Variation in species interactions and their evolutionary consequences

Scott Alan ChamberlainPh.D. Thesis Defense

Wedenesday, 30 May, 2012

Species interactions are important across domains of ecology

• Species interactions contribute to:– Population dynamics– Formation and dynamics of food webs– Evolutionary change through natural selection

Variation in species interactions

• Not error variation in outcome in one context

• The variation in outcomes among more than one context

Site 1

Site 1 Site 2

Variation in outcomes is common

Palmer et al. 2008, Callaway et al. 2002

Competition/FacilitationMutualism

More E.G.: Cushman and Whitham 1989, Thompson and Cunningham 2002, Pennings and Silliman 2005, Navarrete and Berlow 2006

Herbivory outcome for Acacia trees varies with ant species identity

Plan

ts p

rodu

cing

flow

ers

or fr

uit (

%)

Low Elevation High Elevation

Outcomes vary from competition at low elevation to facilitation at high elevation

With Ants Without Ants

Variation in outcomes is important

Miller et al. 2009

Populations Communities Evolution

Kokkoris et al. 2002

Interaction Strength Mean

Inte

racti

on S

tren

gth

Varia

nce

Rudgers & Strauss 2004

Greater StabilityHigher species richness

Less StabilityLower species richness

Effec

t of H

erbv

iore

s on

Ca

ctus

Pop

ulati

on G

row

th

Elevation

Low Mid High Sele

ction

Str

engt

h

Site

Questions

1. What are the evolutionary consequences of variation in species interactions?

2. How do types of species interactions differ in variation?

3. How do gradients differ in importance for variation in species interactions?

What are the evolutionary consequences of variation in species interactions?

• Variation in abundance and community structure lead to variation in species interactions

• How is natural selection altered in response to these variable interactions?

How do types of species interactions differ in variation?

- / - + / - + / +

Competition Predation Mutualism

How do gradients differ in importance for variation in species interactions?

?

Space Time

Outline

• Part I : What are the evolutionary consequences of agriculturally altered species interactions?

• Part II: How variable are species interaction outcomes?

Part I--

What are the evolutionary consequences of agriculturally altered species interactions?

12Foley et al. 2005 Science

Mechanisms for altered evolution in agricultural landscapes

• Gene flow from crops to wild/weed plants• Evolution of resistance to genetically modified

crops (e.g., Bt cotton)• Evolution of resistance in plant weeds to

chemical herbicides

• Yet, little examination of altered natural selection via altered species interactions

13Ellstrand et al. 1999

14

Mutualists

VS.

Natural selectionMeehan et al. 2011, Devictor et al. 2008, Ekroos et al. 2010, Dormann et al. 2007

Antagonists

Spatial variation in importance of mutualists and antagonists on selection

15Gomez et al. 2009 Ecol. Monog.

Flower Traits

Mutualists

Plant Fitness

Flower Traits

Mutualists

Plant Fitness

Antagonists

Site 1 Site 2

Abundance and community structureAbundance Community Structure

Questions

Does proximity to crops:1. Alter abundance of mutualists and

antagonists?2. Alter community structure of mutualists and

antagonists? 3. Affect selection on native plant floral traits?4. Alter contribution of mutualists and

antagonists to selection on native plant floral traits?

Study System: Helianthus annuusMutualistPollinators

Halictus ligatus

Megachile spp.

Apis mellifera

AntagonistsSeed predators

Neolasioptera (Diptera)

Isophrictis (Lepidoptera)

Wild sunflower(H. annuus texanus)

Crop sunflower(H. annuus)

Folivores

Smicronyx (Coleoptera)

Study Design

Sunflower Crop

Near

Far

Distance≤ 10 m

Distance~ 2.5 km

Agricultural landscape

Other Crop[corn/sorghum/wheat/cotton]

Natural habitat

Proximity to sunflowers (2 levels)X

Seed source (2 levels)

@ 5 sites in ‘10, @ 2 sites in ‘11

Data Collected• Pollinators: pollinator observations• Seed predators: counted damaged seeds• Folivores: leaf damage

