Climate Change and Wildfire Effects in Aridland Riparian Ecosystems: An Examination of Current and Future Conditions

Abstract

Aridland riparian ecosystems are limited, the climate is changing, and further hydrological change is likely in the American Southwest. To protect riparian ecosystems and organisms, we need to understand how they are affected by disturbance processes and stressors such as fire, drought, and non-native plant invasions. Riparian vegetation is critically important as foraging, resting, migrating, and breeding habitat to birds and other animal species in the southwestern United States. Fremont cottonwood (Populus fremontii), Arizona sycamore (Platanus wrightii), and other woody species provide birds with nesting sites and foraging opportunities, some of which are absent or rare in adjacent plant communities. The structurally diverse, species-rich vegetation along many southwestern streams supports high densities of territories and nest sites for a variety of birds including several species of high conservation priority. Survival and reproduction of woody riparian plants is largely determined by periodic floods and droughts. As in other regions, rivers and streams of the American Southwest have been heavily altered by human activity , resulting in significant changes to disturbance regimes. Hydrological models, incorporating greenhouse gas emission scenarios, project that these changes will be exacerbated by climate change. In this report, we review the ecohydrology of southwestern streams and share results from our study sites along the Middle Rio Grande to describe effects of hydrological changes, wildfire, and invasions on plant communities and riparian-nesting birds. We also examine climate change projections and output from population models to gauge the future of aridland riparian ecosystems in an increasingly arid Southwest.

Full text

United States Department of Agriculture
Forest Rocky Mountain General Technical Report June 2017
Service Research Station RMRS-GTR-364
Climate Change and Wildfire Effects in Aridland
Riparian Ecosystems: An Examination of
Current and Future Conditions
D. Max Smith and Deborah M. Finch

Smith, D. Max; Finch, Deborah M. 2017. Climate change and wildfire effects in aridland riparian
ecosystems: An examination of current and future conditions. Gen. Tech. Rep. RMRS-
GTR-364. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain
Research Station. 65 p.
Abstract
Aridland riparian ecosystems are limited, the climate is changing, and further hydrological
change is likely in the American Southwest. To protect riparian ecosystems and organisms, we
need to understand how they are affected by disturbance processes and stressors such as fire,
drought, and non-native plant invasions. Riparian vegetation is critically important as foraging,
resting, migrating, and breeding habitat to birds and other animal species in the southwestern
United States. Fremont cottonwood (Populus fremontii), Arizona sycamore (Platanus wrightii),
and other woody species provide birds with nesting sites and foraging opportunities, some of
which are absent or rare in adjacent plant communities. The structurally diverse, species-rich
vegetation along many southwestern streams supports high densities of territories and nest sites
for a variety of birds including several species of high conservation priority. Survival and reproduc-
tion of woody riparian plants is largely determined by periodic floods and droughts. As in other
regions, rivers and streams of the American Southwest have been heavily altered by human ac-
tivity, resulting in significant changes to disturbance regimes. Hydrological models, incorporating
greenhouse gas emission scenarios, project that these changes will be exacerbated by climate
change. In this report, we review the ecohydrology of southwestern streams and share results
from our study sites along the Middle Rio Grande to describe effects of hydrological changes,
wildfire, and invasions on plant communities and riparian-nesting birds. We also examine climate
change projections and output from population models to gauge the future of aridland riparian
ecosystems in an increasingly arid Southwest.
Keywords: breeding birds, climate change, Middle Rio Grande, riparian, woody vegetation,
wildfire
Authors
D. Max Smith is a Research Associate contracted with the U.S. Forest Service, Rocky Mountain
Research Station in Albuquerque, New Mexico.
Dr. Deborah M. Finch is a Supervisory Biologist and Program Manager with the U.S. Forest
Service, Rocky Mountain Research Station in Albuquerque, New Mexico.
Acknowledgments
We thank the Desert and Southern Rockies Landscape Conservation Cooperatives for funding
this research. Additional funding was provided by the Rocky Mountain Research Station’s Climate
Change Research Program. Dave Hawksworth provided help in the field, and the Middle Rio
Grande Conservancy District granted us access to our study sites. We received wildfire data from
Xavier Anderson and Rob Barr. Logistical support was provided by Kal Louks, Yasmeen Najmi,
and Yancey Ranspot. Dan Auerbach, David Lytle, and David Merritt helped us construct our popu-
lation models. Jonathan AuBuchon, David Merritt, and Robert Padilla provided helpful comments
on an earlier draft.
Cover photos: Clockwise from the top, Tonto Creek near Roosevelt, AZ; a wildfire in
the Middle Rio Grande riparian forest; Mourning Dove (Zenaida macroura) nestlings;
the Rio Grande stream bed near Albuquerque, NM. All photos by D.M. Smith.

All Rocky Mountain Research Station publications are published by U.S. Forest Service
employees and are in the public domain and available at no cost. Even though U.S. Forest
Service publications are not copyrighted, they are formatted according to U.S. Department
of Agriculture standards and research findings and formatting cannot be altered in reprints.
Altering content or formatting, including the cover and title page, is strictly prohibited.
Contents
Introduction.................................................1
Chapter 1. Dynamics of Aridland Riparian Ecosystems.............3
Purpose and Methodology .................................3
Characteristics of Southwestern Streams ......................5
Ecohydrology of Woody Vegetation .........................11
Current State of Riparian Ecosystems .......................14
Use of Riparian Vegetation by Breeding Birds .................14
Chapter 2. Response of Woody Riparian Plants to Wildfire.........18
Introduction: Wildfire and Woody Vegetation ..................18
PostFire Dynamics Along the Middle Rio Grande ...............19
Summary: Impacts of Wildfire on Aridland Riparian Ecosystems ...30
Chapter 3. Use of Woody Plants by Riparian-Nesting Birds in
Unburned Plots and Post-Wildfire Sites ..................31
Introduction: Effects of Wildfire on Breeding Birds ..............31
Nest-Plant Use in Unburned and Postfire Riparian Sites .........31
Summary: Impacts of Wildfire on Riparian-Nesting Birds .........40
Chapter 4. Climate change, Wildfire, and the Future of
Aridland Riparian Ecosystems ..........................41
Introduction: Climate Change and Hydrology of the American
Southwest.........................................41
Hydrological Projections ..................................43
Modeling Changes in Cottonwood Populations.................46
Management Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
References ................................................51
Appendix A—Hydrological Projections .........................59
Appendix B—Application of the Cottonwood Population Model.....61

USDA Forest Service RMRS-GTR-364. 2017. 1
Introduction
A frequently discussed function of aridland riparian ecosystems is the contribution of
woody riparian plants to breeding bird habitat. Fremont cottonwood (Populus fremon-
tii), Arizona sycamore (Platanus wrightii), and other species provide birds with nesting
sites and foraging opportunities, some of which are absent or rare in adjacent plant com-
munities (fig. 1); (Bock and Bock 1984; Carothers et al. 1974; Hunter et al. 1987). The
structurally diverse, species-rich vegetation along many southwestern streams supports
high densities of territories and nest sites for a variety of birds including several species
of high conservation priority (fig. 2; Smith and Finch 2014; Smith et al. 2009a; Stoleson
and Finch 2003).
Survival and reproduction of woody riparian plants is largely determined by periodic
disturbances such as flood and drought. As in other regions, streams of the American
Southwest have been heavily affected by human activity, resulting in significant chang-
es to disturbance regimes (Meritt and Poff 2010; Shafroth et al. 2002). Hydrological
models, incorporating greenhouse gas emission scenarios, predict that these changes
will be exacerbated by climate change (Seager et al. 2013). Given the limited extent of
aridland riparian ecosystems and likelihood of further hydrological change, an under-
standing of current and future effects of disturbance processes on populations of riparian
plants is needed to protect breeding bird communities in the American Southwest. In
Figure 1—The mixture of native and nonnative woody vegetation along the San Juan River has greater struc-
tural diversity than the adjacent plant community.

2 USDA Forest Service RMRS-GTR-364. 2017.
this report, we review the ecohydrology of southwestern streams and share results from
our studies at the Middle Rio Grande to describe effects of hydrological changes and
wildfire on woody plants and riparian-nesting birds. We also examine climate change
projections and output from population models to gauge the future of aridland riparian
ecosystems in an increasingly arid Southwest.
Figure 2—The riparian vegetation along the upper Gila River in southwestern New Mexico has high species
richness of woody plants and extremely high densities of nesting birds including the Federally endangered
Southwestern Willow Flycatcher (Empidonax traillii extimus) and the Federally threatened Yellow-Billed
Cuckoo (Coccyzus americanus).

USDA Forest Service RMRS-GTR-364. 2017. 3
Chapter 1. Dynamics of Aridland Riparian Ecosystems
Purpose and Methodology
To maintain the function of aridland riparian ecosystems, an understanding of the
dynamics of streamside plant and animal communities is needed. In this chapter, we ex-
amine current variation in hydrological characteristics among sites distributed across the
American Southwest, review the ecohydrology of riparian plant species, and describe
use of riparian vegetation by breeding birds.
For the purposes of this report, we focus on semiarid-to-arid portions of Utah,
Colorado, New Mexico, and Arizona that are within the Colorado River and Rio Grande
basins. We reviewed peer-reviewed studies of riparian vegetation and riparian-nesting
birds conducted along streams in this region. We also summarized data from gauge
sites along nine streams in the Colorado River basin and two in the Rio Grande basin
to illustrate hydrological differences among these streams (table 1). We classified the
stream gauge sites into two geographical groups. Six of the sites, referred to hereafter
as “Rocky Mountain sites,” were along streams headwatered in the Rocky Mountains
of Wyoming and Colorado. Five sites, referred to as “Central Highland sites,” were
headwatered in the Central Highland ranges of Arizona and New Mexico (fig. 3). The
Rocky Mountain stream sites are on the Colorado River, the Green River, the Gunnison
River, The San Juan River, the Rio Chama, and the Rio Grande. The Central Highland
stream sites are on the Gila River, the Salt River, the San Francisco River, Tonto Creek,
and the Verde River. Gauge sites were between 600 and 2,000 m in elevation. Based
on long-term temperature and precipitation records, the Rocky Mountain stream gauge
sites were cooler and drier than the Central Highland sites (table 1).
We examined patterns of hydrologic variables that characterize streams and affect
survival and reproduction of riparian vegetation. These variables, defined here, were:
• annual discharge (in million cubic meters) = the total volume of water measured
at a stream gauge site each year;
• mean daily discharge (in cubic meters per second) = the mean discharge volume
for each day of the year;
• peak discharge magnitude (in cubic meters per second) = the maximum mean
daily discharge value measured each year; and
• peak discharge date = the day of each year that the peak discharge occurred.
We obtained discharge data for the period of 1/1/1960 to 12/31/2011 recorded at the
11 stream gauge sites by the U.S. Geological Survey and made available online by the
National Water Information System Database (http://waterdata.usgs.gov/nwis). We esti-
mated the annual discharge volume and determined the magnitude and date of the peak
discharge at each stream site for each year of the historical period.

4 USDA Forest Service RMRS-GTR-364. 2017.
Table 1—Geographic and climatic characteristics of the 11 stream gauge sites examined in this chapter. Precipitation and temperature data were obtained from weather
stations near stream gauge sites, made available the Western Regional Climate Center (http://www.wrcc.dri.edu/Climsum.html).
Mean annual Mean min Mean max
precipitation temperature temperature
Gauge site Gauge number Basin Headwaters Elevation (m) (cm) (C) (C)
Colorado River at Cameo 09095500 Colorado River Rocky Mountains 1,467 25.1 4.9 19.6
Gunnison River at Grand Junction 09152500 Colorado River Rocky Mountains 1,411 22.4 4.7 18.9
Green River at Greendale 09234500 Colorado River Rocky Mountains 1,705 30.2 –1.4 15.3
San Juan River at Bluff 09379500 Colorado River Rocky Mountains 1,234 19.8 3.9 21.3
Rio Grande at Otowi 08313000 Rio Grande Rocky Mountains 1,673 25.1 1.4 20.3
Rio Chama at Abiquiu 08286500 Rio Grande Rocky Mountains 1,914 24.9 2.9 18.2
Salt River at Roosevelt 09498500 Colorado River Central Highlands 664 33.8 12.7 27.2
Verde River above Horseshoe Dam 09508500 Colorado River Central Highlands 618 36.1 13.5 29.3
San Francisco River at Clifton 09444500 Colorado River Central Highlands 1,047 34.0 11.0 27.2
Gila River at Gila 09430500 Colorado River Central Highlands 1,419 36.6 3.1 23.6
Tonto Creek near Roosevelt 09499000 Colorado River Central Highlands 769 44.5 9.4 27.5

USDA Forest Service RMRS-GTR-364. 2017. 5
Characteristics of Southwestern Streams
Hydrologic Patterns
Our analysis of stream gauge data shows that there are consistent differences in
characteristics between Rocky Mountain sites and the Central Highland sites. Of
the sites examined, mean annual discharge and mean peak discharge were generally
greater for Rocky Mountain sites, but maximum peak discharges were greater at
most of the Central Highland sites (figs. 4 and 5, table 2). Timing of peak discharge
was more consistent at Rocky Mountain sites, with peaks that generally occurred in
May or June (table 2). On average, peak discharge at Central Highland sites occurred
in February or March (table2), but these peaks also occurred during other times
of the year (figs. 5 and 6). These differences in discharge patterns result from a
number of factors. With their high elevation (>3,000 m) and latitudinal position, the
Rocky Mountain ranges accumulate heavy snowpacks during the winter months,
which contribute to the peak discharges that predictably occur during the spring and
summer in the Southwest. The Central Highland ranges are lower in elevation and
latitude than the Rocky Mountain headwaters, so a combination of snowmelt and
rain contributes to peak discharges (Neary et al. 2012; Webb et al. 2007). Central
Highland ranges intercept extremely large amounts of rainfall from winter Pacific
frontal storms, fall tropical storms, and summer monsoons (Stromberg et al. 2007).
Figure 3—The gauge sites examined were along streams headwatered in the Rocky Mountains and Central
Highland ranges.

6 USDA Forest Service RMRS-GTR-364. 2017.


Figure 4—Annual discharge measured at stream gauge sites at the Rio Grande at Otowi, NM, (A) and Salt
River at Roosevelt, AZ (B). The range of annual discharge is greater for the Salt River and similar Central
Highland streams than for the Rio Grande and other Rocky Mountain streams. The red line indicates the long-
term average.

USDA Forest Service RMRS-GTR-364. 2017. 7


Figure 5—Magnitude and timing of peak discharges measured at stream gauge sites at the Gunnison River
at Grand Junction, CO, (A) and the Verde River above Horseshoe Dam, AZ (B). Maximum peak discharges
are greater at the Verde and other Central Highland streams than at the Gunnison and other Rocky Mountain
streams. Timing of peak discharge varies among the streams as well. The red line indicates the long-term
average.

8 USDA Forest Service RMRS-GTR-364. 2017.
Table 2—Streamflow characteristics recorded at stream gauge sites used in this analysis. Data
were obtained from the National Water Information System Database (http://waterdata.usgs.
gov/nwis).
Mean annual
discharge Maximum peak
1960–2011 discharge Peak
(million cubic 1960–2011 discharge
Gauge site meters) (cms) month
Colorado River at Cameo 3394.0 1076.0 May
Gunnison River at Grand Junction 2194.1 656.9 May
Green River at Greendale 1741.0 365.2 June
San Juan River at Bluff 1799.2 982.5 June
Rio Grande at Otowi 1235.1 339.8 May
Rio Chama at Abiquiu 414.3 183.4 May
Salt River at Roosevelt 797.1 2525.8 March
Verde River above Horseshoe Dam 538.2 3114.8 March
San Francisco River at Clifton 211.9 1478.1 March
Gila River at Gila 157.9 662.6 March
Tonto Creek near Roosevelt 151.1 1039.2 March

Figure 6—Mean daily discharge for each day of the year from 1960 to 2011 at the Colorado River at Cameo,
CO, (A) and Tonto Creek at Roosevelt, AZ (B). Peak discharges of the Colorado and other Rocky Mountain
streams typically occur in spring. Peak discharges of Tonto Creek and other Central Highland streams can
occur from late summer through early spring.

