al. (2003). "Plant diversity, soil microbial communities, and ecosystem function: Are there any links?” Ecology 84(8): 2042-2050.
Zelles, L. (1996). Community Structure of Soil Microorganismss. Methods in Soil Biology Berlin Heidelberg, Springer 76-92.
Zelles, L. (1997). "Phospholipid fatty acid profiles in selected members of soil microbial communities.” Chemosphere 35(1/2): 275-294.
Zelles, L. (1999). "Fatty acid patterns of phospholipids and lipopolysaccharides in the characterisation of microbial communities in soil: a review." Biology and Fertility of Soils 29: 111-129.
Zhang, Y.-M., N. Wu, et al. (2005). "Changes in enzyme activities of spruce (Picea balfouriana) forest soil as related to burning in the eastern Qinghai-Tibetan Plateau." Applied Soil Ecology 30(3): 215-225.
Zimmerman, A. R. and M.-Y. Ahn (2011). Organo-mineral–enzyme interaction and soil enzyme activity Soil Enzymology. Berlin Heidelberg Springer 271-292.
73
CHAPTER 2 (SHORT COMMUNICATION):
Microbial community composition rapidly responds to changes in aboveground succession
Abstract: Plant-soil-microorganism interactions shape both above- and belowground communities
structure and function. Understanding the functional significance of plant community
composition on the structure of soil microorganisms and vice versa is challenged by complex
interactions with both biotic and abiotic environment that vary with time and space. This study
tracks an in situ compositional shift in soil microbial communities with a change in aboveground
vegetation over several seasons and years. Using phospholipid fatty acid analysis (PLFA), we
repeatedly measured soil microbial community composition along a replicated chronosequence
of natural forest regeneration of abandoned pastures in Puerto Rico to characterize belowground
function with successional changes in plant species. During the duration of the study, one of the
replicate active pasture sites was abandoned and began experiencing woody plant encroachment.
Within 1-2 years of woody plant regeneration, the microbial community structure shifted from
that associated with the other active pasture sites to a community structure found under early
secondary forest cover. Our results indicate similar directional trajectories with succession of
above and belowground communities. We propose that the microbial community is responding
to changes in plant composition, specifically woody species establishment on pasture. Our long-
term, repeated sampling allowed us to observe direct links between above and belowground
communities and document rapid changes in microbial composition. Our findings have
implications for predicting rapid ecological responses to land-cover change and forest recovery.
74
1. Introduction
Plant-microbe interactions and the resulting effects on ecological succession are of key interest
to researchers working both above and below the soil interface (Kardol and Wardle 2010).
Understanding how the relationship between above and belowground communities drive overall
ecosystem succession and development is important in restoration ecology and our
understanding of how ecosystems recover. Shifts in vegetation have been shown to alter biomass,
activity and composition of the soil microbial community (Bardgett et al. 1998, Wardle et al.
1999, Zak et al. 2003, Carney and Matson 2006). In turn, shifts in soil microbial community
composition can alter plant composition (Bradford et al. 2002, Wardle 2004, van der Putten et al.
2013 and many others), especially in regards to presence or absence of mycorrhizal fungi and
other symbionts (Chapin et al. 1994, van der Heijden et al. 2008). Soil microbes influence plant
dynamics via symbiosis (both mutalistic & parasitic), as well as through their effects on nutrient
cycling. Soil microbes release plant available nutrients through mineralization activities, and also
immobilize nutrients through assimilation.
However, generalizations of how plant-microbe interactions influence ecological
succession are complicated by the complexity of interactions with the soil physical and chemical
components and the scales at which these interactions occur (Porazinska et al. 2003, Bardgett et
al. 2006). Soils represent a diverse and heterogeneous matrix of interacting physical, chemical
and biological variables that vary over space and time, challenging our ability to identify specific
factors and mechanisms that drive changes in aboveground or belowground communities during
succession (Wardle et al. 2004). Understanding the mechanisms from plant-microbe interactions
that shape community succession is important for understanding and predicting ecosystem
recovery and response to disturbance.
75
While plant succession has been studied for more than a century, belowground microbial
community succession is much less understood (Schmidt et al. 2007), especially during
secondary forest succession (Jia et al. 2005, Banning et al. 2011). Evidence supporting parallel
successional trajectories in soil microbial and plant communities is even more rare, yet
theoretically very likely (Harris 2003, Jangid et al 2011). In those cases where it parallel
succession is documented, whether or not microbes facilitate or merely follow plant succession is
still debated (Harris 2009).
In this study, we used a well-replicated secondary forest chronosequence to examine
links between aboveground and belowground succession. Successional stages in tree species
have been well established at these sites, with distinct communities in young secondary forests,
late successional forests, and primary forests (Marín-Spiotta et al. 2007). Recently, we have also
observed successional stages in soil microbial phospholipid fatty acid analysis (PLFA)
composition between active pastures, young secondary forests, and older forests (Smith et al. In
prep. see Chapter 1). The successional patterns in microbial composition persist across inter- and
intra-annual temporal variability. Here, we explore whether soil microbial succession follows or
precedes aboveground succession to better understand the plant-microbe mechanisms driving
ecosystem succession. This study compares microbial community composition across the entire
chronosequence, as well as composition between pasture sites. Out of the three replicate pasture
sites, one site began experiencing woody plant encroachment between summer 2010 and summer
2011 (Figure 1). While soils were collected biannually (in both the wet and dry season) from
2010-2012, we focus only on data collected during the wet season from 2010-2012 to illustrate
the rapid succession of microbial community composition. This study is unique in its long-term
repeated sampling approach, which was designed to account for seasonal variability in microbial
76
dynamics, but allowed us the opportunity to observe in situ microbial community composition
change in real time.
2. Methods
To investigate the effects of secondary forest succession on soil microbial community
composition, we measured phospholipid fatty acid (PLFA) biomarkers along a chronosequence
of sites consisting of: active pastures, secondary forest regenerated on abandoned pastures 20, 30,
40, 70 and 90 years old and remnant primary forests. All land cover types had three replicate
sites except for the 40-year-old sites where the third replicate site was recently deforested. All
sites are located on private land, 580-700m above seal level and within approximately five km of
each other in the Sierra de Cayey in southeastern Puerto Rico (18°01´ N, 66°05´ W). Mean
annual temperature and precipitation is approximately 21.5ºC and 1261mm from 1971-2000
(SERCC, 2013). Across all sites, soils are characterized as very-fine, kaolinitic, isothermic
Humic Hapludox in the Los Guineos soil series (Soil Survey Staff, 2008).
Soils were collected to 20 cm depth from each site using a 4 mm diameter soil core up to
20 cm depth. Several soil cores (5-8) were collected and combined, making one composite soil
sample per subplot for a total of three replicate samples per site. For a complete description of
sampling and experimental design, see Smith et al. (In prep see Chapter 1). Here we describe
data from soils collected in July 2010, August 2011 and July 2012. We restrict our analyses to
the wet seasons as we have an additional year of microbial measurements during the wet season.
Inter- and intra-annual variability in the data is described in Smith et al. (In prep, see Chapter 1).
PLFAs were extracted using a modified fatty acid methyl ester and PLFA methods
(described in Smithwick et al. 2005) on 3g of freeze-dried and ball-milled (SPEX Sample Prep
77
mixer, Metuchen, New Jersey) homogenized soils. The following PLFA biomakers were used to
repsent indicator species: 15:0iso (Gram-positive bacteria, Kaur et al. 2005, Zelles 1997, 1999),
16:0 10methyl (Actinobacteria, Ratledge and Wilkinson 1988), 16:1 w7c (Gram-negative
bacteria, Ratledge and Wilkinson 1988, Zelles 1999), 16:1 w5c (Arbuscular Mycorrhizal Fungi
(Olsson et al. 1995, Olsson 1999), 18:1 w9c (Saprotrophic Fungi, Bardgett et al. 1996,
Frostegard et al. 2011), 18:2 w6,9c (Saprotrophic Fungi, Frostegard and Baath 1996, Joergensen
and Wichern 2008, Kaiser et al. 2010) and 19:0cyclo (Anaerobic, gram-negative bacteria, Vestal
and White 1989).
Total C and N concentrations were determined on ground, air-dried soil using a Flash
2000 NC Analyzer (Thermo Scientific, Wilmington, Delaware) at University of Wisconsin-
Madison. Soil pH was measured on dried and ground samples using a Sartorius PP-20
professional pH reader in a 1:1 (by volume) 1 M KCl slurry (Sparks, 1996). Soil moisture
content was determined gravimetrically on freshly sampled, field moist soils.
Principal Component Analyses was performed on the arcsine-square root transformed,
relative abundance of PLFA biomarkers (Ramette 2007). Only indicator species were used in
principal component analyses of soil communities for Figure 2, whereas all PLFA biomarkers (<
20.5 C chain length, Vestal and White 1989, Zelles 1999) were analyzed for Figure 3
communities in order to better illustrate the shift in microbial structure at pasture site one
compared to the other pasture sites. Statistical analysis was performed using JMP Pro Version 10
(SAS Inst. Inc., Cary, NC, USA).
78
3. Results
In 2010, when all pasture sites were composed of grassland (non-woody) vegetation, including
pasture 1 (Figure 1a), principal components analysis of the structure of microbial indicator
species showed distinct communities based on land use and forest age (Figure 2a). Microbial
community composition in the pastures were separated from communities associated with early
secondary forests (20, 30 and 40 years-old, left side of figure) and microbial communities
associated with late secondary (70 and 90 years-old) and primary forests (right side of the figure).
Within a year, woody vegetation began to colonize one of the replicate pasture sites (Figure 1b)
and the corresponding principal components analysis showed microbial community composition
of the pastures moving closer (or becoming more similar) to the cluster of early secondary forest-
associated communities (Figure 2b). By the summer of 2012, woody vegetation began to
dominate pasture site 1 (Figure 1c) and microbial community composition further merged with
communities associated with early secondary forests (Figure 2c).
When the principal component scores for land use and forest age (pastures, secondary
forests 20, 30, 40, 70, and 90 years old, and primary forests) were averaged across sites (Figure
2abc), no detectable changes were observed in microbial community composition between
pastures and early successional secondary forests during 2011-2012. However, when analyzing
the trajectory of individual sites, microbial communities in pasture sites not experiencing woody
plant encroachment remain distinct from the early secondary forests (Figure 3). The one pasture
site undergoing forest regeneration shifted from a pasture-associated community in 2010 (Figure
2a) to an early successional secondary forest-associated community in 2011 (Figure 3). The rapid
change in microbial community composition with woody regeneration was not driven by
79
changes in soil C, N, moisture or pH, as there was no significant changes in these variables in
pasture site 1 from 2010-2012 (data not shown).
4. Discussion
4.1 Microbial Community Succession
Our results show that microbial community structure succession tracks successional changes in
aboveground forest species composition. Microbial community structure rapidly succeeds. Our
results also show that successional shifts in microbial community structure occurs quite rapidly
(within a year) following initial shifts in aboveground forest succession (i.e. forest regeneration).
Evidence that microbial community succession can occur in soils over extremely short time
scales, within several months or a year, has only recently appeared (Schmidt et al. 2007, Fierer et
al. 2010), despite theories regarding microbial community succession existing for more than half
a century (Frankland 1998). However, much of the literature documenting microbial succession
is focused on litter decomposition (Kendrick and Burges 1962, DeAngelis et al. 2013) or primary
ecosystem succession with recent deglaciation (Nemergut et al. 2007, Knelman et al. 2012).
Microbial succession in liter decomposition can occur on timescales of months to years, while
studies investigating microbial successional responses to shifts in vegetation have only measured
change on decadal timescales (Kendrick and Burges 1962, Nemergut et al. 2007, Knelman et al.
2012, DeAngelis et al. 2013). However similar to our study, in recently deglaciated soils,
microbial community composition underwent succession more rapidly than plant community
succession (Nemergut et al. 2007, Schmidt et al. 2008).
Microbial succession and feedback interactions with aboveground plant colonization is
thought to fuel and facilitate aboveground vegetation succession via immobilization and net
80
mineralization processes which alter nutrient pools for plants (Wardle et al. 2004, Van Der
Heijden et al. 2008, Harris 2009). Soil microbial succession and turnover can alter soil N cycling
processes, provide pulses of DON and DIN via necromass, and as a result, influence plant
productivity (Schmidt et al. 2007). The field of restoration ecology has documented multiple
instances where microbial communities do not only facilitate, but are essential for aboveground
ecosystem development (Harris 2008, 2009). In fact, soil microbial community composition is
often used as an indicator for the success of ecological restoration or to assess the effects of
management practices on ecosystem recovery (Harris 2003).
At the same time, microbial communities are also theorized to ‘follow’ aboveground
development rather than ‘facilitate’ it (Harris 2009, Banning et al. 2011, Williams et al. 2013).
For example, soil bacterial composition recovered with forest recovery along a bauxite mining
rehabilitation chronosequence due to changes in soil pH, C, N and P concentrations (Banning et
al. 2011). Along a sand-dune-soil chronosequence, Williams et al. (2013) reported that bacterial
community development mimicked primary plant succession (i.e. tracking community turnover
and steady-state climax conditions). At the same time, the authors suggested possible plant-
microbe feedback interactions that may have facilitated overall ecosystem succession (Williams
et al. 2013).
Our results suggest that the soil microbial community responded to the changing
vegetation, likely via a change in plant-based energy and nutrient inputs and thus, were acting
more as ‘followers’ with ecosystem development. In regards to potential feedback mechanisms,
prior results show that potential extracellular enzyme activity did not change across the years at
pasture site 1 (Smith et al. In prep. see Chapter 1). However, extracellular enzyme activities
represent potential activities and do not measure in situ activities (Burns and Dick 2002). Further
81
in our Fe oxide and cay-rich soils, measured enzyme activities may represent enzymes stabilized
by mineral interactions rather than a microbial functional response to a change in vegetation
(Burns 1982, Quiquampoix et al. 2002).
The effects of plant-microbe feedbacks on plant community composition depend on the
type and scale of the response from the microbial community. The change in microbial
composition with the encroachment of woody vegetation that we have documented may have
altered the availability essential plant nutrients such as N and P in the soil. We did not, however,
measure these properties in our study. Thus, we are unable to discredit the potential for
microbial-mediated changes to the soil-plant environment that may enhance forest successional
development. The soil microbial community, in our study, may thus be acting as both followers
and facilitators of forest succession.
4.2 Drivers of Microbial Community Composition
The shift in microbial community structure from a pasture-associated structure to an early
secondary forest structure was initiated by woody plant colonization in the abandoned pasture
site. This is most likely attributed to a change in soil inputs from the shift in aboveground
vegetation (Bardgett and Wardle 2010). Plant colonization can modify soil properties in a variety
of ways. Increases in plant cover can increases soil moisture and reduce soil temperature
(Belesky et al. 1989), resulting in a change in the microclimate of microbial habitat (Kirchmann
and Eklund, 1994). The importance of soil properties, such as precipitation and moisture
(Drenovsky et al. 2010, Evans and Wallenstein 2011, Bouskill et al. 2012), soil C and N (Fierer
and Jackson 2006, Paul 2006), soil type and texture (Bossio et al. 2005, Lauber et al. 2008, Wu
et al. 2008, Bach et al. 2010, Jangid et al. 2010), and pH (Fierer and Jackson 2006, Lauber et al.
82
2009, Rousk et al. 2010) on shaping microbial community composition has been well
documented across a variety of ecosystems.
Tree species composition, diversity and richness can also alter soil pH, C and N
concentrations, thus indirectly altering microbial community composition (Zak et al. 2003,
Balser and Firestone 2005, Ushio et al. 2008). Additionally, shifts in vegetation can affect energy
and nutrient inputs into the soil ecosystem and thereby influence microbial composition and
activity (Bardgett and Wardle 2010). In our study, the rapid successional shift in microbial
community composition was likely due to a change in leaf litter and root inputs from the recently
regenerated woody vegetation (Figure 4), rather than a change in the soil physical environment
or total C, N as soil moisture, pH, %C and %N did not vary in pasture site 1 from 2010-2012.
Prior results show that litter and SOM quality and quantity differs between the forests and the
pastures (Marín-Spiotta et al. 2008, Ostertag et al. 2008). Differences in the chemistry and
quantity of leaf and root litter and root exudates can influence the composition and activity of the
microbial community (Wardle and Lavelle 1997, Zak et al. 2003, Carney and Matson 2006, de
Graaff et al. 2010, Potthast et al. 2010, Talbot and Treseder 2012, Ushio et al. 2012). Figure 4
illustrates plant-soil microbe interactions with initiation of woody plant growth; microbial
community composition shifts in response to new litter and root inputs that differ in quality and
quantity from the grassland vegetation.
Marín-Spiotta et al. (2008) showed that leaf litter inputs in the pasture sites had relatively
greater concentrations of carbohydrates and greater amounts of acid soluble C and glucose than
the forest sites using 13C NMR spectroscopy and wet chemistry, respectively. Forest litter inputs,
on the other hand, were composed of relatively more lipid and carbonyl C compounds and non-
polar extractables, Klason lignin and water-soluble phenols (or tannins) (Marín-Spiotta et al.
83
2008). The chemical composition suggests that grassland litter inputs are more easily degraded
by the soil microbial community compared to forest leaf litter (Horner et al. 1988, Hobbie 1996,
Kraus et al. 2003).
Forest regeneration also altered the distribution and turnover of SOM physical fractions
across our chronosequence. The pasture sites had lower concentrations of physically unprotected
particulate organic matter than the forested sites, which were attributed to differences in the
decomposability of litter inputs (Marín-Spiotta et al. 2008). Depletion of unprotected, recent
plant derived SOM in the pastures resulted in greater radiocarbon-based mean residence times
(Marín-Spiotta et al. 2008). While a change in the quantity and quality of root inputs with forest
regeneration has not yet been examined at our sites, roots likely played a role in influencing
microbial composition (de Graaff et al. 2010). Overall, these results support the hypothesis that
microbial succession was driven by a shift in litter inputs that occurred during woody
regeneration.
5. Conclusion
Through this study we showed that soil microbial community succession tracks changes in
aboveground succession. We also showed that the soil microbial community responds rapidly (in
less than a year) to a shift in aboveground vegetation. New technologies in identifying microbial
community composition and function may offer better insight into how and if plant-mediated
changes in microbial composition in turn affect aboveground succession. Increasing evidence
points to the importance of above and belowground interactions in driving community
succession across a wide variety of ecosystems (van der Putten 2009, 2013). Our long-term,
repeated sampling allowed us to observe direct links between above and belowground
84
communities and document rapid changes in microbial composition. However, the specific
feedback mechanisms shaping plant-microbe mediated succession are to a large extent unknown
(van der Putten et al. 2013). At our sites, a change in litter inputs associated with forest
regeneration of an abandoned pasture appears to be driving rapid successional changes in
belowground microbial community structure. Understanding how above and belowground
community interactions drive ecosystem development is important for understanding and
predicting overall ecosystem recovery from land use change.
85
6. Figures
Figure 1. Woody plant encroachment of pasture site one from (a) 2010 with no woody biomass in active pasture, (b) 2011, early colonization by woody biomass, and (c) 2012 woody biomass becomes more dense and forest development becomes more evident.
86
Figure 2. Principal components analysis of PLFA indicator species by land use and land cover type (pasture, secondary forest and primary forest) from (a) 2010, the pasture community is separate from both forest types, (b) 2011, pasture community moves closer to the early secondary forest sites, and (c) 2012, the pasture community remains clustered with the early secondary forest communities.
87
Figure 3. Principal components analysis of all PLFAs (<20.5 C atoms long) by site for active pasture, 20 year old secondary forests, secondary forests 30-90 years old and primary forests. By 2011, pasture site one (PS1) has shifted away from other pasture sites and towards earliest secondary forest sites (20 year old secondary forest).
88
Figure 4. Plant-soil-microbe interactions for (a) grasslands, forests and (b) woody encroachment on grasslands. Woody plant regeneration on pasture grasslands result in microbial succession from pasture-associated community composition to early secondary forest community composition via differences in woody leaf and root inputs.
89
7. References Bach, E. M., S. G. Baer, et al. (2010). "Soil texture affects soil microbial and structural recovery during grassland restoration." Soil Biology & Biochemistry 42: 2182-2191. Balser, T. C. and M. K. Firestone (2005). "Linking microbial community composition and soil processes in a California annual grassland and mixed-conifer forest." Biogeochemistry 73(2): 395-415. Banning, N. C., D. B. Gleeson, et al. (2011). "Soil Microbial Community Successional Patterns during Forest Ecosystem Restoration." Applied and Environmental Microbiology 77(17): 6158-6164. Bardgett, R. D., P. J. Hobbs, et al. (1996). "Changes in soil fungal: bacterial biomass ratios following reductions in the intensity of management of an upland grassland" Biology and Fertility of Soils 22(3 ): 261-264. Bardgett, R. D. and A. Shine (1999). "Linkages between plat litter diversity, soil microbial biomass and ecosystem function in temperate grasslands." Soil Biology and Biochemistry 31: 317-321. Bardgett, R. D. and D. A. Wardle (2010). Aboveground-Belowground Linkages: Biotic Interactions, Ecosystem Processes, and Global Change Oxford, UK, Oxford University Press. Bardgett, R. D., D. A. Wardle, et al. (1998). "Linking above-ground and below-ground interactions: how plant responses to foliar herbivory influence soil organisms." Soil Biology and Biochemistry 30(1867–1878). Bossio, D. A., M. S. Girvan, et al. (2005). "Soil Microbial Community Response to Land Use Change in an Agricultural Landscape of Western Kenya." Microbial Ecology 49(1): 50-62. Bouskill, N. J., H. C. Lim, et al. (2012). "Pre-exposure to drought increases the resistance of tropical forest soil bacterial communities to extended drought." The ISME Journal 7(2): 384-394. Bradford, M. A., T. H. Jones, et al. (2002). "Impacts of soil faunal community composition on model grassland ecosystems." Science 298(5593): 615-618. Burns, R. G. (1982). "Enzyme activity in soil: Location and a possible role in microbial ecology” Soil Biology and Biochemistry 14: 423-427. Burns, R. G. and R. P. Dick (2002). Enzymes in the environment: activity, ecology, and applications., Marcel Dekker, Inc. Carney, K. M. and P. A. Matson (2006). "The influence of tropical plant diversity and composition on soil microbial communities." Microbial Ecology 52: 226-238.