Abundance - mutualistsGreater Near vs. Far

Far

Near

Proximity to cropsunflowers

Abun

danc

e

Far Near

Abun

danc

e

Visi

ts in

flore

scen

ce-1

min

-1

Abundance - antagonistsGreater Far vs. Near

Far

Near

Proximity to cropsunflowers

Abun

danc

e

Far Near

Abun

danc

e

Smic

rony

xIs

ophr

ictis

Neo

lasi

opte

ra

Abundance - antagonistsGreater Far vs. Near

Far Near

Abun

danc

e

Suck

ing

foliv

ore

abun

danc

eCh

ewin

g fo

livor

e ab

unda

nce

Community structure - mutualists & antagonistsDiffers Near vs. Far for both M and A

Far

Near

Mutualists

Antagonists

Proximity to cropsunflowers

Abun

danc

e

Beta-diversity

• This pattern may be due to large crop sunflower resource pulse driving greater diversity among sites

* No difference for antagonists

Proximity, P = 0.004

Abun

danc

e

Proximity to cropsunflowers

Far

Near

How does proximity to crop sunflowers affect selection on H.a. texanus flower traits?

-> Five of nine heritable in narrow-sense (sire-offspring regression)

Throat width

Floral tube size

Proximal throat size

Throat length

Petal size

Disk diameter

Ray width

Ray length

Number of rays

Phenotypic selection analysis

• Total selection (s’)– Measures direct + indirect selection– Simple regression measures calculates covariance

between standardized trait (mean=0, sd=1) and relative fitness

• Direct selection (β)– Measures direct selection on a trait by removing

indirect selection on all other traits in a multiple regression

– Multiple regression with standardized traits (mean=0, sd=1) and relative fitness

Testing for differences in selection by proximity

• Analysis of Covariance– 2010: five sites

• Model: relative fitness ~ site * proximity * trait• trait * proximity• trait * site * proximity

– 2010 & 2011: two sites• Model: relative fitness ~ year * site * proximity * trait• trait * year * proximity• trait * year * site * proximity

– Total selection• Separate models for each floral trait

– Direct selection• One model including all floral traits

ANCOVAtrait * proximity

trait * site * proximitytrait * year * proximity

trait * year * site * proximity

Natural selection Differs Near vs. Far in some traits

Throat width

Floral tube size

Proximal throat size

Throat length

Petal size

Disk diameter

Ray width

Ray length

Number of rays

Total Selection (s’) Direct Selection (β)

NS

NS

NS

NS

NS

NS

NS

NS

NS

Far Near Far Near

Natural selection – Dispersion

Far NearProximity to crop sunflowers

Sele

ction

(s’ o

r β)

• This pattern may be due to large crop sunflower resource pulse driving greater diversity among sites

Do mutualists and antagonists contribute to selection on floral traits differently?

Antagonists Mutualists

Flower traits

Plant fitness

Floral Traits Inflorescence Traits

W = Relative Plant Fitness

Neolasioptera helianthiIsophrictis sp Pollen

Seed predators andpollen depositionstandardized to mean = 0, sd = 1

Multi-group analysis to compare pathsbetween treatments (Near vs. Far)

Principal componentsAnalysis: reduced dimensionality, using just PC1 for each

Site 1 2011

Far Near

Site 22011

-0.07 0.02-0.06 0.004

-0.010.002

Conclusions Part I

• Sunflower mutualists more abundant near, antagonists more abundant far from crops

• Beta-diversity of mutualists greater near crops• Natural selection altered by proximity to sunflower

crops• Changes in mutualist/antagonist communities drive

differences in selection near vs. far from crops• This is one of few studies to show agricultural

effects on natural selection across a landscape in a native plant species

Implications

• Mutualist-antagonist framework may be useful in understanding agricultural effects on plant evolution

• Natural selection altered in agricultural landscapes, BUT contrary to expectation

• These results may not be found in non-intensive agricultural landscapes

Questions

1. What are the evolutionary consequences of variation in species interactions?

2. How do types of species interactions differ in variation?

3. How do gradients differ in importance for variation in species interactions?

Part II--

How variable are species interaction outcomes?

Questions

• A) How do different species interaction types differ in variation in outcomes?