USDA Forest Service RMRS-GTR-364. 2017. 9
Given the localized nature of these storms and annual variation in their occur-
rence, there is substantial variation in peak discharge at Central Highland streams
(Stromberg et al. 2007).
The differences in peak discharge are reflected in the phenology of woody vegeta-
tion. At Rocky Mountain streams, cottonwoods and willows release seeds during the
spring and summer when snowmelt-driven floods typically subside (Cooper et al. 1999;
Molles et al. 1998). At Central Highland streams, cottonwood and willow seed dispersal
occurs during the late winter and early spring, typically coinciding with the drawdown
of high flows that result from the combination of snowmelt and rain (Beauchamp and
Stromberg 2007).
Anthropogenic Changes
In recent centuries, hydrology, geomorphology, and ecology of streams have
been affected by factors that include changes in climate and anthropogenic land use
(Scurlock 1998; Stromberg et al. 2010a). Among the most significant of the latter is reg-
ulation of streams to control flood risk and provide water for agricultural, industrial, and
municipal purposes. This regulation includes the construction of dams and reservoirs,
channelization of streams, withdrawal of groundwater, and diversion of surface flows
(fig. 7; Phillips et al. 2011; Summit 2013). Peak discharge magnitude, timing, and dura-
tion are now altered from historical conditions at many streams and some sections that
once had perennial flows now run dry apart from periods of heavy runoff (White and
Stromberg 2009). Other sections are inundated by reservoir pools behind dams while


10 USDA Forest Service RMRS-GTR-364. 2017.
below many dams, magnitude and timing of peak flows have been altered. In addition,
sediment accumulates upstream from dams, while sediment-poor water causes degrada-
tion below dams, disconnecting floodplains and increasing the depth to groundwater
(Novack 2006).
The level of regulation differs among streams’ gauge sites we examined. Dams were
constructed upstream from all of the Rocky Mountain gauge sites, with the largest dams
and reservoirs (those exceeding 1 million acre feet capacity) upstream from the gauges
on the Green and San Juan rivers. Large dams and reservoirs are located downstream
from these sites as well, with major dams on the Colorado River below its junctions
with the Green, San Juan, and Gunnison Rivers (Summit 2013). There are also large
dams and reservoirs downstream from the gauge sites on the Rio Chama and Rio
Grande (Phillips et al. 2011). In addition to dams, trans-basin diversions lie upstream
from several Rocky Mountain stream sites. These include the Colorado-Big Thompson
Project, which transfers flows from the upper Colorado River over the Continental
Divide to eastern Colorado, augmenting flow in the Big Thompson River (Dewine and
Cooper 2007). The San Juan-Chama Project transfers flows from the San Juan River
to the Rio Chama for municipal and agricultural use in central New Mexico (Flanigan
and Haas 2008). There are only minor dams and diversions above the Central Highland
stream gauge sites examined, so the streams are largely unregulated at these locations.
Large dams, reservoirs, and diversion projects are located downstream from the gauge
Figure 7—Anthropogenic influences on streams include (A) diversion of surface flows, (B) groundwater
pumping, (C) channelization, and (D) construction of large dams and reservoirs.

USDA Forest Service RMRS-GTR-364. 2017. 11
sites examined on the Verde River, Salt River, and Gila River (Webb et al. 2007).
Though flows are generally perennial at each of the Central Highland stream sites, dams
and diversions prevent the lower Salt River from flowing to its confluence with the Gila
River apart from periods of heavy precipitation (White and Stromberg 2009). Likewise,
the lower Gila River rarely reaches its confluence with the Colorado River in Arizona
(Summitt 2013).
Ecohydrology of Woody Vegetation
Opportunities for reproduction of woody plants are limited along aridland streams
(Bock and Bock 1989; Cooper et al. 1999). Pioneering species such as cottonwoods,
willows (Salix spp.), and saltcedars (Tamarix spp.) have short-lived seeds that will
not establish unless they settle on damp and exposed substrates (fig. 8). Other riparian
trees, including boxelders (Acer negundo) and Russian olives (Elaeagnus angustifolia),
have large seeds with long viability periods and the ability to establish in shaded sites
with ground cover, though damp conditions are required to induce germination (fig. 9;
Dewine and Cooper 2007; Katz and Shafroth 2003).
Figure 8—The small seeds dispersed by (A) Rio Grande cottonwood, (B) Goodding’s willow, and
(C) saltcedar are transported by wind and water and require exposed, damp surfaces for germination.
Figure 9—The larger seeds of (A) boxelder and (B) Russian olive are dispersed by wind and by animals.
These seeds can germinate on covered surfaces if moisture is present.

12 USDA Forest Service RMRS-GTR-364. 2017.
Along aridland streams, peak discharges can produce floods that create conditions
required for germination of woody plants. Periods of heavy precipitation or snowmelt
result in flows that scour vegetation and litter, re-route stream channels, and deposit
sediment (fig. 10). In the wake of these floods lie sites that are devoid of competing
vegetation and ideal for germination of pioneering species (fig. 11; Auble and Scott
1998). As flood waters recede, they leave behind soil moisture that triggers germination
and promotes seedling survival (Bhattacharjee et al. 2008). Seed dispersal of cotton-
wood, willow, and saltcedar typically coincides with the drawdown of spring floods,
when these sites are left exposed (Braatne et al. 1996; Sher et al. 2002). Following
establishment, phreatophytic species such as Fremont cottonwood and Goodding’s
willow (Salix gooddingii) require a connection between their roots and the groundwater
table to ensure growth and survival (Busch et al. 1992; Snyder and Williams 2000). At
many streams, high peak discharges are needed to recharge aquifers and maintain this
connection (Stromberg 2001). Reduction in magnitude of peak discharge can reduce
recharge rates, causing dieback and drought mortality (fig. 12). Through these influ-
ences on reproduction and survival of woody vegetation, stream characteristics such
as magnitude and timing of peak discharge exert great control over the composition of
riparian ecosystems (Brand et al. 2008; Merritt and Bateman 2012).
Figure 10—High flows as seen here on the Animas River in northwestern New Mexico are important for pro-
viding germination sites and recharging groundwater aquifers, thereby promoting reproduction and survival of
woody riparian vegetation.

USDA Forest Service RMRS-GTR-364. 2017. 13
Figure 11—These cottonwood (yellow arrow) and saltcedar (red arrow) seedlings established following the
drawdown of high springtime flows near the active channel of the Middle Rio Grande.
Figure 12—During periods of low flow, woody riparian plants are vulnerable to drought stress and mortality if
depth to groundwater exceeds root lengths, as shown by dead shrub in lower left of figure.

14 USDA Forest Service RMRS-GTR-364. 2017.
Current State of Riparian Ecosystems
Changes in discharge brought upon by large-scale regulation have had varying ef-
fects on woody riparian vegetation. Dams limit the extent of flooding and germination
within the floodplain and lower the water table in areas away from the active channel,
thereby increasing mortality of established trees (Coble and Kolb 2013; Dewine and
Cooper 2007; Molles et al. 1998). Changes to peak discharge timing also limit repro-
duction of species such as cottonwoods and willows (Cooper et al. 1999; Mortenson
and Weisberg 2010). Along many regulated southwestern streams, conditions have
become more suitable for nonnative species, such as saltcedar and Russian olive, that
have longer seed dispersal periods, longer seed viability, and greater resistance to
drought than certain native species (Birken and Cooper 2006; Busch and Smith 1995;
Mortenson and Weisberg 2010).
Following the alteration of streamflow and floodplain dynamics, wildfire emerged as
a significant disturbance agent at the end of the 20th century. Following the reduction in
frequency and magnitude of floods, litter and debris accumulated in the forest understo-
ry. This accumulation, along with increased density of native and nonnative vegetation,
resulted in fuel loads, fire size, and fire intensity that are greater than existed prior to
stream modification (Bêche et al. 2005; Stuever et al. 1995). Wildfire is thought to be
a historically rare occurrence in riparian zones and little is known about its effects on
riparian organisms (Bock and Block 2005). Information from postfire sites is therefore
needed to project long-term changes in aridland riparian ecosystems.
Use of Riparian Vegetation by Breeding Birds
The trees and shrubs growing along aridland streams are used by scores of bird
species during the nesting season. Both native and nonnative plants serve as nest sites
and foraging substrates for birds occupying multiple niches (Smith and Finch 2014).
Though riparian-nesting birds use a variety of plants for nesting and foraging, results
from studies along southwestern streams highlight differences among woody species in
the resources they provide.
Cottonwoods, Goodding’s willow, boxelder, and Arizona sycamore are riparian trees
that are frequently used as nest sites for birds along aridland streams (fig. 13). Under
typical conditions, cottonwoods and sycamores are the tallest species in alluvial stretch-
es, capable of forming stands supporting greater densities and richness of birds than
other vegetation types (Carothers et al. 1974; Merritt and Bateman 2012; Powell and
Steidl 2001; Strong and Bock 1990). Cottonwoods and sycamores are used by a greater
number of bird species than other trees because they have large branches to support
heavy nests, they have dead and decaying limbs that are excavated by woodpeckers
for food and nest sites, and they feature substantial canopies used by foliage gleaners
(Bock and Bock 1984; Hunter et al. 1987; Smith and Finch 2014; Stoleson et al. 2000).

USDA Forest Service RMRS-GTR-364. 2017. 15
Range-restricted species including Common Black-Hawk (Buteogallus anthracinus),
Violet-crowned Hummingbird (Amazilia violiceps), and Gila Woodpecker (Melanerpes
uropygialis) construct most, if not all of their nests in cottonwoods and sycamores at
their riparian breeding grounds (Hunter et al. 1987; Smith and Finch 2014; Wethington
2002). In addition to nest sites, cottonwoods maintain populations of arthropods, such
as cicadas (Tibicen marginatus) and floodplain crickets (Gryllus alogus), by providing
food, water, and oviposition sites (Sabo et al. 2008; Smith et al. 2006a). These arthro-
pods are critical sources of food for adult and juvenile birds (Rosenberg et al. 1982),
making cottonwoods a key component of the riparian food web (fig. 14).
Fewer birds are known to nest in Goodding’s willow and boxelder as compared to
cottonwoods and sycamores, but these trees are important to certain species and the
breeding bird community as a whole. Mature boxelder and Goodding’s willow trees
are excavated by woodpeckers and they are used as nest plants by the range-restricted,
secondary cavity-nesting Lucy’s Warbler (Oreothlypis luciae; Stoleson et al. 2000).
Figure 13—Native trees of aridland riparian ecosystems include (A) cottonwood, (B) Goodding’s willow,
(C) boxelder, and (D) Arizona sycamore.

16 USDA Forest Service RMRS-GTR-364. 2017.
The trunks of Goodding’s willows often grow horizontally, making them ideal nest
sites for Mourning Doves (Zenaida macroura) and other large understory-nesters
(Smith et al. 2012). Along with cottonwood, the presence of Goodding’s willow
increases the abundance of breeding birds in riparian patches (Brand et al. 2008).
Boxelder is as an important habitat component for birds along the upper Gila River
because it is preferred as a nest plant over other trees by the Federally endangered
Southwestern Willow Flycatcher (Empidonax traillii extimus; Stoleson and Finch
2003), and is also used by the Federally threatened Yellow-Billed Cuckoo (Coccyzus
americanus). In addition, Willow Flycatcher nests in boxelder are less vulnerable
to Brown-headed Cowbird (Molothrus ater) parasitism than nests in other species
(Brodhead et al. 2007). Where it co-occurs with larger trees, boxelder forms a woody
subcanopy layer that is absent in many western riparian forests, effectively increasing
potential nesting opportunities (Knopf and Olson 1984; Smith and Finch 2014).
Figure 14—Floodplain cottonwoods take in groundwater, CO2, and solar energy to create carbohydrates.
They also store water in the roots and transport tissue. Cicadas obtain food and water from the roots as
nymphs and the branches as adults; floodplain crickets obtain food and water from green leaves that are
blown off by the wind. These arthropods are in turn consumed by breeding birds and other insectivores.

USDA Forest Service RMRS-GTR-364. 2017. 17
Russian olive and saltcedar are widespread nonnative species that form a considerable
component of riparian-nesting bird habitat in the Southwest (Friedman et al. 2005).
These species can grow as small trees (fig. 15) but are typically not large enough
to support cavity and canopy-nesting birds (Smith and Finch 2014; Stoleson and
Finch 2001). Russian olive and saltcedar do, however, provide nest sites for shrub
and subcanopy-nesting birds including Willow Flycatcher and Yellow-billed Cuckoo
(Brown 1992; Stoleson and Finch 2001). Densities of some breeding birds are greater
in stands of saltcedar than in native stands (Brand et al. 2009). For some species, in-
cluding Black-chinned Hummingbird (Archilochus alexandri) and Willow Flycatcher,
rates of nesting success in these plants is equal to or exceeds rates measured in native
vegetation (Smith et al. 2009b; Sogge et al. 2008). In some areas, however, Willow
Flycatchers, Bell’s Vireo, and possibly other species, have greater rates of cowbird
parasitism in nonnative vegetation than in native vegetation (Brand et al. 2009; Stoleson
and Finch 2001). In addition to providing nest sites, nonnative woody plants contain
densities of arthropod prey similar to those of native riparian plants (Durst et al. 2008;
Mund-Meyerson 1991). Russian olive and saltcedar can therefore provide foraging op-
portunities for certain foraging guilds, but birds in canopy gleaning, bark gleaning, and
excavating guilds require larger trees such as cottonwoods and sycamores (Bock and
Bock 1984; Ellis 1995).
Results from the studies above have shown that composition of riparian vegetation
influences habitat suitability for certain riparian-nesting birds. Changes in hydrological
conditions and disturbance regimes can therefore influence not only woody plants, but
also their associated animal communities.
Figure 15—(A) Russian olive and (B) saltcedar grow as small trees in aridland floodplain, but they typically do
not grow large enough to support the nests of canopy- and cavity-nesting birds. These birds prefer trees that
support nest heights of 10 m or taller.

18 USDA Forest Service RMRS-GTR-364. 2017.
Chapter 2. Response of Woody Riparian Plants to Wildfire
Introduction: Wildfire and Woody Vegetation
Wildfire influences the composition and structure of woody plant communities,
primarily as an agent of mortality. When the aboveground tissues of trees and shrubs
are killed by wildfire (hereafter referred to as “topkilled”), ecosystems are affected
by the loss of services including canopy shading and vertical transport of water and
nutrients (Smith et al. 2006a; Whelan 1995). These losses in turn alter the availability
of resources to animal communities. Fire also creates snags and woody debris, which
provide habitat for terrestrial and aquatic animals and help to control hydrologic and
geomorphic processes (Brown 2002; Minckley and Rinne 1985; Smith et al. 2012;
Thomas et al. 1979).
Woody plants use a variety of mechanisms to regenerate after fire. Though fire
can result in topkill of broadleaved trees and shrubs, individuals of certain species
recover vegetatively by producing basal sprouts or root suckers from underground
buds (Kramer and Kozlowski 1979). If these sprouts experience conditions ideal for
growth and survival, top-killed stands can replace themselves after fire (Bond and
Midgley 2001; Cocking et al. 2014; Keyser et al. 2005). In addition, certain species
can rapidly reestablish a forest canopy with epicormic sprouts that arise from tree boles
and branches (Clarke et al. 2012). Woody species can also recover from fire through
seedling establishment. For some species in frequently burned vegetation types, fire
stimulates seed dispersal by opening serotinous cones and, through mechanisms such as
scarification and soil warming, encourages germination (Bonnet et al. 2005; Enright et
al. 1996; Keeley 1987; Lotan 1976;).
On stream sections such as the Middle Rio Grande in central New Mexico, fires
often burn with high severity (fig. 16) fueled by litter, debris, and vegetation that has
accumulated in the absence of high-magnitude floods (Drus 2013; Ellis 2001; Stuever
1997). Few studies have examined response of woody plants in riparian areas to wild-
fire relative to other southwestern vegetation types. However, it is widely perceived
that on regulated streams, native woody plants do not recover from fire as well as
nonnative species such as Russian olive and saltcedar, which resprout more vigorously
and are more flexible in their germination requirements (Busch 1995). To address the
consequences of a changing disturbance regime, a systematic understanding of the ef-
fects of wildfire on structure and composition of riparian ecosystems is needed. In this
section, we report wildfire effects measured along the Middle Rio Grande in central
New Mexico and review results from previous studies at this and other streams.