90
Chapin, F. S., III, L. R. Walker, et al. (1994). "Mechanisms of primary succession following deglaciation at Glacier Bay, Alaska." Ecological Monographs 64(149–175). de Graaff, M.-A., A. T. Classen, et al. (2010). "Labile soil carbon inputs mediate the soil microbial community composition and plant residue decomposition rates." New Phytologist 188(4): 1055-1064. DeAngelis, K. M., D. Chivian, et al. (2013). "Changes in microbial dynamics during long-term decomposition in tropical forests." Soil Biology and Biochemistry 66: 60-68. Drenovsky, R. E., K. L. Steenwerth, et al. (2010.). "Land use and climatic factors structure regional patterns in soil microbial communities" Global Ecology and Biogeography 19( 27-39). Eom, A., D. C. Hartnett, et al. (2000). "Host plant species effects on arbuscular mycorrhizal fungal communities in tallgrass prairie." Oecologia 122(435–444) Evans, S. E. and M. D. Wallenstein (2011). "Soil microbial community response to drying and rewetting stress: does historical precipitation regime matter?" Biogeochemistry 109(1-3):101-116. Fierer, N. and R. B. Jackson (2006). "The diversity and biogeography of soil bacterial communities." Proceedings of the National Academy of Sciences of the United States of America 103: 626-631. Fierer, N., D. Nemergut, et al. (2010). "Changes through time: integrating microorganisms into the study of succession." Research in Microbiology 161(8): 635-642. Frankland, J. C. (1998). "Fungal succession — unravelling the unpredictable." Mycological Research 102(1): 1-15. Frostegård, A. and E. Bååth (1996). "The use of phospholipid fatty acid analysis to estimate bacterial and fungal biomass in soil " Biology and Fertility of Soils 22(1-2 ): 59-65. Frostegård, Å., A. Tunlid, et al. (2011). "Use and misuse of PLFA measurements in soils." Soil Biology and Biochemistry 43(8): 1621-1625. García-Palacios, P., M. A. Bowker, et al. (2011). "Early-successional vegetation changes after roadside prairie restoration modify processes related with soil functioning by changing microbial functional diversity." Soil Biology and Biochemistry 43(6): 1245-1253. Harris, J. (2003). "Measurements of the soil microbial community for estimating the success of restoration " European Journal of Soil Science 54(4 ): 801-808. Harris, J. (2008 ). "Soil Microbial Communities and Restoration." Oikos 117: 1833.
91
Harris, J. (2009). "Soil Microbial Communities and Restoration Ecology: Facilitators or Followers?" Science 325(5940): 573-574. Hobbie, S. E. (1996). "Temperature and plant species control over litter decomposition in Alaskan arctic tundra. ." Ecological Monographs 66: 503–522. Hollister, E. B., C. W. Schadt, et al. (2010). "Structural and functional diversity of soil bacterial and fungal communities following woody plant encroachment in the southern Great Plains." Soil Biology and Biochemistry 42(10): 1816-1824. Holtkamp, R. (2010). Changing soils: Belowground food webs during secondary succession. environmental science department. Wageningen, Wageningen University. PhD: 111. Horner, J. D., J. Gosz, et al. (1988). "The role of carbon-based plant secondary metabolites in decomposition in terrestrial ecosystems." American Naturalist 132(869–883). Jangid, K., M. A. Williams, et al. ( 2010). "Development of soil microbial communities during tallgrass prairie restoration." Soil Biology and Biochemistry 42(302-312). Jangid, K., M. A. Williams, et al. (2011). "Land-use history has a stronger impact on soil microbial community composition than aboveground vegetation and soil properties." Soil Biology and Biochemistry 43(10): 2184-2193. Jia, G.-m., J. Cao, et al. (2005). "Microbial biomass and nutrients in soil at the different stages of secondary forest succession in Ziwulin, northwest China." Forest Ecology and Management 217(1): 117-125. Joergensen, R. and F. Wichern (2008). "Quantitative assessment of the fungal contribution to microbial tissue in soil." Soil Biology and Biochemistry 40(12): 2977-2991. Kaiser, C., A. Frank, et al. (2010). "Negligible contribution from roots to soil-borne phospholipid fatty acid fungal biomarkers 18:2ω6,9 and 18:1ω9." Soil Biology and Biochemistry 42(9): 1650-1652. Kardol, P., N. J. Cornips, et al. (2007). "Microbe-mediated plant-soil feedback causes historical contingency effects in plant community assembly." Ecological monographs 77(2): 147-162. Kardol, P. and D. A. Wardle (2010). "How understanding aboveground–belowground linkages can assist restoration ecology." Trends in Ecology & Evolution 25(11): 670-679. Kaur, A., A. Chaudhary, et al. (2005). "Phospholipid fatty acid – A bioindicator of environment monitoring and assessment in soil ecosystem." Current Science 89(7): 1103-1112. Kendrick, W. B. and A. Burges (1962). "Biological aspects of the decay of Pinus sylvestris leaf litter." Nova Hedwigia 4: 313-342.
92
Kirchmann, H. and M. Eklund (1994). "Microbial biomass in a savanna-woodland and an adjacent arable soil profile in Zimbabwe." Soil Biology and Biochemistry 26(9): 1281-1283. Knelman, J. E., T. M. Legg, et al. (2012). "Bacterial community structure and function change in association with colonizer plants during early primary succession in a glacier forefield." Soil Biology and Biochemistry 46: 172-180. Kraus, T. E. C., R. A. Dahlgren, et al. (2003). "Tannins in nutrient dynamics of forest ecosystem: a review." Plant and Soil 256: 41–66. Lauber, C. L., M. Hamady, et al. (2009). "Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial community structure at the continental scale. ." Applied and Environmental Microbiology 75: 5111. Lauber, C. L., M. S. Strickland, et al. (2008). "The influence of soil properties on the structure of bacterial and fungal communities across land-use types." Soil Biology and Biochemistry 40:2407-2415. Liao, J. D. and T. W. Boutton (2008). "Soil microbial biomass response to woody plant invasion of grassland." Soil Biology and Biochemistry 40(5): 1207-1216. Macdonald, C. A., N. Thomas, et al. (2009). "Physiological, biochemical and molecular responses of the soil microbial community after afforestation of pastures with Pinus radiata." Soil Biology and Biochemistry 41(8): 1642-1651. Marín-Spiotta, E., R. Ostertag, et al. (2007). "Long-term patterns in tropical reforestation: Plant community composition and aboveground biomass accumulation." Ecological Applications 17(3): 828-839. Marin-Spiotta, E., W. L. Silver, et al. (2009). "Soil organic matter dynamics during 80 years of reforestation of tropical pastures." Global change biology 15(6): 1584-1597. Marín-Spiotta, E., C. W. Swanston, et al. (2008). "Chemical and mineral control of soil carbon turnover in abandoned tropical pastures." Geoderma 143(1-2): 49-62. McCulley, R. L., S. R. Archer, et al. (2004). "Soil respiration nand nutrient cycling in wooded communities developing in grassland." Ecology 85(10): 2804-2817. Nemergut, D. R., S. P. Anderson, et al. (2006). "Microbial Community Succession in an Unvegetated, Recently Deglaciated Soil." Microbial Ecology 53(1): 110-122. Nemergut, D. R., S. P. Anderson, et al. (2007). "Microbial community succession in an unvegetated, recently deglaciated soil." Microbial Ecology 53(1): 110-122. Olsson, P. A. (1999). "Signature fatty acids provide tools for determination of the distribution and interactions of mycorrhizal fungi in soil." FEMS Microbiology Ecology 29: 303-310.
93
Olsson, P. A., E. Baath, et al. (1995). "The use of phospholipid and neutral lipid fatty acids to estimate biomass of arbuscular mycorrhizal fungi in soil." Mycological Restoriation 5: 623-629. Ostertag, R., E. Marín-Spiotta, et al. (2008). "Litterfall and Decomposition in Relation to Soil Carbon Pools Along a Secondary Forest Chronosequence in Puerto Rico." Ecosystems 11(5): 701-714. Paul, E. A. and F. E. Clark (1996). Soil Microbiology and Biochemistry. San Deigo, CA, Academin Press. Porazinska, D. L., R. D. Bardgett, et al. (2003). "Relationships at teh aboveground-belowground interface: plants, soil biota, and soil processes." Ecological Monographs 73(3): 377-395. Potthast, K., U. Hamer, et al. (2010). "Impact of litter quality on mineralization processes in managed and abandoned pasture soils in Southern Ecuador." Soil Biology and Biochemistry 42(1): 56-64. Quiquampoix, H., S. Servagent-Noinville, et al. (2002). Enzyme adsorption on soil mineral surfaces and consequences for the catalytic activity Enzymes in the Environment R. P. D. Richard G. Burns. New York, NY, Marcel Dekker New York 285-306. Ramette, A. (2007). "Multivariate analyses in microbial ecology." FEMS Microbiology Ecology 62(2): 142-160. Ratledge, C. and S. G. Wilkinson (1988). Microbial Lipids. London, Academic Press. Reynolds, H. L., A. Packer, et al. (2003). "Grassroots ecology: Plant-microbes-soil interactions as drivers of plant community structure and dynamics." Ecology 84(9): 2281-2291. Rousk, J., E. Bååth, et al. (2010). "Soil bacterial and fungal communities across a pH gradient in an arable soil." International Society for Microbial Ecology 4(1340-1351). Schmidt, S. K., E. K. Costello, et al. (2007). "Biogeochemical consequences of rapid microbial turnover and seasonal succession in soil." Ecology 88(6 ): 1379-1385. Schmidt, S. K., S. C. Reed, et al. (2008). "The earliest stages of ecosystem succession in high-elevation, (5000 metres above sea level), recently deglaciated soils." Proc. Royal Soc. B Biol. Sci. 275(2793-2802). SERCC (2013). "Southeast Regional Climate Center mean temperature and precipitation from 1971-200 at Guavate camp climate station, Puerto Rico." 2013, from www.sercc.com/cgi-bin/sercc/cliMAIN.pl?pr4867.
94
Smithwick, E. A. H., M. G. Turner, et al. (2005). "Variation in NH4+ mineralization and
microbial communities with stand age in lodgepole pine (Pinus contorta) forests, Yellowstone National Park (USA)." Soil Biology and Biochemistry 37(8): 1546-1559. Soil Survey Staff. (2008). Official Soil Series Descriptions Lincolcn, NE, USDA-NRCS. Sparks, D. L. et al. (1996). Methods of soil analysis. Part 3-Chemical methods., Soil Science Society of America Inc. Talbot, J. M. and K. K. Treseder (2012). "Interactions among lignin, cellulose, and nitrogen drive litter chemistry–decay relationships." Ecology 93(2): 345-354. Ushio, M., T. C. Balser, et al. (2012). "Effects of condensed tannins in conifer leaves on the composition and activity of the soil microbial community in a tropical montane forest." Plant and Soil 365(1-2): 157-170. Ushio, M., R. Wagai, et al. (2008). "Variations in the soil microbial community composition of a tropical montane forest ecosystem: Does tree species matter?" Soil Biology and Biochemistry 40(10): 2699-2702. van der Heijden, M. G. A., R. D. Bardgett, et al. (2008). "The unseen majority: soil microbes as drivers of plant diversity and productivity in terrestrial ecosystems." Ecology Letters 11(3): 296-310. van der Putten, W. H., R. D. Bardgett, et al. (2013). "Plant-soil feedbacks: the past, the present and future challenges." Journal of Ecology 101(2): 265-276. van der Putten, W. H., R. D. Bardgett, et al. (2009). "Empirical and theoretical challenges in aboveground–belowground ecology." Oecologia 161(1): 1-14. Wardle, D. A. (2004). "Ecological Linkages Between Aboveground and Belowground Biota." Science 304(5677): 1629-1633. Wardle, D. A., K. I. Bonner, et al. (1999). "Plant removals in perennial grassland: vegetation dynamics, decomposers, soil biodiversity, and ecosystem properties." Ecological Monographs 69 535–568. Williams, M. A., K. Jangid, et al. (2013). "Bacterial communities in soil mimic patterns of vegetative succession and ecosystem climax but are resilient to change between seasons." Soil Biology and Biochemistry 57: 749-757. Wu, T., D. O. Chellemi, et al. (2008). "Comparison of soil bacterial communities under diverse agricultural land management and crop production practices." Microbial Ecology 55: 293-310
95
Zak, D. R., W. E. Holmes, et al. (2003). "Plant diversity, soil mirobial communities and ecosystem funtion: Are there any links?" Ecology 84(8): 2042-2050. Zak, D. R., D. Tilman, et al. (1994). "Plant Production and Soil Microorganisms in Late-Successional Ecosystems: A Continental-Scale Study." Ecology 75(8): 2333-2347. Zelles, L. (1997). "Phospholipid fatty acid profiles in selected members of soil microbial communities." Chemosphere 35(1/2): 275-294. Zelles, L. (1999). "Fatty acid patterns of phospholipids and lipopolysaccharides in the characterisation of microbial communities in soil: a review." Biology and Fertility of Soils 29: 111-129.
96
CHAPTER 3:
Linking microbial ecology and soil organic matter aggregate stabilization with tropical land cover change
Abstract:
Soil microorganisms control multiple input and loss pathways of soil carbon (C); thus changes in
microbial communities are likely to affect soil organic matter (SOM) cycling and storage. This
study aimed to identify links between microbial community composition and the distribution of
SOM among soil aggregate fractions to answer the following research questions: (1) Are
different microbial groups associated with different SOM pools? (2) How do these relationships
differ with changes in vegetation during tropical forest succession? We measured microbial
composition, via phospholipid fatty acid (PLFA) analysis, and C and nitrogen (N) concentrations
on physically separated aggregate fractions of soils from pastures, secondary forests (40 and 90
years old) naturally regenerated on abandoned pastures, and primary forests in Puerto Rico. Our
study yielded three main results: (1) The majority of C and N (relative to bulk soil C and fraction
mass) was isolated in the macroaggregate-occluded silt and clay-sized fractions; (2) Microbial
community composition varied by aggregate fraction, with a smaller fungal-to-bacterial ratio in
smaller-sized aggregates and a greater gram-positive to gram-negative bacterial ratio in the silt
and clay fractions compared with the macroaggregate and microaggregate fractions; and (3)
Microbial composition varied by land cover type and forest successional stage, with the greatest
differences in community structure between the pastures and early secondary forests and the late
secondary and primary forests. Our results indicate that association with mineral surfaces in the
clay and silt-sized fractions contained within macroaggregates is the dominant stabilization
mechanism for SOM in these highly-weathered, fine-textured soils. This study also shows that
97
the soil matrix plays an important role in the spatial distribution of fungal and bacterial
dominated communities, and that this distribution is sensitive to changes in vegetation, with
potential implications for SOM storage and turnover. Understanding how microbial communities
respond to disturbance and ecosystem recovery is important for predicting effects of changes in
land use and land cover on belowground C pools and nutrient availability.
Keywords: Soil aggregates: Microbial community: PLFA: Tropical forest recovery: Soil carbon
98
1. Introduction
Soils are one of the largest reservoirs of global carbon (C), comprising more than 60% of
terrestrial C in the form of soil organic matter (SOM)(Houghton, 2007; Trumbore, 2009).
Changes in the storage of soil C driven by land use and land cover change are thought to have
contributed to rising atmospheric CO2 concentrations, with consequences for the global climate
(Houghton et al., 2000). Microorganisms largely determine the fate of C in soils, through their
control over SOM decomposition processes (McGuire and Treseder, 2009; Wieder et al., 2013).
Soils provide a heterogeneous environment for microorganisms, with a non-uniform distribution
of C substrates and nutrients (Balser et al., 2006; Ettema and Wardle, 2002) and this
heterogeneity is influenced by soil aggregation processes (Chenu et al., 2001). Different
microbial groups preferentially use different sources and quantities of C (Kramer and Gleixner,
2006, 2008; Paterson et al., 2008), and therefore changes in the relative abundance of key
microbial groups in soils, such as fungi, gram-negative and gram-positive bacteria, may
significantly alter SOM cycling and storage (Six et al., 2006). For example, fungal-dominated
communities are believed to enhance soil C sequestration due to increased biomass and higher
growth efficiencies compared to bacterial-dominated communities (Bailey et al., 2002; Holland
and Coleman, 1987; Jastrow et al., 2007; Rousk and Bååth, 2007; Six et al., 2006; Zhao et al.,
2005). Fungal byproducts and necromass are considered to have slower rates of decomposition
relative to bacterial biomass and residues (Guggenberger et al., 1999; Martin and 1986, 1986;
Six et al., 2006). An increase in soil fungal-to-bacterial abundance is therefore expected to lead
to greater soil C accumulation and reduced CO2 loss. However, many of these theories have not
been examined under field conditions (Six et al., 2006).
99
Interactions between SOM and the soil matrix are fundamental in the protection of
organic compounds from microbial decomposition and mineralization to CO2, primarily via
sorption or other interactions with mineral surfaces and occlusion within soil aggregates (Lützow
et al., 2006; Marschner et al., 2008; Schmidt et al., 2011; Tisdall and Oades, 1982). Soil
aggregation promotes the protection of soil C in several ways: SOM can become incorporated
into a hierarchical architecture of soil aggregates as a binding agent (Oades and Waters, 1991;
Tisdall and Oades, 1982); it can be stabilized on the surfaces of the inorganic constituents of soil
aggregates (Golchin et al., 1994) and it may become incorporated into anaerobic or inaccessible
cores of microaggregates where microbial activity is slowed (Sexstone et al., 1985). Fungal
hyphae, microbes and plant roots can aid in the stabilization of soil aggregates through the
production of exudates, secondary metabolites and organic inputs that act as glue between
organic and inorganic constituents of soil (Tisdall and Oades 1982; Jastrow and Miller 1998,
Wright and Upadhyaya 1998, Six et al. 2006). While the importance of organic materials as
primary binding agents in the development of soil aggregation was originally thought to be
important only in temperate soils characterized by high activity clays (Oades and Waters, 1991),
SOM can also be protected from decomposition through its association with soil aggregates in
highly weathered tropical soils, although the mechanisms may differ (Lehmann et al., 2001;
Shang and Tiessen, 1998). However, specific microbially-mediated processes controlling the use
and turnover of soil C within soil aggregates is largely unknown (Schimel and Schaeffer, 2012),
especially in tropical soils. This is especially true when it comes to understanding the
relationship between microbial composition and SOM pools. The spatial distribution of
microbial communities relative to their C substrates can improve our understanding of how
microbial composition mediates SOM distribution and stabilization.
100
Macroaggregates (2000– 250 µm size class) typically have short residence times in the
field and generally contain relatively labile C and more recent C inputs, and thus are not
considered stabilized pools of SOM (Elliott, 1986; Tisdall and Oades, 1982). On the other hand,
microaggregates (250 - 53 µm) and the silt- and clay-sized aggregates (< 53 µm) are recognized
for their contribution to long-term C storage due to physical occlusion and mineral protection
from adsorption interactions with clay, respectively (Six et al., 2000). Soil aggregates can also
contribute to the persistence of organic compounds through their role in creating complex soil
structure and limiting accessibility between decomposers and C substrates.
Much of the research on microbial abundance among aggregates and particle-size
fractions have shown conflicting results in the spatial distribution of microbial biomass and
composition (Torsvik and Øvreås, 2002). In the majority of studies, microbial biomass is greatest
in the smallest-sized fractions (silt and clay) (Kandeler et al., 1999; Kandeler et al., 2000;
Monrozier et al., 1991; Poll et al., 2003; Qin et al., 2010). At the same time, however, other
studies reported greater biomass or microbial abundance in coarse or larger-sized fractions (Briar
et al., 2011; Chiu et al., 2006; Huygens et al., 2008). Fungal abundance (Chiu et al., 2006;
Huygens et al., 2008) and the fungal-to-bacterial ratio (Kandeler et al. 2000, Poll et al. 2003,
Briar et al. 2011) typically decrease with diminishing particle size, but some have reported no
change in the fungal to bacterial ratio among fractions (Huygens et al. 2008, Chiu et al. 2006).
The variability in results among studies may be due to soil-specific processes such as differences
in soil C and N (Chiu et al., 2006) as well as to differences in soil mineralogy and to differences
in the methods used to characterize microbial biomass and community composition.
While these studies, conflicting as they are, offer insight into relationships between
microbial communities and the distribution of soil aggregate size classes, most are focused on
101
agricultural treatments in temperate soils. We are not aware of any studies investigating
microbial composition in soil fractions in highly weathered tropical soils. Tropical soils store
large amounts of the world’s terrestrial C and year-round productivity and fast decomposition
rates make them important contributors to the global C cycle. Furthermore, the response of soil
microbes and their association with SOM pools during tropical reforestation has not been
studied, despite the large potential for tropical secondary forests to act as a C sink. Despite the
broad geographic expansion of secondary forests across the tropics (Grau and Aide, 2008;
Meiyappan and Jain, 2012), large uncertainties surround the fate of soil C and the role of
microorganisms in forests growing on disturbed land.
The aim of our research was to evaluate how microbial community composition interacts
with soil aggregates to shape soil C dynamics during tropical secondary forest succession on
abandoned pastures. To test the relationship among microbial composition, soil aggregate
fractions and soil C and N pools, we measured microbial biomass and composition via
phospholipid fatty acid-fatty acid methyl ester analysis-(PLFA) and C and N concentrations in
the following physically-separated soil aggregate fractions: (1) macroaggregates (2000 - 250
µm), microaggregates (250 - 53 µm), (2) macroaggregate-occluded microaggregates (250 - 53
µm), (3) macroaggregate-occluded and silt and clay (< 53 µm) fractions, and (4) free silt and
clay-sized fractions (< 53 µm) in surface soils (0-20 cm) collected from active pastures,
secondary forests and primary forests along a reforestation chronosequence on highly-weathered
soils.