• B) What are relative importance of drivers of variation in outcomes?

Meta-analysis Web of Science search Experimental studies only Interaction outcome w/ & w/o competitor, predator, or mutualist Error estimates & sample sizes available Response variables: abundance, population growth,

reproduction, etc. Responses measured over >1 year, population, or species, etc.

Final dataset 353 papers

Variation in Interaction Outcome Magnitude

Mean Interaction Outcomefrom Armas et al. (2004)

Site A

Site C Site D Site E

Change in sign of Interaction Outcome

0

0

-1-11

Site B

Site BSite A

Site D

Site C

Site E

Site BSite A

Site D

Site C

Site E

0 or 1

Negative RII = better w/o herbivoryPositive RII = better with herbivory

Gradients that drive variation in interaction outcome

Time

Species identity

3rd party presence

Abiotic

Space

Nutrients

Across hours, days, years

Sp. A interacts with either sp. B or sp. C

Two species w/ or w/o 3rd species

Across sites

How do different species interaction types differ in variation in outcomes?

• Mean strength– Mutualisms weaker than antagonisms (Morris et al. 2007)– General sense in literature that mutualisms less important

because so variable (Sachs & Simms 2006)– Weak interactions the most variable (Berlow et al. 1999)

• Interaction complexity– Predation more specialized than mutualism (Gomez et al.

2010)– Strength greater with fewer interactions (Edwards et al.

2010)

How do different species interaction types differ in variation in outcomes?

Predation

Competition

Mutualism

Interaction ComplexityMean Strength

More specialized

??

More generalized

Expected Varation

Low

High

Strong

Weak

?? ??

How do different species interaction types differ in variation in outcomes?

p c m

Pro

port

ion

of s

tudi

es

0.0

0.2

0.4

0.6

CV

* ef

fect

siz

e

0

50

100

150

A

bB

(120) (143) (90)

c

a

< <

= =

Prop

ortio

n of

stud

ies

with

sig

n ch

ange

CV o

f Effe

ct S

ize

Predation Competition Mutualism

CV

* e

ffe

ct s

ize

0

50

100

150

200

250

(53) (46) (117) (97) (40)

ab

ab

a

bb

What are relative importance of drivers of variation in outcomes?

Species id

entity

AbioticSpace

CV o

f Effe

ct S

ize

3rd party

presence

Time

Variation highly dependent on context in which the interaction occurs

CV

* ef

fect

siz

e

0

100

200

300

400

aba

b

p c m

Pro

port

ion

of s

tudi

es

0.0

0.2

0.4

0.6

0.8

p c m p c m p c m p c m

a a

b

(21) (26) (6) (11) (16) (19) (53) (37) (27) (26) (47) (24) (9) (17) (14)

a a

b

a

bb a

b

c

a

b b

abiotic spatial species identity temporal third party presence

Prop

ortio

n of

stud

ies

with

sig

n ch

ange

CV o

f Effe

ct S

ize

Variation highly dependent on context in which the interaction occurs

CV

* ef

fect

siz

e

0

100

200

300

400

aba

b

p c m

Pro

port

ion

of s

tudi

es

0.0

0.2

0.4

0.6

0.8

p c m p c m p c m p c m

a a

b

(21) (26) (6) (11) (16) (19) (53) (37) (27) (26) (47) (24) (9) (17) (14)

a a

b

a

bb a

b

c

a

b b

abiotic spatial species identity temporal third party presence

Prop

ortio

n of

stud

ies

with

sig

n ch

ange

CV o

f Effe

ct S

izeC

V*

effe

ct s

ize

0

100

200

300

400

aba

b

p c m

Pro

port

ion

of s

tudi

es

0.0

0.2

0.4

0.6

0.8

p c m p c m p c m p c m

a a

b

(21) (26) (6) (11) (16) (19) (53) (37) (27) (26) (47) (24) (9) (17) (14)

a a

b

a

bb a

b

c

a

b b

abiotic spatial species identity temporal third party presence

Predation

Mutualism

- Opposite of prediction that specialized predation may lead to less variation

- Instead, when you interact with more species, each interaction is more equivalent, and are not that variable