USDA Forest Service RMRS-GTR-364. 2017. 19
Postfire Dynamics Along the Middle Rio Grande
Study Area and Methods
The Middle Rio Grande is the section of the Rio Grande that flows through central
New Mexico between Cochiti Dam and Elephant Butte Reservoir (fig. 17). Alterations
to the river’s hydrology have occurred over several centuries, but large-scale changes
began in the 1920s when, to increase agriculture production and reduce flood damage,
agencies constructed a network of diversion dams, levees, irrigation canals, and drains
(Phillips et al. 2011; Scurlock 1998). The levees currently prevent the river from mean-
dering across its natural floodplain. The stream bank has been stabilized by vegetation
and structures, known as jetty jacks that limit the movement of the active channel
within the floodway. A canopy of mature Rio Grande cottonwoods (Populus deltoides
ssp. wislizenii) established in the 1950s when alluvium was stabilized by jetty jacks and
colonized by seedlings (Crawford et al. 1993). The river became fully regulated by the
closure of Cochiti Dam in 1973. Since completion of the dam, peak flood volume has
been reduced and the stream channel has been narrowed and incised to disconnect the
floodplain from streamflow in many locations (Novack 2006).
Figure 16—The trees in this photo, taken along the Middle Rio Grande 2 weeks after a high-severity wildfire,
were burned and topkilled.

20 USDA Forest Service RMRS-GTR-364. 2017.
During years of low discharge, the active channel runs dry in certain areas where
water is diverted for agriculture. In wet years with heavy runoff, flood pulses, released
from Cochiti Dam, inundate some portions of the floodway, but floods lack the shear
force needed to scour away woody vegetation. In addition, sediment transport is often
confined to the active channel, where any woody plants that are established post-flood
are vulnerable to subsequent high flows (Howe and Knopf 1991; Molles et al. 1998).
As a result, cottonwoods, which once occurred in varied stage classes scattered across
the Middle Rio Grande floodplain, are largely confined to dense, senescent stands
located between levees and the stream channel (Whitney 1996).
As the role of flooding has diminished along the Middle Rio Grande, wildfire has
become an increasingly common disturbance (Stuever et al. 1995; Williams et al.
2007). To measure postfire recovery of riparian trees, we collected data in two sections
of riparian forest that differ in characteristics of hydrology and wildfire. The north
section, located between the towns of Los Lunas and Belen in Valencia County, is
larger than the south section and is bordered to the east by the active stream channel
and to the west by agricultural fields and residential areas (fig. 17). The south section
is south of Belen in Socorro County in a more rural area and is bordered to the west by
agricultural fields. Based on our observations during the high runoff years of 2005 and
2008, the discharge threshold above which flooding occurs is lower at the north section
than at the south section (table 3; Smith et al. 2009a). In addition, the typical depth to
Figure 17—Our Middle Rio Grande study area contained the Banco wildfire site in the north section and the
3-4-6 and Sevilleta wildfire sites in the south section.

USDA Forest Service RMRS-GTR-364. 2017. 21
groundwater, which influences cottonwood drought mortality, is shallower in the north
than in the south (Smith et al. 2009a). These sections also differ in the amount of area
typically burned by wildfire (table 3). Though fewer fires per year are ignited in the ru-
ral south section, average fire size is larger because of longer firefighting response times
than in the north section (Williams et al. 2007).
In 2013, we visited sites burned by three wildfires. The Sevilleta Fire was the largest
of the three, burning 154 ha of the forest on the west side of the river on April 2 and 3,
2011. The 3-4-6 Fire was intermediate in size, burning 38.5 ha on June 26, 2011, and
the Banco Fire was the smallest, burning 0.5 ha on March 26, 2008. The Sevilleta and
3-4-6 sites were located south of Belen in Socorro County and the Banco site was south
of Los Lunas and north of Belen in Valencia County. We selected these sites because
they were the most recent fires to occur in our study area during our 2013 sampling
period. Prior to burning, each site consisted of a cottonwood canopy and an understory
composed of native and nonnative vegetation, with Russian olive and saltcedar the most
abundant woody species (Smith et al. 2007).
At each site, we established at least one transect that extended from the levee road
to the river channel. We established a sampling point every 40 m along each transect,
until we reached the stream channel or left the burned area. We marked a 12.6-m radius
sampling plot at each point. Within the sampling plot, we measured bole condition (live
or topkilled), diameter at breast height (d.b.h.), and sprouting status of each tree greater
than 5 cm d.b.h. We assigned each tree a basal sprout status based on condition of trees
and postfire sprouts. The statuses were: “bole live, no sprout,” “bole dead with live
sprout,” “bole dead with no sprout,” and “bole dead with dead sprout.” Sprouting trees
typically had one basal sprout clump, so we did not count the number of basal sprouts
per tree. In addition to basal sprouts, we checked every tree for epicormic sprouts along
the bole and in the canopy. At half of the sampling points, we measured abundance of
regenerating shoots within a 5-m radius subplot centered on the sampling point. We
identified the species of each shoot and determined whether it was a basal sprout, a root
sucker, or a sapling that had germinated.
Table 3—Section-specific variables at our Middle Rio Grande
study area. Bankfull discharge threshold refers to discharge
volume required for some flooding to occur, drought discharge
refers to discharge volume below which drought mortality
is observed, and floodway inundation threshold refers to
discharge volume required for floods to inundate the entire
area between the stream channel and the levees.
Section
Variable North South
Total area (ha) 411 321
Bankfull discharge threshold (cms) 142 198
Drought discharge threshold (cms) 42 57
Floodway inundation threshold (cms) 170 227
Area burned 2002–2012 (ha) 3 253

22 USDA Forest Service RMRS-GTR-364. 2017.
Topkill
The death of aboveground tissue is largely determined by the severity with which
a woody plant is burned. If intensity of fire is too low to penetrate the outer layers of
bark or scorch crown foliage, a woody plant may retain viable aboveground tissue.
The ability to withstand fire, however, varies among size and species of woody plants
(Gignoux et al. 1997; Hoffman and Solbrig 2003). Over 90 percent of cottonwoods,
saltcedars, and Russian olives were topkilled at the sites we monitored (table 4). All
of the cottonwoods examined at the 3-4-6 site and Banco site were topkilled. Ninety-
five percent were topkilled at the Sevilleta Fire site (fig. 18), which appeared burned
Table 4—Woody plant species observed regenerating at post-wildfire fire sites in our Middle Rio Grande
study area.
Number Percent Percent with Number of Number of
of trees Percent with basal epicormic root suckers saplings
Species examined topkilled sprouts sprouts observed observed
Rio Grande cottonwood 296 92 47 1.7 79 5
Saltcedar 96 94 90 0 371 0
Russian olive 52 93 80 0 4 0
Goodding’s willow 9 89 89 0 0 0
Screwbean mesquite 2 100 100 0 0 0
White mulberry 1 100 100 0 0 11
Siberian elm 4 25 0 0 2 2
Figure 18—The cottonwoods in this photo, taken at the Sevilleta site 2 years after the fire, were burned by
a mixed-severity fire and topkilled, but they retained bark and fine branches. The understory vegetation in-
cludes native and nonnative woody plants that resprouted after the fire.

USDA Forest Service RMRS-GTR-364. 2017. 23
with less severity than the 3-4-6 site (fig. 19) and the Banco site. The sample sizes for
Goodding’s willow, screwbean mesquite (Prosopis pubescens), white mulberry (Morus
alba), and Siberian elm (Ulmus pumila) were too small to compare rates of topkill
among sites. Earlier studies along the Middle Rio Grande found that topkill of cot-
tonwoods was 100 percent where trees are burned with high severity, 78 percent to 100
percent under moderate severity, and 52 percent to 70 percent under low severity (Ellis
2001; Stuever 1997). Topkill of cottonwoods ranged from 52 percent to 89 percent at
four additional wildfire sites monitored by Johnson and Merritt (2009) along the Middle
Rio Grande. Stromberg and Rychener (2010) estimated that 58 percent of cottonwoods
were topkilled by fire along the unregulated San Pedro River. These estimates indicate
that wildfire will result in topkill of 50 to 100 percent of cottonwoods, with the per-
centage largely determined by fire intensity, which is itself influenced by a number of
factors (Ellis 2001).
Figure 19—The cottonwoods in this photo, taken at the 3-4-6 site 2 years after the fire,
were burned by a high-severity fire and topkilled and quickly shed their bark.

24 USDA Forest Service RMRS-GTR-364. 2017.
Basal Sprouting
At our Middle Rio Grande study sites, basal sprouting was the primary recovery
mechanism for cottonwoods, saltcedars, and Russian olives (figs. 20 and 21). Russian
olives and saltcedars were more likely to have live basal sprouts than were cottonwoods
(table 4). A higher percentage of cottonwoods had basal sprouts at the Banco site than
at the 3-4-6 and Sevilleta sites (table 5). However, we examined a much larger sample
of cottonwoods at the 3-4-6 and Sevilleta sites (n = 108 and 192, respectively) than
at the Banco site (n = 4). Cottonwoods at the Sevilleta site were more likely to have
live basal sprouts than cottonwoods at the 3-4-6 site, but there were similar percent-
ages of trees with live or dead basal sprouts (58.5 percent at 3-4-6 and 63.5 percent at
Sevilleta). Short-term survival of cottonwood basal sprouts was therefore greater at
the Sevilleta fire (83.8 percent) than at the 3-4-6 fire (36 percent). Factors that varied
between these sites, such as fire intensity, tree size, and season of fire, may explain this
difference in short-term survival.
The percentage of cottonwoods with live basal sprouts at the Sevilleta site
(53 percent) was within the range previously reported at fire sites to the south at
Bosque Del Apache National Wildlife Refuge (45 percent and 61 percent; Ellis
2001) and to the north in Albuquerque (32 percent and 35 percent, Stuever 1997).
The percentage with sprouts at the 3-4-6 site (21 percent) was lower than all sites.
Figure 20—Many top-killed trees, such as this cottonwood (right foreground) at the Sevilleta site, produced
basal sprouts.

USDA Forest Service RMRS-GTR-364. 2017. 25
Figure 21—Russian olives (A) and saltcedars (B) recovered from wildfire via basal sprouting at the
3-4-6 and Sevilleta sites.
A.
B.
Table 5—Percentage of top-killed trees that had live basal
sprouts in 2013 at three post-wildfire sites in our Middle
Rio Grande study area.
Site
Species Banco 3-4-6 Sevilleta
Rio Grande cottonwood 100 21 57
Saltcedar 96 85 na
Russian olive 86 76 75

26 USDA Forest Service RMRS-GTR-364. 2017.
This variation among sites may be explained by a number of variables including
hydrological characteristics, such as depth to groundwater, flood frequency (Smith
et al. 2009a), and fire intensity (Ellis 2001). Hydrological characteristics were
similar between the Sevilleta and 3-4-6 sites, but fire intensity differed. Prior to the
fire, the 3-4-6 fire had a higher density of large saltcedar, woody debris, and litter
(Smith et al. 2007), which likely increased fire intensity relative to the Sevilleta
site. The fires also burned during different times of the year, with the 3-4-6 burning
in June and the Sevilleta burning in April. The trees at the 3-4-6 site were larger
than those at the Sevilleta site, indicating that they were older. Large, older trees
may allocate more resources to reproduction and less to sprouting, resulting in
lower basal sprouting frequency and lower survival of basal sprouts (Hodgkinson
1998). Because basal sprout production was the primary mechanism of postfire
recovery we observed, additional work is necessary to isolate the effects of hydrol-
ogy, fire intensity, and tree age on basal sprouting by cottonwoods and other woody
plants.
Epicormic Sprouting
Postfire epicormic sprouting has been observed in several broadleaved species, most
notably among Eucalyptus species in Australia, but has been reported for fewer species
in the United States (Meier et al. 2012). Five of the 192 top-killed cottonwoods exam-
ined at the Sevilleta site produced epicormic sprouts in the branches of their canopies.
These trees retained intact bark and fine branches, indicating that they were burned
with low severity. Two of the epicormic sprouting cottonwoods also produced basal
sprouts (fig. 22). Cottonwoods burned in the Banco and 3-4-6 fires produced basal
sprouts and root suckers, but not epicormic sprouts, probably because individuals were
burned with greater severity than those at the Sevilleta Fire. The higher-severity fires
likely destroyed aboveground meristematic tissues or depleted stores of non- structural
carbohydrates, preventing trees at these sites from producing epicormic sprouts
(Clarke et al. 2012). Stromberg and Rychener (2010) noted epicormic sprouting among
Fremont cottonwoods along the San Pedro River, where, as at our Sevilleta site, most
of the postfire sprouts were basal and a small percentage were epicormic. We do not
know if the epicormic sprouts appeared in 2011 or 2012 at the Sevilleta site, but the
amount of time between top-kill and epicormic sprouting should be measured so that
removal of snags for salvage or firewood can be delayed long enough to prevent loss
of canopy that may be restored by epicormic sprouting. Additional monitoring is also
needed to identify characteristics of cottonwoods that produce epicormic sprouts and
determine if sprouts on cottonwoods become reproductively mature faster than basal
sprouts. If epicormic sprouts produce seeds within a few years of fire, these trees may
aid the recovery of cottonwood forests.

USDA Forest Service RMRS-GTR-364. 2017. 27
Root Suckering
Root suckers were present at each post-wildfire site (fig. 23). Overall, root sucker
abundance was greatest for saltcedar, intermediate for cottonwood, and lowest for
Russian olive (table 4). Cottonwood root sucker density was greatest at the Banco
site, saltcedar root sucker density was greatest at the 3-4-6 site, and Russian olive root
suckers were present only at the 3-4-6 site (table 6). Root sucker production is often an
effective postfire regeneration mechanism because underground meristematic tissues
and carbohydrate reserves are protected from fire and can produce suckers that will
outgrow competing seedlings (Bond and Midgley 2001; Keyser et al. 2005). Unlike
basal sprouts, which emerge in the first months after a fire, root suckering may continue
in subsequent seasons and may be stimulated by precipitation or flood (Ellis 2001).
Figure 22—This cottonwood was topkilled by the Sevilleta fire in 2011, but had pro-
duced (A) epicormic sprouts and (B) basal sprouts by the time this photo was taken in
2013.