2. Methods
2.1 Field site description
102
This study was conducted on an established long-term replicated successional chronosequence
consisting of active pastures, secondary forests growing on abandoned pastures, and primary
forest sites that have not been under pasture or agricultural use in the Sierra de Cayey of
southeastern Puerto Rico (18°01´ N, 66°05´ W), (Marin-Spiotta et al., 2007; Marin-Spiotta et al.,
2009). Forest vegetation differed along the reforestation chronosequence, with the first 30-40
years of succession on abandoned pastures dominated by the early successional tree Tabebuia
heterophylla. Late secondary forests had a mixed-species canopy, while the primary forests had
high abundances of Dacryodes excelsa and the palm Prestoea acuminata var. montana. All sites
were located within 5 km of each other, between 580-700 m elevation and experience a mean
annual temperature of 21.5 ºC (from 1971-2000, Southeast Regional Climate Center) and mean
annual precipitation of 2000 mm (from 2010 to 2012, Caribbean Atmospheric Research Center,
University of Puerto Rico-Mayagüez). Soils were characterized as very-fine, kaolinitic,
isothermic Humic Hapludox in the Los Guineos soil series (Soil Survey Staff, 2008). Soils are
strongly aggregated, acidic (mean pH of 4.0 ± 0.5) and rich in clay, and aluminum and iron
oxides and hydroxides.
2.2 Field sampling
Three replicate soil samples were collected in March 2012 from three replicate sites each of
active pastures, early secondary forests (40 years old), late secondary forests (90 years old), and
primary forest sites, with the exception of the early secondary forests, which only two replicate
sites as the third site was recently deforested. Approximately 20 cm3 of soil was collected from
0-20 cm depth using a shovel to minimize disruption of aggregate structure. Fresh soils were
shipped to the University of Wisconsin-Madison within 48 hours of collection and gently sieved
103
through a 2 mm sieve to remove roots, rocks and litter. Approximately 200 g of sieved soils were
then frozen at -20 °C before being shipped to Boise State University for aggregate fractionation
in August 2012. Soil samples were thawed at 4 °C for 24 hours prior to fractionation.
2.3 Soil aggregate fractionation
Soils were fractionated using a modified wet sieving and aggregate isolation protocol for
microbiological analysis as described below (modified from Allison and Jastrow, 2006; Six et
al., 2000) producing the following fractions: small macroaggregates (2000 – 250 µm), free
microaggregates (250 - 53 µm), free silt and clay (< 53 µm), microaggregates occluded within
macroaggregates (250 - 53 µm), and silt and clay occluded within macroaggregates (< 53 µm)
(Figure 1). Macroaggregate-occluded fractions were separated using a microaggregate isolator
(Six et al., 2000). Coarse particulate organic matter (POM) was also isolated using the
microaggregate isolator, but often consisted of roots and gravel that were contained within the
macroaggregates and was therefore excluded from microbial and C analysis. Each soil sample
was fractionated in duplicate and fractions were composited to produce enough material for
chemical and microbiological analysis.
Briefly, a starting weight of 80-100 g of soil were slaked, or submerged in 1 cm of de-
ionized water, for 5 minutes prior to performing the wet-sieving process. Soils were wet-sieved
by moving the sieve in and out of the water 50 times with an amplitude of approximately 3 cm
for 2 minutes using a digital metronome. Macroaggregates (2000 - 250 µm), microaggregates
(250 - 53 µm) and a silt and clay fraction (< 53 µm) were separated via wet-sieving and
subsampled for microbial analyses and for dry-weight conversion. Subsamples for microbial
analyses were stored at -80° C in sterilized whirl-pak bags until they could be shipped to UW-
104
Madison and freeze-dried. Due to the silt and clay fraction being in solution at the end of the
wet-sieving process, a portion of the slurry was centrifuged at 2500 rpm for 2 minutes, decanted
and then subsampled for microbiological analysis and dry-weight conversion. Subsamples (10 -
15 g) of macroaggregates were placed into the microaggregate isolator for separation of
macroaggregate-occluded fractions. Macroaggregates in the microaggregate isolator were shaken
for 5 minutes on high and 20 minutes on low to break apart the aggregates. Macroaggregate-
occluded coarse POM was retained in the 250 µm sieve attached to the shaker. The material <
250 µm was collected on a 53 µm sieve and was wet-sieved using the same procedure to separate
macroaggregate-occluded microaggregates and macroaggregate-occluded silt and clay.
The dry-weight equivalent of each soil fraction subsampled for microbiological analyses
was calculated using 2-10 g subsamples oven-dried at 100 °C for 48 hours. The remaining
portion of each recovered fraction was weighed after being dried at 60 °C for 48-72 hours (or
until dry weights stabilized) and was then added to the dry-weight equivalent of subsamples for
microbiological analyses and dry weight calculation subsamples. The subsampling and potential
heterogeneity in soil moisture content in these high clay and strongly aggregated soils resulted in
recovery weights totaling > 100% for 75 % of the samples. To correct for this error, we report
contributions of fractions to the total C pool and soil mass based on the final recovery sum of the
free fraction masses. Furthermore, the contribution of macro-aggregate occluded fractions to the
bulk soil was normalized for mass relative to the total recovered mass of the macroaggregate
fractions, as they were released by disruption of the macroaggregates.
2.4 Soil carbon and nitrogen
105
Total C and N concentrations were determined on finely ground, oven-dried (60 °C) soil physical
fractions using a Flash 2000 NC Analyzer (Thermo Scientific, Wilmington, Delaware) at
University of Wisconsin-Madison. Soil samples were ball-milled using a SPEX Sample Prep
mixer/Mill (Metuchen, New Jersey). All samples were run in duplicate with replicate error < 10
% using aspartic acid and Montana soil reference material as calibration and internal standards.
As this soil contains no inorganic carbon, total C concentrations (% C) represent organic C. Soil
C:N ratios were calculated as molar ratios. Fraction C and N per bulk C and N were calculated as
the proportion of each fraction (by weight) multiplied by the % C, N normalized to the % C, N of
the bulk sample. Proportions of the macroaggregate-occluded fractions were calculated as both
proportions by mass to the overall sum of free aggregates and as a proportion of macroaggregate
mass.
2.5 Microbial community composition
Microbial community composition was measured using a hybrid phospholipid fatty acid and
fatty acid methyl ester analysis protocol, hereby referred to as simply PLFA (Smithwick et al.,
2005). Briefly, PLFAs were extracted from freeze-dried and homogenized soil (1.0 g) using a
specific ratio (1:2:0.9) of chloroform, methanol, and a phosphorus-buffer. We used 0.5 g for
fractions with < 1.0 g of soil material recovered during the fractionation procedure. Insufficient
material and a difficulty in PLFA extraction limited the number of macroaggregate-occluded
fractions that could be used in later analyses. After isolating and concentrating the extracted
PLFAs, they were saponified, methylated, transferred to an organic phase and then washed with
a basic NaOH solution. Lipid stock standards (9:0 and 19:0) (Sigma-Aldrich, St. Louis, MO,
USA) of known concentrations (7.08 µg/ml for 19:0 and 9 µg/ml for 9:0) were added to each
106
sample. Samples were run on a Hewlett-Packard 6890 Gas Chromatograph equipped with a
flame ionization detector and an Ultra 2 capillary column (Agilent Technologies Inc., Santa
Clara, CA, USA). Peaks were identified using bacterial fatty acid standards and peak
identification software (MIDI Inc, Newark, DE, USA). Final volumes of PLFAs of low-mass
samples were dried down and reconcentrated for optimal performance of the MIDI peak
identification software. Peak areas were converted to µmol PLFA g soil -1 (absolute abundance)
using internal standard peaks (9:0, 19:0).
Microbial biomass was calculated as the sum of all peaks (as µmol PLFA g soil -1)
identified < 20.5 C atoms long (Vestal and White, 1989; Zelles, 1999). Select PLFAs were used
as indicator species for all other analyses: 15:0iso (Gram-positive bacteria; Kaur et al., 2005;
Zelles, 1997, 1999), 16:0 10methyl (Actinobacteria; Ratledge and Wilkinson, 1988); 16:1 w7c
(Gram-negative bacteria; Ratledge and Wilkinson, 1988; Zelles, 1999), 16:1 w5c (Arbuscular
Mycorrhizal Fungi; Olsson, 1999; Olsson et al., 1995), 18:1 w9c (Saprotrophic Fungi; Bardgett
et al., 1996; Frostegård et al., 2011), 18:2 w6,9c (Saprotrophic Fungi; Frostegård and Bååth,
1996; Joergensen and Wichern, 2008; Kaiser et al., 2010) and 19:0cyclo (Anaerobic, gram-
negative bacteria; Vestal and White, 1989). Fungal abundance was calculated as the sum of the
relative abundance of 18:1 w9c and 18:2 w6,9c (Frostegård and Bååth, 1996).
2.7 Statistical analyses
Principal Component Analysis (PCA) of select indicator PLFA biomarkers was performed on the
arcsine-square root transformed relative abundances (Ramette, 2007), while untransformed
relative abundance values of indicator PLFAs were used in all other analyses. Standard error
107
calculations for PCA figures were pooled for fixed treatment effects, while propagated standard
error (SE) was calculated for reported means to account for within and among site variability.
A split-plot, random effects standard least square model was used to analyze both
chemical (C and N concentrations, C:N ratios, as well as the distribution of C and N across soil
fractions relative to bulk soil) and microbiological variables (PLFA biomass, indicator species,
fungal-to-bacterial ratio, fungal biomass) across the chronosequence and among physical
fractions. Restricted maximum likelihood (REML) models were run on site means and weighted
by within site replicates to account for uneven replication. Because the fractionation procedure
produced very little material for certain fractions (especially the macroaggregate-occluded
fraction), the number of soil samples was reduced to account for balanced statistical analyses.
For all analyses, at least two replicate samples obtained from at least two replicate sites per land
cover type were used, when all three replicate samples per sites did not yield enough material.
Statistical analyses testing for differences in microbial PLFA composition among all fractions
(both free and macroaggregate-occluded fractions) were only performed on early and late
secondary forest and primary forest soils as there were not enough replicates in macroaggregate-
occluded fractions from the pasture soils due to insufficient material recovery and unsuccessful
PLFA extractions on pasture soil fractions. Statistical analyses using all fractions (both
macroaggregate-occluded and free) should be interpreted with caution, as fractions are
operationally not independent of each other. Statistical analysis was performed using JMP Pro
Version 10 (SAS Inst. Inc., Cary, NC, USA). Relationships are reported as significant at p <
0.05, unless noted otherwise in the text.
108
3. Results
3.1 Physical fraction carbon and nitrogen
There were significant differences in the carbon concentration, carbon-to-nitrogen ratio (C:N),
and the distribution of C and N among soil fractions (Table 1, 2). Averaged across land cover
types, the concentration of C was greatest in the free microaggregate and macroaggregate-
occluded microaggregate fractions (4.16 ± 1.32 and 4.14 ± 1.82 g C/100 g soil, respectively)
compared to the macroaggregate-occluded silt and clay fractions (3.45 ± 1.26 g C/100 g soil)
(Table 1).
The distribution of C among fractions (or the relative amount of bulk soil C recovered in
each fraction) differed among fraction types (p < 0.0001, Table 2) with the macroaggregates
containing the greatest proportion of C relative to all other fraction types (74.6 ± 23.2 % of bulk
soil C, Table 1). The majority of C held within the macroaggregates (52.6 ± 21.5% of
macroaggregate C) was contained in the occluded silt and clay fractions. Relative to the bulk
soil, the macroaggregate-bound silt and clay fractions contained the second largest proportion of
C relative to all other fraction types (38.9 ± 22.6% of bulk soil C). Other fractions did not differ
in their contribution to bulk soil C. Therefore, while microaggregates had greater C
concentrations than all other fractions, the greater proportion of bulk C was contained in the
macroaggregate-occluded silt and clay fraction.
N concentrations, which averaged 0.31 ± 0.25 g N/100 g soil, did not differ by soil
fraction type (see Table 1, 3 for means of individual fractions and land cover). However, the
distribution of N (% of bulk soil N recovered in each fraction) and the C:N ratio differed among
soil physical fractions (p < 0.0001, Table 2). Similar to the distribution of C, the macroaggregate
and macroaggregate-occluded silt and clay fractions contained the greatest proportion of N
109
relative to the other fractions (74.3 ± 23.4 % and 44.3 ± 25.7 %, respectively) (Table 1). Other
physical fractions showed no difference in N distribution. Carbon-to-N ratios were lower in the
silt and clay fractions (both free and macroaggregate-occluded) compared to all other fractions
(Table 1).
3.2 Microbial community composition in aggregate fractions
3.2.1 Microbial composition from pastures and forests in free aggregate fractions
Microbial community structure varied among aggregate fractions (Figure 2). Community
structure among the macroaggregates, microaggregates, and the free silt and clay fractions
differed along the first and second principal component axes (p < 0.0001 for both PC1 and PC2,
Table 4). Principal component one was negatively correlated with C and N concentrations (Table
4). The PLFA biomarkers for gram-positive bacteria (15:0 iso) and fungi (18:1w9c) were highly
correlated (85.9% and 76.8%) with PC1, explaining the majority of variation in community
structure along that axis. Along PC2, the PLFA biomarkers for actinobacteria (16:0 10 Methyl)
and methanotrophic, gram-negative bacteria (18:1w7c) explained the greatest proportion of
variation in community structure (70.0 % and 69.3 %, respectively). Additionally, the indicator
PLFA for gram-positive bacteria was highly correlated with the indicators for actinobacteria
(73.3 %) and anaerobic, gram-negative (19:0 cyclo) bacteria (77.7 %).
While total microbial biomass did not differ by soil fraction (Table 1, 3), the fungal-to-
bacterial ratio (F:B) differed among fractions (p = 0.023) (Figure 3). F:B was highest in the
macroaggregate fractions (1.07 ± 0.28), followed by the microaggregate (0.85 ± 0.21), and the
silt and clay fractions (0.71 ± 0.13). The pattern in F:B ratios was largely driven by shifts in the
relative abundance of fungi as the relative abundance of bacteria remained constant (data not
110
shown). Fungal-to-bacterial ratio was positively but weakly (R2 = 0.11) correlated with the C:N
ratio. Similarly, microbial biomass was positively but weakly (R2 = 0.094) correlated to % C.
Biomass normalized by % C did not differ among fractions (data not shown). The gram-positive-
to-gram-negative bacterial ratio varied by soil fraction (p < 0.0001) and the interaction between
soil fraction and land cover type (p = 0.005). The gram-positive-to-gram-negative bacterial ratio
was highest in the silt and clay fractions compared to the macroaggregate and microaggregate
fractions.
The abundance of PLFA indicator biomarkers varied among fractions (Table 5). The
relative abundance of indicator PLFAs for fungi (16:1w5c, 18:1w9c, 18:2w6,9c), gram-negative
bacteria (16:1w7c) and methanotrophic bacteria (18:1w7c) all differed among fractions. Fungal
abundance was greatest in the macroaggregate fractions, while gram-negative bacterial
abundance was greatest in the microaggregate and silt and clay fractions. The abundance of the
indicator PLFA for methanotrophic bacteria was greater in the macroaggregate and
microaggregate fractions compared to the silt and clay fractions. The interaction between land
cover and fraction was statistically significant for PLFAs 18:1w7c, 18:1w9c, and 18:2w6,9c.
3.2.2 Microbial composition from forests in free and occluded aggregate fractions
Since the pasture soils yielded insufficient material for microbial analyses of fractions within the
macroaggregates, comparisons among occluded fractions are reported for the forest sites only.
Mean indicator PLFA biomarkers representing microbial community structure differed among
aggregate fractions across all forest types (Figure 5). Soil aggregate fraction and the interaction
between fraction and forest age had a significant effect on both PC1 and PC2 (Table 6). The
PLFA biomarker for actinobacteria (16:0 10 Methyl), anaerobic, gram-negative bacteria (19:0
111
cyclo) and methanotrophic bacteria (18:1w7c) were highly correlated (85.3 %, 69.2 % and 69.0
%) with PC1, explaining the majority of variation in community structure along that axis. Along
PC2, the PLFA biomarkers indicating fungi (18:2w6,9c) and gram-positive bacteria (15:0 iso)
explained the greatest amount of variation in community structure (70.2 % and 65.2 %,
respectively).
Microbial PLFA biomass, fungal-to-bacterial ratio, gram-positive – to- gram-negative
ratio, and the relative abundance of PLFA indicators for arbuscular mycorrhizal fungi,
actinobacteria, methanotrophic bacteria and anaerobic bacteria significantly differed by soil
fraction (see Table 7 for p-values). The fungal-to-bacterial ratio decreased from macroaggregates
> free microaggregates > free silt and clay, with the macroaggregate-occluded fractions being
intermediate between macroaggregates and free microaggregates. The gram-positive – to- gram-
negative ratio was significantly higher in the free and macro-aggregate occluded silt and clay
fraction compared to the macroaggregates and the macroaggregate-occluded microaggregate
fraction, with the free microaggregate fraction being intermediate. The relative abundance of
actinobacteria indicator PLFAs was reduced in the macroaggregate-occluded fractions compared
to the free macroaggregate, microaggregate and silt and clay fractions.
3.3 Land cover effects on aggregate fractions, carbon and nitrogen, and microbial community
composition
The carbon-to-nitrogen ratio, concentration of C, N, and the distribution of C and N (% of bulk
soil C and N, respectively) did not vary among pastures, secondary forests (early and late) and
primary forests (Table 2). There was a significant interaction between land cover and fraction on
the concentration of C (Table 2). This effect was mostly driven by differences between the
112
macroaggregate-occluded silt and clay fractions and the microaggregate fractions (both free and
macroaggregate-occluded) from the primary forests (Table 3).
Land cover did play a significant role, however, in determining microbial community
composition. In the principal components analysis of free fractions (macroaggregate,
microaggregate, and the silt and clay aggregate fractions – Figure 2), PC1 values differed
between land cover types with the pasture and early secondary forest communities being
significantly different from those of the late secondary and primary forests (Table 4). In addition,
the abundance of PLFA indicator biomarkers varied among different forest types (Table 5). The
biomarker indicating gram-positive bacteria (15:0 iso) and anaerobic, gram-negative (19:0 cyclo)
bacteria differed by land cover, but not by fraction. The relative abundance of 15:0 iso was
significantly reduced in the pastures relative to the late secondary and primary forests. The
fungal-to-bacterial ratio (F:B) also differed across land cover types with greater values in the
pasture and early secondary forest relative to the late secondary and primary forests (p < 0.0001,
Figure 3).
When analyzing microbial community structure from both macroaggregate-occluded and
free fractions, PC 2 values differed significantly across forest types (e.g. early secondary, late
secondary and primary forests), (Figure 5, Table 6). Land cover also had a significant effect on
the relative abundance of PLFA biomarkers for saprotrophic fungi and gram-positive bacteria,
with fungal abundance decreasing with forest age and gram-positive bacteria highest in the late
secondary forests (data not shown). There was a significant interaction between soil fraction and
land cover for all indicator species, which was mostly due to variations in the macroaggregate-
occluded silt and clay fractions in the early secondary forests (data not shown).
113
4. Discussion
4.1 Carbon is preferentially associated with clay and silt-size fractions within aggregates
Our physical fractionation approach recovered greater amounts of total soil C (about 75 %) in the
largest aggregate size classes (2000 – 250 µm). These findings are consistent with models of
aggregate hierarchy that predict greater C content with increasing aggregate size as the larger
size classes will encompass C within and in between smaller aggregates making up the
macroaggregates (Oades and Waters, 1991; Six et al., 2000). However, most of the
macroaggregate-C was associated with the silt and clay size fractions released from within the
larger fractions, which made up the bulk of the macroaggregate mass. These results suggest that
soil C in our highly weathered, fine textured soils is stabilized by organo-mineral interactions,
rather than by the occlusion of particulate organic matter inside soil aggregates. Similarly,
previous research at the same sites reported > 80 % of total soil C in the heavy density, or
mineral-associated pool (Marin-Spiotta et al., 2009).
Elemental composition of macroaggregates, macroaggregate-occluded microaggregates
and free microaggregates were consistent with patterns of greater transformation of plant inputs
with decreasing fraction size reported in the literature (Baisden et al., 2002; Marin-Spiotta et al.,
2009; Marín-Spiotta et al., 2008; Six et al., 1998). For example, the silt and clay fractions (both
free and macroaggregate-occluded) had significantly lower C:N ratios compared to all other
aggregate fractions, suggesting differences in the degree of alteration of SOM in the different
physical fractions and preferential enrichment of microbial products and microbially-processed
SOM in mineral fractions (Kaiser and Kalbitz, 2012; Kramer et al., 2003; Miltner et al., 2012;
Oades, 1988).
Overall, the importance of macroaggregates on C stabilization has been mainly attributed
114
to their influence on microaggregate formation and turnover (Denef et al., 2007; Six et al., 1998;
Six et al., 2000). Macroaggregates are thought to be less stable than microaggregates, thus prone
to increased turnover and release of more stable microaggregates (Six and Jastrow, 2002; Tisdall
and Oades, 1982). However, across our pasture to forest chronosequence, the macroaggregate
fraction was resistant to slaking and extra measures had to be used in order to separate them from
other free fractions, and to release macroaggregate-occluded aggregates in the microaggregate
isolator. In these highly weathered, oxide-rich soils, the enhanced stability of macroaggregates
and protection of SOM from further decay is chiefly controlled by mineral surface interactions.
While clay minerals and Fe and Al oxides are recognized to stabilize aggregates in soils with 1:1
mineralogy, current models of soil aggregate dynamics do not predict C enrichment with
increasing aggregate size for these soils (Oades and Waters, 1991). Here we show that
aggregation of C-rich silt and clay-size fractions results in the accumulation of SOM with
increasing aggregate size, even when organic C does not act as the primary binding agent.
