Fire regime shift linked to increased forest density in a piñon-juniper

savanna ecotone landscape

Ellis Margolis

USGS

New Mexico Landscapes Field Station

Santa Fe, New Mexico

International Journal of Wildland Fire (2014)

• Romme et al. (2009) “…piñon-juniper [PJ] is a diverse vegetation type for which a simple model of historical structure and dynamics is inadequate.”

• 3 different PJ types

1. Persistent woodland – trees dominant, lots of bare soil or rock

2. Wooded shrubland - shrubs dominant w/ variable tree cover

3. PJ Savanna – more open w/ grass in understory

Background

“Spreading, low-intensity surface fires had a very limited role in molding stand structure and dynamics of many or most piñon-juniper woodlands” Romme et al. (2009)

• Fire scars “rare or absent” (Baker & Shinneman 2004)

• Rare scarred junipers may be undatable due to lobate growth, missing and false rings

State of knowledge: PJ fire regimes

Pinus edulis (PIED) Juniperus scopulorum (JUSC)

• Increased PJ extent and density since mid-1800s

• Multiple causes:

– Favorable climate

– Post-disturbance “recovery”

– Livestock grazing

– Natural range expansion

– Fire exclusion (not well tested)

State of knowledge: PJ forest structure

Goal: reconstruct the fire regime and forest structure in a PJ-dominated savanna landscape

1. Did frequent, low-severity fire historically burn in PJ savannas?

2. Is there evidence in PJ savannas that fire exclusion following Euro-American settlement is a mechanism driving increases in tree density?

Study area - Rowe Mesa North-central New Mexico

Area: 31,000 ha Elevation: 1963 m - 2455 m (Mean) Slope: 4° Aspect: generally south facing Temp. avg. Tmax 18.4°C, avg. Tmin 0.8°F Precip. - annual avg. 442mm Summer (monsoon) dominated

Vegetation (ground corrected Landfire EVT):

• 47% - Piñon-juniper (savanna) -

Pinus edulis, Juniperus scopulorum, J. monosperma

• 29% - Grassland (short grass) - Bouteloua gracilis

• 24% - Ponderosa pine - Pinus ponderosa, Quercus undulata

Sample design 7 km grid – sampled the nearest forest patch (> 1 ha) with pre-settlement trees 7 sites (2-5 ha)

Tree-ring sampling Fire history - At each site we sampled all live & dead scarred trees & logs Historical forest structure 2 Veg plots (0.2 ha) at each site (a) cored 15 largest dbh trees

(JUSC, PIED, PIPO) and (b) measured diameter, species, & condition of trees > 5cm dbh (n = 3,023 trees)

Cored 2 largest dbh trees within 20 m of fire scars

• Tree establishment dates: estimated pith and corrected for age-to-sample height

• Fire frequency: landscape, site, multi-site fires

• Regime shift analysis (running t-test):

(a) fire (b) age structure (c) moisture-sensitive tree-ring series

• Used local size-age relationships to determine if tree establishment was pre- or post-1879

Tree-ring and data analysis

PIED & JUSC with multiple fire scars

NO FIRE

Dated 112 fire-scarred trees 17% PIED, 8% JUSC, 74% PIPO Fires – 630 fire scars, 87 fire years (1547-1899) 1879 railroad arrival No fire recorded at multiple sites after 1879 Variability in: •Fire freq. between sites •Synchrony within & between sites

PJ–dominated site with multi-scarred PIED and JUSC

No obvious widespread severe fires - trees survived multiple fires - no synchronous death dates

Fire interval statistics for all trees at all 7 sites (1601-2011)

a WMPI is not shown where the Weibull model did not fit the fire interval data (Kolmogorov-Smirnov test, P > 0.05). b Statistics can’t be calculated.