28 USDA Forest Service RMRS-GTR-364. 2017.
Long-term studies, however, are needed to verify these suggestions. We found root
suckers at each site, but basal sprout density was greater than root sucker density at two
of the three sites. Root sucker density was greatest at the mesic Banco site, where ideal
hydrological conditions may have encouraged continuous sprouting and growth in the
5 years after the fire occurred. Ellis (2001) also found greater numbers of cottonwood
root suckers at a mesic wildfire site that burned with mixed intensity. Closer monitoring
of additional wildfire sites is needed to better understand the phenology of root sucker
production and survival.
Figure 23—In addition to basal sprouts, we found root suckers produced by cottonwoods, saltcedars, and
other woody species at post-wildfire sites.
Table 6—Density (number per m2) of root suckers at three
post-wildfire sites in our Middle Rio Grande study area.
Site
Species Banco 3-4-6 Sevilleta
Rio Grande cottonwood 0.09 0.01 0.004
Saltcedar 0.02 0.08 0.01
Russian olive 0 0.01 0

USDA Forest Service RMRS-GTR-364. 2017. 29
Postfire Germination
The Banco site, which was burned in March of 2008 and partially flooded in June
of that year, was the only location where we observed postfire germination of woody
plants in either section. We found cottonwood saplings in a sampling plot approximately
10 m from the active channel. These saplings were growing in a band parallel to the ac-
tive channel and were more than 5 m away from the nearest adult cottonwoods. During
an earlier survey in June of 2008, 2 months after the burn, we observed cottonwood
seeds settling in moist soil at this site (fig. 24). At this time, a flood pulse had been
released from Cochiti Reservoir north of Albuquerque (fig. 25). The source of seeds
germinating in the post-wildfire site was an adjacent unburned stand of mature cotton-
woods. The only other woody species that had apparently germinated at this site were
white mulberry (Morus alba) and Siberian elm (Ulmus pumila; table 4).
Figure 24—We observed postfire germination of cottonwoods at the Banco wildfire site after cottonwood
seeds settled in moist soil 3 months after the fire in 2008 (A). We found cottonwood saplings in the same area
in 2009 (B) and in 2012 (C).
Figure 25—Discharge of the Middle Rio Grande, measured at the Albuquerque
gauge site, exceeded a flood threshold after the Banco Fire occurred, coinciding
with the dispersal of cottonwood seeds we observed at the site in 2008. Data
were obtained from the USGS National Water Information System database
(http://waterdata.usgs.gov/nwis).

30 USDA Forest Service RMRS-GTR-364. 2017.
We have never observed cottonwood germination in unburned areas of the north and
south study sections. In this portion of the Middle Rio Grande, the continuous canopy,
dense understory, and dampened flood pulse prevent seedling establishment (Howe and
Knopf 1991; Molles et al. 1998). Ellis (2001) also observed saplings in post-wildfire
sites along the Middle Rio Grande that were flooded within 2 years of being burned.
Our observations of saplings in post-wildfire sites indicate that fire stimulates cotton-
wood recruitment by opening the canopy and removing vegetation and litter from the
soil surface. This disturbance creates sites where, if flooded, germination will occur. A
combination of fire and low-magnitude flooding can therefore act as a replacement for
high-magnitude flooding and stimulate cottonwood germination along this regulated
stream.
Summary: Impacts of Wildfire on Aridland Riparian Ecosystems
We demonstrated that cottonwoods are vulnerable to topkill but can regenerate
through multiple mechanisms under ideal conditions. In most respects, however, non-
native Russian olives and saltcedars were superior to cottonwood in their ability to
recover from fire. In xeric portions of the forest, such as the Sevilleta and 3-4-6 sites,
low basal sprouting and root suckering rates of cottonwood will result in much of the
area currently occupied by cottonwoods being replaced by saltcedar and Russian olive.
In mesic areas, such as the Banco site, cottonwoods may increase in number because
of heavy root suckering and postfire germination. Mesic areas like this are rare in our
study reach (Molles et al. 1998), however, and continued occurrence of wildfire will
likely reduce the extent of cottonwood stands in xeric locations. Wildfire may acceler-
ate a conversion from native to nonnative cover along regulated sections of streams,
but effects could differ along unregulated sections where timing and magnitude of peak
discharges are not as heavily altered.

USDA Forest Service RMRS-GTR-364. 2017. 31
Chapter 3. Use of Woody Plants by Riparian-Nesting Birds
in Unburned Plots and Post-Wildfire Sites
Introduction: Effects of Wildfire on Breeding Birds
With their dependence on vegetation for nesting and foraging sites, and relative ease
of monitoring, breeding birds are an ideal group through which we can evaluate the
effects of disturbances on animal communities. Wildfire disturbance in particular has a
profound influence on breeding bird habitat in the western United States. In ponderosa
pine (Pinus ponderosa) forests, for example, wildfire changes the composition and
structure of woody vegetation by increasing snag density, opening space in the forest
canopy, and stimulating growth of shrubs (Chambers and Mast 2005; Hutto et al. 2008;
Kotliar et al. 2007). These changes benefit snag- and shrub-associates and aerial insecti-
vores, but reduce the abundance of birds that require vegetation in the canopy and litter
on the forest floor (Saab and Powell 2005).
In the American Southwest, wildfire effects have been documented for birds nesting
in coniferous forests, grasslands, and shrublands, but few studies have been conducted
in floodplain riparian forests (Bock and Block 2005). As we have shown in the previ-
ous chapter, wildfire has the potential to dramatically influence the composition and
structure of riparian forests. These changes undoubtedly affect riparian vertebrate
communities, such as breeding birds, but the nature of these effects remains largely
unknown.
The riparian forest along the Middle Rio Grande in central New Mexico forms
an extensive zone of productive habitat for riparian-nesting birds (Farley et al. 1994;
Smith et al. 2009a). Though short-term effects of fire have been described for arthropod
communities and some birds in this forest (Bess et al. 2002; Smith et al. 2006a,b, 2007,
2012), less is known about response to fire by the greater bird community. A deeper
understanding of responses to fire in floodplain systems is necessary to help managers
identify habitat features and sensitive species that require protection under the current
disturbance regime. In this exploratory analysis, we document the use of nest plants by
riparian-nesting birds in unburned and post-wildfire sites to determine which species
may be positively or negatively affected by fire-induced changes.
Nest-Plant Use in Unburned and Postfire Riparian Sites
Study Sites and Methods
We searched for nests along the Middle Rio Grande in central New Mexico as part of
a wider study examining effects of fuel reduction and wildfire on riparian organisms.
For this comparison, we analyzed data from nine unburned plots and six post-
wi ldfi re sites (fig. 26). Ten of the sites were on land managed by the Middle Rio Grande
Conservancy District and two were on land managed by the U.S. Fish and Wildlife Service.

32 USDA Forest Service RMRS-GTR-364. 2017.
Our unburned plot data were collected from control plots or fuel reduction plots prior
to their treatments. Control and fuel reduction plots were clustered in three blocks,
with plots as close to one another in each block as possible to ensure that vegetation
was similar prior to treatments (Smith et al. 2007). Post-wildfire data were collected in
study sites that were established after portions of the forest were accidentally burned
(Smith et al. 2007, 2012). Mean plot size was 22.9 ha for unburned plots and 18.4 ha
for post-wildfire sites (table 7).
Figure 26—We searched for nests in (A) un-
burned plots, in (B) post-wildfire sites including
the Chavez wildfire site shown here 1 year post-
burn, and in (C) 2 years postburn.
Table 7—Characteristics of plots searched for nests along the Middle Rio Grande.
Year Years Area
Plot Type burned searched searched (ha) Ownershipa
Middle 1 Unburned NA 2000–2003 19.4 MRGCD
Middle 2 Unburned NA 2000–2004 29.2 MRGCD
Middle 3 Unburned NA 2000–2003 13.2 MRGCD
Middle 4b Unburned/Wildfire 2002 2000–2008 9.4 MRGCD
Middle 5 Wildfire 2000 2000–2003 22.3 MRGCD
Middle 6 Wildfire 2000 2000–2008 31.4 MRGCD
Middle 7b Unburned NA 2002–2008 35.0 MRGCD
San Francisco Wildfire 2003 2005–2008 19.0 MRGCD
South 1 Unburned NA 2000–2004 28.9 MRGCD
South 2 Unburned NA 2000–2002 15.6 MRGCD
South 3 Unburned NA 2000–2002 26.7 USFWS
South 4 Unburned NA 2000–2002 15.5 USFWS
South 5 Wildfire 1996 2000–2007 22.1 USFWS
South 6 Wildfire 1996 2000–2006 6.1 USFWS
a Ownership: MRGCD = Middle Rio Grande Conservancy District, USFWS = U.S. Fish and Wildlife Service.
b Middle 4 was initially established as a control plot but was monitored as a wildfire site after burning in 2002.
Middle 7 was established in 2002 as a control plot replacement for Middle 4.
B.
C.
A.

USDA Forest Service RMRS-GTR-364. 2017. 33
Crews visited unburned plots and wildfire sites daily to systematically search for
nests of all species encountered. Nest searches were conducted from late April through
August each year from 2000 to 2008. At least once per week, crew members walked
throughout each plot to locate nests by following adults carrying material or food,
incidentally flushing adults from nests, or listening for begging sounds of nestlings.
We recorded nest locations with a handheld GPS and revisited each nest when it was
no longer active to record nest plant species and condition (live or snag) and measure
nest height, using a clinometer where necessary. We described nest substrate as “fallen
debris” if the nest was constructed on fallen trunks or branches.
We compared nest substrate use among birds grouped into three nesting guilds. We
assigned species that excavate their own nest cavities into the excavator guilds. Species
that do not excavate their own nest cavities were assigned to the secondary cavity-
nesting guild. We separated open-nesting species into two guilds based on their nest
placement within riparian forest strata. If a species’ mean nest height was lower than
10 m and minimum nest height was lower than 3 m, we assigned that species to the sub-
canopy/shrub guild. We assigned species to the canopy guild if their mean nest height
was higher than 10 m and minimum nest height was higher than 3 m.
General Nest Plant Use
We summarized data from 1,321 nests of 39 landbird species (table 8). In unburned
plots, 40 percent of nests were constructed in live cottonwoods, with another 40 per-
cent in the two most abundant nonnative trees, Russian olive and saltcedar (fig. 27).
Following wildfire, cottonwood and Russian olive were used for smaller percentages
of nests, but saltcedar use was similar between unburned plots and post-wildfire sites
(fig. 27). Wildfire appeared to increase the use of cottonwood snags, but it did not affect
use of Russian olive snags or saltcedar snags (fig. 27).
Specific Nest-Plant Use
In post-wildfire sites, the species that nested in snags most frequently was Western
Kingbird (scientific names in table 8), which constructed 65 percent of its open-cup
nests in the leafless canopies of postfire cottonwood snags (fig. 28). This was also the
only canopy-nesting species to use snags in unburned plots. Kingbirds prefer to nest in ex-
posed sites, from which they can see potential competitors or predators and actively defend
their territories (Gamble and Bergin 2012), making cottonwood snags an ideal substrate
for this species. The frequent use of nest sites with high exposure to the sun by Western
Kingbird indicates physiological adaptations to heat by adults, eggs, and nestlings. Another
four canopy-nesting species and six subcanopy/shrub species nested in post-wildfire snags,
showing that snags are a useful resource for these guilds following fire.

34 USDA Forest Service RMRS-GTR-364. 2017.
Table 8—Bird species monitored at unburned (UB) and post-wildfire (WF) sites.
Nests
monitored Cottonwood Russian olive Saltcedar Snags Fallen debris
number use use use use use
Species Nest guild UB WF UB WF UB WF UB WF UB WF UB WF
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Percent - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Cooper’s Hawk
Accipiter cooperii Canopy 18 4 100 100 0 0 0 0 0 0 0 0
Swainson’s Hawk
Buteo swainsoni Canopy 4 3 100 67 0 0 0 0 0 33 0 0
Great Horned Owl
Bubo virginianus Canopy 3 3 100 67 0 0 0 0 0 33 0 0
Western Wood-Pewee
Contopus sordidulus Canopy 7 10 100 50 0 0 0 0 0 50 0 0
Western Kingbird
Tyrannus verticalis Canopy 8 53 63 23 13 0 0 0 13 76 0 0
American Crow
Corvus brachyrhynchos Canopy 1 0 100 — 0 — 0 — 0 — 0 —
Common Raven
Corvus corax Canopy 2 2 100 50 0 0 0 0 0 50 0 0
Summer Tanager
Piranga rubra Canopy 13 6 77 100 8 0 0 0 0 0 0 0
Ladder-backed Woodpecker
Picoides scalaris Excavator 3 1 67 0 0 0 0 0 33 100 0 0
Downy Woodpecker
Picoides pubescens Excavator 28 7 86 43 0 0 0 0 14 57 0 0
Hairy Woodpecker
Picoides villosus Excavator 7 5 71 20 0 0 0 0 29 80 0 0
Northern Flicker
Colaptes auratus Excavator 17 24 59 21 0 0 0 0 41 79 0 0
Black-capped Chickadee
Poecile atricapillus Excavator 5 0 100 — 0 — 0 — 0 — 0 —
Barn Owl
Tyto alba Secondary cavity 0 1 — 0 — 0 — 0 — 100 — 0
American Kestrel
Falco sparverius Secondary cavity 4 7 75 0 0 0 0 0 25 100 0 0
Ash-throated Flycatcher
Myiarchus cinerascens Secondary cavity 28 13 61 15 4 0 0 0 36 85 0 0
White-breasted Nuthatch
Sitta carolinensis Secondary cavity 4 7 50 0 0 0 0 0 50 100 0 0
Bewick’s Wren
Thryomanes bewickii Secondary cavity 28 13 46 15 4 0 4 0 32 85 4 0
Eastern Bluebird
Sialia sialis Secondary cavity 0 1 — 0 — 0 — 0 — 100 0 0

USDA Forest Service RMRS-GTR-364. 2017. 35
European Starling
Sturnus vulgaris Secondary cavity 6 4 67 0 0 0 0 0 33 100 0 0
Lucy’s Warbler
Oreothlypis luciae Secondary cavity 1 0 100 — 0 — 0 — 0 — 0 —
Mourning Dove
Zenaida macroura Subcanopy/shrub 128 235 17 19 31 15 24 22 11 15 10 23
Greater Roadrunner
Geococcyx californianus Subcanopy/shrub 2 1 0 0 0 100 100 0 0 0 0 0
Black-chinned Hummingbird
Archilochus alexandri Subcanopy/shrub 256 111 32 22 40 16 20 16 3 9 0 1
Bushtit
Psaltriparus minimus Subcanopy/shrub 2 1 100 100 0 0 0 0 0 0 0 0
American Robin
Turdus migratorius Subcanopy/shrub 1 2 100 100 0 0 0 0 0 0 0 0
Gray Catbird
Dumetella carolinensis Subcanopy/shrub 1 2 0 0 100 50 0 0 0 0 0 0
Northern Mockingbird
Mimus polyglottos Subcanopy/shrub 6 2 0 50 17 0 50 0 0 0 0 0
Phainopepla
Phainopepla nitens Subcanopy/shrub 27 2 74 100 0 0 22 0 0 0 0 0
Common Yellowthroat
Corvus corax Subcanopy/shrub 0 1 — 0 — 0 — 0 — 0 — 0
Yellow-breasted Chat
Icteria virens Subcanopy/shrub 5 26 0 0 20 27 60 23 0 0 0 0
Spotted Towhee
Pipilo maculatus Subcanopy/shrub 15 7 0 0 0 0 7 29 0 0 0 0
Black-headed Grosbeak
Pheucticus melanocephalus Subcanopy/shrub 27 22 26 5 19 9 48 50 4 9 0 0
Blue Grosbeak
Passerina caerulea Subcanopy/shrub 29 31 7 3 7 13 86 68 0 0 0 0
Lazuli Bunting
Passerina amoena Subcanopy/shrub 0 1 — 0 — 0 — 0 — 0 — 0
Indigo Bunting
Passerina cyanea Subcanopy/shrub 1 1 0 0 0 0 0 100 0 0 0 0
Bullock’s Oriole
Icterus bullockii Subcanopy/shrub 1 10 100 80 0 0 0 0 0 20 0 0
House Finch
Haemorhous mexicanus Subcanopy/shrub 5 1 80 0 0 0 0 0 20 100 0 0
Lesser Goldfinch
Spinus psaltria Subcanopy/shrub 6 2 17 50 0 0 83 0 0 50 0 0
Table 8—(Continues).
Nests
monitored Cottonwood Russian olive Saltcedar Snags Fallen debris
number use use use use use
Species Nest guild UB WF UB WF UB WF UB WF UB WF UB WF
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Percent - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

36 USDA Forest Service RMRS-GTR-364. 2017.
Figure 27—The percentage of nests found live cottonwoods was greater in unburned plots but
the percentage in cottonwood snags was greater in wildfire sites. In addition, the percentage of
nests found in Russian olive trees was greater in unburned plots.
Figure 28—Most of the Western Kingbird nests we observed were constructed in the canopies of cottonwood
snags. This nest was constructed in a cottonwood topkilled by fire.