4.2 Microbial composition as a function of aggregate structure
Overall, shifts in microbial composition among aggregate fractions suggest that the microbial
abundance of key indicator groups is linked to the content and composition of soil C within
aggregate fractions. Differences in the quantity and potential bioavailability of soil C with
differences in the contribution of fungi, gram-positive and gram-negative bacteria in the silt and
clay fractions also indicate the importance of clay mineral-interactions in influencing microbial
communities and C stabilization processes.
Decreasing fungal abundance with soil particle size has been consistently documented
across soil and vegetation types (Briar et al., 2011; Chiu et al., 2006; Kandeler et al., 1999;
115
Kandeler et al., 2000; Poll et al., 2003; Schutter and Dick, 2002). Greater fungal abundance in
macroaggregates and larger particle size fractions has been attributed to increased availability of
C substrates (Briar et al., 2011; Chiu et al., 2006). Further, fungi may play a chief role in
stabilizing soil aggregates and have therefore been associated with macroaggregates (Tisdall and
Oades, 1982). The theory of aggregate hierarchy predicts that fungal hyphae, plant roots and
other organic materials bind smaller aggregates into larger aggregates, thus their contribution
increases with increasing aggregate size (Briar et al., 2011; Huygens et al., 2008; Tisdall and
Oades, 1982). Greater fungal abundance in larger or coarser-grained particle fractions has also
been attributed to fungal cell disruption during sieving methods (Chiu et al., 2006).
The greater recovery of organic C in the silt and clay-sized fractions within
macroaggretates in these highly-weathered soils suggests that the greater abundance of fungal
biomarkers in the larger size classes is not likely explained by the hierarchical contribution of
fungal hyphae and other organic binding agents to aggregate stability (sensu Oades and Waters,
1991). Instead, the higher relative abundance of fungi in the macroaggregates could be due to
higher SOM C:N ratios, which are known to favor fungal colonization (Bossuyt et al., 2005;
Eiland et al., 2001; Six et al., 2006), more favorable substrate properties (Huygens et al., 2008),
fungal physiology and restricted access to smaller pore sizes (Killham, 1994; Six et al., 2006).
On the other hand, bacterial enrichment in smaller particle sizes can be explained by pore-size
exclusion of fungi and macroorganisms (Heijnen and Van Veen, 1991), reduced bacterial
predation (Elliott and Coleman, 1988; Ladd et al., 1996) and greater nutrient availability in
smaller aggregate fractions (Van Gestel et al., 1996). Clay minerals have also been noted for
their protection of bacterial biomass against predation (Elliott et al., 1980; Ladd et al., 1996;
116
Rutherford and Juma, 1992), desiccation (Huang, 2004), fluctuating physicochemical
environments (Stotzky, 1986; Theng et al., 1995) and inhibitory compounds (Filip et al., 1972).
The greater relative abundance of fungi versus bacteria in the larger aggregate fractions is
consistent with fungal ecology and behavior. Fungal-dominated communities are generally
associated with enhanced C stabilization relative to bacterial-dominated communities as fungal
biomass, residues and secondary metabolites are more resistant to decomposition, which leads to
greater soil C accumulation (Bailey et al., 2002; Six et al., 2006). We might therefore assume
that higher F:B ratios would be associated with the fractions with more stable C. In our study, the
silt and clay-sized fraction contained a greater proportion of bulk soil C, and contained more
microbially-processed SOM (i.e. greater amounts of mineral-stabilized C and lower C:N). While
the silt and clay-sized fractions protected within macroaggregates contained more fungi than
bacteria in the forest soils, the free silt and clay fraction had the lowest fungal-to-bacterial ratio
relative to other fractions in both the forest and pasture soils.
The increasing gram-positive – to – gram-negative ratio in the silt and clay fraction is
also consistent with our general understanding of gram-positive and gram-negative ecology.
Gram-positive bacteria are well adapted to using older and more microbially-processed forms of
SOM (Fierer et al., 2003; Griffiths et al., 1999; Kramer and Gleixner, 2006). Gram-negative
bacteria, on the other hand, are more adapted to using fresh plant inputs as a C source, which is
why they are often more abundant in rhizosphere soils (Griffiths et al., 1999; Kramer and
Gleixner, 2006; Potthoff et al., 2006). Thus, the increase in gram-positive bacteria relative to
gram-negative bacteria in the silt and clay fraction corresponds well with the lower C:N ratios in
the silt and clay fractions. A higher relative abundance of gram-positive bacteria in the clay
fraction may indicate mineralization of more decomposed C, whereas a lower gram-positive – to
117
– gram-negative ratio in larger aggregate fractions may indicate decomposition of more recently
derived, labile plant material (Kramer and Gleixner, 2006).
4.3 Land cover influences microbial composition, not C, N or aggregate dynamics
While differences in the abundance of microbial indicator species were greatest among
soil fractions, land cover type also had an effect on soil microbial community composition with
different microbial community composition between pastures and forests and with forest
succession. This successional pattern in microbial community structure in the soil physical
fractions is consistent with data for bulk soils collected across several years and seasons at the
same sites (Smith et al., in prep.), suggesting similar controls operating at different spatial scales
in the soil matrix.
In general, the fungal-to-bacterial ratio is expected to increase with reforestation of
agricultural land as fungi dominate decomposition of lignin and hemi-cellulose, which are
commonly found in greater concentrations in forest litter than in forage grasses (Marín-Spiotta et
al., 2008; Paul, 2006). In contrast our results showed a greater abundance of mycorrhizal fungi,
saprotrophic fungi and a higher fungal-to-bacterial ratio in physical fractions from pastures and
early secondary forest sites compared to the late secondary and primary forest sites in our study.
This pattern, while not consistent with trends in litter chemistry for our sites (Marín-Spiotta et
al., 2008), is supported by data measured for bulk soils from these sites in both wet and dry
seasons during the period of this study (Smith et al., in prep.). Many field studies similarly report
that fungal-to-bacterial ratios are not consistently greater in forests compared to pastures or
agricultural soils, illustrating that the relationship between land cover change and microbial
118
composition remains unclear (Burke et al., 2003; Chaer et al., 2009; Potthast et al., 2012;
Strickland and Rousk, 2010).
Few studies have measured changes in microbial communities and their relationships to
aboveground succession during forest regeneration (but see Hedlund, 2002; Macdonald et al.,
2009). Data on the response of belowground communities to land cover change is especially
sparse in the tropics (Acosta-Martínez et al., 2007; Waring et al., 2013). Knowledge of the
drivers, response rates and successional trajectories of individual microbial groups, such as fungi
and bacteria, to changes in vegetation and land use is highly uncertain, limiting our ability to
predict ecosystem recovery and the fate of soil C during land use transitions. Furthermore,
interactions among microbial decomposer groups and SOM pools and the consequences for
changes in community composition on C cycling are not well understood. Our data suggests that
differences in soil aggregate structure may influence changes in microbial community
composition and function with land use change.
5. Conclusions
Soil aggregation is an important mechanism influencing the spatial distribution and stabilization
of C and microorganisms within the soil matrix (Six et al., 2002). Our study revealed the
importance of organo-mineral interactions in defining the relationship between soil aggregates,
microbial communities and SOM storage in highly weathered tropical soils. The silt and clay-
sized fractions contributed to the accumulation of C inside large soil aggregates and we expect
that binding between oxide-rich minerals played an important role in macroaggregate stability.
Although organic C did not appear to be an important aggregate-binding agent in these fine-
textured Oxisols, larger aggregate sizes did contain greater proportions of the bulk soil C pool,
119
consistent with predictions of aggregate hierarchy. However, this enrichment of C in the larger
size classes was due to C association with occluded silt and clay-size mineral fractions. Different
biomass ratios of fungi to bacteria and gram-positive to gram-negative bacteria among physical
soil fractions indicate interactions between microbial community composition and the
biochemistry and spatial distribution of SOM pools. Differences in the association of microbial
groups with physical soil fractions during reforestation suggest that changes in vegetation or soil
structure during land cover change can ultimately affect soil biogeochemical processes. Our
results demonstrate that while clay and silt-size organo-mineral interactions primarily drive the
accumulation and stabilization of SOM in highly weathered mineral soils, spatial heterogeneity
in microbial composition among aggregates can have important implications for decomposer
activity and soil C turnover.
120 6. Tables and Figures Table 1. Mean and propagated standard error for mass recovery, carbon (C), nitrogen (N) concentrations, C:N ratios, contributions to bulk C and N and microbial biomass in soil aggregate fractions from 0–20 cm depth averaged across all land covers (pasture, early secondary forest, late secondary forest and primary forest). There are two replicate sites per land cover class, except in the 90-year forests where there were three replicate sites. Different letters down each column represent significant differences among physical fractions using Tukey’s method of mean comparisons.
Fraction % of bulk soil* gC/100g soil gN/100g soil C:N Distribution of C (% of bulk soil C)
Distribution of N (% of bulk soil N)
Biomass (µmol PLFA/g soil)
Macroaggregate 77.3 ± 23.4 a 3.76 ± 1.94 ab 0.28 ± 0.13 15.7 ± 2.7 a 74.6 ± 23.2 a 74.3 ± 23.4 a 0.224 ± 0.14
Macroaggregate-occluded Microaggregate**
n.a. (24.7 ± 13.8) c
4.14 ± 1.82 a 0.31 ± 0.13 16.0 ± 1.9 a 18.8 ± 12.7 c (25.6 ± 15.2)
18.5 ± 13.2 c (25.2 ± 16.0)
0.302 ± 0.24
Macroaggregate-occluded Silt and clay**
n.a. (58.7 ± 19.6) b
3.45 ± 1.26 b 0.30 ± 0.11 13.7 ± 1.8 b 38.9 ± 22.6 b (52.6 ± 21.5)
44.3 ± 25.7 b (60.0 ± 25.4)
0.250 ± 0.09
Free microaggregate 12.0 ± 6.1 d 4.16 ± 1.32 a 0.32 ± 0.08 15.4 ± 2.5 a 12.5 ± 6.2 c 12.7 ± 6.7 c 0.222 ± 0.11
Free silt and clay 11.7 ± 10.3 d 3.64 ± 1.17 ab 0.33 ± 0.12 13.0 ± 1.3 b 9.7 ± 8.4 c 11.7 ± 10.3 c 0.200 ± 0.13
* Bulk soil weight = sum of the mass of the macroaggregate, microaggregate, and silt and clay fraction. See section 2.3 for details.
**Numbers in parentheses for macroaggregate-occluded fractions represent % of macroaggregate mass. See section 2.4 for details.
121
Table 2. Restricted maximum likelihood results for effects of land cover and fraction on carbon and nitrogen contents.
F-Value
Effect df gC/ 100g soil
gN/ 100g soil
C:N Distribution of C (% of bulk soil C)
Distribution of N (% of bulk soil N)
Land Cover 3,5 1.91 1.38 0.54 0.75 0.72
Fraction 4, 20 6.21** 2.67 42.82*** 107.17*** 94.69***
Land Cover * Fraction
12, 20 2.39* 1.81 0.79 0.43 0.40
*p < 0.05, **p < 0.01, ***p < 0.0001
122 Table 3. Mean and SE for mass recovery, carbon (C), nitrogen (N) concentrations, carbon-to-nitrogen ratio (C/N), contributions to bulk C and N, and microbial biomass in soil aggregate fractions from 0–20 cm depth averaged across two replicate sites per land cover class, except in the 90-year forests where there were three replicate sites. Different letters down each column represent significant differences among land cover types and aggregate fractions using Tukey’s method of mean comparisons.
Soil Fraction & Land Cover Type % of bulk soil* gC/100g soil gN/100g soil C/N
Distribution of C (% of bulk
soil C) Distribution of N (% of bulk
soil N)
Biomass (µmol PLFA/g
soil)
Macroaggregate
Pasture 73.13 ± 0.06 3.41± 1.01 ab 0.28 ± 0.03 15.11 ± 0.08 0.67 ± 0.18 0.67 ± 0.59 0.225 ± 0.08
Early secondary forest 80.95 ± 0.05 2.72 ± 0.30 ab 0.2 ± 0.01 16.21 ± 0.03 0.78 ± 0.10 0.77 ± 1.38 0.208 ± 0.10
Late secondary forest 74.01 ± 0.03 4.38 ± 1.52 ab 0.34 ± 0.00 15.12 ± 0.09 0.76 ± 0.05 0.75 ± 1.44 0.225 ± 0.05
Primary forest 75.69 ± 0.03 4.51± 0.58 ab 0.32 ± 0.01 16.45 ± 0.01 0.77 ± 0.10 0.78 ± 1.72 0.239 ± 0.09
Macroaggregate-occluded microaggregate**
Pasture
12.69 ± 0.03 (0.16 ± 0.03) 4.39 ± 0.46 ab 0.33 ± 0.00 16.32 ± 0.05 0.16 ± 0.03
(20.8 ± 4.2) 0.15 ± 0.61 na
Early secondary forest
25.92 ± 0.02 (0.33 ± 0.03) 2.39 ± 0.16 ab 0.17 ± 0.00 16.18 ± 0.02 0.22 ± 0.05
(27.9 ± 7.3) 0.22 ± 1.01 0.220 ± 0.08
Late secondary forest
21.41 ± 0.02 (0.29 ± 0.03) 4.61 ± 1.55 ab 0.36 ± 0.01 14.93 ± 0.11 0.23 ± 0.10
(31.2 ± 12.0) 0.23 ± 0.78 0.307 ± 0.10
Primary forest
12.51 ± 0.02 (0.20 ± 0.02) 5.17 ± 0.81 a 0.37 ± 0.00 16.47 ± 0.04 0.14 ± 0.05
(22.7 ± 4.1) 0.14 ± 1.25 0.378 ± 0.21
Macroaggregate-occluded silt and clay**
Pasture
48.04 ± 0.03 (0.69 ± 0.08) 3.37 ± 0.43 ab 0.29 ± 0.00 13.59 ± 0.04 0.43 ± 0.06
(58.3 ± 3.1) 0.48 ± 0.63 na
Early secondary forest
33.16 ± 0.03 (0.41 ± 0.03) 2.97 ± 0.53 ab 0.25 ± 0.01 14.03 ± 0.05 0.35 ± 0.09
(42.5 ± 8.2) 0.39 ± 0.86 0.249 ± 0.06
Late secondary forest
43.46 ± 0.05 (0.59 ± 0.06) 3.88 ± 0.99 ab 0.35 ± 0.02 12.94 ± 0.08 0.41 ± 0.14
(55.1 ± 17.5) 0.47 ± 0.91 0.210 ± 0.05
123
Primary forest 45.72 ± 0.07 (0.69 ± 0.03) 3.57 ± 0.39 b 0.3 ± 0.03 14.09 ± 0.03 0.37 ± 0.18
(54.4 ± 8.8) 0.43 ± 1.16 0..291 ± 0.05
Free microaggregate
Pasture 9.62 ± 0.01 3.82 ± 0.40 ab 0.31 ± 0.00 14.99 ± 0.04 0.10 ± 0.01 0.1 ± 0.31 0.215 ± 0.03
Early secondary forest 14.24 ± 0.03 3.11± 0.14 ab 0.23 ± 0.00 15.59 ± 0.02 0.14 ± 0.05 0.14 ± 1.24 0.181 ± 0.07
Late secondary forest 15.63 ± 0.02 4.49 ± 1.01 ab 0.35 ± 0.00 14.78 ± 0.06 0.16 ± 0.02 0.16 ± 0.73 0.234 ± 0.06
Primary forest 9.09 ± 0.01 5.22 ± 0.74 a 0.37 ± 0.00 16.33 ± 0.03 0.1 ± 0.04 0.1 ± 1.98 0.257 ± 0.06
Free silt and clay
Pasture 17.25 ± 0.05 3.29 ± 0.46 ab 0.31 ± 0.00 12.74 ± 0.05 0.12 ± 0.02 0.14 ± 0.26 0.206 ± 0.02
Early secondary forest 4.81 ± 0.01 3.07 ± 0.21 ab 0.28 ± 0.00 12.86 ± 0.02 0.05 ± 0.04 0.07 ± 0.28 0.177 ± 0.10
Late secondary forest 10.36 ± 0.02 4.24± 0.96 ab 0.42 ± 0.00 12.02 ± 0.09 0.10 ± 0.04 0.13 ± 0.79 0.191 ± 0.03
Primary forest 15.22 ± 0.04 3.96 ± 0.45 ab 0.32 ± 0.00 14.5 ± 0.04 0.12 ± 0.08 0.13 ± 0.89 0.226 ± 0.08
* Bulk soil weight = sum of the mass of the macroaggregate, microaggregate, and silt and clay fraction. See section 2.3 for details. **Numbers in parentheses for macroaggregate-occluded fractions represent % of macroaggregate mass. See section 2.4 for details.
124
Table 4. Restricted maximum likelihood and analysis of variance results for effects of land cover and fraction on principal component one and principal component two (from Figure 5).
F-Value Rsq (RMSE) Effect df*
PC1 PC2 PC1 PC2
Land Cover 3,6 18.499** 3.8550 0.9595 (0.7788) 0.9611 (0.6641)
Fraction 2, 11 30.765*** 37.094***
Land Cover * Fraction 6, 11 2.836 1.6035
%C 1, 25 14.7126** ns 0.3705 (2.1293) ns
%N 1, 25 11.2208* ns 0.3098 (2.2297) ns
C:N 1, 25 1.9636 ns 0.0728 (2.5842) ns
df* degrees of freedom for REML represents df and den df, while for ANOVAs represents df of the effect, and df of the error.
*p < 0.05, **p < 0.01, ***p < 0.0001
ns indicates that there were no significant effects of carbon and nitrogen chemistry for PC2
125
Table 5. Microbial community indicators as determined by PLFA analysis of soil aggregate fractions across land cover types: pastures, early secondary forests (40yr old), late secondary forests (90yr old), and primary forests. Mean and propagated standard errors of two replicate sites per land cover class, except in the 90-year forests where there were three replicate sites are reported for the relative abundance (%) of indicator PLFAs. Lowercase letters (abc) indicate significant differences among aggregate fractions averaged across land cover types, while capital letters (ABC) under each column indicate significant differences among land cover types averaged across fractions.
Land Cover Microbial Indicator Soil Fraction
Pasture Early Secondary Forest
Late Secondary Forest Primary Forest
Macroaggregate 3.29 ± 0.12 3.10 ± 0.18 5.30 ± 0.84 4.78 ± 0.64
Microaggregate 3.25 ± 0.21 3.63 ± 0.14 5.22 ± 0.51 4.59 ± 0.45
Silt and clay 3.40 ± 0.21 3.61 ± 0.23 5.13 ± 0.50 4.47 ± 0.70
Gram-positive bacteria
(15:0 iso)
A AB C BC
Macroaggregate 2.37 ± 0.07 2.58 ± 0.15 3.53 ± 0.29 2.95 ± 0.28
Microaggregate 2.22 ± 0.09 2.92 ± 0.39 3.29 ± 0.39 2.74 ± 0.25
Silt and clay 2.50 ± 0.14 2.96 ± 0.33 3.27 ± 0.23 2.75 ± 0.16
Actinobacteria
(16:0 10 Methyl)
Macroaggregate a 4.70 ± 0.73 6.35 ± 1.91 5.01 ± 1.13 3.75 ± 0.55
Microaggregate b 4.78 ± 0.83 4.90 ± 1.16 4.20 ± 0.66 3.29 ± 0.22
Silt and clay c 3.22 ± 0.28 4.04 ± 0.52 3.13 ± 0.39 2.58 ± 0.12
AMF
(16:1w5c)
Macroaggregate a 1.72 ± 0.10 1.30 ± 0.12 1.73 ± 0.14 1.46 ± 0.18
Microaggregate b 1.99 ± 0.17 1.75 ± 0.28 1.75 ± 0.21 1.82 ± 0.38
Silt and Clay b 1.70 ± 0.17 1.97 ± 0.25 1.77 ± 0.57 1.71 ± 0.14
Gram-negative bacteria
(16:1w7c)
Macroaggregate a 3.11 ± 0.52 4.65 ± 1.63 3.73 ± 2.11 3.25 ± 0.97
Microaggregate a 3.13 ± 0.41 4.27 ± 1.49 3.74 ± 1.40 2.85 ± 0.61
Silt and Clay b 2.88 ± 0.37 3.17 ± 0.58 3.24 ± 0.59 2.71 ± 0.69
Methanotrophic
bacteria
(18:1w7c)
Macroaggregate a 12.15 ± 0.37 14.60 ± 0.58 10.48 ± 0.59 8.94 ± 0.59
Microaggregate b 10.92 ± 0.52 12.45 ± 1.63 10.00 ± 2.11 8.59 ± 0.97
Silt and Clay c 10.10 ± 0.41 8.94 ± 1.49 8.82 ± 1.40 8.67 ± 0.61
Fungi: Ecto or SF
(18:1w9c)
Macroaggregate a 8.01 ± 0.13 3.55 ± 0.19 3.23 ± 0.31 2.68 ± 0.24
Microaggregate b 2.11 ± 0.82 3.21 ± 0.73 2.32 ± 0.95 1.58 ± 0.76
Silt and Clay b 1.61 ± 0.22 2.05 ± 0.49 1.71 ± 0.39 1.45 ± 0.23
SF
(18:2w6,9c)
Macroaggregate 3.53 ± 0.24 3.18 ± 0.60 6.14 ± 1.14 4.64 ± 0.57
Microaggregate 3.47 ± 0.25 3.47 ± 0.42 6.02 ± 1.42 4.25 ± 0.66
Silt and Clay 3.90 ± 0.21 3.35 ± 0.46 5.86 ± 1.22 4.30 ± 0.61
Anaerobic, gram-negative
bacteria
(19:0 cyclo) AB B A AB
126
Table 6. Restricted maximum likelihood results for effects of forest age and fractions (both free and macroaggregate-occluded fractions) on principal component one and principal component two (from Figure 8), n = 2 (sites) for all forest age classes (early secondary, late secondary and primary forests). Pastures were excluded from this analysis because many pasture samples did not yield enough macroaggregate-occluded fraction mass for microbiological analyses.
