DISTRIBUTION AND RELATIVE ABUNDANCE OF ARIZONA FOREST BIRDS IN RELATION TO BURN SEVERITY

Chris Kirkpatrick1, Courtney J. Conway2, and Patricia B. Jones1

1 University of Arizona; 2 USGS Arizona Cooperative Fish and Wildlife Research Unit

Introduction • Many montane forest birds in the southwestern U.S. (especially those inhabiting pine and pine-oak woodlands) evolved in areas that historically experienced recurrent (≥1 fire every decade), low- to moderate-severity surface fires. • During the last century, grazing of surface fuels by introduced livestock and widespread fire suppression have greatly reduced the frequency of surface fires in the region.

• Marshall (1957, 1963) speculated that fire suppression and subsequent fuels accumulation had reduced the distribution and abundance of many “open-woodland” birds in the “Sky Island” mountain ranges of southeastern Arizona (Fig. 1).

• Increased fuel loads have resulted in a recent increase in the frequency of large, often severe wildfires (Fig. 2) and prompted the use of low-severity prescribed fires by land management agencies to reduce fuel loads.

• Despite the increasing frequency of wild and prescribed fires, few studies have examined the effect of fires (especially fires of different severities) on the unique avian community inhabiting montane forests of southeastern Arizona.

Conclusions• Most (73%) of the species analyzed were positively associated with recently burned areas and displayed stronger associations (i.e., more extreme odds ratios) with survey points that had evidence of severe as opposed to less-severe fire.

• Positive associations were particularly strong for western wood-pewee and house wren and negative associations were particularly strong for warbling vireo and red-breasted nuthatch.

• None of the 16 species identified by Marshall (1963) as “open-woodland” birds showed positive associations with burned areas.

• Our results suggest that recent fires have had a positive effect on the distribution/relative abundance of several montane forest bird species in the region. However, these species are not the “open-woodland” birds we would have expected to have benefited from fire based on research by Marshall (1967, 1963).

• Because some bird species appear to respond differently to different levels of burn severity, future studies should examine how birds respond to fire across a range of burn severities instead of examining simple responses of birds to “fire” per se.

Red-faced warblerRed-faced warbler Photo credit: Brian Small Photo credit: Brian Small

++ = Strong, positive association (P ≤ 0.05 and odds ratio ≥3.0) + = Positive association (P ≤ 0.10 and odds ratio 1.5-3.0) -- = Strong, negative association (P ≤ 0.05 and odds ratio ≤0.3) - = Negative association (P ≤ 0.10 and odds ratio 0.3-0.6) 0 = No observed association.

Presence/Absence Relative Abundance

Species Severe Fire

Less-severe Fire

Severe Fire

Less-severe Fire

Hairy woodpecker + + n/a n/a

Greater Pewee + + n/a n/a

Western Wood-Pewee ++ 0 ++ 0

White-breasted Nuthatch

+ + n/a n/a

Red-breasted Nuthatch -- -- n/a n/a

House Wren ++ + ++ +

Warbling Vireo -- -- -- 0

Yellow-rumped Warbler 0 - n/a n/a

Grace’s Warbler + + n/a n/a

Virginia’s Warbler + + n/a n/a

Spotted Towhee 0 - + -

Table 1. Direction and strength of associations between bird presence/absence and relative abundance and two levels of burn severity using data collected during montane forest bird surveys in southeastern Arizona, April to July, 2000.

Results • We observed evidence of fire at 406 of 1,513 survey points: 300 survey points had evidence of “less severe fire” and 106 survey points had evidence of “severe fire”.

• Median time since fire was 6 years (range 2-47 years) for 82% of 406 burned survey points.

• We detected a total of 10,473 individuals of 97 species and found associations between fire and bird presence/absence for 17% of 65 species analyzed and between fire and bird relative abundance for 25% of 16 species analyzed (Table 1).

Fig. 1. “Sky Island” mountain ranges of southeastern AZFig. 1. “Sky Island” mountain ranges of southeastern AZ

Fig. 2. Start of the 34,000-ha Aspen Fire, Catalina Mountains, AZFig. 2. Start of the 34,000-ha Aspen Fire, Catalina Mountains, AZ

• TUCSON

NE

W

ME

XIC

O

MEXICO

AR

IZO

NA

Methods Continued• To control for elevation, we included elevation as a covariate in the regression models and restricted data for each species to the elevational range in which we observed the species.

• We included route as an independent variable in the regression models to control for the lack of independence between survey points along routes.

• We based our conclusions on the combined evidence of P-values from hypothesis testing and magnitudes of differences (odds ratios) generated from parameter estimation.

Objectives • To identify effects of recent fires (both severe and less-severe) on the distribution and relative abundance of forest birds in mountains of southeastern Arizona.

00 11 22 33 44

0) No evidence of fire1) Evidence of low-severity surface fire (e.g., fire-charring roughly 0-0.3 m above

ground on a few trees)2) Evidence of moderate-severity surface fire (e.g., fire-charring roughly 0.3-1.5 m

above ground on most trees; a few small oaks or pines killed in understory)3) Evidence of high-severity surface fire (e.g., fire charring often >1.5 m above

ground on trees; almost all oaks or pines killed in understory)4) Evidence of high-severity crown fire (e.g., all above-ground vegetation killed)

Fig. 3. Burn severity index (BSI):

Methods• From April to July 2000, we conducted 3-min point count surveys at 1,513 points along 84 routes (located primarily in pine and pine-oak woodlands) in the Chiricahua, Huachuca, Santa Catalina, Rincon, Santa Rita, Pinaleno, Galiuro, and Santa Teresa mountains (Fig. 1).

• We quantified burn severity within 100 m of each survey point using a burn severity index (BSI; Fig. 3). Based on available fire history data, we estimated median time since fire for the majority of the burned survey points.

• We collapsed BSI values into 3 general categories to increase power of statistical tests:

1) no evidence of fire (BSI 0)2) evidence of “less-severe fire” (BSI 1-2)3) evidence of “severe fire” (BSI 3-4)

• To test for associations between birds and recent fire, we used: 1) logistic regression to examine whether bird presence/absence was associated with burn severity for 65 bird species, and 2) ordinal logistic regression to examine whether bird relative abundance was associated with burn severity for a subset of 16 bird species.

Virginia’s warblerVirginia’s warbler Photo credit: Brian Small Photo credit: Brian Small

Literature Cited Marshall, J. T. Jr. 1957. Birds of pine-oak woodland in southern Arizona and adjacent

Mexico. Pacific Coast Avifauna Number 32.Marshall, J. T. Jr. 1963. Fire and birds in the mountains of southern Arizona. Proceedings of the Tall Timbers Fire Ecol. Conf. 2:135-141.