Site/

Analysis period

Filter Intervals

(#)

Mean fire

interval (yrs)

Median fire

interval (yrs)

Weibull median

probability interval (yrs)

Min

(yrs)

Max

(yrs)

North All scars 8 22.8 17.5 19 3 57

1711-2011 10% scarred 2 10 10 * 7 13

25% scarred 2 10 10 * 7 13

Northwest All scars 7 29.0 19.0 23.2 4 83

1685-2010 10% scarred 5 31.8 19.0 24.6 4 83

25% scarred 5 31.8 19.0 24.6 4 83

Northeast All scars 17 16.2 14.0 13.4 2 52

1604-2011 10% scarred 11 25.1 24.0 22.4 4 69

25% scarred 10 27.6 25.0 25.8 10 69

Central All scars 30 11.7 9.0 9.3 2 54

1547-2010 10% scarred 16 16.4 10.0 13.8 4 53

25% scarred 11 20.1 14.0 17.2 2 53

East All scars 31 9.8 8.0 8.8 2 26

1580-1926 10% scarred 20 15.0 12.5 13.1 2 36

25% scarred 20 15.0 12.5 13.1 2 36

South-central All scars 22 8.0 6.5 7.1 2 25

1702-1947 10% scarred 17 9.9 8.0 8.4 2 27

25% scarred 17 9.9 8.0 8.4 2 27

South All scars 26 10.6 9 8.6 1 36

1612-2011 10% scarred 9 21.6 13 12.6 3 116

25% scarred 6 32.3 18 21.8 7 116

Site fire interval statistics

Pre-settlement 1879 Contemporary 2011

Species Mean SE Mean SE

JUMO 0.0 --- 15.4 6.6

JUSC 21.8 6.7 131.8 27.7

PIED 20.4 7.5 397.1 128.3

PIPO 92.9 16.6 336.8 108.3

All species 135.1 15.3 881.1 127.0

Contemporary tree density (trees ha-1) is

600% of the reconstructed density

Sample size, n = 14, 0.2 ha plots.

SE = standard error of the mean.

r = -0.74

Fire exclusion linked to increased tree establishment

- Variables plotted in 20-yr bins on last year of bin (n=206 trees, 442 fire scars) - Crosshatched area indicates when age structure wasn’t adequately sampled

Conclusions • Evidence of frequent, low severity fire in a PJ savanna

• Evidence supporting fire exclusion as a mechanism for increased PJ tree density in a PJ savanna

Fire scar dates -

ROW 801

Year Feature

1651 Pith

1676 Fire scar

1725 Fire scar

1752 Fire scar

1772 Fire scar

1792 Fire scar

1805 Fire scar

1842 Fire scar

1857 Fire scar

1878 Fire scar

1936 Bark

Grazing & fire exclusion drives dramatic ecosystem change

This is dirt. Dirt doesn’t burn.

Thanks!

Funding from U.S. Forest Service CFRP; Support from Tom Swetnam and the UofA Tree-Ring Lab; Field and lab help from Craig Allen, Melissa Savage, Collin Haffey, Atticus Zavelle, John Danloe, James Johnston, Miguel Villarreal, & Reese Brown

This is dirt. Dirt doesn’t burn.

Goal: reconstruct the fire regime and forest structure in a PJ-dominated landscape

1. Did frequent, low-severity fire historically burn in PJ savannas?

2. Is there evidence in PJ savannas that fire exclusion following Euro-American settlement is a mechanism driving increases in tree density?

Guidance for fire management (at multiple scales)

• Fire on Rowe Mesa every 3 yrs on avg. (max 12 yrs)

• Fire spreading within 2-5ha forest patches 10 – 30 yrs (avg.)

• Fire recorded at 2 or more sites (separated by 7 km), every 7 yrs on avg. (max 25 yrs)

• Fires recorded at >50% of sites every 20 yrs on avg. , (min 4 yrs, max 72 yrs)

PJ–dominated site with JUSC fire scars

Conditional burn probability • Fire spread model FlamMAP • 1000 random fire ignitions • Conditional burn probability (P) • P = # times burned/# of fires • 97th percentile weather

• Useful to prioritize areas to treat • Low burn probability = Low

priority for treatment

• Does the tree-ring fire-scar record agree with this map?

Some evidence of frequent fire at PIPO/PJ ecotone

• Allen (1989) – Jemez Mtns, NM – fire-scarred PIPO adjacent to PJ, but no PJ fire scars or PJ age structure

• Muldavin et al (2003) – south-central, NM – fire-scarred piñon at PJ/PIPO ecotone, but insufficient tree age data to assess severity in PJ woodland

• Huffman et al. (2008) – Colorado plateau, AZ & NM – fires recorded in PIPO generally did not spread into adjacent PJ based on PJ age structure

• Poulous et al. (2009) – W Texas – PJ woodland with fire-scarred P. cembroides – fire extent and relationship with PJ establishment not clear

• Colorado Plateau: (e.g., Mesa Verde, CO; Floyd et al. 2000, 2004) indicate a low frequency, high severity fire regime, but there is some evidence of frequent, low severity fire (Brown et al. 2001; Muldavin et al. unpublished).