F-Value Rsq (RMSE) Effect df*
PC1 PC2 PC1 PC2
Forest Age 2,3 2.6517 14.856* 0.9575 (0.796) 0.9675 (0.514)
Fraction 4, 12 32.128*** 18.619***
Forest Age * Fraction 8, 12 9.214** 3.692*
*p < 0.05, **p < 0.01, ***p < 0.0001
127 Table 7. Microbial community composition of free and occluded soil aggregates. Mean and propagated standard errors of sites (SE; n = 2) are presented for biomass, PLFA ratios and the relative abundance (%) of indicator PLFAs. Lowercase letters (abc) indicate significant differences using Tukey’s mean comparisons among soil aggregate fractions averaged across land cover types. Letters not shared among columns denote significant differences.
Soil Fraction Microbial PLFA composition
Macro-
aggregates
Microaggregates in
macroaggregates
Silt and clay in macroaggregates
Free micro-aggregates
Free silt and clay
PLFA Biomass*
(µmol PLFA/g soil) 0.22 ± 0.13 ab 0.30 ± 0.24 a 0.25 ± 0.09 ab 0.22 ± 0.10 ab 0.20 ± 0.13 b
F:B*** 1.00 ± 0.32 a 0.88 ± 0.30 bc 0.94 ± 0.83 ab 0.84 ± 0.27 c 0.69 ± 0.11 d
Gm+:Gm-*** 0.81 ± 0.20 c 0.80 ± 0.14 c 0.88 ± 0.14 ab 0.82 ± 0.10 bc 0.92 ± 0.20 a
Actinobacteria † ***
(16:0 10 Methyl) 2.90 ± 0.55 a 2.49 ± 0.68 b 2.21 ± 0.51 b 2.90 ± 0.72 a 2.90 ± 0.45 a
Mycorrhizal† ***
(16:1w5c) 4.98 ± 2.62 a 3.93 ± 1.78 bc 2.24 ± 0.89 d 4.06 ± 1.59 b 3.21 ± 0.81 c
Methanotrophic † **Bacteria (18:1w7c) 3.73 ± 0.79 a 2.48 ± 0.45 c 2.74 ± 0.88 bc 3.44 ± 0.51 ab 2.83 ± 0.61 bc
Anaerobic Bacteria† *
(19:0 cyclo) 4.63 ± 1.31 a 4.34 ± 1.37 ab 3.51 ± 0.58 ab 4.54 ± 1.35 a 4.45 ± 0.90 ab
*p < 0.05, ** p < 0.01, ***p < 0.0001,
† relative abundance (%)
128
Figure 1. Aggregate fractionation sequence.
129
Figure 2. Principal component analysis of microbial community structure using PLFA indicator species. Points represent averages of site means for land cover and fractions; macroaggregate, microaggregate, and silt and clay fraction. Standard error bars represent pooled error across all site and sample replicates.
Pasture Silt and Clay
Pasture Macros
Pasture Micros
40 yr Clay
40 yr Macros
40 yr Micros
90 yr Silt and Clay
90 yr Macros
90 yr Micros
Primary Silt and Clay
Primary Macros
Primary Micros
-‐2.5
-‐2
-‐1.5
-‐1
-‐0.5
0
0.5
1
1.5
2
2.5
3
-‐3 -‐2 -‐1 0 1 2 3 4
PC1 (38.6%)
PC2 (25.6%)
130
Figure 3. PLFA fungal-to-bacterial ratio across land cover types and fractions (F:B = fungal-to-bacterial ratio). Both land cover type (pasture, early and late secondary forest, and primary forest) and soil fraction had a significant effect (p = 0.0226, <0.0001, respectively) on the fungal-to-bacterial ratio.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
Pasture Early secondary forest
Late secondary forest
Primary forest
F:B ratio
macroaggregates
microaggregates
silt and clay
131
Figure 4. Gram-positive – to – gram-negative bacterial ratio across land cover types and fractions (Gm+:Gm- = gram-positive – to – gram-negative bacterial ratio). Soil fraction and the interaction between soil fraction and land cover type had a significant effect (p <0.0001, = 0.0046, respectively) on the gram-positive – to – gram-negative bacterial ratio.
0
0.2
0.4
0.6
0.8
1
1.2
Pasture Early secondary forest
Late secondary forest
Primary forest
Gm+:Gm- ratio
Macroaggregates
Microaggregates
Silt and Clay
132
Figure 5. Principal component analysis of microbial community structure using PLFA indicator species. Points represent averages of site means for all forest ages and all fractions, n = 2 (sites) for all forest age classes (early secondary, late secondary and primary forests). Standard error bars represent pooled error across all site and sample replicates.
40 yr
90 yr primary
40 yr
90 yr primary
40 yr
90 yr
primary
40 yr
90yr
primary
40yr
90 yr
primary
-‐3
-‐2
-‐1
0
1
2
3
4
-‐6 -‐5 -‐4 -‐3 -‐2 -‐1 0 1 2 3
PC2 (25.8%)
PC1 (39.0%)
Macroaggregates Microaggregates Silt and Clay Macro (microaggregates) Macro (silt and clay)
133
7. References
Acosta-‐Martínez, V., L. Cruz, et al. (2007). "Enzyme activities as affected by soil properties and land use in a tropical watershed." Applied Soil Ecology 35(1): 35-‐45. Allison, S. D. and J. D. Jastrow (2006). "Activities of extracellular enzymes in physically isolated fractions of restored grassland soils." Soil Biology and Biochemistry 38(11): 3245-‐3256. Bailey, V. L., J. L. Smith, et al. (2002). "Fungal-‐to-‐bacterial ratios in soils investigated for enhanced C sequestration." Soil Biology & Biochemistry 34: 997–1007. Baisden, W. T., R. Amundson, et al. (2002). "Turnover and storage of C and N in five density fractions from California annual grassland surface soils." Global Biogeochemical Cycles 16(4): 1117. Balser, T. C., K. D. McMahon, et al. (2006). "Bridging the gap between micro -‐ and macro-‐scale perspectives on the role of microbial communities in global change ecology." Plant and Soil 289(1-‐2): 59-‐70. Bardgett, R. D., P. J. Hobbs, et al. (1996). "Changes in soil fungal: bacterial biomass ratios following reductions in the intensity of management of an upland grassland" Biology and Fertility of Soils 22(3 ): 261-‐264. Bossuyt, H., J. Six, et al. (2005). "Protection of soil carbon by microaggregates within earthworm casts." Soil Biology & Biochemistry 37: 251–258. Briar, S. S., S. J. Fonte, et al. (2011). "The distribution of nematodes and soil microbial communities across soil aggregate fractions and farm management systems." Soil Biology and Biochemistry 43(5): 905-‐914. Burke, R. A., M. Molina, et al. (2003). "Stable carbon isotope ratio and composition of microbial fatty acids in tropical soils." Journal of Environmental Quality 32(1): 198-‐206. Chaer, G., M. Fernandes, et al. (2009). "Comparative Resistance and Resilience of Soil Microbial Communities and Enzyme Activities in Adjacent Native Forest and Agricultural Soils." Microbial Ecology 58(2): 414-‐424. Chenu, C., G. Stotzky, et al. (2001). "Interactions between microorganisms and soil particles: an overview " Interactions between soil particles and microorganisms: impact on the terrestrial ecosystem: 3-‐40. P.M. Huang, J-M Bollag, N Senesi. John Wiley & Sons.Eds Chiu, C.-‐Y., T.-‐H. Chen, et al. (2006). "Particle size fractionation of fungal and bacterial biomass in subalpine grassland and forest soils." Geoderma 130(3-‐4): 265-‐271.
134
Denef, K., L. Zotarelli, et al. (2007). "Microaggregate associated C as a diagnostic fraction for management-‐induced changes in soil organic carbon in two Oxisols." Soil Biology & Biochemistry 39(1165–1172). Eiland, F., M. Klamer, et al. (2001). "Influence of initial C/N ratio on chemical and microbial composition during long term composting of straw. ." Microbial Ecology 41(4): 272-‐280. Elliott, E. T. (1986). "Aggregate structure and carbon, nitrogen, and phosphorus in native and cultivated soils. ." Soil Science Society of America Journal 50: 627-‐633. Elliott, E. T., R. V. Anderson, et al. (1980). "Habitable pore space and microbial trophic interactions." Oikos 35(3): 327-‐335. Elliott, E. T. and D. C. Coleman (1988). "Let the soil work for us." Ecological Bulletins 39: 23-‐32. Ettema, C. H. and D. A. Wardle (2002). "Spatial soil ecology." Trends in Ecology & Evolution 17(4): 177-‐183. Fierer, N., J. P. Schimel, et al. (2003). "Variations in microbial community composition through two soil depth profiles." Soil Biology and Biochemistry 35(1): 167-‐176. Filip, Z., K. Haider, et al. (1972). "Influence of clay minerals on growth and metabolic activity of Epicoccum nigrum and Stachybotrys chartarum." Soil Biology and Biochemistry 4(2): 135-‐145. Frostegård, A. and E. Bååth (1996). "The use of phospholipid fatty acid analysis to estimate bacterial and fungal biomass in soil " Biology and Fertility of Soils 22(1-‐2 ): 59-‐65. Frostegård, Å., A. Tunlid, et al. (2011). "Use and misuse of PLFA measurements in soils." Soil Biology and Biochemistry 43(8): 1621-‐1625. Golchin, A., J. M. Oades, et al. (1994). "Soil structure and carbon cycling." Australian Journal of Soil Research 32: 1043-‐1068. Grau, H. and M. Aide (2008). Globalization and land-‐use transitions in Latin America. Ecology and Society. 13. Griffiths, B. S., M. Bonkowski, et al. (1999). "Changes in soil microbial community structure in the presence of microbial-‐feeding nematodes and protozoa." Pedobiologia 43(4): 297-‐304. Guggenberger, G., S. D. Frey, et al. (1999). "Bacterial and fungal cell-‐wall residues in conventional and no-‐tillage agroecosystems." Soil Science Society of America Journal 63: 1188-‐1198.
135
Hedlund, K. (2002). "Soil microbial community structure in relation to vegetation management on former agricultural land." Soil Biology & Biochemistry 34(9): 1299-‐1307. Heijnen, C. E. and J. A. Van Veen (1991). "A determination of protective microhabitats for bacteria introduced into soil." FEMS Microbiology Letters 85(1): 73-‐80. Holland, E. A. and D. C. Coleman (1987). " Litter placement effects on microbial and organic matter dynamics in an agroecosystem." Ecology 68: 425–433. Houghton, R. A. (2007). "Balancing the global carbon budget." Annual Review of Earth and Planetary Science 35: 313-‐347. Houghton, R. A., D. L. Skole, et al. (2000). "Annual fluxes of carbon from deforestation and regrowth in the Brazilian Amazon." Nature 403: 301-‐303. Huang, P. M. (2004). Soil Mineral–Organic Matter-‐-‐Microorganism Interactions: Fundamentals and Impacts. Advances in Agronomy, Academic Press. Volume 82: 391-‐472. Huygens, D., K. Denef, et al. (2008). "Do nitrogen isotope patterns reflect microbial colonization of soil organic matter fractions?" Biology and Fertility of Soils 44(7): 955-‐964. Jastrow, J. D., J. E. Amonette, et al. (2007). "Mechanisms controlling soil carbon turnover and their potential application for enhancing carbon sequestration " Climate Change 80: 5-‐23. Joergensen, R. and F. Wichern (2008). "Quantitative assessment of the fungal contribution to microbial tissue in soil." Soil Biology and Biochemistry 40(12): 2977-‐2991. Kaiser, C., A. Frank, et al. (2010). "Negligible contribution from roots to soil-‐borne phospholipid fatty acid fungal biomarkers 18:2ω6,9 and 18:1ω9." Soil Biology and Biochemistry 42(9): 1650-‐1652. Kaiser, K. and K. Kalbitz (2012). "Cycling downwards -‐ dissolved organic matter in soils." Soil Biology & Biochemistry 52(0): 29-‐32. Kandeler, E., S. Palli, et al. (1999). "Tillage changes microbial biomass and enzyme activities in particle-‐size fractions of a Haplic Chernozem." Soil Biology and Biochemistry 31: 1253-‐1264. Kandeler, E., E. D. Tscherko, et al. (2000). "Structure and function of the soil microbial community in microhabitats of a heavy metal polluted soil." Biol Fertil Soils 32(390-‐400). Kaur, A., A. Chaudhary, et al. (2005). "Phospholipid fatty acid – A bioindicator of environment monitoring and assessment in soil ecosystem." Current Science 89(7): 1103-‐1112.
136
Killham, K. (1994). Soil Ecology Cambridge, UK, Cambridge University Press. Kramer, C. and G. Gleixner (2006). "Variable use of plant-‐ and soil-‐derived caron by microorganisms in agricultural soils." Soil Biology & Biochemistry 38: 3267-‐3278. Kramer, C. and G. Gleixner (2008). "Soil organic matter in soil depth profiles: Distinct carbon preferences of microbial groups during carbon transformation." Soil Biology & Biochemistry 40: 425-‐433. Kramer, M. G., P. Sollins, et al. (2003). "N Isotope Fractionation and Measures of Organic Matter Alteration during Decomposition." Ecology 84(8): 2021-‐2025. Ladd, J. N., R. C. Foster, et al. (1996). "Soil structure and biological activity." Soil biochemistry 9: 23-‐78. Lehmann, J., M. da Silva Cravo, et al. (2001). "Organic matter stabilization in a Xanthic Ferralsol of the central Amazon as affected by single trees: chemical characterization of density, aggregate, and particle size fractions." Geoderma 99(1): 147-‐168. Lützow, M. V., I. Kögel-‐Knabner, et al. (2006). "Stabilization of organic matter in temperate soils: mechanisms and their relevance under different soil conditions–a review." European Journal of Soil Science 57(4): 426-‐445. Macdonald, C. A., N. Thomas, et al. (2009). "Physiological, biochemical and molecular responses of the soil microbial community after afforestation of pastures with Pinus radiata." Soil Biology and Biochemistry 41(8): 1642-‐1651. Marin-‐Spiotta, E., R. Ostertag, et al. (2007). "Long-‐term patterns in tropical reforestation: Plant commnity composition and aboveground biomass accumulation." Ecological Applications 17(3): 828-‐839. Marin-‐Spiotta, E., W. L. Silver, et al. (2009). "Soil organic matter dynamics during 80 years of reforestation of tropical pastures." Global change biology 15(6): 1584-‐1597. Marín-‐Spiotta, E., C. W. Swanston, et al. (2008). "Chemical and mineral control of soil carbon turnover in abandoned tropical pastures." Geoderma 143(1-‐2): 49-‐62. Marschner, B., S. Brodowski, et al. (2008). "How relevant is recalcitrance for the stabilization of organic matter in soils?" Journal of Plant Nutrition and Soil Science 171(1): 91-‐110. Martin, J. P. and K. H. 1986 (1986). Influence of mineral colloids on turnover rates of soil organic carbon. Interactions of soil minerals with natural organics and microbes. P. M. Huang and M. Shnitze. Madison, SSSA. SSSA Special Pub. No. 17: 283–304.
137
McGuire, K. and K. Treseder (2009). "Microbial communities and their relevance for ecosystem models: Decomposition as a case study." Soil Biology & Biochemistry 42(4): 529-‐535. Meiyappan, P. and A. K. Jain (2012). "Three distinct global estimates of historical land-‐cover change and land-‐use conversions for over 200 years." Frontiers of Earth Science 6: 122-‐139. Miltner, A., P. Bombach, et al. (2012). "SOM genesis: microbial biomass as a significant source." Biogeochemistry 111(1-‐3): 41-‐55. Monrozier, L. J., J. N. Ladd, et al. (1991). "Components and microbial biomass content of size fractions in soils of contrasting aggregation." Geoderma 49: 37-‐62. Oades, J. M. (1988). "The retention of organic matter in soils." Biogeochemistry 5(1): 35-‐70. Oades, J. M. and A. G. Waters (1991). "Aggregate hierarchy in soils." Australian Journal of Soil Research 29: 815–828. Olsson, P. A. (1999). "Signature fatty acids provide tools for determination of the distribution and interactions of mycorrhizal fungi in soil." FEMS Microbiology Ecology 29: 303-‐310. Olsson, P. A., E. Baath, et al. (1995). "The use of phospholipid and neutral lipid fatty acids to estimate biomass of arbuscular mycorrhizal fungi in soil." Mycological Restoration 5: 623-‐629. Paterson, E., G. Osler, et al. (2008). "Labile and recalcitrant plant fractions are utilised by distinct microbial communities in soil: Independent of the presence of roots and mycorrhizal fungi." Soil Biology & Biochemistry 40: 1103-‐1113. Paul, E. A. (2006). Soil microbiology, ecology and biochemistry, Academic Press. Poll, C., A. Theide, et al. (2003). "Micro-‐scale distribution ofmicroorganisms and microbial enzyme activities in a soilwith long-‐term organic amendment." European Journal of Soil Science 54: 715–724. Potthast, K., U. Hamer, et al. (2012). "Land-‐use change in a tropical mountain rainforest region of southern Ecuador affects soil microorganisms and nutrient cycling." Biogeochemistry 111(1-‐3): 151-‐167. Potthoff, M., K. L. Steenwerth, et al. (2006). "Soil microbial community composition as affected by restoration practices in California grassland." Soil Biology and Biochemistry 38(7): 1851-‐1860.
138
Qin, S., C. Hu, et al. (2010). "Soil organic carbon, nutrients and relevant enzyme activities in particle-‐size fractions under conservational versus traditional agricultural management." Applied Soil Ecology 45(3): 152-‐159. Ramette, A. (2007). "Multivariate analyses in microbial ecology." FEMS Microbiology Ecology 62(2): 142-‐160. Ratledge, C. and S. G. Wilkinson (1988). Microbial Lipids. London, Academic Press. Rousk, J. and E. Bååth (2007). "Fungal biomass production and turnover in soil estimated using the acetate-‐in-‐ergosterol technique " Soil Biology & Biochemistry 39: 2173–2177. Rutherford, P. M. and N. G. Juma (1992). "Influence of texture on habitable pore space and bacterial-‐protozoan populations in soil." Biology and fertility of soils 12(4): 221-‐227. Schimel, J. P. and S. M. Schaeffer (2012). "Microbial control over carbon cycling in soil." Frontiers in Microbiology 3. Schmidt, M. W. I., M. S. Torn, et al. (2011). "Persistence of soil organic matter as an ecosystem property." Nature 478: 49–56. Schutter, M. E. and R. P. Dick (2002). "Microbial community profiles and activities among aggregates of winter fallow and cover-‐cropped soil." Soil Science Society of America Journal 66(1): 142-‐153. Sexstone, A. J., N. P. Revsbech, et al. (1985). "Direct measurement of oxygen profiles and denitrification rates in soil aggregates." Soil Science Society of America Journal 49(3): 645-‐651. Shang, C. and H. Tiessen (1998). "Organic matter stabilization in two semiarid tropical soils: size, density, and magnetic separations." Soil Sci. Soc. Am. J. 62(5): 1247-‐1257. Six, J., R. T. Conant, et al. (2002). "Stabilization mechanisms of soil organic matter: implications for C-‐saturation of soils." Plant and Soil 241(2): 155-‐176. Six, J., E. T. Elliott, et al. (1998). "Aggregation and Soil Organic Matter Accumulation in Cultivated and Native Grassland Soils." Soil Sci. Soc. Am. J. 62: 1367-‐1377 Six, J., S. D. Frey, et al. (2006). "Bacterial and fungal contributions to carbon sequestration in agroecosystems." Soil Science Society of America Journal 70(2): 555. Six, J. and J. D. Jastrow (2002). Organic matter turnover. Encyclopedia of soil science. New York, NY, Marcel Dekker 936-‐942. Six, J., R. Merckx, et al. (2000). "A re-‐evaluation of the enriched labile soil organic matter fraction." European Journal of Soil Science 51: 283-‐293.
139
Smith, A. P., E. Marin-‐Spiotta, et al. (in prep.). "Seasonal and successional changes in soil microbial community structure during reforestation of a tropical post-‐agricultural landscape." Smithwick, E. A. H., M. G. Turner, et al. (2005). "Variation in NH4+ mineralization and microbial communities with stand age in lodgepole pine (Pinus contorta) forests, Yellowstone National Park (USA)." Soil Biology and Biochemistry 37(8): 1546-‐1559. Staff, S. S. (2008). Official soil series descriptions Lincolcn, NE, USDA-‐NRCS. Stotzky, G. (1986). Influence of soil mineral colloids on metabolic processes, growth, adhesion, and ecology of microbes and viruses. Interactions of soil minerals with natural organics and microbes. P. M. Huang and M. Schnitzer. Madison, WI, Soil Science Society of America. sssa special publication 17: 305-‐428. Strickland, M. S. and J. Rousk (2010). "Considering fungal:bacterial dominance in soils – Methods, controls, and ecosystem implications." Soil Biology and Biochemistry 42(9): 1385-‐1395. Theng, K. G., V. A. Orchard, et al. (1995). Interactions of clays with microorganisms and bacterial survival in soil: a physicochemical perspective. Environmental impact of soil component interactions: Volume 2: metals, other inorganics, and microbial activities.: 123-‐143. Tisdall, J. M. and J. M. Oades (1982). "Organic-‐matter and water-‐stable aggregates in soils." Journal of Soil Science 33: 141–163. Torsvik, V. and L. Øvreås (2002). "Microbial diversity and function in soil: from genes to ecosystems." Current Opinion in Microbiology 5: 240-‐245. Trumbore, S. (2009). "Radiocarbon and soil carbon dynamics." Annual Review of Earth and Planetary Sciences 37(1): 47-‐66. Van Gestel, M., R. Merckx, et al. (1996). "Spatial distribution of microbial biomass in microaggregates of a silty-‐loam soil and the relation with the resistance of microorganisms to soil drying." Soil Biology and Biochemistry 28(4): 503-‐510. Vestal, J. R. and D. C. White (1989). "Lipid analysis in microbial ecology." Bioscience 39(8): 535-‐541. Waring, B. G., S. R. Weintraub, et al. (2013). "Ecoenzymatic stoichiometry of microbial nutrient acquisition in tropical soils." Biogeochemistry. Wieder, W., G. B. Bonan, et al. (2013). "Global soil carbon projections are improved by modellig microbial processes." Nature Climate Change 3: DOI: 10.1038/NCLIMATE1951.
