Perennial Agroenergy Feedstocks as En Route Habitat
for Spring Migratory Birds
Bruce A. Robertson & Douglas A. Landis &
T. Scott Sillett & Elizabeth R. Loomis & Robert A. Rice
Published online: 30 September 2012
# Springer Science+Business Media New York 2012
Abstract Increased production of bioenergy crops in North
America is projected to exacerbate already heavy demands
upon existing agricultural landscapes with potential to im-
pact biodiversity negatively. Grassland specialist birds are
an imperilled avifauna for which perennial-based, next-
generation agroenergy feedstocks may provide suitable hab-
itat. We take a multi-scaled spatial approach to evaluate the
ability of two candidate second-generation agroenergy feed-
stocks (switchgrass, Panicum virgatum, and mixed grass?
forb plantings) to act as spring migratory stopover habitat
for birds. In total, we detected 35 bird species in mixed
grass?forb plantings and switchgrass plantings, including
grassland specialists and species of state and national con-
servation concern (e.g., Henslow?s Sparrow, Ammodramus
henslowii). Some evidence indicated that patches with
higher arthropod food availability attracted a greater diver-
sity of migrant bird species, but species richness, total bird
abundance, and the abundance of grassland specialist spe-
cies were similar in fields planted with either feedstock.
Species richness per unit area (species density) was relative-
ly higher in switchgrass fields. The percent land cover of
forest in landscapes surrounding study fields was negatively
associated with bird species richness and species density.
Habitat patch size and within-patch vegetation structure
were unimportant in predicting the diversity or abundance
of spring en route bird assemblages. Our results demonstrate
that both switchgrass and mixed grass?forb plantings can
attract diverse assemblages of migrant birds. As such, in-
dustrialized production of these feedstocks as agroenergy
crops has the potential to provide a source of en route habitat
for birds, particularly where fields are located in relatively
unforested landscapes. Because industrialization of cellulos-
ic biomass production will favor as yet unknown harvest
and management regimes, predicting the ultimate value of
perennial-based biomass plantings for spring migrants
remains difficult.
Keywords Biofuels . Agriculture . Grassland birds .Habitat
selection . Agroecology . Switchgrass
Introduction
Demand for new sources of sustainable energy has spurred
interest in the production of dedicated field-grown biomass
(hereafter ?agroenergy?) crops. Such crops can be converted
to liquid fuels via cellulosic technology or to electricity by
combustion. The perceived potential for agroenergy crops to
provide a significant fraction of future liquid fuel demand is
sufficiently high that European and American governments
have set mandated production goals [1, 2]. Such policies are
projected to expand greatly the acreage of land under culti-
vation and intensify the management of land currently under
cultivation [3]. Together, land-use change and agricultural
intensification are currently regarded as the greatest current
global threat to the maintenance of biodiversity and the
ecosystem services it supports [4, 5]. In North America,
land-use changes associated with the expansion of contem-
porary agroenergy crops (e.g., corn) are generally expected
B. A. Robertson : T. S. Sillett : R. A. Rice
Migratory Bird Center, Smithsonian Conservation Biology
Institute, National Zoological Park,
Washington, DC 20013, USA
B. A. Robertson (*)
Division of Science Mathematics and Computing, Bard College,
30 Campus Drive,
Annandale-on-Hudson, NY 12504, USA
e-mail: broberts@bard.edu
B. A. Robertson : D. A. Landis : E. R. Loomis
DOE Great Lakes Bioenergy Research Center,
Michigan State University,
East Lansing, MI 48824, USA
Bioenerg. Res. (2013) 6:311?320
DOI 10.1007/s12155-012-9258-3
Author's personal copy
to reduce biodiversity [6?8], yet next-generation perennial-
based crops like switchgrass (Panicum virgatum) or mixed
grass?forb combinations [9] may provide new acreage of
suitable habitat for animals native to North American tem-
perate grasslands [6, 8].
North American grassland specialist birds have experi-
enced dramatic long-term population declines [10] and rep-
resent an important component of native biodiversity likely
to be impacted by the expansion of agroenergy [6, 8, 11,
12]. Research into halting population declines has focused
heavily on understanding the factors shaping the demogra-
phy of breeding and wintering populations [reviewed in 13],
while the migratory ecology and habitat requirements of this
imperiled avifauna have received scant attention.