USDA Forest Service RMRS-GTR-364. 2017. 37
Figure 29—In unburned plots, mature cottonwoods frequently had nesting cavities excavated in dead por-
tions of trunks and branches.
Snags have been identified as important nesting sites for cavity-nesting birds
(Thomas et al. 1979). Though no cavity-nesting species responded to the postfire
increase in snag density as strongly as Western Kingbird, the excavator and secondary
cavity-nester guilds constructed a greater percentage of their nests in snags at post-
wildfire sites than in unburned plots. Post-wildfire snags were used by all cavity-nesting
species except for Black-capped Chickadee and Lucy’s warbler, both of which forage
by gleaning arthropods from live foliage (Foote et al. 2010; Johnson et al. 2012). Snag-
associate species will likely occupy post-wildfire sites for 10 to 12 years after a fire, by
which time most snags will have fallen to the ground (D.M. Smith, personal observa-
tion). In addition to snags, live cottonwoods were important nest sites for cavity-nesting
birds. Unburned riparian plots had mature cottonwoods with dead branches and trunks
that were frequently excavated as nest sites (fig. 29) and later used by secondary cavity
nesters (Sedgwick 1997). By consuming dead wood and existing cavities, wildfire may
actually reduce nest site availability for cavity-nesters in this forest.

38 USDA Forest Service RMRS-GTR-364. 2017.
Wildfire can reduce live canopy of Middle Rio Grande forest plots by up to 80
percent (Johnson and Merritt 2009). Cooper’s Hawk, Summer Tanager, and other
canopy-nesting birds require large cottonwoods or similar trees that support their nests
(fig. 30) and meet their nest height preferences (Hunter et al. 1987; Smith and Finch
2014). The low percentages of canopy nests constructed in Russian olive and saltcedar
in both unburned and post-wildfire plots (table 9) indicate that these trees will not pro-
vide adequate nests for most canopy nesters if cottonwood cover is reduced following
fire. Canopy nesters that are unable to nest in snags will therefore avoid post-wildfire
sites if there is no regeneration of the cottonwood canopy.
Figure 30—Cooper’s Hawks constructed their large nests in the cottonwood canopy of unburned plots.
Table 9—Number of species and percentage of nests constructed in live trees, snags, and fallen debris in
unburned (UB) and post-wildfire (WF) sites by birds in each nesting guild.
Number of
bird Russian Fallen
species Cottonwood Saltcedar olive Snags debris
Guild UB WF UB WF UB WF UB WF UB WF UB WF
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Percent
Canopy 8 7 89 40 0 0 4 0 2 59 0 0
Excavator 5 4 77 24 0 0 0 0 23 76 0 0
Secondary cavity 6 7 56 9 1 0 3 0 34 91 1 0
Subcanopy/shrub 16 19 28 19 27 24 30 15 5 11 3 12

USDA Forest Service RMRS-GTR-364. 2017. 39
Figure 31—Blue Grosbeaks and other shrub-nesting birds constructed many nests in postfire saltcedar
resprouts.
Though wildfire removes canopy nest sites, abundance of understory nesting sites
quickly increases after fire as a result of postfire sprouting and deposition of woody de-
bris (Smith et al. 2012). Species in the shrub/subcanopy guild used cottonwood, Russian
olive, and saltcedar with similar frequency in unburned plots but used saltcedar more
frequently than cottonwood or Russian olive in post-wildfire sites (table 9). The higher
use of saltcedar in post-wildfire sites occurred because vigorous sprouting of saltcedar
produced dense stands that appealed to shrub-nesting birds (fig. 31). Shrub/subcanopy
nesters used fallen debris more frequently in the post-wildfire sites, where top-killed
trees begin to shed branches, trunks, and pieces of bark within the first years after fire
(fig. 32). The species that used fallen debris most frequently was Mourning Dove,
which constructed many nests on trunks, branches, and pieces of bark that had fallen
from cottonwoods into resprouting trees and shrubs (Smith et al. 2012). Our observa-
tions of nest plant use suggest that Mourning Doves and other generalists appear to
benefit the most from changes in forest structure that occur following wildfire. To fully
understand the influence of wildfire on reproductive success, however, additional infor-
mation is needed including effects of wildfire on food availability, thermal regimes, nest
predation, and nest parasitism.

40 USDA Forest Service RMRS-GTR-364. 2017.
Summary: Impacts of Wildfire on Riparian-Nesting Birds
At the Middle Rio Grande and other streams, wildfire removes cottonwood canopy,
creates snags and fallen debris, and induces resprouting of woody plants, especially
saltcedar. These changes to forest structure and composition create nest sites for some
bird species but also make the forest unsuitable for nesting by others. As postfire suc-
cession occurs, density of canopy-nesting birds will remain low if cottonwoods do not
recover. In addition, cavity-nesting species will lose nest sites if snags are not replaced
by mature trees. Projections of postfire cottonwood dynamics are therefore necessary to
link characteristics of disturbances to habitat suitability for breeding birds.
Figure 32—Within the first years following fire, top-killed cottonwood shed large pieces of bark and branches.
The fallen debris was often suspended in resprouting vegetation, forming ideal nesting sites for Mourning
Doves and other species. Nests were placed on top of pieces of debris and were obscured by the sprouted
vegetation.

USDA Forest Service RMRS-GTR-364. 2017. 41
Chapter 4. Climate change, Wildfire, and the Future of
Aridland Riparian Ecosystems
Introduction: Climate Change and Hydrology of the American
Southwest
Streamflows are susceptible to alterations resulting from anthropogenic climate
change (fig. 33), the effects of which are expected to be especially severe in the
American Southwest (Garfin et al. 2014; Seager et al. 2007). Under current and pro-
jected levels of CO2 emissions, climatologists predict that increased temperatures and
reduced snowpack will decrease the amount of snowmelt runoff that enters streams
(Pierce et al. 2008; Seager and Vecchi 2010; Seager et al. 2013). Summer monsoons
also cause flood events in many of the region’s streams (Stromberg et al. 2010b).
Changes in patterns of these storms are more difficult to model than changes in winter
precipitation, but there is evidence that greenhouse warming will force monsoons to
occur later in the year and with greater severity, further changing streamflow dynamics
(Cook and Seager 2013; Serrat-Capdevila et al. 2013). Droughts, which are currently
a fixture of the Southwest’s climate, will continue to occur, but with increasing sever-
ity, resulting in further reductions in discharge volume (Cayan et al. 2010; Gutzler and
Robbins 2011; Woodhouse et al. 2010). These changes will undoubtedly affect riparian
ecosystems by limiting germination and increasing mortality of species dependent on
floods and groundwater, to the benefit of generalist species, including nonnative trees
Figure 33—Mean projections from global climate models show increases in maximum air tem-
perature and decreases in snow water equivalent and runoff for (A) the Colorado River Basin, (B)
the Gila River Basin, and (C) the Rio Grande Basin (Reclamation 2016a,b).
A.

42 USDA Forest Service RMRS-GTR-364. 2017.
B.
C.
and shrubs, with characteristics such as seed viability and drought resistance that dif-
fer from native riparian obligates (Merritt and Poff 2010; Perry et al. 2012; Stromberg
et al. 2010b). Because characteristics of discharge can determine rates of reproduction
and survival of woody vegetation, we examined projected changes in peak magnitude
and timing at the stream gauge sites described in Chapter 1. We then incorporated
projections for the Rio Grande into a cottonwood population model. Using output from
this model, we can predict the effects of climate change on riparian forest structure and
breeding bird habitat.
Figure 33—(Continued).

USDA Forest Service RMRS-GTR-364. 2017. 43
Hydrological Projections
Methodology
We obtained projections of mean daily discharge for the stream gauge sites, made
available by the U.S. Bureau of Reclamation (Miller et al. 2011). The projections
use bias-corrected, spatial-downscaled precipitation and temperature data from the
World Climate Research Programme Coupled Model Intercomparison Project phase 3
(CMIP3; Appendix A). We calculated averages from 36 projections of mean daily
discharge from 15 global climate models run under the A2 emissions scenario for each
of the 11 sites. We projected changes in magnitude and timing of peak discharge for the
period of 2010 to 2099. To do this, we calculated the mean of each variable across the
36 model runs for the periods of 1980 to 2009, 2010 to 2039, 2040 to 2069, and 2070
to 2099. We then calculated the departure from the 1980 to 2009 mean for each of the
three periods with future years. Using these values, we adjusted long-term observed
means to reflect the projected changes.
Changes in Magnitude and Timing of Peak Discharge
Overall, models predicted greater changes in peak discharge date than in peak
discharge magnitude (fig. 34). Though averaged model output predicted decreases
in peak discharge magnitude at the Rio Grande, Rio Chama, and San Juan River, the
Rio Grande was the only stream in which these decreases were projected by more than
50 percent of the model runs (table 10). At each Central Highlands stream, more than
50 percent of model runs predicted no change in peak discharge magnitude (table 10).
Figure 34—Projections of
magnitude and timing of peak
mean daily discharge under
the A2 emissions scenario
showed decreases in vol-
ume and increasingly early
peak dates at several Rocky
Mountain streams. Changes
were less apparent at the
Central Highland streams.
Data are averaged for three
30-year periods. Black lines
represent averages from 36
model runs; blue lines repre-
sent individual model runs.
Methods for calculating these
projections are described in
Appendix A.

44 USDA Forest Service RMRS-GTR-364. 2017.
Figure 34—(Continued).
Figure 34—(Continued).

USDA Forest Service RMRS-GTR-364. 2017. 45
At least 80 percent of the model runs predicted an earlier peak discharge date at the
Colorado River, Gunnison River, Green River, and San Juan River. Greater than 50 per-
cent of model runs indicated no change in date for the Rio Grande basin streams and all
of the Central Highland streams.
Peak discharges in Rocky Mountain streams typically result from snowpack runoff,
which is strongly influenced by temperature. The predicted increase in temperatures
would shrink the snowpack and accelerate snowmelt (Garfin et al. 2014; Pierce et al.
2008). As a result, peak discharges will on average occur earlier and with lower magni-
tude (Hurd and Coonrod 2008; Stewart et al. 2005). There is less model agreement on
changes in amount of rainfall, which has a greater impact on timing and magnitude of
peak discharge in the Central Highland streams (Ellis et al. 2008). Greenhouse warm-
ing may result in heavier storms occurring during the summer, fall, and winter periods,
Table 10—Percentage of 36 model projections indicating changes in peak discharge volume and date
(ordinal day) of peak discharge at stream gauge sites.
Peak discharge volume Peak discharge date
(m3/second) (ordinal day)
Gauge site Decrease Increase Null Decrease Increase Null
Colorado River at Cameo 28 14 58 100 0 0
Gunnison River at Grand Junction 33 3 64 83 0 17
Green River at Greendale 14 25 61 81 0 19
San Juan River at Bluff 47 3 50 94 0 6
Rio Grande at Otowi 53 3 44 36 6 58
Rio Chama at Abiquiu 25 3 72 31 0 69
Salt River at Roosevelt 14 8 78 8 3 89
Verde River above Horseshoe Dam 19 11 70 11 14 75
San Francisco River at Clifton 6 19 75 6 6 88
Gila River at Gila 5 17 78 8 8 84
Tonto Creek near Roosevelt 8 8 84 3 19 78
Figure 34—(Continued).

46 USDA Forest Service RMRS-GTR-364. 2017.
which would increase variability of peak discharges from year to year, especially at
Central Highland streams, which flood readily when heavy rainfall occurs (Christensen
et al. 2004; Hawkins et al. 2015). The models we examined did not show a directional
change in peak discharge volume or timing at Central Highland streams, but variability
in peak discharge volume will likely occur as a result of increased intensity of mon-
soons and Pacific frontal storms (Cook and Seager 2013; Garfin et al. 2014).
Modeling Changes in Cottonwood Populations
Stochastic Population Model Development
To make an initial assessment of how the above projected changes in hydrology will
affect riparian ecosystems, we modified the stochastic cottonwood population model
developed by Lytle and Merritt (2004) to predict future populations of cottonwoods
at our study sections along the Middle Rio Grande (Appendix B). Lytle and Merritt
(2004) used historical hydrograph data to simulate occurrence of floods and droughts,
which are stochastic events that determine survival and reproduction. The original
stage-structured model contained six classes: seedling, 1-year-old sapling, 2-year-old
tree, 3-year-old tree, 4-year-old tree, non-reproductive adult (5- to 10-year-old tree),
and reproductive adult (6 years or older). The model also contained stage-specific tran-
sition probabilities and a fecundity term, which was greater than 0 only in years when
the hydrograph peaks above a flood threshold and the declining limb coincides with the
seed dispersal period (fig. 35). Transition probabilities are based on survival, which is
reduced when peak discharge volume is below a drought threshold (Lytle and Merritt
2004). Our modified model includes wildfire as an additional stochastic event, as well
as the effects of regulation and climate change on cottonwood populations. We applied
the model to our Middle Rio Grande study area by fitting it with vital rates, including
top-kill and resprout probability, estimated from our observations and those from previ-
ous studies (Ellis 2001; Stuever 1997; Appendix B).
Figure 35—Life cycle model for cottonwoods along the Middle Rio Grande. Stages are seedlings (N1), 2-
to 4-year-olds (N2-N4), non-reproductive adults (N5), and reproductive adults (N6). N is density, G is the
probability of transition to the next stage, P is the probability of remaining in that stage, R is probability of
retrogression from one stage to another as a result of postfire resprouting, F is fecundity through seedling
germination, and S is fecundity through root suckering.

USDA Forest Service RMRS-GTR-364. 2017. 47
We used the model to project populations of trees over a 100-year period in each
section, with flood, fire, and drought occurring as stochastic events. For each projection
year, we adjusted magnitude and timing of peak discharge to reflect projected changes
under the A1B, A2, and B1 greenhouse gas emission scenarios. We calculated 95
percent confidence intervals and range of stage class densities from 1,000 simulations
under each emissions scenario at both sections of forest. We report changes in density
of the youngest and oldest stage classes because mature cottonwoods are important nest
sites for riparian-nesting birds and the presence of seedlings represents the occurrence
of a reproductive event.
Cottonwood Projections
Our model predicted a sharp decline in density of mature cottonwoods under
each emissions scenario in both sections (figs. 36 and 37). In most of the simulations
(Appendix B), the decline in density was steeper for the south section where fire size
was greater and flood frequency was lower than in the north section. There was a
wider range of projected densities in the south section because of the greater number of
seedling germination events that occurred following wildfire in the simulations. Range
of projected seedling densities was also greater in the south section. Though typically
Figure 36—Density of mature cottonwoods decreased through the projection period,
as shown by the 95 percent confidence intervals and ranges from simulated projec-
tions modeled under three emission scenarios in the north section of our Middle Rio
Grande study area. The A1B scenario represented moderate increases in greenhouse
gas emissions, A2 represented large increases, and B1 represented small increases.