140
Zelles, L. (1997). "Phospholipid fatty acid profiles in selected members of soil microbial communities " Chemosphere 35(1/2): 275-‐294. Zelles, L. (1999). "Fatty acid patterns of phospholipids and lipopolysaccharides in the characterisation of microbial communities in soil: a review." Biology and Fertility of Soils 29: 111-‐129. Zhao, X. R., Q. Lin, et al. (2005). "Does soil ergosterol concentration provide a reliable estimate of soil fungal biomass?" Soil Biology & Biochemistry 37: 311–317.
141
CHAPTER 4: Shifts in the functional capacity of soil microbial communities with tropical forest
regeneration on abandoned pastures Abstract Land cover change, such as deforestation and reforestation, can alter ecosystem processes
controlling soil carbon storage and loss, and contribute to processes of global change. As soil
microorganisms play a central role in regulating soil organic matter and carbon (C) cycling, it is
important to understand how land cover change affects microbial function. This research
investigates the role of microbial activity and community functional diversity in SOM dynamics
during secondary forest succession on former pastures. We measured soil microbial community
functional gene abundance and diversity using GeoChip 3.0 across a land-use chronosequence
where data on aboveground plant communities and trends in SOM dynamics have already been
established. Results show that bacteria dominated carbon fixation and degradation processes,
with the exception of lignin decomposition genes, which were primarily derived from fungi.
Functional genes involved in methane oxidation were also chiefly derived from bacteria, while
methane production genes were mostly derived from archaea.
Using nonmetric multidimensional scaling (NMS), functional gene composition for C, N
and P cycling genes correlated more with microbial composition than extracellular activities, and
C and N concentrations. The phospholipid fatty acid (PLFA) fungal-to-bacterial ratio correlated
with axis two for individual NMS analyses of C, N and P cycling genes (Pearson correlation
coefficient of 27.1%, 25.4%, and 31.4%, respectively). While functional gene composition did
not vary by land cover type for C, N and P cycling genes, the relative abundance of select
functional genes involved in C, N and P cycling differed between early and late secondary
142
forests. The relative abundance of the majority of genes involved in cellulose and chitin
degradation, phosphorus acquisition and ammonification significantly decreased (p < 0.10) with
forest age. Our results indicate that for the majority of genes involved in SOM cycling, there is a
certain amount of functional redundancy between microbial communities associates with
pastures and forests. However, our investigation also shows shifts in the functional capacity of
soil microorganisms between early and late secondary forests in specific processes regulating
cellulose, chitin and starch degradation, ammonification and phosphorus acquisition. These
differences in functional capacity appear to be linked with differences in microbial composition.
This provides evidence for the importance of understanding structure-function relationships as it
identifies how shifts in microbial community composition directly influences carbon cycling
processes. Additionally, it further justifies the need for continued investigations into
compositional changes with land cover change and ecosystem recovery.
Keywords: land cover change, soil microbial community, tropical forests, Geochip, functional genes
143
1. Introduction Land use and land cover change can alter ecosystem processes controlling soil carbon (C)
storage and loss and contribute to climate change (Houghton and Goodale 2004). In Latin
America and the Caribbean, one of the most common land cover changes is associated with a
gain or loss in forest cover. Secondary forests, or those naturally regenerating or planted as
plantations, are increasing in many areas in the tropics (Aide and Grau 2004, Aide et al. 2012)
and are estimated to cover 0.8 - 1.25 million km2 in Latin America alone (Meiyappan and Jain
2012). Tropical forests represent a substantial proportion of the terrestrial C sink, holding more
than 500 Pg of C in both above- and belowground stocks (Houghton et al. 1993, Prentice 2001).
Soil microorganisms play an integral part in regulating biogeochemical cycling of C, N
and P as living biomass through their respiration and extracellular enzyme activities. They also
contribute to soil organic matter (SOM) pools through their necromass (Glaser et al. 2004).
Despite their importance, understanding of how the microbial communities respond to
disturbance and ecosystem recovery (Kuramae et al 2010, Banning et al. 2011) and how their
response affects SOM formation and stabilization (Six et al. 2002, Wixon and Balser 2009) is
limited. Microbes make up the majority of global diversity (Torsvik and Øvreås 2002) with
populations of 2000 - 8.3 million species and up to 10 billion organisms inhabiting 1 g soil
(Roselló-Mora and Amann 2001). Recent metagenomic technologies now provide the
opportunity to sequence and characterize microbial phylogenetic diversity across a variety of
environmental samples, but provide little information on the functional capacity of the microbial
community (He et al. 2010, Zhou et al. 2010).
One tool developed to characterize the functional gene composition of the microbial
community is Geochip, a high-throughput, gene-based metagenomic functional gene microarray.
144
GeoChip contains probes that specifically target genes coding for enzymes involved in
microbially-mediated environmental` processes, such as biogeochemical cycling and
contaminant biodegradation (He et al. 2007, 2010). Geochip has been applied to evaluate
microbial functional diversity across a variety of ecosystems; antarctic soils, subtropical
mangroves, mixed grass savannas, hydrothermal vents and contaminated soils and sediments
(Liang et al. 2009, Van Nostrand et al. 2009, Wang et al. 2009, Hollister et al. 2010, Bai et al.
2012, Chan et al. 2013, Zhang et al. 2013). Zhang et al. (2007) linked functional gene diversity
to soil organic carbon in a subalpine primitive forest and plantations, while Reich et al. (2004)
used GeoChip to link above and belowground functional diversity in a temperate grassland.
In this study, we used GeoChip 3.0 to assess the effects of changes in land cover on the
functional capacity in SOM cycling processes of the soil microbial community along a
successional chronosequence of tropical forests growing on former pastures in Puerto Rico.
Puerto Rico provides an opportune environment to study reforestation effects on ecological
processes and biogeochemical cycling. From 1937 to 1995, the Sierra de Cayey region, where
our research takes place, increased forest cover from less than 20% to 62% due to emigration to
urban areas (Pascarella et al. 2000), resulting in a highly fragmented landscape of land use and
cover in various ages and stages of agricultural use, abandonment and forest regrowth (Helmer et
al. 2002, Grau et al. 2003).
We have identified a replicated chronosequence of active pastures, primary forests, and
secondary forests of different ages where prior work has characterized changes in aboveground
species composition (Marín-Spiotta et al. 2007), litterfall and decomposition (Ostertag et al.
2008), SOM chemistry and turnover (Marín-Spiotta et al. 2008, 2009) and more recently,
microbial community composition via phospholipid fatty acid analysis and extracellular enzyme
145
activities (Smith et al. In prep. see Chapter 1). The distribution and turnover of SOC in physical
fractions varied among pastures, secondary forests and primary forests, indicating different soil
C dynamics with reforestation (Marín-Spiotta et al. 2008). The soil microbial community
composition varied seasonally and with forest succession but microbial function via extracellular
enzyme activities did not change with land use or land cover types (Smith et al. In prep. see
Chapter 1). Extracellular enzymes, however, often do not provide the most consistent or
comprehensive evaluation of microbial function as measurements of enzyme activities represents
a ‘potential activity’ and also cannot represent in situ conditions (Sinsabaugh et al. 2012, German
et al. 2011). GeoChip provides us with a more comprehensive measurement of microbial
functional capacity to better understand the belowground response to land use change and
ecosystem recovery. The 3.0 version of GeoChip has 9,558 oligonucleotide probes for detecting
60 genes involved in C, N, and P cycling. Sequence-specific and microbial group-specific probes
in GeoChip 3.0 target 3,172 different archaeal, bacterial and fungal organisms (He et al. 2010).
To our knowledge, no published data exists on extensive functional gene diversity in tropical,
terrestrial ecosystems. The use of ecological replication in our study also makes it unique from
other studies using GeoChip functional gene analysis.
2. Methods
2. 1 Site description
This study was conducted on previously established chronosequence plots (Marín-Spiotta et al.
2007) consisting of active pasture, secondary forests growing on pastures abandoned 40 and 90
years ago, and primary forest sites that have not been under pasture or agricultural use. All sites
were located within approximately five km of each other, on private land, between 580 and 700
146
m above seal level in the Sierra de Cayey in southeastern Puerto Rico (18°01´ N, 66°05´ W).
Mean air temperature was 21.5ºC (Southeast Regional Climate Center, 2013) with little seasonal
variation (Daly et al. 2003). Monthly precipitation was approximately 56.35 mm during the time
we sampled, with a daily mean of 1.82 ± 0.39 mm (Jajome Alto climate station, 2013). All soils
were characterized as very-fine, kaolinitic, isothermic Humic Hapludox in the Los Guineos soil
series (Soil Survey Staff, 2008). Forest tree species composition varies among early successional
secondary forests, late successional secondary forests and primary forests and is described in
detail in Marín-Spiotta et al. (2007).
2. 2 Sample collection
Soils were collected from two replicate sites for each land cover type: active pastures, early
secondary forests (40 years old), late secondary forests (90 years old), and remnant primary
forests in January 2012. Two replicate soil samples were collected from each site by compositing
4-5 soil core samples (4 mm diameter soil core to 20 cm depth). Approximately 15 g of soil was
subsampled into a sterile whirl-pak bag and immediately frozen with dry ice for GeoChip
analyses. Soils were stored on dry ice until shipped within 24 hours of collection to the Institute
for Environmental Genomics at the University of Oklahoma for DNA extractions and GeoChip
functional gene possessing. A total of 16 soil samples were analyzed, representing two
composite soil samples from two replicate sites from four different land cover types; active
pasture, early secondary forest, late secondary forest and primary forest.
2. 3 GeoChip functional gene processing
A detailed description of GeoChip functional gene possessing can be found in He et al. (2010)
and He et al. (2007), including information of retrieval and verification of functional gene
147
sequences, oligonucleotide probe design and synthesis, microarray fabrication, and the pipeline
for data analysis. All steps associated with GeoChip measurements and data normalizations were
performed at the Institute for Environmental Genomics at the University of Oklahoma.
Briefly, microbial community DNA was extracted from 5 grams of soil using a modified
freeze-thaw-grind method, an EDTA, NaCl and CTAB extraction buffer and Wizard (Promega,
Madison, WI) genomic purification kit (described in Zhou et al. 1996). DNA concentrations
were quantified with a NanoDrop ND-1000 spectrophotometer (NanoDrop technology,
Rockland, DE). DNA was then labeled with a fluorescent dye via random priming (Van
Nostrand et al. 2009). A hybridization buffer with an oligonucleotide reference standard was
added to the samples for signal normalization (Liang et al. 2009) and placed onto a microarray.
All hybridizations were performed in duplicate. Microarrays were scanned with an MS 200
Microarray Scanner (NimbleGen) and quantified using ImaGene 6.0 software (Biodiscovery
Inc., El Segundo, CA). Signal intensity data was normalized using a data analysis pipeline
described in He et al. (2007); poor quality spots and spots with low signal intensities were
removed based on a signal-to-noise ratio of 2.0 (Wu et al. 2006), normalized signal intensities
values were averaged across technical replicates (hybridization duplicates), and outliers with
signal-means greater than three times the standard deviation were also removed.
Overall, the data received includes mean signal intensity values for a list of multiple gene
probes (specific to sequences or groups of various archaeal, bacterial and fungal organisms) for
each gene detected per sample. For this study, we chose to focus on gene probes detected only
for genes involved in C, N and P cycling. Unfortunately due to difficulties in DNA isolation and
purification, two out of three soil samples from one of our replicate primary forest sites were
unable to be processed. Therefore, our study was limited to functional gene information from
148
active pastures, early secondary forests and late secondary forests. The primary forest samples
were removed entirely as there were not enough samples for sound statistical analysis.
2.4 Soil environmental and microbial parameters To assess any relationships between functional gene capacity and soil abiotic and biotic
properties, we measured soil pH, field moisture, total C and N, and C-to-N ratios. Soil pH was
measured on dried and ground samples using a Sartorius PP-20 professional pH reader in a 1:1
(by volume) 1 M KCl slurry (Sparks, 1996). Moisture content was determined gravimetrically on
freshly sampled, field moist soils. Total C and N concentrations were determined on ground, air-
dried soil using a Flash 2000 NC Analyzer (Thermo Scientific, Wilmington, Delaware) at
University of Wisconsin-Madison. Given the low pH (4± 0.5) and mineralogy of our soils, there
is no inorganic C so total C is interpreted as organic C. Soil C-to-N ratios were calculated as
molar ratios (Cleveland and Liptzin 2007).
Microbial biomass and composition was measured using a hybrid phospholipid fatty acid
(PLFA) and fatty acid methyl ester (FAME) analysis protocol (Smithwick et al. 2005). Microbial
biomass is calculated as the sum of all peaks (as µmol PLFA g soil -1) identified less than 20.5 C
atoms long (Vestal and White 1989; Zelles 1999). Select PLFAs were used as indicator species;
15:0iso (Gram-positive bacteria, Kaur et al. 2005, Zelles 1997, 1999), 16:0 10methyl
(Actinobacteria, Ratledge and Wilkinson 1988), 16:1 w7c (Gram-negative bacteria, Ratledge and
Wilkinson 1988, Zelles 1999), 16:1 w5c (arbuscular mycorrhizal fungi (Olsson et al. 1995,
Olsson 1999), 18:1 w9c (saprotrophic fungi, Bardgett et al. 1996, Frostegard et al. 2011), 18:2
w6,9c (saprotrophic fungi, Frostegard and Baath 1996, Joergensen and Wichern 2008, Kaiser et
al. 2010) and 19:0cyclo (anaerobic, gram-negative bacteria, Vestal and White 1989). The fungal-
to-bacterial ratio was calculated as the sum of 18:1 w9c and 18:2 w6,9c over the sum of 15:0iso,
149
15:0anteiso, 16:1w7c, 17:0anteiso, 17:0iso,17:0cyclo, 18:1w5c, 18:1w7c, 19:0cyclo (Frostegard
and Baath 1996, Zelles 1997, 1999).
2.5 Statistical analysis To examine the effects of land cover change on functional gene composition, a subset (consisting
of genes involved in C, N and P cycling) of normalized GeoChip 3.0 signal intensity data was
analyzed for gene and gene probe richness, relative abundance and composition of C, N and P
cycling functional genes. Gene and gene probe richness was calculated as the number of genes or
gene probed detected per sample. Relative abundance of gene probes and genes detected,
referred to as relative percent, was calculated as the number of gene (or gene probes) detected
per sample over the total number of genes (or gene probes) from that gene category detected
across all samples. Samples from only the pastures, early secondary forests and late secondary
forests were analyzed statistically as difficulties in extracting sufficient DNA at one of the
primary forests sites resulted in a loss of replication for that land cover type.
Changes in overall functional gene composition were analyzed across land cover types
using nonmetric multidimensional scaling (NMS) of normalized signal intensities (see GeoChip
functional gene processing) of each sample. Sørensen (Bray-Curtis) distance measures were used
to calculate dissimilarity distance matrices and compared to a Monte Carlo randomization test.
Two-dimensional solutions were chosen based on low final stress values from a real run
compared to the randomized runs. A Mantel Test was used to evaluate correlations between
distance matrices of GeoChip data and soil environmental and microbiological properties.
Pearson (linear, r2) and Kendall (rank, tau) correlation coefficients are calculated as a result of
the Mantel Test. All NMS, Monte Carlo and Mantel test analyses were performed using PC-
ORD 6 (MjM software, Gleneden Beach, OR).
150
Restricted maximum likelihood (REML) models and mean comparisons using Student’s
t-test were performed to compare gene and gene probe richness, relative abundance and NMS
scores across land cover types via JMP Pro Version 10 software (SAS Inst. Inc., Cary, NC,
USA). REML models included the fixed effect of land cover type in addition to random effects
of site and sample replication. However, due to our experimental design of nested replication
(two replicate sites per treatment and two samples per site) leading to a final n = 2, it was
difficult to tease apart significant differences between land cover types. Thus, the random effect
of site was eliminated when noted. Identifying significant differences was further complicated by
the large amount of data associated with GeoChip functional gene information. At the same time,
our study is one of the first to measure GeoChip function gene information using several levels
of ecologically relevant replication. Due to the large cost of running GeoChip analyses, many
investigations use few samples and pseudoreplication, if any replication is used at all.
P-values between 0.05-0.10 were considered marginally significant, while p-values
reported as significant are < 0.05 (unless otherwise noted).
3. Results 3.1 Microbial composition of gene probes detected
The phylogenetic composition of genes involved in C, N and P cycling spanned across archaea,
prokaryotes (bacteria) and eukaryotes (from the kingdom: fungi), however the majority are
derived from bacteria; 81% of C cycling, 95% of N cycling and 95% of P cycling genes (Figure
1). Fungi represented a small proportion of genes detected in C and P cycling (15% and 4%,
respectively), but very little (0.005%) were detected in N cycling. The fungal-derived N cycling
gene probes were mostly detected in ammonification genes (16 detected probes, Figure 2).
151
Overall, the greatest amount of fungal-derived gene probes were detected in decomposition, or C
degradation (671 detected probes, Figure 2). The only SOM cycling process where fungal-
associated gene probes were relatively greater than bacterial-associated probes was in the probes
detected for phytase, a phosphorus acquisition gene that produces phosphatases (Figure 2).
Archaea and bacteria were distributed equally across methane processing gene probes detected
(Figure 2), however bacterial-associated gene probes made up the majority (96%) of those
involved in methane oxidation, while archaeal-derived gene probes dominated (91%) methane
production genes (data not shown).
Out of all the gene probes detected in C, N and P cycling genes, approximately 25% were
associated with unculturable microorganisms (data not shown). The majority (81%) of those
gene probes derived from sequences of unculturable microorganisms were involved in N cycling
processes. Additionally, over half of those probes were associated with genes involved in
denitrification. Only 8% of genes detected for C cycling processes were associated with
unculturable microorganisms, with a third of those being involved in C fixation, a third in C
degradation and a third in methane production and oxidation. In phosphorus cycling, 11% of
gene probes were derived from unculturable microorganisms and were only associated with the
phosphorus acquisition gene, polyphosphate kinase (ppk), which produces polyphosphate from
adenosine triphosphate (ATP) (data not shown).
3.2 Functional genes and gene probes detected
Overall, a total of 38,017 gene probes were detected across all samples. The mean number of
gene probes detected per sample was 24,368 (± 3307). Functional gene probes detected averaged
25,406 (± 1217) in the pasture sites, 25,845 (± 900) in the early secondary forest sites, and
152
21,855 (± 2940) in the late secondary forest sites. There was, however, no significant effect of
land cover or site on total number of gene probes detected. Out of the total number of gene
probes detected per sample, 4,346 were associated with carbon cycling, 3,128 with nitrogen
cycling and 547 with phosphorus cycling. There was no effect of site on total number and
proportion of detected gene probes involved in C, N or P cycling. Land cover has a marginally
significant effect on both the total and relative percent of detected gene probes associated with C
and P cycling (p = 0.083, 0.077 for C, P respectively, Table 1). Early and late secondary forests
differed in the total and relative percent of gene probes involved in C and P cycling.
3.2.1 C cycling genes
The majority (76.8%) of C cycling gene probes detected across all samples were involved in C
degradation processes. More than 50% of total genes were detected in all land cover types with
the exception of a few genes (xylanase, assA, LMO, mmoX, pmoA, and mcrA). Out of the 39
genes detected (~ 4, 275 gene probes), 41% (16 genes) were affected by land cover change
(Table 2). The early secondary forest had a significantly greater percent of genes detected, or
marginally significant, compared to the late secondary forest in the majority of genes affected by
land cover change (approx. 81%, Table 2).
Soil microbial community functional gene composition for C cycling genes as measured
by NMS ordination analysis was significantly related to soil moisture and the PLFA fungal-to-
bacterial ratio along axis two: 52.7% and 27.1%, respectively (Figure 3). Soil total C and N
concentration, pH, PLFA biomass and extracellular enzyme activities (for betaglucosidase,
alphaglucosidase, cellobiohydrolase, NAGase, xylosidase and phosphatase) were not correlated
with ordination axis one or two. There was also no significant relationship between PLFA
153
indicator biomarkers for gram-positive bacteria (15:0 iso), arbuscular mycorrhizal fungi
(16:1w5c), gram-negative bacteria (16:1w7c), actinobacteria (16:0 10 methyl), methanotrophic
bacteria (18:1w7c), saprotrophic fungi (18:1w9c, 18:2w6,9c), and anaerobic, gram-negative
bacteria (19:0cyclo) and NMS ordination axes one and two. In addition, there was no significant
effect of land cover on microbial gene composition of C cycling genes (i.e. NMS scores for axis
one and two) when site was added as a random effect (n = 2). However, when samples from
different sites were pooled across land cover types (n = 4), there was a significant effect of land
cover on NMS axis two scores (p = 0.0076) with the late secondary forest functional gene
composition differing from the early secondary forest and pasture functional gene composition.
Carbon degradation genes amyA, endochitanase, and C fixation genes, pcc, had the highest
Pearson correlation coefficients (94.5%, 94.6% and 90.5%, respectively) associated with axis 1
(Table 3), while aceA and aceB (involved in carbon degradation) had the highest Pearson
correlation coefficients (88.9%, 66.7%, respectively) associated with axis two; meaning they
represent the greatest proportion of variation along NMS axes one and two.