Prop

ortio

n of

stud

ies

with

sig

n ch

ange

Variation highly dependent on context in which the interaction occurs

CV

* ef

fect

siz

e

0

100

200

300

400

aba

b

p c m

Pro

port

ion

of s

tudi

es

0.0

0.2

0.4

0.6

0.8

p c m p c m p c m p c m

a a

b

(21) (26) (6) (11) (16) (19) (53) (37) (27) (26) (47) (24) (9) (17) (14)

a a

b

a

bb a

b

c

a

b b

abiotic spatial species identity temporal third party presence

CV

* ef

fect

siz

e

0

100

200

300

400

aba

b

p c m

Pro

port

ion

of s

tudi

es

0.0

0.2

0.4

0.6

0.8

p c m p c m p c m p c m

a a

b

(21) (26) (6) (11) (16) (19) (53) (37) (27) (26) (47) (24) (9) (17) (14)

a a

b

a

bb a

b

c

a

b b

abiotic spatial species identity temporal third party presence

CV o

f Effe

ct S

ize - In predation studies, species were largely animals, which are more mobile than plants

- In competition and mutualism studies, species were largely plants, which are immobile

- Interactions involving immobile plants may be more variable along abiotic gradients as they cannot escape them

Prop

ortio

n of

stud

ies

with

sig

n ch

ange

Conclusions• Types of species interactions differed in outcome variation

– Implications: • We can’t treat different species interactions as equivalent• In interaction webs, it may be most important to understand variation in

mutualistic links

• Types of gradients differed in outcome variation

– Implications: • Some sources of variation in species interactions should be given priority (i.e.,

species identity), especially in new study systems

>

SpaceSpecies identity Abiotic> > >

>

Time 3rd party presence

Future work• Add other species interaction types: herbivory,

parasitism, facilitation

• Do any variables correlate with variation in species interaction outcomes?– Do body size ratios predict variable outcomes?

Questions

1. What are the evolutionary consequences of variation in species interactions?

2. How do types of species interactions differ in variation?

3. How do gradients differ in importance for variation in species interactions?

Thanks to• Committee

– Jennifer Rudgers– Ken Whitney– Volker Rudolf– Dennis Cox

• Help– Toby Liss– Wael Al Wawi– Charles Danan– Yosuke Akiyama– Neha Deshpande– Rohini Sigireddi– Prudence Sun– Morgan Black– Edward Realzola

• Microscopy– John Slater– Robert Langsner

• Meta-analysis– Tens of authors who

provided data

• Discussion– The R-W lab– Steve Hovick– Tom Miller

• Of course: Katherine Horn

Predation

Competition

Mutualism/Facilitation

Mean / Variation Argument

Interaction ComplexityArgument

+

-

- -

+

+-

OK, BUT WHAT ARE THE CONSEQUENCES?

Consequences of Variation in OutcomeEcological

Outcomes between membracids and antsvaried with:

- Time (among years)- Membracid life stage- Membracid abundance

And these likely will influence population dynamics of the interaction

Cushman & Whitham (1989)

Consequences of Variation in OutcomeEvolutionary

Thompson (2005), Bronstein (1994)

Interaction Outcome CV

Distributed OutcomesRaw Material for Evolution of Species Interactions

β

1

0

-10 100

John Thompson - Distributed Outcomes

% o

f Int

erac

tions

Antagonistic Mutualistic (+)(-)

Population 1

Population 2

Drivers of Variation in Outcome?-An example of species identity variation

Moth attack

(% of fruits of Opuntia imbricata)

Miller (2007)

What are the consequences of agriculture

• Populations– ????????

• Communities– Communities often simplified, made more similar

across sites (decreased beta-diversity)– Interaction networks are simplified in agricultural

landscapes• Evolution

– Antagonists (predators, competitors) often XXXX– Mutualists often XXXX

Ekroos et al. 2010, Tylianakis et al. 2007

Predation

Mutualism

Presence of both mutualists and antagonists may increase trait diversity

59Siepielski & Benkman 2010

Principal Component Axis (cone & seed traits)

Num

ber o

f Tre

es