48 USDA Forest Service RMRS-GTR-364. 2017.
rare, seedlings were present in at least one simulation of each projection year at the
north section under all three emission scenarios (fig. 38). At the south section, the
range of seedling densities decreased with time and, under the A2 emissions scenario,
were absent after 80 years (fig. 39). These model results show that reproduction of
cottonwoods will be possible under certain conditions at both sections, but decreases in
discharge will make reproduction increasingly unlikely at the south section.
Figure 37—Density of mature cottonwoods decreased through the projection period
under most simulations modeled under three emission scenarios in the south sec-
tion of our Middle Rio Grande study area as shown by the 95 percent confidence
intervals. The range of densities from simulated projections fluctuated because of
occasional reproduction following wildfire and flood. The A1B scenario represented
moderate increases in greenhouse gas emissions, A2 represented large increases,
and B1 represented small increases.

USDA Forest Service RMRS-GTR-364. 2017. 49
Figure 38—Density of cotton-
wood seedlings were similar
under most simulations mod-
eled under three emission
scenarios in the north section
of our Middle Rio Grande
study area. The upper range
of densities indicated that
seedlings were present in at
least one simulation in each
projection year. The A1B sce-
nario represented moderate
increases in greenhouse gas
emissions, A2 represented
large increases, and B1 rep-
resented small increases.
Figure 39—Projections of
cottonwood seedling density
differed among three emission
scenarios in the south sec-
tion of our Middle Rio Grande
study area as shown by the 95
percent confidence intervals
and range of densities. The
upper ranges indicated that,
as greenhouse warming in-
creased, fewer seedlings were
present in simulations, espe-
cially under the A2 scenario.
The A1B scenario represented
moderate increases in green-
house gas emissions, A2
represented large increases,
and B1 represented small
increases. The range of densi-
ties was greater in the south
section than in the north.

50 USDA Forest Service RMRS-GTR-364. 2017.
Management Implications
Our model output supports the contention that Middle Rio Grande cottonwood
forests are in decline and will be largely replaced by other woody species by the end
of this century (Howe and Knopf 1991; Molles et al. 1998). The decline in cottonwood
density will be more rapid in the south section of our study area, which had lower flood
probability and higher wildfire probability than the north section. Nonnative woody
species, such as Russian olive and saltcedar, are present throughout our study area and
will likely increase in abundance as cottonwood declines. Replacement of cottonwood
by Russian olive and saltcedar will change the structure of the Middle Rio Grande ri-
parian forest by increasing the density of low-stature vegetation and decreasing canopy
height. Riparian-nesting birds will be affected as a result, with canopy-nesting birds
including Cooper’s Hawk and Summer Tanager immediately affected by loss of nest
sites, followed by cavity nesters and snag-associates including Western Kingbird and
Lucy’s Warbler. Loss of cottonwoods would profoundly change the unique composition
of riparian-nesting birds at the Middle Rio Grande and other streams, as most of the
remaining birds will be widespread and generalist species such as Spotted Towhee and
Blue Grosbeak (Smith and Finch 2014).
One of our most significant observations was the germination of cottonwoods fol-
lowing high severity fire and low-magnitude flooding in 2008. During years of heavy
snowpack, managers can plan releases from Cochiti Dam to induce germination of cot-
tonwoods in portions of the forest that have been burned or otherwise disturbed. Based
on our observations from 2008, peak flows at Albuquerque should exceed 140 cms
(5,000 cfs) and the hydrograph peak should occur in late May or early June. Additional
monitoring of fire-and-flood events will be necessary to refine prescriptions for cotton-
wood germination and identify germination requirements for additional native species.
Though cottonwoods are a major element of this forest, multi-species or multi-
guild models should be implemented to fully anticipate changes in forest structure and
composition at this and other streams (Merritt et al. 2010). Arizona sycamore is not
as sensitive to fire as cottonwood (Bock and Bock 2015), so changes in disturbance
regimes will have effects that differ from the Rio Grande where this species is a com-
ponent of the forest canopy. Additional information about the response to flood, fire,
and drought by native and nonnative species is needed to better predict changes in the
quality of habitat for riparian-nesting birds.
We found that there is substantial variation among southwestern streams in their
natural hydrology, extent of regulation, and vulnerability to climate change. For this
reason, projections of cottonwood populations must incorporate data specific to indi-
vidual stream sites or groups of similar sites. In addition to considering climate change
effects, models should also consider how hydrological patterns will be influenced

USDA Forest Service RMRS-GTR-364. 2017. 51
by changes in water use by agricultural, municipal, and industrial sectors. Following
the example of carbon emission scenarios, multiple water use scenarios should be
developed to reflect the changes in the human landscape of the Southwest. Output from
these models can then be used to set standards for surface flows necessary to maintain
aridland riparian ecosystems in the region.
References
Auble, G.T.; Scott, M.L. 1998. Fluvial disturbance patches and cottonwood recruitment
along the upper Missouri River, Montana. Wetlands. 18: 546–556.
Bêche, L.A.; Stephens, S.L.; Resh, V.H. 2005. Effects of prescribed re on a Sierra
Nevada (California, USA) stream and its riparian zone. Forest Ecology and
Management. 218: 37–59.
Bess, E.C.; Parmenter, R.P.; McCoy, S.; [et al.]. 2002. Responses of a riparian forest-
oor arthropod community to wildre in the Middle Rio Grande Valley, New
Mexico. Environmental Entomology. 31: 774–784.
Beauchamp, V.B.; Stromberg, J.C. 2007. Flow regulation of the Verde River, Arizona
encourages Tamarix recruitment but has minimal effects on Populus and Salix stand
density. Wetlands. 27: 381–389.
Bhattacharjee, J.; Taylor, J.P.; Smith, L.M.; [et al.]. 2008. The importance of soil
characteristics in determining survival of rst-year cottonwood seedlings in altered
habitats. Restoration Ecology. 16: 563–571.
Birken, A.S.; Cooper, D.J. 2006. Processes of Tamarix invasion and oodplain
development along the lower Green River, Utah. Ecological Applications. 16:
1103–1120.
Bock, C.E.; Block, W.M. 2005. Fire and birds in the southwestern United States.
Studies in Avian Biology. 30: 14–32.
Bock, C.E.; Bock, J.H. 1984. Importance of sycamores to riparian birds in southeastern
Arizona. Journal of Field Ornithology. 55: 97–103.
Bock, J.H.; Bock, C.E. 1989. Factors limiting sexual reproduction in Platanus wrightii
in southeastern Arizona. Aliso. 12: 295–301.
Bond, W.J.; Midgley, J.J. 2001. Ecology of sprouting in woody plants: The persistence
niche. Trends in Ecology and Evolution. 16: 45–51.
Bonnet, V.H.; Schoettle, A.W.; Shepperd, W.D. 2005. Postre environmental conditions
inuence the spatial pattern of regeneration for Pinus ponderosa. Canadian Journal
of Forest Research. 35: 37–47.
Braatne, J.H.; Rood, S.B.; Heilman, P.E. 1996. Life history, ecology, and conservation
of riparian cottonwoods in North America. In: Stettler, R.F.; Bradshaw, H.D., Jr.;
Heilman, P.E.; [et al.], eds. Biology of populus and its implications for management
and conservation. Ottawa, Ontario, Canada: National Research Council of Canada:
57–85.

52 USDA Forest Service RMRS-GTR-364. 2017.
Brand, L.A.; Cerasale, D.J.; Rich, T.D.; [et al.]. 2009. Breeding and migratory
birds: Patterns and processes. In: Stromberg, J.C.; Tellman, B., eds. Ecology and
conservation of the San Pedro River. Tucson, AZ: University of Arizona Press:
153–174.
Brand, L.A.; White, G.C.; Noon, B.R. 2008. Factors inuencing species richness and
community composition in a desert riparian corridor. Condor. 110: 199–210.
Brodhead, K.M.; Stoleson, S.H.; Finch, D.M. 2007. Southwestern Willow Flycatchers
(Empidonax traillii extimus) in a grazed landscape: Factors affecting brood
parasitism. Auk. 124: 1213–1228.
Brown, B.T. 1992. Nesting chronology, density, and habitat use of black-chinned
hummingbirds along the Colorado River, Arizona. Journal of Field Ornithology. 63:
393–506.
Brown, T.K. 2002. Creating and maintaining wildlife, insect, and sh habitat structures
in dead wood. In: Laudenslayer, William F., Jr.; Shea, Patrick J.; Valentine, Bradley
E.; [et al.], tech. coords. 2002. Proceedings of the symposium on the ecology and
management of dead wood in western forests; 1999 November 2–4; Reno, NV. Gen.
Tech. Rep. PSW-GTR-181. Albany, CA: U.S. Department of Agriculture Forest
Service, Pacic Southwest Research Station: 883–892.
Burnham, K.P.; Anderson, D.R. 2002. Model selection and multimodel inference: A
practical information-theoretic approach. New York: Springer-Verlag Press. 496 p.
Busch, D.E. 1995. Effects of re on southwestern riparian plant community structure.
The Southwestern Naturalist. 40: 259–267.
Busch, D.E.; Smith, S.D. 1995. Mechanisms associated with decline of woody species
in riparian ecosystems of the southwestern U.S. Ecological Monographs. 65:
347–370.
Busch, D.E.; Ingraham, N.L.; Smith, S.D. 1992. Water uptake in woody riparian
phreatophytes of the southwestern United States: A stable isotope study. Ecological
Applications. 2: 450–459.
Carothers, S.W.; Johnson, R.R.; Atchison, S.W. 1974. Population structure and social
organization of southwestern riparian birds. American Zoologist. 14: 97–108.
Cayan, D.R.; Das, T.; Pierce, D.E.; [et al.]. 2010. Future dryness in the southwest US
and the hydrology of the early 21st century drought. Proceedings of the National
Academy of Science. 107: 21271–21276.
Chambers, C.L.; Mast, J.N. 2005. Ponderosa pine snag dynamics and cavity excavation
following wildre in northern Arizona. Forest Ecology and Management. 216:
227–240.
Christensen N.S.; Wood, A.W.; Voisin, N.; [et al.]. 2004. The effects of climate change
on the hydrology and water resources of the Colorado River Basin. Climate Change.
62: 337–363.
Clarke, P.J.; Lawes, M.J.; Midgley, J.J.; [et al.]. 2012. Resprouting as a key functional
trait: How buds, protection and resources drive persistence after re. New
Phytologist. 197: 19–35.
Coble, A.P.; Kolb, T.E. 2013. Native riparian tree establishment along the regulated
Dolores River, Colorado. Western North American Naturalist.73: 41–53.

USDA Forest Service RMRS-GTR-364. 2017. 53
Cocking M.I.; Varner, J.M.; Knap, E.E. 2014. Long-term effects of re severity on oak-
conifer dynamics in the southern Cascades. Ecological Applications. 24: 94–107.
Cook, B.I.; Seager, R. 2013. The response of the North American Monsoon to increased
greenhouse gas forcing. Journal of Geophysical Research: Atmospheres. 118:
1690–1699.
Cooper, D.J.; Merritt, D.M.; Andersen, D.C.; [et al.]. 1999. Factors controlling the
establishment of Fremont cottonwood seedlings on the upper Green River, U.S.A.
Regulated Rivers: Research and Management. 15: 418–440.
Crawford, C.S.; Cully, A.C.; Leutheuser, R.; [et al.]. 1993. Middle Rio Grande
Ecosystem: Bosque biological management plan. Albuquerque, NM: University of
New Mexico, Center for Southwest Research. 320 p.
Dewine, J.M.; Cooper, D.J. 2007. Effects of river regulation on riparian box elder (Acer
negundo) forests in canyons of the upper Colorado River basin, USA. Wetlands. 27:
278–289.
Drus, G.M. 2013. Tamarisk (Tamarix spp.) and Desert Riparian Ecosystem Change.
Dissertation. University of California Santa Barbara.
Durst, S.L.; Theimer, T.C.; Paxton, E.H.; [et al.]. 2008. Temporal variation in the
arthropod community of desert riparian habitats with varying amounts of saltcedar.
Journal of Arid Environments. 72: 1644–1653.
Ellis, A.W.; Hawkins, T.W.; Balling, R.C., Jr.; [et al.]. 2008. Estimating future runoff
levels for a semi-arid uvial system in central Arizona, USA. Climate Research. 35:
227–239.
Ellis, L.M. 1995. Bird use of saltcedar and cottonwood vegetation in the Middle Rio
Grande Valley of New Mexico, U.S.A. Journal of Arid Environments. 30: 339–349.
Ellis, L.M. 2001. Short-term response of woody plants to re in a Rio Grande riparian
forest, central New Mexico, U.S.A. Biological Conservation. 97: 159–170.
Enright, N.J.; Lamont, B.B.; Marsula, R. 1996. Canopy seed bank dynamics and
optimum re regime for the highly serotinous shrub, Banksia hookeriana. Journal of
Ecology. 84: 9–17.
Farley, G.H.; Ellis, L.M.; Stuart, J.N.; [et al.]. 1994. Avian species richness in
different-aged stands of riparian forest along the Middle Rio Grande, New Mexico.
Conservation Biology. 8: 1098–1108.
Flanigan, K.G.; Haas, A.I. 2008. The impact of full benecial use of San Juan-Chama
Project water by the city of Albuquerque on New Mexico’s Rio Grande Compact
obligations. Natural Resources Journal. 48: 371–405.
Foote, J.R.; Mennill, D.J.; Ratcliffe, L.M.; [et al.]. 2010. Black-capped Chickadee
(Poecile atricapillus). In: Poole, A., ed. The birds of North America online. Ithaca,
NY: Cornell Lab of Ornithology. https://birdsna.org/Species-Account/bna/species/
bkcchi/introduction.
Friedman, J.M.; Auble, G.T.; Shafroth, P.B.; [et al.]. 2005. Dominance of non-native
trees in western USA. Biological Invasions. 7: 747–751.
Gamble, L.R.; Bergin, T.M. 2012. Western Kingbird (Tyrannus verticalis). In: Poole,
A., ed. The birds of North America online. Ithaca, NY: Cornell Lab of Ornithology.
http://bna.birds.cornell.edu/bna/species/227doi:10.2173/bna.227

54 USDA Forest Service RMRS-GTR-364. 2017.
Garn, G.; Franco, G.; Blanco, H.; [et al.]. 2014. Southwest. In: Melillo, J.M.;
Richmond, T.C.; Yohe, G.W., eds. Climate change impacts in the United States: The
third national climate assessment. U.S. Global Change Research Program: 462–486.
doi:10.7930/J08G8HMN.
Gignoux, J.; Clobert, J.; Menaut, J.-C. 1997. Alternative re resistance strategies in
savanna trees. Oecologia. 110: 576–583.
Gutzler, D.S.; Robbins, T.O. 2011. Climate variability and expected change in the
western United States: Regional downscaling and drought statistics. Climate
Dynamics. 37: 835–849.
Hawkins, G.A.; Vivoni, E.R.; Robles-Morua, A.; [et al.]. 2015. A climate change
projection for summer hydrologic conditions in a semiarid watershed of central
Arizona. Journal of Arid Environments. 118: 9–20.
Hodgkinson, K.C. 1998. Sprouting success of shrubs after re: Height-dependent
relationships for different strategies. Oecologia. 115: 64–72.
Hoffman, W.A.; Solbrig, O.T. 2003. The role of topkill in the differential response of
savanna woody species to re. Forest Ecology and Management. 180: 273–286.
Howe, W.H.; Knopf, F.L. 1991. On the imminent decline of Rio Grande cottonwoods in
central New Mexico. The Southwestern Naturalist. 36: 218–224.
Hunter, W.C.; Ohmart, R.D.; Anderson, B.W. 1987. Status of breeding riparian-obligate
birds in southwestern riverine systems. Western Birds. 18:10–18.
Hutto, R.L.; Conway, C.J.; Saab, V.A.; [et al.]. 2008. What constitutes a natural re
regime? Insight from the ecology and distribution of coniferous forest birds. Fire
Ecology. 4: 115–132.
Johnson, B.; Merritt, D. 2009. The effects of wildre on native tree species in the
Middle Rio Grande bosques of New Mexico. Fort Collins, CO: Colorado State
University. 43 p.
Johnson, R.R.; Yard, H.K.; Brown, B.T. 2012. Lucy’s Warbler (Oreothlypis luciae).
In: Poole, A., ed. The birds of North America online. Ithaca, NY: Cornell Lab of
Ornithology. http://bna.birds.cornell.edu/bna/species/318doi:10.2173/bna.318.
Katz, G.L.; Shafroth, P.B. 2003. Biology, ecology, and management of Elaeagnus
angustifolia (Russian olive) in western North America. Wetlands. 23: 763–777.
Keeley, J.E. 1987. Role of re in seed germination of woody taxa in California
chaparral. Ecology. 68: 434–443.
Keyser, T.L.; Smith, F.W.; Shepard, W.D. 2005. Trembling aspen response to a mixed-
severity wildre in the Black Hills, South Dakota, USA. Canadian Journal of Forest
Research. 35: 2679–2684.
Knopf , F.L.; Olson, T.E. 1984. Naturalization of Russian olive: Implications to Rocky
Mountain wildlife. Wildlife Society Bulletin. 12: 289–298.
Kotliar, N.B.; Kennedy, P.L.; Ferree, K. 2007. Avifaunal responses to re in
southwestern montane forests along a burn severity gradient. Ecological
Applications. 17: 491–507.
Kramer P.J.; Kozlowski, T.T. 1979. Physiology of woody plants. New York: Academic
Press.