3.2.2 N cycling genes
A total of 3,128 nitrogen cycling gene probes were detected across all samples (Table 1). There
was no difference in the number (1957 ± 267) or ratio (62.56 ± 8.53%) of gene probes detected
across land cover types. There was a marginal effect of land cover on the ratio of genes; gdh,
ureC (involved in ammonification), nirK, and nosZ (involved in denitrification) (Table 4). For
these four genes, the early secondary forest had a greater proportion of genes detected compared
to the late secondary forest.
154
Microbial community functional gene composition for N cycling genes was significantly
related to PLFA fungal biomarkers 18:1w9c, 18:2w6,9c and the PLFA fungal-to-bacterial ratio:
40.4%, 30.5% and 25.4%, respectively (Figure 4). Soil C, N, pH, PLFA biomass and
extracellular enzyme activities (for betaglucosidase, alphaglucosidase, cellobiohydrolase,
NAGase, xylosidase and phosphatase) were not correlated with ordination axis 1 or 2. There was
also no significant relationship between PLFA indicator biomarkers for gram-positive bacteria
(15:0 iso), arbuscular mycorrhizal fungi (16:1w5c), gram-negative bacteria (16:1w7c),
actinobacteria (16:0 10 methyl), methanotrophic bacteria (18:1w7c), and anaerobic, gram-
negative bacteria (19:0cyclo) and NMS ordination axes one and two. In general, genes involved
in denitrification (nirS, narG, and nosZ) explained the greatest proportion of variation along
NMS ordination axes one and two (Table 5). The gene probes for nirS, nosZ (denitrification) and
nifH (N fixation) were all derived from uncultured bacteria (data not shown).
There was no significant effect of land cover change on microbial community function
gene composition (i.e. NMS scores for axis one and two). This was driven by large variation
within replicate sites for land cover type. The sample representing the pasture that is far removed
from the other pasture sites along both ordination axes in the NMS plot of nitrogen cycling
functional genes (Figure 4) is the same sample that shows a similar spatial separation in the NMS
plot of carbon cycling function genes (Figure 3). It is difficult, however, to discern what is
driving such large variation between this sample and the rest of the samples from the pasture
sites due to high variability among soil replicate samples and the large quantity of data
associated with each sample (3,128 gene probes detected in N cycling alone).
3.2.3. Phosphorus cycling genes
155
Phosphorus gene probes detected across all samples represent only 1% of all gene probes
detected (Table 1). However, there are not as many phosphorus cycling genes and gene probes
identified. A total of 547 gene probes representing 3 phosphorus cycling genes (phytase, ppk and
ppx) were identified across all samples with 64% of that total identified in the active pastures,
65% in early secondary forests and 55% in the late secondary forests. These three genes, phytase,
ppk, and ppx are all involved in P acquisition and utilization. The phytase gene produces
phosphatases that catalyze reactions producing bioavailable forms of P. Both the mean number
and relative percent of phosphorus cycling gene probes detected varied with land cover type,
with more phosphorus cycling genes and gene probes in the early secondary forest compared to
the late secondary forest (p = 0.077). This was also true for the mean ratios of genes ppk and ppx
(Figure 5). Land cover type (pasture, early and late secondary forest) had a marginally significant
effect on the relative percents of ppk (p = 0.061) and ppx (p = 0.054), with the late secondary
forests having less ppk and ppx genes being detected compared to the early secondary forest.
Microbial community functional gene composition for phosphorus cycling genes was
significantly related to PLFA fungal biomarkers 18:1w9c, 18:2w6,9c and the PLFA fungal-to-
bacterial ratio (F/B): 43.8%, 30.6% and 31.4%, respectively (Figure 6), as observed for the N
cycling genes. In NMS analyses of C, N and P cycling function genes, F:B significantly
corresponded to NMS axis two. Also similar to the NMS analyses for C and N cycling genes,
soil C, N, and pH, and PLFA biomass, PLFA biomarkers (15:0 iso, 16:1w5c, 16:1w7c, 16:0 10
methyl, 18:1w7c, and 19:0cyclo) and extracellular enzyme activities did not correlate with either
ordination axis. Gene probes from each phosphorus utilization gene (phytase, ppk and ppx) were
significantly correlated with NMS ordination axis one, explaining 82.5%, 81.4% and 89.1% of
the variation (Table 6). Land cover type had no effect on microbial community function gene
156
composition for P cycling genes. This was also driven by large variation within replicate field
sites. Similar to the NMS plots for C and N cycling function genes, the same sample from one of
the pasture sites showed greater variation in P cycling gene composition from the other pasture
samples.
4. Discussion 4. 1 Microbial functional redundancy in SOM cycling Microbial functional capacity of SOM cycling did not reveal strong variations with land use
change when comparing signal intensity data (used in NMS analyses). While there were
differences in the amount of genes and gene probes detected with land use change, it was usually
marginally significant (p < 0.10). Further, differences in the composition of bacterial, fungal and
archaeal genes detected did not change among the pastures, early and late secondary forests.
These results may imply a high level of functional redundancy in the microbial communities
examined in our study. Functional redundancy, or similarity, is defined as the ability of a species
to perform similar ecological roles in different environments or of different communities
performing similar ecological roles in the same or in different environments (Lawton and Brown
1993, Rosenfield et al. 2002, Allison and Martiney 2008). In our study, functional redundancy is
shown as similar functional gene potentials across the chronosequence, in which shifts in
microbial community composition was previously assessed via PLFA (Smith et al. In prep. see
Chapter 1). While GeoChip functional gene information has revealed differences in the
functional gene structure between microbial communities in pastures compared to early
recovered grasslands (grazing excluded for 3 years) in a Tibetan alpine meadow (Yang et al.
2013), it has also supported high functional redundancy between grasslands and woody plant
157
encroachment of grasslands (Hollister 2008), and with forest succession (Zhang et al. 2007).
Functional redundancy is commonly attributed to microbial communities due to high levels of
diversity among microorganisms, universal abilities in decomposition and SOM transformations
across communities and a high potential of microbial adaptation, acclimation and gene transfer
(Lawton and Brown 1993, Rosenfield et al. 2002). Therefore, it is not impossible that the
functional capacity of the microbial community as measured by gene expression does not change
with land cover change or forest age at our sites, even if the community composition does.
Ordination analyses (NMS, principal components analysis, correspondence analyses, etc)
are often used to explain variations in microbial community ecology (McCune and Grace, 2002)
and are widely used in the literature to assess differences in GeoChip functional gene
information (Bai et al. 2012, Chan et al. 2013, Wakelin et al. 2013, Yang et al. 2013, Zhang et al.
2013). However, many authors report differences in functional gene composition across
experimental treatments based on the spatial variation shown in ordination analyses, but fail to
statistically test the significance of the spatial variation (i.e. distance measurements for each axis)
shown in the ordination analysis. While our data show distinct clusters of pasture, early and late
secondary forest points in the NMS plots of C, N and P cycling functional genes, the variation
between specific sites within a land cover type limited our ability to detect an effect of land
cover type on functional gene capacity. High spatial heterogeneity in microbial community
structure and function is common in soils (Ettema and Wardle 2004), especially tropical forest
soils (Pett-Ridge and Firestone 2005, Smith et al. In prep. see Chapter 1). Therefore, sampling
for and identifying within and between site variability is important in describing microbial
functional diversity of soils. As new technologies emerge in identifying microbial function, we
hope that costs for such analyses become more affordable in order to account for spatial
158
heterogeneity via multiple ecological replicates.
4. 2 Microbial functional recovery in SOM cycling
In contrast, land cover type had a significant effect on the total number of gene probes detected,
implying that land cover change does indeed influence the functional gene capacity of the
microbial community. Overall, the amount of genes and gene probes detected decreased in the
late secondary forest compared to the early secondary forest. A decrease in microbial community
structure or function with forest succession or age is not uncommon (Waldrop et al. 2000, Jai et
al. 2005). Microbial biomass-C and –N rapidly accumulated in early stages of secondary forest
succession and then decreased and maintained constant with forest development (Jia et al. 2005).
This could account for the higher functional genes associated with the early secondary forest
compared to the late secondary forest. Contrary to our results of the ordination analyses, the
difference in the amount of genes and gene probes detected in C, N and P cycling between the
early and late secondary forest suggests that there may not be a large degree of functional
redundancy between the different microbial communities associated with the early and late
secondary forest (Smith et al. In prep. see Chapter 1). Instead, GeoChip gene detection narrates a
parallel story to microbial community composition via PLFA; secondary forest development
affects microbial community dynamics. This supports the case for direct links between microbial
structure (composition) and function (SOM cycling potential) (Tilman et al. 1997, Waldrop et al.
2000, Zak et al. 2003). Both differences between young and old forest plantations and a strong
link between microbial PLFA composition and C cycling function (defined as extracellular
enzyme activities) were reported in study on tropical land use and land cover change (Waldrop et
al. 2000). Continued focus on relationships between microbial composition and functional
159
capacity in SOM cycling, especially in the tropics, will enhance our understanding and
predictions of SOM transformations.
The differences in genes and gene probes detected between the young and old forests
may also support the notion of microbial community resilience, or recovery with forest
development. Prior results show that while microbial community composition (via PLFA)
differed between young secondary forests (20, 30 and 40 years old) and older secondary forests
(70 and 90 years old), composition did not vary between the older secondary forests and primary
forest remnants (or forests that have not been under pasture or agricultural cultivation). This
suggests that microbial community composition recovers to its original state with secondary
forest development. It further pinpoints a tipping point of this recovery between 40 and 70 years.
Aboveground tree species (greater than 10cm dbh) nearly matched primary forest composition
after 60 years of secondary forest regeneration at the same sites (Marin-Spiotta et al. 2007).
Recovery times for turnover rates of mineral associated C occurred more rapidly across the
chronosequence, with rates recovering by only 20 years of forest regrowth to primary forest
levels (Marín-Spiotta et al. 2008). While we were unable to statistically assess primary forest
functional capacity (due to methodological difficulties in extraction of sufficient DNA),
preliminary data explorations using data from just one primary forest site revealed similarities in
GeoChip functional gene information between the late secondary forest and primary forest.
Means of gene detection ratios and signal intensity values for the primary forest sites were also
lower than those of the pasture and early secondary forest sites. Thus, the results of microbial
function gene capacity may agree with our prior results: that the microbial community is resilient
to historical land use, recovering to primary forest levels with secondary forest succession.
4. 3 Challenges in applying metagenomic technologies to ecological questions
160
Advances in the tools and technologies used to investigate microbial community structure and
function have exploded over the last several decades. These methods are largely based on the
analysis of genomic material, or the sequencing of DNA and RNA. We can now identify
unculturable and novel microorganisms and genes (Zhou et al. 2010). These new technologies
have provided an enormous amount of insight into microbial structural and functional diversity
(Torsvik and Øvreås 2002). However, there are still many limitations to these new
technologies, both in methodology and in interpretation (Kozdrój and van Elsas 2001, Hollister
2008). Further, both measuring and defining microbial functional structure and diversity remains
a challenge in microbial ecology. While GeoChip avoids the pitfalls of many genome-based
technologies, such as the bias associated with PCR (Zhou et al. 2010), its interpretation as an
accurate portrayal of microbial functional capacity should be taken with caution (Hollister 2008,
2010). GeoChip functional gene identification detects gene sequences for both active and
dormant microorganisms and does not differentiate between genes that are actively being used
and those that are not. Therefore, it is better referred to as a ‘potential’ for microbial function
versus a representation of in situ microbial function. So, while the functional ‘potential’ may not
change across our sites, the ‘in situ’ activities and process rates could vary across
chronosequence. This may have implications for overall ecosystem processes and SOM
transformations. Presently, there is no technology that exists that provides such comprehensive
gene information in addition to whether or not the gene is actively being used.
GeoChip, and other genomic technologies, are further limited by the quantity and quality
of information that are already known, meaning we are unable to identify functional genes whose
sequences are not already identified. While we can now identify unculturable microorganisms
(previously unknown and unidentifiable), we have not identified (genomically) even a majority
161
of the microorganisms in soils (Gentry et al. 2006). Therefore, while GeoChip can detect an
enormous number of genes from a large variety of organisms, it is unable to detect genes from
unsequenced organisms (Loy et al. 2006). GeoChip also does not contain gene probes for every
identified microbial sequence in existence. Therefore, our ability to detect changes in functional
gene potential is limited by not just the extent of known and sequenced gene probes, but also by
the range of probes included in the GeoChip microarray. For example, GeoChip was designed
with more bacterial-associated gene probes over fungal or archaeal ones (He et al. 2007). Prior
work at our site revealed significant differences in fungal abundance and the fungal-to-bacterial
ratio, but not bacterial abundance with forest recovery (Smith et al. In prep. see Chapter 1).
Perhaps a greater inclusion of fungal-derived gene probes in the GeoChip microarray would
provide more insight into land cover change effects on microbial functional capacity.
Overall while GeoChip was helpful in describing the functional potential in SOM cycling
of soil microbial communities at our sites, it could not explain SOM dynamics or the activities of
extracellular enzymes involved in organic matter degradation. It did, however, provide valuable
evidence for the link between microbial community structure and function, hence, justifying the
need for continued explorations into identification of how microbial communities respond to
ecosystem disturbance and recovery as compositional changes may have significant implications
for C cycling and nutrient availability.
5. Conclusion GeoChip functional gene information gave us the opportunity to comprehensively evaluate
functional diversity and composition of microbial communities from tropical pastures and
secondary forests. Shifts in microbial composition (via PLFA, Smith et al. In prep. See Chapter
162
1) and functional gene detection for cellulose, chitin and starch degradation, ammonification and
phosphorus acquisition genes between early and late secondary forests suggests that: (1)
microbial community composition is linked to microbial function and shifts in microbial
composition can be used as an indicator for a change in microbial function, and (2) microbial
community recovery of both structure and function occurs sometime between 40-70 years of
forest regrowth as function gene measurements from the late secondary forest were closer to
values associated with the primary forest soil communities. This may indicate a tipping point for
ecosystem recovery after at least 40 years of forest regeneration. Overall, this research provides
evidence for the importance of understanding structure-function relationships to predict how
shifts in microbial composition may influence C and nutrient cycling.
163
6. Tables and Figures
Table 1. Total and relative percentage of genes detected by land cover in C, P and N cycling. The relative percent of genes detected is the mean percent of genes detected for each sample (averaged by land cover type) over the total number of genes in that gene category detected across all samples. Numbers in parenthesis represents a propagated standard error (SE), which was calculated as the square root of the sum of standard error values squared for each site.
Land Cover Gene Category
Total†
Active pastures Early secondary forest (40 yr old)
Late secondary forest (90 yr old)
total C gene probes* 4346 2892 (160) ab 2926 (102) a 2449 (345) b Carbon Cycling
% C gene probes* 11% 67% (3.67) ab 67% (2.35) a 56% (7.93) b total P gene probes* 547 348 (492) ab 357 (504) a 299 (424) b Phosphorus Cycling
% P gene probes* 1% 64% (17.53) ab 65% (12.65) a 55% (38.56) b
total N gene probes 3128 2051 (105) 2066 (236) 1753 (68) Nitrogen Cycling
% N gene probes 8% 66% (3.35) 66% (7.54) 56% (2.18) Mean comparisons using Student’s t-test are represented by lowercase letters. Land cover types not connected by the same letter are significantly different. †The relative percent of gene probes for all samples represents the ratio of gene probes detected in each gene category over all gene probes detected. *Land Cover effect marginally significant (p < 0.10)
164
Table 2. The relative percentage of genes detected per sample to total carbon cycling genes detected across all samples, averaged by land cover type. Numbers in parenthesis represents a propagated standard error (SE), calculated as the square root of the sum of standard error values squared for each site.
Land Cover Gene Function Gene
Active Pasture Early Secondary Forest (40 yr old)
Late Secondary Forest (90 yr old)
CDH** 76.83 (8.79)ab 81.71 (7.71)a 72.56 (8.18)b
cellobiase* 64.54 (3.98)ab 71.17 (4.7)a 55.1 (11.94)b
endoglucanase 73.38 (2.6) 74.03 (1.84) 61.04 (5.35)
Cellulose
exoglucanase* 61.51 (2.37)ab 66.12 (1.97)a 53.62 (8.55)b
acetylglucosaminidase** 69.05 (3.87)a 72.43 (2.68)a 58.51 (7.49)b
endochitinase* 65.58 (3.93)a 64.42 (3.67)a 53.17 (6.33)b
Chitin
exochitinase 66.25 (7.29) 60.63 (6.73) 56.25 (9.01)
ara* 68.72 (4.85)ab 70.81 (4.3)a 55.05 (5.62)b
ara_fungi 67.37 (5.68) 69.92 (4.24) 57.63 (6.45)
mannanase 68.06 (5.93) 66.2 (2.07) 58.8 (6.74)
xylA** 66.98 (8.41)a 68.89 (1.15)a 57.63 (5.12)b
Hemicellulose
xylanase 60.53 (6.33) 60.53 (4.47) 49.12 (8.32)
glx* 66.36 (9.09)ab 70.45 (3.28)a 60.91 (6.43)b
lip 74.22 (9.5) 79.69 (2.21) 68.75 (13.26)
mnp 58.78 (8.22) 58.78 (5.57) 53.38 (4.87)
Lignin
phenol oxidase 65.54 (5.41) 66.1 (1.64) 53.72 (10.33)
AceA* 66.14 (5.04)a 66.63 (2.33)a 55.48 (9.03)b
AceB 68.38 (4.14) 67.5 (4.24) 59.26 (8.13)
AssA 50.00 (25) 56.25 (12.5) 37.5 (17.68)
camDCAB 100.00 (0) 75.00 (50) 100.00 (0)
limEH 68.27 (7.93) 68.27 (9.62) 66.35 (4.3)
LMO 45.83 (18.63) 37.5 (8.33) 45.83 (8.33)
vanA** 73.73 (1.85)a 76.74 (0.71)a 64.56 (6.65)b
Others
vdh 76.67 (6.67) 75.83 (1.67) 63.33 (10.54)
pulA** 66.35 (5.18)a 66.11 (5.11)a 53.37 (5.48)b
amyA* 66.94 (4.18)a 67.22 (2.87)a 56.95 (9.53)b
amyX** 68.75 (12.5)a 50.00 (0)b 43.75 (12.5)b
apu 65.00 (22.36) 55.00 (22.36) 45.00 (10)
cda 63.08 (2.18) 62.5 (3.29) 51.15 (4.25)
glucoamylase 59.65 (7.55) 62.72 (3.62) 55.26 (10)
isopullulanase 75.00 (25) 75.00 (25) 50.00 (0)
Carbon degradation
Starch
nplT* 67.14 (2.02)a 60.71 (3.03)ab 51.07 (9.29)b
aclB 56.82 (3.21) 59.09 (3.21) 54.55 (9.09)
CODH** 68.43 (2.58)a 70.51 (3.9)a 60.1 (8.63)b
pcc 68.17 (2.14) 69.3 (2.17) 58.75 (8.69)
Carbon fixation
rubisco 61.85 (5.69) 65.46 (3.23) 54.34 (9.75)
mmoX 58.33 (7.45) 53.33 (9.43) 43.33 (4.71) Oxidation
pmoA 60.59 (2.54) 59.32 (2.4) 48.73 (12.82)
Methane
Production mcrA* 55.71 (7.31)a 51.09 (5.16)ab 39.95 (13.76)b Lowercase letters indicate student's t-test mean comparisons across land cover types (pasture, early and late secondary forest). Land cover types not connected by letter are significantly different. Absence of letter indicates no significant difference between land cover types. * Effect of land cover marginally significant (p < 0.10), **significant (p < 0.05)
165
Table 3. Pearson and Kendall correlation coefficients of gene composition to ordination axes shown in Figure 3.
Correlation Coefficients
NMS Axis 1
Gene
Function
Pearson Kendall
amyA Carbon degradation 94.9% 84.8% Endochitinase Carbon degradation 94.6% 81.8% pcc Carbon fixation 90.5% 90.9%
NMS Axis 2
Pearson Kendall
aceA Carbon degradation 88.9% 78.8% aceB Carbon degradation 66.7% 60.6%
166
Table 4. The relative percentage of genes detected per sample to total nitrogen cycling genes detected across all samples, averaged by land cover type. Numbers in parenthesis represents a propagated standard error (SE), calculated as the square root of the sum of standard error values squared for each site.
Land Cover
Gene Category
Gene Active pasture Early secondary forest
(40 yr old) Late secondary forest
(90 yr old) Nitrogen fixation nifH 62.55 (5.45) 59.05 (2.60) 49.18 (9.47)
amoA 65.50 (64.9) 67.28 (65.54) 57.54 (54.97) Nitrification hao 50.00 (68.78) 58.82 (70.72) 45.59 (60.96)
gdh* 59.56 (63.2) a 61.03 (61.4) a 45.59 (51.4) b Ammonification ureC* 68.78 (6.39) ab 70.72 (8.09) a 60.96 (2.77) b narG 71.41 (0.00) 73.66 (5.88) 63.98 (21.41) nirK* 60.10 (5.52) ab 63.88 (5.47) a 51.43 (6.68) b nirS 61.73 (11.59) 61.83 (11.59) 51.06 (20.95) norB 63.60 (2.72) 60.96 (5.77) 52.63 (10.36)
Denitrification
nosZ* 64.90 (4.32) ab 65.54 (2.04) a 54.97 (9.70) b nasA 67.35 (31.82) 65.29 (4.55) 59.41 (4.55) nirR 73.44 (2.85) 74.61 (3.32) 66.02 (6.57) nirA 72.73 (8.00) 77.27 (1.66) 62.50 (6.71)
Assimilatory N reduction
nirB 69.23 (4.3) 64.42 (0.85) 54.81 (8.45) napA 57.81 (7.35) 54.02 (7.35) 45.54 (6.24) Dissimilatory N reduction
nrfA 63.20 (7.66) 61.40 (1.64) 51.40 (8.90) Lowercase letters indicate student's t-test mean comparisons. Land cover types not connected by letter are significantly different * Effect of land cover marginally significant (p < 0.10)
167
Table 5. Pearson and Kendall correlation coefficients of gene composition to ordination axes in NMS of nitrogen cycling function genes (shown in Figure 4).