• A recent review by Romme et al. (2009) states: “Spreading, low-intensity surface fires had a very limited role in molding stand structure and dynamics of many or most piñon-juniper woodlands in the historical landscape.”

What we do know about PJ fire regimes and forest structure

What we know about PJ – Fire

– Romme et al. (2009) “Spreading, low-intensity surface fires had a very limited role in molding stand structure and dynamics of many or most piñon-juniper woodlands in the historical landscape.”

– “Applicability and Confidence. This statement is well supported by empirical research in persistent woodlands and wooded shrublands, …”

BUT – “Pre-1900 disturbance regimes in piñon-juniper

savannas are not well understood; consequently, we have only low or moderate confidence that this applies to savannas.”

• PJ fire scars – “rare or absent” (Baker & Shinneman 2004)

• Scarred juniper may be undatable or challenging to date due to lobate growth, missing and false rings

Challenges of tree-ring reconstructions in PJ

Open Pipo

Young PJ savanna

Dense PJ

Dense Pipo

Study area - Rowe Mesa North-central New Mexico

Area: 31,000 ha Elevation: 1963 m - 2455 m (Mean) Slope: 4° Aspect: generally south facing Temp. avg. Tmax 18.4°C, avg. Tmin 0.8°F Precip. - annual avg. 442mm Summer (monsoon) dominated

Old open PJ savanna w/grass understory

• PIED or JUSC trees with multiple fire scars were crossdated at four of the seven sites (Figs. 2 and 3).

• Seventy one percent (20 of 28) of fire-scarred PIED or JUSC had multiple fire scars. The oldest fire-scarred trees at three of the seven sites were PIED or JUSC.

• Twenty of the 30 fire-scarred trees at site S were PIED or JUSC (Fig. 2). • Thirty six of the 87 unique fire years reconstructed in the study area were recorded

as fire scars by PIED or JUSC trees (range of scar dates, 1601 – 1888). • Of the fire years recorded by PIED or JUSC, 85% were recorded by at least one

other tree at the same site and by additional trees at other sites separated by an average of 7.4 km (i.e. within- and between-site replication). For example, the 1812 fire was recorded by 22 trees at two sites, 5 of which were PIED at site S (Fig. 2).

• PJ-dominated sites generally had a greater difference between MFIALL and MFI25%

(e.g. MFIALL = 10.6 years and MFI25% = 32.3 years, site S), compared to PIPO-dominated sites (e.g. MFIALL = 8.0 years and MFI25% = 9.9 years, site SC, Table 3).

Background

• Piñon and juniper (PJ) ecosystems: ~ 30 million ha

• Ponderosa pine: ~ 15 million ha, but more uncertainty about ecological processes in PJ

• Fire scars “rare or absent” (Baker & Shinneman 2004)

• Rare scarred junipers may be undatable due to lobate growth, missing and false rings

Pinus edulis (PIED) Juniperus scopulorum (JUSC)

P. ponderosa (PIPO)

PIPO fire scars 1. In ephemeral drainage bottoms where debris would accumulate on “upstream” side of tree 2. On the few slopes 3. Not on trees out in the flat grasslands *Emphasized the importance of the scarring process: fuel for initial scar is key

Conclusions • Evidence of frequent, low severity fire in PJ savanna

(but less frequent than pure PIPO ecosystem)

• Evidence supporting fire exclusion as a mechanism for increased PJ tree density in PJ savanna

Recent fires have been high severity

Romme et al. (2009):

Fire - “Spreading, low-intensity surface fires had a very limited role in molding stand structure and dynamics of many or most piñon-juniper woodlands in the historical landscape.”

Forest structure - Increased PJ extent and density: – Favorable climate

– Post-disturbance “recovery”

– Livestock grazing

– Natural range expansion

– Fire exclusion (not well tested)

What do we know about PJ fire regimes and forest structure?

PJ tree-ring fire history methods background

• Fire scars “rare or absent” (Baker & Shinneman 2004)

• Rare scarred junipers may be undatable due to lobate growth, missing and false rings

Pinus edulis (PIED) Juniperus scopulorum (JUSC)

It is logical that PJ savannas could have burned frequently at low severity, because of the grass component

“History never repeats itself,

but it does tend to rhyme.” Mark Twain