USDA Forest Service RMRS-GTR-364. 2017. 55
Lotan, J. 1976. Cone serotiny—Fire relationships in lodgepole pine. In: Tall Timbers
Fire ecology conference proceedings 14. Tallahassee, FL: Tall Timbers Research
Center: 267–278.
Lytle, D.A.; Merritt, D.M. 2004. Hydrologic regimes and riparian forests: A structured
population model for cottonwood. Ecology. 85: 2493–2503.
Meier, A.R.; Saunders, M.R.; Michler, C.H. 2012. Epicormic buds in trees: A review
of bud establishment, development and dormancy release. Tree Physiology. 32:
565–584.
Meehl, G.A.; Covey, C.; Delworth, T.; [et al.]. 2007. The WCRP CMIP3 multi-
model dataset: A new era in climate change research. Bulletin of the American
Meteorological Society. 88: 1383–1394.
Merritt, D.M.; Bateman, H.L. 2012. Linking stream ow and groundwater to avian
habitat in a desert riparian system. Ecological Applications. 22: 1973–1988.
Merritt, D.M.; Poff, N.L. 2010. Shifting dominance of riparian Populus and Tamarix
along gradients of ow alteration in western North American rivers. Ecological
Application. 20: 135–152.
Merritt, D.M.; Scott, M.L.; Poff, N.L.; [et al.]. 2010. Theory, methods and tools, for
determining environmental ows for riparian vegetation: Riparian vegetation-ow
response guilds. Freshwater Biology. 55: 206–225.
Miller, W.P.; Piechota, T.C.; Gangopadhyay, S.; Pruitt, T. 2011. Development of
streamow projections under changing climate conditions over Colorado River basin
headwaters. Hydrological Earth System Sciences. 15: 2145–2164.
Minckley, W.L.; Rinne, J.N. 1985. Large woody debris in hot-desert streams: An
historical review. Desert Plants. 7: 142–153.
Molles, M.C., Jr.; Crawford, C.S.; Ellis, L.M.; [et al.]. 1998. Managed ooding for
riparian ecosystem restoration. Bioscience. 48: 749–756.
Mortenson, S.G.; Weisberg, P.J. 2010. Does river regulation increase the dominance of
invasive woody species in riparian landscapes? Global Ecology and Biogeography.
19: 562–574.
Mund-Meyerson, M.J. 1991. Arthropod abundance and composition on native versus
exotic vegetation in the Middle Rio Grande riparian forest as related to avian
foraging. Thesis. Albuquerque: University of New Mexico.
Neary, D.G.; Medina, A.L.; Rinne, J.N. eds. 2012. Synthesis of Upper Verde River
research and monitoring 1993–2008. Gen. Tech. Rep. RMRS-GTR-291. Fort Collins,
CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research
Station. 296 p.
Novack, S.J. 2006. Hydraulic modeling analysis of the Middle Rio Grande River from
Cochiti Dam to Galisteo Creek, New Mexico. Thesis. Fort Collins, CO: Colorado
State University.
Perry, L.G.; Andersen, D.C.; Reynolds, L.V.; [et al.]. 2012. Vulnerability of riparian
ecosystems to elevated CO2 and climate change in arid and semiarid North America.
Global Change Biology. 18: 821–842.
Phillips, F.M.; Hall, G.E.; Black, M.E. 2011. Reining in the Rio Grande: People, land,
and water. Albuquerque, NM: University of New Mexico Press.

56 USDA Forest Service RMRS-GTR-364. 2017.
Pierce, D.W.; Barnett, T.P.; Hidalgo, H.G.; [et al.]. 2008. Attribution of declining
western U.S. snowpack to human event. Journal of Climate. 21: 6425–6444.
Powell, B.F.; Steidl, R.J. 2001. Habitat selection by riparian songbirds breeding in
southern Arizona. Journal of Wildlife Management. 66: 1096–1103.
R Development Core Team. 2011. R: A language and environment for statistical
computing. Vienna, Austria: R Foundation for Statistical Computing. ISBN
3-900051-07-0. URL. http://www.R-project.org.
Reclamation. 2016a. SECURE Water Act Section 9503—Reclamation climate
change and water 2016. Denver, CO: U.S. Department of the Interior, Bureau of
Reclamation. 307 p.
Reclamation. 2016b. West-wide climate risk assessments: Hydroclimate projections.
Denver, CO: U.S. Department of the Interior, Bureau of Reclamation. 154 p.
Rosenberg, K.V.; Ohmart, R.D.; Anderson, B.W. 1982. Community organization of
riparian breeding birds: Response to an annual resource peak. Auk. 99: 260–274.
Saab, V.A.; Powell, H.D.W. 2005. Fire and avian ecology in North America: Process
inuencing patterns. Studies in Avian Biology. 30: 1–13.
Sabo, J.L.; McCluney, K.E.; Marusenko, Y.; [et al.]. 2008. Greenfall links groundwater
to aboveground food webs in desert river oodplains. Ecological Monographs. 78:
615–631.
Scurlock, D. 1998. From the Rio to the Sierra: An environmental history of the Middle
Rio Grande Basin. Gen. Tech. Rep. RMRS-GTR-5. Fort Collins, Colorado: U.S.
Department of Agriculture, Forest Service, Rocky Mountain Research Station. 440 p.
Seager, R.; Vecchi, G.A. 2010. Greenhouse warming and the 21st century hydroclimate
of southwestern North America. Proceedings of the National Academy of Science.
107: 21277–21282.
Seager, R.; Ting, M.; Li, C.; [et al.]. 2013. Projections of declining surface water
availability in the southwestern United States. Nature Climate Change. 3: 482–486.
Seager, R.M.; Ting, I.; Held, Y.; [et al.]. 2007. Model projections of an imminent
transition to a more arid climate in southwestern North America. Science. 316:
1181–1184.
Sedgwick, J.A. 1997. Sequential cavity use in a cottonwood bottomland. Condor. 99:
880–887.
Serrat-Capdevila, A.; Valdes, J.B.; Dominguez, F.; [et al.]. 2013. Characterizing
the water extremes of the new century in the US Southwest: A comprehensive
assessment from state-of-the-art climate model projections. International Journal of
Water Resources Development. 29: 152–171.
Shafroth, P.B.; Stromberg, J.C.; Patten, D.T. 2002. Riparian vegetation response to
altered disturbance and stress regimes. Ecological Applications. 12: 107–123.
Sher, A.A.; Marshall, D.L.; Taylor, J.P. 2002. Establishment patterns of native Populus
and Salix in the presence of nonnative Tamarisk. Ecological Applications. 12:
760–772.

USDA Forest Service RMRS-GTR-364. 2017. 57
Smith, D.M.; Finch, D.M. 2014. Use of native and nonnative nest plants by riparian‐
nesting birds along two streams in New Mexico. River Research and Applications.
30: 1134–1145.
Smith, D.M.; Finch, D.M.; Gunning, C.; [et al.]. 2009a. Post-wildre recovery of
riparian vegetation during a period of water scarcity in the southwestern U.S. Fire
Ecology. 5: 38–55.
Smith, D.M.; Finch, D.M.; Hawksworth, D.L. 2009b. Black-chinned Hummingbird
nest-site selection and nest survival in response to fuel reduction in a southwestern
riparian forest. Condor. 111: 641–652.
Smith, D.M.; Finch, D.M.; Hawksworth, D.L. 2012. Nesting characteristics of
mourning doves in central New Mexico: Response to riparian forest change. Journal
of Wildlife Management. 76: 382–390.
Smith, D.M.; Kelly, J.F.; Finch, D.M. 2006a. Cicada emergence in southwestern
riparian forest: Inuences of wildre and vegetation composition. Ecological
Applications. 16: 1608–1618.
Smith, D.M.; Kelly, J.F.; Finch, D.M. 2006b. Inuences of disturbance and vegetation
on abundance of native and exotic detritivores in a southwestern riparian forest.
Environmental Entomology. 35: 1525–1531.
Smith, D.M.; Kelly, J.F.; Finch, D.M. 2007. Avian nest box selection and nest success in
burned and unburned southwestern riparian forest. Journal of Wildlife Management.
71: 411–421.
Snyder, K.A.; Williams, D.G. 2000. Water sources used by riparian trees varies among
stream types on the San Pedro River, Arizona. Journal of Agricultural and Forest
Meteorology. 105: 227–240.
Sogge, M.K.; Sferra, S.J.; Paxton, E.H. 2008. Tamarix as habitat for birds: Implications
for riparian restoration in the southwestern United States. Restoration Ecology. 16:
146–154.
Stewart, I.T.; Cayan, D.R.; Dettinger, M.D. 2005. Changes toward earlier streamow
timing across western North America. Journal of Climate. 18: 1136–1155.
Stoleson, S.H.; Finch, D.M. 2001. Breeding bird use of and nesting success in exotic
Russian olive in New Mexico. Wilson Bulletin. 113: 452–455.
Stoleson, S.H.; Finch, D.M. 2003. Microhabitat use by breeding southwestern willow
ycatchers on the Gila River, New Mexico. Studies in Avian Biology. 26: 91–95.
Stoleson, S.H.; Shook, R.S.; Finch, D.M. 2000. Breeding biology of Lucy’s Warbler in
southwestern New Mexico. Western Birds. 31: 235–242.
Stromberg, J. 2001. Inuence of stream ow regime and temperature on growth rate of
the riparian tree, Platanus wrightii, in Arizona. Freshwater Biology. 46: 227–239.
Stromberg, J.C.; Rychener, T.J. 2010. Effects of re on riparian forests along a free-
owing dryland river. Wetlands. 30: 75–86.
Stromberg, J.C.; Beuchamp, V.B.; Dixon, M.D.; [et al.]. 2007. Importance of low-ow
and high-ow characteristics to restoration of riparian vegetation along rivers in arid
south-western United States. Freshwater Biology. 52: 651–679.

58 USDA Forest Service RMRS-GTR-364. 2017.
Stromberg, J.C.; Lite, S.L.; Dixon, M.D. 2010b. Effects of stream ow patterns on
riparian vegetation of a semiarid river: Implications for a changing climate. River
Research and Applications 26: 712–729.
Stromberg, J.C.; Tluczek, M.G.F.; Hazelton, A.F.; [et al.]. 2010a. A century of riparian
forest expansion following extreme disturbance: Spatial-temporal change in
Populus/Salix/Tamarix forests along the San Pedro River, Arizona, USA. Forest
Ecology and Management. 259: 1181–1189.
Strong, T.E.; Bock, C.E. 1990. Bird species distribution patterns in riparian habitats in
southeastern Arizona. Condor. 92: 866–885.
Stuever, M.C. 1997. Fire-induced mortality of Rio Grande cottonwood. Thesis.
Albuquerque, NM: University of New Mexico.
Stuever, M.C.; Crawford, C.S.; Molles, M.C.; [et al.]. 1995. Initial assessment of the
role of re in the middle Rio Grande Bosque. In: Greenlee, J., ed. Fire effects on rare
and endangered species and habitats. Coeur d’Alene, ID: International Association of
Wildland Fire: 275–283.
Summitt, A.R. 2013. Contested waters: An environmental history of the Colorado
River. Boulder, CO: University of Colorado Press.
Thomas, J.W.; Anderson, R.G.; Maser, C.; [et al.]. 1979. Snags. In: Thomas, J.W.,
ed. Wildlife habitats in managed forests: The Blue Mountains of Oregon and
Washington. Agric. Handb. 553. Washington, DC: U.S. Department of Agriculture:
60–77.
Webb, R.H.; Leake, S.A.; Turner, R.M. 2007. The ribbon of green: Change in riparian
vegetation on the southwestern United States. Tucson, AZ: University of Arizona
Press.
Wethington, S.M. 2002. Violet-crowned Hummingbird (Amazilia violiceps). In: Poole,
A., ed. The birds of North America online. Ithaca, NY: Cornell Lab of Ornithology.
http://bna.birds.cornell.edu/bna/species/688.
Whelan, R.J. 1995. The ecology of re. Cambridge: Cambridge University Press.
White, J.M.; Stromberg, J.C. 2009. Resilience, restoration, and riparian ecosystems:
Case study of a dryland, urban river. Restoration Ecology. 19: 101–111.
Whitney, J.C. 1996. The Middle Rio Grande: Its ecology and management. In:
Shaw, D.W.; Finch, D.M., tech. cords. Desired future conditions for Southwestern
riparian ecosystems: Bridging interests and concerns together; 1995 Sept. 18–22;
Albuquerque, NM. Gen. Tech. Rep. RM-GTR-272. Fort Collins, CO: U.S.
Department of Agriculture, Forest Service, Rocky Mountain Forest and Range
Experiment Station: 281–283.
Williams, V.; David, P.; Fluder, J.J., III. 2007. Middle Rio Grande Bosque Community
Wildre Protection Plan. Unpublished report on le with: Middle Rio Grande
Conservancy District, Albuquerque, NM.
Woodhouse, C.A.; Meko, D.M.; MacDonald, G.M.; [et al.]. 2010. A 1,200-year
perspective of 21st century drought in southwestern North America. Proceedings of
the National Academy of Science. 107: 21283–21288.