Correlation Coefficients
NMS Axis 1
Gene
Function
Pearson Kendall
nirS Denitrification 92.5% 78.8% narG Denitrification 91.6% 84.8% nosZ Denitrification 89.1% 78.8%
NMS Axis 2
Pearson Kendall
narG Denitrification 89.1% 72.7% nifH Nitrogen fixation 83.5% 75.8% narG Denitrification 81.4% 78.8%
168
Table 6. Pearson and Kendall correlation coefficients of gene composition to ordination axes in NMS of nitrogen cycling function genes (shown in Figure 6).
Correlation Coefficients
NMS Axis 1
Gene
Function
Pearson Kendall*
phytase Phosphorus utilization 82.5% 78.8% (-) ppk Phosphorus utilization 81.4% 66.7% ppx Phosphorus utilization 89.1% 63.6% (-)
NMS Axis 2
Pearson Kendall
ppx Phosphorus utilization 67.2% 67.9% * negative sign in parentheses indicates correlation with variation associated with the left side of the NMS plot (Figure _).
169
Figure 1. Microbial composition of genes involved in C, N and P cycling totaled across all land cover types.
170
Figure 2. Microbial composition of genes detected in C cycling processes: degradation, fixation and methane cycling, N cycling processes: fixation, ammonification, nitrification, assimilatory and dissimilatory reduction, and denitrification, and P acquisition genes; phytase, ppk and ppx.
171
Figure 3. Nonmetric multidimensional scaling (NMS) plot of Bray-Curtis (Sorensen) similarities for carbon cycling functional gene profiles (Geochip hybridization signal intensities) and soil environmental and microbial variables (2D stress value = .0986). Pearson correlations of environmental and microbial variables to ordination axes show soil moisture and PLFA fungal-to-bacterial ratio (F/B) explaining (52.7% and 27.1%) of the variation along ordination axis 2.
172
Figure 4. Nonmetric multidimensional scaling (NMS) plot of Bray-Curtis (Sorensen) similarities for nitrogen cycling functional gene profiles (Geochip hybridization signal intensities) and soil environmental and microbial variables (2D stress value = .0916). Pearson correlations of environmental and microbial variables to ordination axes show PLFA fungal biomarkers 18:1w9c, 18:2w6,9c, and the PLFA fungal-to-bacterial ratio (F/B) explaining 40.4%, 30.5% and 25.4% of the variation along ordination axis 2.
173
Figure 5. The relative percentage of genes detected per sample to total phosphorus cycling genes detected across all samples, averaged by land cover type. Error bars represent the propagated standard error (SE), calculated as the square root of the sum of standard error values squared for each site. Lowercase letters represent mean comparison using student’s t tests. Land cover types not sharing letters are significantly different. * Indicates marginal significance (p < 0.10).
ab a a a
b b
0 10 20 30 40 50 60 70 80
phytase ppk* ppx*
%
Pasture Early secondary forest Late secondary forest
174
Figure 6. Nonmetric multidimensional scaling (NMS) plot of Bray-Curtis (Sorensen) similarities for phosphorus cycling functional gene profiles (Geochip hybridization signal intensities) and soil environmental and microbial variables (2D stress value = .0920). Pearson correlations of environmental and microbial variables to ordination axes show PLFA fungal biomarkers 18:1w9c, 18:2w6,9c, and the PLFA fungal-to-bacterial ratio (F/B) explaining 43.8%, 30.6% and 31.4% of the variation along ordination axis 2.
175
7. References Aide, T. and H. Grau (2004). "Globalization, migration, and Latin American ecosystems." Science 305: 1915–1916. Aide, T. M., M. L. Clark, et al. (2012). "Deforestation and Reforestation of Latin America and the Caribbean (2001–2010)." Biotropica: 1-‐10. Allison, S. D. and J. B. H. Martiny (2008). "Colloquium Paper: Resistance, resilience, and redundancy in microbial communities." Proceedings of the National Academy of Sciences 105(Supplement 1): 11512-‐11519. Bai, S., J. Li, et al. (2012). "GeoChip-‐based analysis of the functional gene diversity and metabolic potential of soil microbial communities of mangroves." Applied Microbiology and Biotechnology 97(15): 7035-‐7048. Banning, N. C., D. B. Gleeson, et al. (2011). "Soil Microbial Community Successional Patterns during Forest Ecosystem Restoration." Applied and environmental microbiology 77(17): 6158-‐6164. Bardgett, R. D., P. J. Hobbs, et al. (1996). "Changes in soil fungal: bacterial biomass ratios following reductions in the intensity of management of an upland grassland." Biology and Fertility of Soils 22(3 ): 261-‐264. Chan, Y., J. D. Van Nostrand, et al. (2013). "Functional ecology of an Antarctic Dry Valley." PNAS 110(22): 8990-‐8995. Cleveland, C. C. and D. Liptzin (2007). "C: N: P stoichiometry in soil: is there a “Redfield ratio” for the microbial biomass?" Biogeochemistry 85(3): 235-‐252. Daly, C., E. H. Helmer, et al. (2003). "Mapping the climate of Puerto Rico, Vieques and Culebra." International Journal of Climatology 23: 1359–1381. Ettema, C. H. and D. A.Wardle (2002). "Spatial soil ecology." Trends in Ecology & Evolution 17(4): 177-‐183. Frostegård, A. and E. Bååth (1996). "The use of phospholipid fatty acid analysis to estimate bacterial and fungal biomass in soil " Biology and Fertility of Soils 22(1-‐2 ): 59-‐65. Frostegård, Å., A. Tunlid, et al. (2011). "Use and misuse of PLFA measurements in soils." Soil Biology and Biochemistry 43(8): 1621-‐1625. Gentry, T. J., G. S. Wickham, et al. (2006). "Microarray Applications in Microbial Ecology Research." Microbial Ecology 52(2): 159-‐175.
176
German, D. P., M. N. Weintraub, et al. (2011). "Optimization of hydrolytic and oxidative enzyme methods for ecosystem studies." Soil Biology and Biochemistry 43(7): 1387-‐1397. Glaser, B., M. a.-‐B. Turrión, et al. (2004). "Amino sugars and muramic acid—biomarkers for soil microbial community structure analysis." Soil Biology and Biochemistry 36(3): 399-‐407. Grau, H. R., T. M. Aide, et al. (2003). "The Ecological Consequences of Socioeconomic and Land-‐Use Changes in Postagriculture Puerto Rico." Bioscience 53(12): 1159-‐1168. He, Z., Y. Deng, et al. (2010). "GeoChip 3.0 as a high-‐throughput tool for analyzing microbial community composition, structure and functional activity." The ISME Journal 4(9): 1167-‐1179. He, Z., T. J. Gentry, et al. (2007). "GeoChip: a comprehensive microarray for investigating biogeochemical, ecological and environmental processes." The ISME Journal 1(1): 67-‐77. He, Z., M. Xu, et al. (2010). "Metagenomic analysis reveals a marked divergence in the structure of belowground microbial communities at elevated CO2." Ecology Letters 13(5): 564-‐575. Helmer, E. H., O.Ramos, et al. (2002). "Mapping the Forest Type and Land Cover of Puerto Rico, a Component of the Caribbean Biodiversity Hotspot." Caribbean Journal of Science 38(3-‐4): 165-‐183. Hollister, E. B. (2008). Land use and land cover change: The effects of woody plant encrahment and prescribed fire on biodiversity and ecosystem carbon dynamics in a southern great plains mixed grass savanna. Molecular and Environmental Plant Sciences, Texas A&M. PhD: 148. Hollister, E. B., C. W. Schadt, et al. (2010). "Structural and functional diversity of soil bacterial and fungal communities following woody plant encroachment in the southern Great Plains." Soil Biology and Biochemistry 42(10): 1816-‐1824. Houghton, R. A. and C. L. Goodale (2004). "Effects of Land-‐Use Change on the Carbon Balance of Terrestrial Ecosystems." Ecosystems and land use change: 85-‐98. Houghton, R. A., J. D. Unruh, et al. (1993). "Current land cover in the tropics and its potential for sequestrating carbon." Global Biogeochem. Cycles 7: 305-‐320. Jia, G.-‐m., J. Cao, et al. (2005). "Microbial biomass and nutrients in soil at the different stages of secondary forest succession in Ziwulin, northwest China." Forest Ecology and Management 217(1): 117-‐125. Joergensen, R. and F. Wichern (2008). "Quantitative assessment of the fungal contribution to microbial tissue in soil." Soil Biology and Biochemistry 40(12): 2977-‐2991.
177
Kaiser, C., A. Frank, et al. (2010). "Negligible contribution from roots to soil-‐borne phospholipid fatty acid fungal biomarkers 18:2ω6,9 and 18:1ω9." Soil Biology and Biochemistry 42(9): 1650-‐1652. Kauffman, J. B., R. F. Hughes, et al. (2009). "Carbon Pool and Biomass Dynamics Associated with Deforestation, Land Use, and Agricultural Abandonment in the Neotropics." Ecological Applications 19(5): 1211-‐1222. Kaur, A., A. Chaudhary, et al. (2005). "Phospholipid fatty acid – A bioindicator of environment monitoring and assessment in soil ecosystem." Current Science 89(7): 1103-‐1112. Kozdroj, J. and J. D. v. Elsas (2001). "Structural diversity of microorganisms in chemically perturbed soil assessed by molecular and cytochemical approaches." Journal of Microbiological Methods 43: 197-‐212. Kuramae, E. E., H. A. Gamper, et al. (2010). "Microbial secondary succession in a chronosequence of chalk grasslands." The ISME Journal 4(5): 711-‐715. Lawton, J. H. and V. K. Brown (1993). Redundancy in Ecosystems. Biodiversity and ecosystem functions. E. D. Schulze and H. A. Mooney. New York, NY, Springer Verlag. Liang, Y., J. D. V. Nostrand, et al. (2009). "Microarray-‐based functional gene analysis of soil microbial communities during ozonation and biodegradation of crude oil." Chemosphere 75(2): 193-‐199. Loy, A., M. W. Taylor, et al. (2006). Applications of nucleic acid microarrays in soil microbial ecology. Molecular Techniques for Soil, Rhizosphere, and Plant Microorganisms. J. E. Cooper and J. R. Rao. Wallingford, U.K., CABI Publishing: 18-‐41. Marín-‐Spiotta, E., R. Ostertage, et al. (2007). "Long-‐term patterns in tropical reforestation: Plant commnity composition and aboveground biomass accumulation " Ecological Applications 17(3): 828-‐839. Marín-‐Spiotta, E., W. L. Silver, et al. (2009). "Soil organic matter dynamics during 80 years of reforestation of tropical pastures." Global change biology 15(6): 1584-‐1597. Marín-‐Spiotta, E., C. W. Swanston, et al. (2008). "Chemical and mineral control of soil carbon turnover in abandoned tropical pastures." Geoderma 143(1-‐2): 49-‐62. McCune, B. and J. B. Grace (2002). Analysis of ecological communities Gleneden Beach, Oregon, MjM software design.
178
Meiyappan, P. and A. K. Jain (2012). "Three distinct global estimates of historical land-‐cover change and land-‐use conversions for over 200 years." Frontiers of Earth Science 6(2): 122-‐139. Olsson, P. A. (1999). "Signature fatty acids provide tools for determination of the distribution and interactions of mycorrhizal fungi in soil." FEMS Microbiology Ecology 29: 303-‐310. Olsson, P. A., E. Baath, et al. (1995). "The use of phospholipid and neutral lipid fatty acids to estimate biomass of arbuscular mycorrhizal fungi in soil." Mycological Restoriation 5: 623-‐629. Ostertag, R., E. Marín-‐Spiotta, et al. (2008). "Litterfall and Decomposition in Relation to Soil Carbon Pools Along a Secondary Forest Chronosequence in Puerto Rico." Ecosystems 11(5): 701-‐714. Pascarella, J. B., T. M. Aide, et al. (2000). "Land-‐Use History and Forest Regeneration in the Cayey Mountains, Puerto Rico." Ecosystems 3(3): 217-‐228. Pett-‐Ridge, J. and M. K. Firestone (2005). "Redox Fluctuation Structures Microbial Communities in a Wet Tropical Soil." Applied and Environmental Microbiology 71(11): 6998-‐7007. Prentice (2001). The Carbon Cycles and Atmospheric Carbon Dioxide. Climate Change 2001: The Scientific Basis. Contributions of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. J. T. Houghton. Cambridge, UK, Cambridge University Press. Reich, P. B., D. Tilman, et al. (2004). "Species and functional group diversity independently influence biomass accumulation and its response to CO2 and N." Proc. Natl. Acad. Sci. USA 101: 10101–10106. Rosenfeld, J. S. (2002). "Functional Redundancy in Ecology and Conservation." Oikos 98(1): 156-‐162. Rossello-‐Mora, R. and R. Amann (2001). "The species concept for prokaryotes." FEMS Microbiology Reviews 25: 39-‐67. SERCC (2013). "Jajome Alto Climate Station precipitation data." From http://atmos.uprm.edu/). Sinsabaugh, R. L. and J. J. Follstad Shah (2012). "Ecoenzymatic stoichiometry and ecological theory." Annual Review of Ecology, Evolution, and Systematics 43: 313-‐343. Six, J., C. Feller, et al. (2002). "Soil organic matter, biota and aggregation in temperate and tropical soils -‐ Effects of no-‐tillage." Agronomie 22(7-‐8): 755-‐775.
179
Smithwick, E. A. H., M. G. Turner, et al. (2005). "Variation in NH4+ mineralization and microbial communities with stand age in lodgepole pine (Pinus contorta) forests, Yellowstone National Park (USA)." Soil Biology and Biochemistry 37(8): 1546-‐1559. Sparks, D. L., A. L. Page, et al. (1996). Methods of soil analysis. Part 3 -‐ chemical methods. . Madison, WI, Soil Science Society of America. Staff, S. S. (2008). Official Soil Series Descriptions Lincolcn, NE, USDA-‐NRCS. Tilman, D. (1997). "The Influence of Functional Diversity and Composition on Ecosystem Processes." Science 277(5330): 1300-‐1302. Torsvik, V. and L. Øvreås (2002). "Microbial diversity and function in soil: from genes to ecosystems." Current Opinion in Microbiology 5: 240-‐245. Van Nostrand, J. D., W.-‐M. Wu, et al. (2009). "GeoChip-‐based analysis of functional microbial communities during the reoxidation of a bioreduced uranium-‐contaminated aquifer." Environmental Microbiology 11(10): 2611-‐2626. Vestal, J. R. and D. C. White (1989). "Lipid Analysis in Microbial Ecology." Bioscience 39(8): 535-‐541. Wakelin, S. A., B. I. P. Barratt, et al. (2013). "Shifts in the phylogenetic structure and functional capacity of soil microbial communities follow alteration of native tussock grassland ecosystems." Soil Biology and Biochemistry 57: 675-‐682. Waldrop, M. P., T. C. Balser, et al. (2000). "Linking microbial community composition to function in a tropical soil." Soil Biology and Biochemistry 32: 1837-‐1846. Wang, F., H. Zhou, et al. (2009). "From the Cover: GeoChip-‐based analysis of metabolic diversity of microbial communities at the Juan de Fuca Ridge hydrothermal vent." Proceedings of the National Academy of Sciences 106(12): 4840-‐4845. Wixon, D. L. and T. C. Balser (2009). "Complexity, climate change and soil carbon: A systems approach to microbial temperature response." Systems Research and Behavioral Science 26(5): 601-‐620. Wu, L., X. Liu, et al. (2006). "Microarray-‐Based Analysis of Subnanogram Quantities of Microbial Community DNAs by Using Whole-‐Community Genome Amplification." Applied and environmental microbiology 72(7): 4931-‐4941. Yang, Y., L. Wu, et al. (2013). "Responses of the functional structure of soil microbial community to livestock grazing in the Tibetan alpine grassland." Global Change Biology 19(2): 637-‐648.
180
Zak, D. R., W. E. Holmes, et al. (2003). "Plant diversity, soil microbial communities, and ecosystem function: Are there any links?” Ecology 84(8): 2042-‐2050. Zelles, L. (1997). "Phospholipid fatty acid profiles in selected members of soil microbial communities.” Chemosphere 35(1/2): 275-‐294. Zelles, L. (1999). "Fatty acid patterns of phospholipids and lipopolysaccharides in the characterisation of microbial communities in soil: a review." Biology and Fertility of Soils 29: 111-‐129. Zhang, Y., Z. Lu, et al. (2013). "Geochip-‐based analysis of microbial communities in alpine meadow soils in the Qinghai-‐Tibetan plateau." BMC Microbiology 13(1): 72. Zhang, Y., X. Zhang, et al. (2007). "Microarray-‐based analysis of changes in diversity of microbial genes involved in organic carbon decomposition following land use/cover changes." FEMS Microbiology Letters 266(2): 144-‐151. Zhou, J., M. A. Bruns, et al. (1996). "DNA recovery from soils of diverse composition." Applied and environmental microbiology 62(2): 316-‐322. Zhou, J., Z. He, et al. (2010). "Applying GeoChip Analysis to Disparate Microbial Communities." Microbe 5(2): 60-‐65. Zhou, J., S. Kang, et al. (2008). "Spatial scaling of functional gene diversity across various microbial taxa." Proceedings of the National Academy of Sciences 105(22): 7768-‐7773.
181
CONCLUSIONS:
Microbial succession, recovery, structure-function links
What began as an initial exploration of the role soil microorganisms play in driving soil organic
carbon (SOC) patterns across a post-agricultural forest regeneration chronosequence in
subtropical Puerto Rico, has resulted in a wealth of novel information regarding soil microbial
ecology, the study of how microorganisms behave and interact with each other and their
environment. Through my dissertation research, I have been able to connect soil microbial
community recovery and succession with aboveground forest community succession in both the
long and short term, identify specific drivers of microbial community dynamics across our sites,
link shifts in microbial community structure with functional potential in carbon, nitrogen and
phosphorus cycling, as well as link microbial community composition with SOM pools in a
relatively underrepresented global change process (tropical forest regeneration).
One of the initial findings illustrated successional changes in microbial community
composition with natural forest regeneration on abandoned pastures. Despite intra- and inter-
annual variation in microbial community composition and extracellular enzyme activities,
microbial composition differentiated into three distinct clusters based on land use and forest age:
pasture-associated communities, early secondary forest communities, and late secondary,
primary forest communities. The shifts in microbial community structure with forest age nearly
paralleled compositional succession in the aboveground tree communities (as describes in
previous studies). While microbial succession has been documented for litter decomposition,
compost age and primary succession, I was unable to locate any studies that show nearly
identical successional patterns in both aboveground and belowground communities.
Further, the importance of plant-soil-microbe interactions in shaping both aboveground
182
and belowground community succession was illustrated when one of the replicate pasture sites
began experiencing land use conversion to a forest during the two and a half years soils were
sampled. Not only was microbial succession directly linked to colonization of pasture-associated
grasslands with forest biomass (Chapter 1), but I also showed how the microbial community
responds quite rapidly, or within 6 months to 1 year following initiation of forest regeneration
(Chapter 2). This understanding of rapid microbial response with ecosystem recovery was only
made available through repeated sampling over several years and seasons. This stresses the
importance of long-term data collection and strength in experimental design. Rapid microbial
response to changes in vegetation and plant-associated inputs has implications for understanding
and predicting belowground nutrient and C cycling processes with ecosystem recovery.
The interactions between clay minerals, and the distribution of SOC and microbial
community structure was another important finding in this study (Chapter 3). Through this study,
I showed how the high clay and iron oxide content of these highly-weathered soils are the main
stabilizing mechanism for SOC and soil aggregates. This also seems to drive the distribution of
microorganisms by providing microbial-specific niches through both its influence on SOC
accumulation and control over the physical and chemical environment of aggregates. Microbial
community composition varied among soil aggregate fractions, with two functional composition
ratios driving shifts in microbial composition with land use and cover change. A greater relative
abundance of fungi compared to bacteria and a smaller relative abundance of gram-positive
bacteria compared to gram-negative bacteria in the larger aggregates may alter SOC
mineralization and stabilization processes among aggregate fractions as each of these functional
groups of microbes are known to preferentially utilize different sources of soil organic matter
(SOM). Defining the relationship between microbial composition and the distribution of C in
183
soils in important for understanding and predicting how future changes in tropical land cover
alters SOM cycling and storage processes.
In addition to linking microbial communities to ecosystem function via SOM dynamics, I
also showed how microbial composition is linked to microbial community function (Chapter 4).
The functional gene diversity of genes involved in C, N and P cycling were all correlated with
the fungal-to-bacterial ratio in ordination analyses. Further, the difference of specific functional
genes involved in specific processes regulating cellulose, chitin and starch degradation,
ammonification and phosphorus acquisition detected between the early and secondary forest
paralleled differences in microbial composition with forest regeneration (Chapter 1).
The marked difference in microbial community composition and function between early
and late secondary forests and the similarities that exist between late secondary and primary
forests, suggests that historical land use legacies are ephemeral, persisting only in earlier stages
of ecosystem succession or development. Through this research, I also suggest that the microbial
community rapidly responds to aboveground recovery and return to nearly ‘original’ (i.e.
primary forest) community structure sometime between 40-70 years following forest
regeneration. This may imply that there is a tipping point for microbial community recovery that
is driven by interactions with overall ecosystem development and recovery.
Despite recent recognition of central role soil microbes play in shaping above and
belowground processes, the specific mechanisms such as the role of microbial composition, are
still unclear. This is especially true when it comes to the tropics. Few studies investigate how
tropical soil microbes respond to changes in land use or cover compared to temperate systems.
Through this study, comprehensive information on ecological linkages between soil microbes
and aboveground communities and how these interactions influence SOM cycling dynamics was
184
gained. As more areas in the tropics experience post-agricultural reforestation, understanding
patterns in belowground community structure and function can improve predictions of the fate of
ecosystem carbon with an increase in forest cover.