USDA Forest Service RMRS-GTR-364. 2017. 59
Appendix A—Hydrological Projections
We obtained projections of mean daily discharge for each of these sites, which were
made available by the USBOR (http://gis.usbr.gov/Streamflow_Projections/). These
were made to project unregulated inflow to reservoirs from 1960 to 2099 (Miller et al.
2011). The projections use bias-corrected spatial downscaled precipitation and tempera-
ture data from the World Climate Research Programme Coupled Model Intercomparison
Project phase 3 (Meehl et al. 2007). The precipitation and temperature data were
incorporated into a National Weather Service River Forecasting System (NWS RFS)
model, along with evapotranspiration demand, which was estimated using the Variable
Infiltration and Capacity model, to produce hydrological projections (Miller et al.
2011). The projections, produced by the NWS RFS model, were bias-corrected by
ensuring that means from the NWS RFS models, forced with observed temperature and
precipitation data, had the same average as models forced with projected data across the
same period. We calculated averages from 36 projections of mean daily discharge from
15 global circulation models (table A1) run under the A2 emissions scenario for each of
the 11 stream sites. We did not use these projections to estimate future daily discharge
because the amount of water that will be diverted for municipal, agricultural, and
industrial use in the future is unknown. We instead examined changes to the volume of
annual discharge, magnitude of peak discharge, and timing of peak discharge. We pro-
jected estimates of these variables for each year from 1980 to 2099. We used Akaike’s
Information Criteria (AIC) to determine whether changes in these variables over the
Table A1—Models used to project streamflow at gauge sites.
Number of
Model name Modeling center or group projections
bccr_bcm2 Bjerknes Centre for Climate Research, Norway 1
cccma_cgcm3 Canadian Centre for Climate Modelling & Analysis 5
cnrm_cm3 Météo-France / Centre National de Recherches
Météorologiques 1
csiro_mk3 CSIRO Atmospheric Research, Australia 1
gfdl_cm2 U.S. Dept. of Commerce / NOAA / Geophysical
Fluid Dynamics Laboratory 2
giss_model_e_r NASA / Goddard Institute for Space Studies 1
inmcm3_0.1 Institute for Numerical Mathematics, Russia 1
ipsl_cm4.1 Institut Pierre Simon Laplace, France 1
miroc3_2_medres Center for Climate System Research (The University of Tokyo),
National Institute for Environmental Studies, and Frontier
Research Center for Global Change (JAMSTEC) 3
miub_echo_g Meteorological Institute of the University of Bonn 3
mpi_echam5 Max Planck Institute for Meteorology, Germany 3
mri_cgcm2_3_2a.1 Meteorological Research Institute, Japan 5
ncar_ccsm3_0.1 National Center for Atmospheric Research 4
ncar_pcm1.1 National Center for Atmospheric Research 4
ukmo_hadcm3.1 Hadley Centre for Climate Prediction and Research / Met Office, UK 1

60 USDA Forest Service RMRS-GTR-364. 2017.
projection period were best described by regression models representing increases
or decreases, or by intercept-only null models representing no change (Burnham and
Anderson 2002). We analyzed output from each climate model at each gauge site
by choosing the regression model with the lowest AIC value as the best representa-
tion of changes in annual discharge, peak discharge, and date of peak discharge. We
determined the percentage of climate models predicting increases, decreases, or no
change in each variable. If output from greater than 50 percent of the climate models
fell into one of the above categories, we interpreted results as a predicted change or no
change. If output from greater than 50 percent did not fall into any of the categories,
we interpreted results as too uncertain to draw conclusions. To display changes in these
rates, we calculated the mean of each variable across the 36 model runs for the periods
of 1980 to 2009, 2010 to 2039, 2040 to 2069, and 2070 to 2099. We then calculated
the departure from the 1980 to 2009 mean for each of the three future periods. We also
estimated CVs of discharge volume, peak discharge magnitude, and peak discharge
timing to evaluate the likelihood that variability of these characteristics will change in
the future.
References
Burnham, K.P.; Anderson, D.R. 2002. Model selection and multimodel inference: A
practical information-theoretic approach. New York: Springer-Verlag Press. 496 p.
Meehl, G.A.; Covey, C.; Delworth, T.; [et al.]. 2007. The WCRP CMIP3 multi-
model dataset: A new era in climate change research. Bulletin of the American
Meteorological Society. 88: 1383–1394.
Miller, W.P.; Piechota, T.C.; Gangopadhyay, S.; Pruitt, T. 2011. Development of
streamow projections under changing climate conditions over Colorado River basin
headwaters. Hydrological Earth System Sciences 15: 2145–2164.

USDA Forest Service RMRS-GTR-364. 2017. 61
Appendix B—Application of the Cottonwood
Population Model
We compiled data from our study and others to modify the Lytle and Merritt (2004)
stochastic cottonwood population model. We constructed the model, imported data into
the model, and analyzed the output using a framework scripted in R statistical software
(R Development Core Team 2011). We estimated cottonwood density in unburned
stands using point center‐quarter data collected by Johnson and Merritt (2009) at sample
points spaced at 40-m intervals along five transects within or near our study sections. To
include rates of postfire sprouting and germination of cottonwoods, we used the rates
we estimated in the previous chapter, as well as rates from older burns along the Middle
Rio Grande (Ellis 2001; Smith et al. 2009; Stuever 1997). We estimated the area of for-
est annually burned in each section by obtaining dates and locations of fires from New
Mexico State Forestry and the Valencia and Socorro County fire departments for the
10-year period of 2002 to 2011. We visited each burn to mark the boundaries with GPS
and we used Google Earth Pro to measure the area within the perimeter to the nearest
0.5 ha.
The original model contained six stage classes: seedling, 1-year-old sapling, 2-year-
old tree, 3-year-old tree, 4-year-old tree, non-reproductive adult (5- to 10-year-old tree),
and reproductive adult (6 years or older). The model also contained stage-specific tran-
sition probabilities and a fecundity term, which was non-zero only in years when floods
of specified magnitude, duration, and time of year occurred. Density dependence is
incorporated into the model through stage-specific self-thinning rates (Lytle and Merritt
2004). We added wildfire disturbance to the model by including two stage transitions
(fig. B1). R1 through R4 represent the probability of basal resprouting after fire. We
built this transition into the model as a retrogression to stage three because, at the end
of the first postfire growing season, basal resprouts resemble 2-year-old cottonwood
trees. The second transition we added was root-suckering, represented by S1 and S2.
Figure B1—Life cycle model for cottonwoods along the Middle Rio Grande. Stages are seedlings (N1),
2- to 4-year-olds (N2-N4), non-reproductive adults (N5), and reproductive adults (N6). N is density, G is
the probability of transition to the next stage, P is the probability of remaining in that stage, R is probabil-
ity of retrogression from one stage to another as a result of postfire resprouting, F is fecundity through
seedling germination, and S is fecundity through root suckering.

62 USDA Forest Service RMRS-GTR-364. 2017.
This transition is a fecundity term in which stage classes five and six produce cloned
offspring in the second age class after fire. We assume that subadult and mature cot-
tonwoods are the only stage classes with root structures capable of producing suckers.
Root suckers begin in stage class two instead of one because they grow more rapidly
than germinated seedlings. We found that basal sprout and root sucker production
varied considerably among studies and study sites, so we programmed the model to
sample rates of percent of trees producing basal sprouts and the number of root suck-
ers per tree from the range of these values observed in our study and previous studies
(table B1).
Table B1—Stage-specific variables that were components of population parameters for cottonwoods along the
Middle Rio Grande in central New Mexico.
Cottonwood stage classes, 1 = youngest stage,
6 = oldest stage
Variable Description 1 2 3 4 5 6
Ni(0) Initial population density (cottonwoods/ha) 0 0 0 0 0 188
bi Self-thinning rate 0.029 0.10 0.91 0.66 0.2 –
ai Baseline transition probability 1 1 1 1 0.06 0.03
Fli Flood mortality in area flooded 0.97 0.33 0.02 0.02 0.0 0.0
HSwfi Top-kill rate in area burned with high
severity 1 1 1 1 1 1
MSwfi Top-kill rate in area burned with medium
severity 1 1 0.96 0.95 0.75 0.85
LSwfi Top-kill rate in area burned with low severity 1 0.9 0.8 0.7 0.4 0.6
Dri Drought mortality in drought year 0.49 0.16 0.083 0.05 0.01 0.01
BSi Range of basal sprout rate for top-killed trees 0 0 0 0.33–1 0.33–1 0.33–1
RSi Range of root suckers produced per tree
top-killed by high severity fire 0 0 0 0 0–6 0–6
We expressed the life cycle model as a projection matrix multiplied by the density
of cottonwoods in each stage class (fig. B2). In the model code, the seedling fecun-
dity value (F)—the number of germinated seeds that enter a study section during the
growing season—is zero unless certain conditions are met. The model assumes that
cottonwood seeds are present every year because seeds are transported throughout the
forest by wind and water. Because of the current lack of scouring flow at our study
sections, germination of these seeds only occurs in areas where litter and vegetation are
removed by high-severity wildfire no more than 3 years before a flood (Ellis 2001). For
each year the model was run, we calculated seedlings per section (F) using the follow-
ing equation:
F(t) = K(seed)*Flood(t)*Fire (t) * Decl(t)
where K(seed) is the maximum seedling abundance, Flood(t) and Fire (t) are the propor-
tions of the floodway that were burned by high-severity fire during the last 3 years and
flooded prior to the end of the current seed dispersal period, and Decl(t) is a function

USDA Forest Service RMRS-GTR-364. 2017. 63
that determines the number of seedlings that germinate and survive the first growing
season, based on timing of peak discharge and drawdown rate (Lytle and Merritt 2004).
We selected flood thresholds based on our observations of flood extent and stream gage
measurements from 2003 to 2013. We did not collect data on age class distributions at
our study section, so we set an initial age distribution of entirely mature trees to reflect
conditions reported in many areas (Howe and Knopf 1991). This distribution, however,
is not necessarily reflective of the entire study area.
We used the model to project populations of trees over a 100-year period in each
section, with flood, fire, and drought occurring as stochastic events that can occur dur-
ing each time step of 1 year. Spatial domain of the model included the north and south
study sections. During each time step, a section could experience drought, and all or
parts of a section could experience flooding, or neither drought nor flooding could oc-
cur. To assign flood and drought status to years of each section’s model projections, we
used streamflow values from U.S. Geological Survey discharge data for the Middle Rio
Grande at Albuquerque (USGS gauge number 08330000). We randomly sampled, with
replacement, peak mean daily discharge from each year between 1975 (the year follow-
ing the completion of Cochiti Dam) and 2012. We then adjusted the sampled magnitude
and timing of peak discharge to reflect the projected change under three CMIP3 carbon
emission scenarios. These scenarios were A1B, which represented moderate increases
in emissions; A2, which represented large increases; and B1, which represented small
increases.
Figure B2—Projection matrix for cottonwoods along the Middle Rio Grande, New Mexico.
Stages are seedlings, 2- to 4-year-olds, non-reproductive adults, and reproductive adults. N
is density, G is the probability of transition to the next stage, P is the probability of remaining
in that stage, R is probability of retrogression from one stage to another as a result of post-
fire resprouting, F is fecundity through seedling germination, and S is fecundity through root
suckering.

64 USDA Forest Service RMRS-GTR-364. 2017.
For a given projection year, the section experienced drought if the sampled dis-
charge value was below the section’s drought threshold. The section would experience
complete flooding if the peak discharge was above the threshold for the floodway to
be fully inundated. Partial flooding of the section occurred if the sampled discharge
was greater than the bankfull discharge threshold (threshold above which portions of
the floodway are inundated), but below the full floodway inundation threshold. In this
situation, we calculated the percent difference between the sampled discharge and the
full inundation threshold to calculate the percentage of the section that was flooded that
year.
In addition to hydrological status, we assigned a wildfire status to each section for
each year of the projection period. In the model, portions of a section could be burned
by wildfire, but when a given percentage of a section was burned, it could not be
burned again until 5 years had passed, allowing fuel to reaccumulate. For each projec-
tion year, we first sampled a percentage of area burned, with replacement, from the
10-year historical record of the section. If a percentage greater than 0 was selected,
we then determined what proportions of this percentage would be burned with high,
medium, and light severity. We randomly generated three values between 0 and 1 that
summed to 1. We multiplied the percent of area burned value by each random propor-
tion to arrive at the percentage of the section to be burned by each severity class. Next,
we multiplied the area available to be burned (this is the area within the section that had
not been burned during the last 5 years) by the three severity percentages to produce
the number of hectares that would be burned with high, medium, and light severity
that year. We assumed that, during projection year one, neither section had burned
during the previous 5 years. During each projection year, F is greater than 0 only in the
portions of a section where high-severity fire occurred during that year or during the
previous 2 years and flooding occurred during the current year. In addition to fecundity,
the values of other stage-specific variables were influenced by flood, fire, and drought
(table B2). During each projection year, the model code calculates the percentage of
trees that will be affected by drought, flood, and fire. Transitions and stage abundances
are then determined for the year. We conducted 1,000 simulated projections for each
section and calculated 95 percent confidence intervals and range for stage-class densi-
ties during each year of the projection.

USDA Forest Service RMRS-GTR-364. 2017. 65
References
Ellis, L.M. 2001. Short-term response of woody plants to re in a Rio Grande riparian
forest, central New Mexico, U.S.A. Biological Conservation. 97: 159–170.
Howe, W.H.; Knopf, F.L. 1991. On the imminent decline of Rio Grande cottonwoods in
central New Mexico. The Southwestern Naturalist. 36: 218–224.
Johnson, B.; Merritt, D. 2009. The effects of wildre on native tree species in the
Middle Rio Grande bosques of New Mexico. Fort Collins, CO: Colorado State
University. 43 p.
Lytle, D.A.; Merritt, D.M. 2004. Hydrologic regimes and riparian forests: A structured
population model for cottonwood. Ecology. 85: 2493–2503.
R Development Core Team. 2011. R: A language and environment for statistical
computing. Vienna, Austria: R Foundation for Statistical Computing. ISBN
3-900051-07-0. URL. http://www.R-project.org.
Smith, D.M.; Finch, D.M.; Gunning, C.; [et al.]. 2009. Post-wildre recovery of riparian
vegetation during a period of water scarcity in the southwestern U.S. Fire Ecology. 5:
38–55.
Stuever, M.C. 1997. Fire-induced mortality of Rio Grande cottonwood. Thesis.
Albuquerque, NM: University of New Mexico.
Table B2—Parameters used in projection models for populations of cottonwoods along the Middle Rio Grande in central New
Mexico.
Parameter Description Components
F Fecundity: Number of seedlings Flood area, flood timing, stage height
established in year t decline, high-severity wildfire area
G Stage-specific progression probability Stage-specific transition rate, self-thinning rate, flood
mortality, wildfire mortality, drought mortality
P Probability of remaining in stage class 5 or 6 Stage-specific transition rate, flood mortality, wildfire
mortality, drought mortality
R Probability of retrogression from stages 3–6 to Wildfire area, fire severity, top-kill rate, basal
stage class 3 after topkill by wildfire sprout rate
S Number of root suckers produced by adult trees Wildfire area, fire severity, root-suckering rate

In accordance with Federal civil rights law and U.S. Department of Agriculture (USDA)
civil rights regulations and policies, the USDA, its Agencies, ofces, and employees, and
institutions participating in or administering USDA programs are prohibited from discrimi-
nating based on race, color, national origin, religion, sex, gender identity (including gender
expression), sexual orientation, disability, age, marital status, family/parental status, income
derived from a public assistance program, political beliefs, or reprisal or retaliation for prior
civil rights activity, in any program or activity conducted or funded by USDA (not all bases
apply to all programs). Remedies and complaint ling deadlines vary by program or incident.
Persons with disabilities who require alternative means of communication for program in-
formation (e.g., Braille, large print, audiotape, American Sign Language, etc.) should contact
the responsible Agency or USDA’s TARGET Center at (202) 720-2600 (voice and TTY) or
contact USDA through the Federal Relay Service at (800) 877-8339. Additionally, program
information may be made available in languages other than English.
To le a program discrimination complaint, complete the USDA Program Discrimination
Complaint Form, AD-3027, found online at http://www.ascr.usda.gov/complaint_ling_
cust.html and at any USDA ofce or write a letter addressed to USDA and provide in the
letter all of the information requested in the form. To request a copy of the complaint form,
call (866) 632-9992. Submit your completed form or letter to USDA by: (1) mail: U.S.
Department of Agriculture, Ofce of the Assistant Secretary for Civil Rights, 1400 Indepen-
dence Avenue, SW, Washington, D.C. 20250-9410; (2) fax: (202) 690-7442; or (3) email:
program.intake@usda.gov.
To learn more about RMRS publications or search our online titles:
www.fs.fed.us/rm/publications
www.treesearch.fs.fed.us