Migration represents a critical portion of the life cycle of
temperate migratory birds because they can spend nearly
one third of their life span in transit with at least one species
apparently experiencing ~85 % of adult mortality during
migration [14]. Recent evidence shows that switchgrass
and mixed grass?forb plantings can provide en route stop-
over habitat for diverse bird assemblages during the fall
migratory period [15]. However, the value of these fields
as en route habitat during spring, when habitat structure and
food availability may differ dramatically, remains unclear.
As next-generation agroenergy crops are poised to transform
North American agricultural landscapes, understanding their
ability to provide stopover habitat for migratory birds will
be central to projecting the impact of expanded production
on the conservation of grassland bird populations.
Our goal is to address this information gap directly by
comparing species richness, species density, and abundance
of the entire migratory bird community and of grassland
specialist species between two candidate agroenergy feed-
stocks likely to provide stopover habitat to migratory grass-
land birds: switchgrass and mixed grass?forb plantings. We
first ask if these planting types differ in the migratory bird
communities they attract during the spring migratory period,
and then investigate how food availability, vegetation struc-
ture, and composition at multiple spatial scales (microhabi-
tat, patch, and landscape) shape the distributions of spring
migrant birds in fields. We base our predictions about the
responses of grassland bird communities to perennial feed-
stocks on bird?habitat associations during the fall migratory
period and in this same study system [15]. Grassland spe-
cialist species exhibit well-understood species-specific pref-
erences for habitat structure during the breeding season
[reviewed in 16] that appear to hold during the fall migra-
tory period when species richness is similar between feed-
stock types and is unrelated to the biomass or diversity of
arthropod food [15]. Following fall patterns, we predict that
grassland specialists will avoid plantings within highly for-
ested landscapes and prefer larger habitat patches [15] af-
fecting overall species richness and species density. Because
the factors shaping habitat use may change seasonally, we also
investigate the following factors known to shape distributions
of migratory birds in other temperate North American ecosys-
tems: food availability [reviewed in 17], habitat complexity
[18], patch size [19], and the structure and composition of the
surrounding landscape [20].
Methods
Study Design and Site Selection
We identified established fields sown with mixed grass?forb
plantings or switchgrass (n012, each) throughout southern
Michigan (Fig. 1). Grassland bird species specialize in hab-
itats differing in their physical structure [16]; however,
agroenergy production systems favor feedstocks and man-
agement techniques that maximize biomass, thereby reduc-
ing variation in habitat structure. Production of perennially
based agroenergy feedstocks (e.g., switchgrass) by farmers
or other land-managers (e.g., switchgrass) is still extremely
rare throughout the Midwest, making it impossible to study
crops intensively managed for biomass production. Instead,
we opted to investigate vegetation structural attributes most
likely to be affected by crop selection and management
(e.g., harvest) and relevant to selection of en route habitat
by migrant birds: (1) vegetation height and density and (2)
within-field variation in structure.
In 2010, we searched for fields planted in switchgrass and
mixed grass?forb plantings throughout the southern peninsula
of the state of Michigan, ultimately locating 65 candidate
fields. Fields were managed primarily as wildlife habitat or
as native community restorations. For this reason, switchgrass
fields were rarely strict monocultures.We visually inspected all
candidate sites prior to the onset of research in order to assess
vegetation heterogeneity. Moreover, we gathered data about
the percent forest vs. row crop in a 1,500-m radius surrounding
each site from previous studies [12, 15]. The two sets of
information were combined in a non-mathematical fashion in
order to select a set of study sites that maximized variation
along these two axes. Four switchgrass fields (two of which
were managed for biomass) and one mixed grass?forb planting
were managed with fire during the previous fall. In order to
evaluate the importance of patch size in shaping avian com-
munities, we also selected patches to vary as widely as possible
in size (mixed grass?forb, 3.2?70.8 ha; switchgrass, 2.0?
36.42 ha). Study fields were located a minimum distance of
7 km from each other.
Bird Surveys
We surveyed the bird community associated with study
fields in the spring of 2010, making three visits to each
312 Bioenerg. Res. (2013) 6:311?320
Author's personal copy
patch: (1) April 1?8, (2) April 14?21, and (3) April 27?May
4. We conducted bird surveys during the first 3 h after
sunrise. Species richness and abundance were estimated
based on two survey techniques: fixed-width transects and
area searches. Fixed-width transects [21] were 100 m long
by 25 m wide. Because grassland birds are difficult to detect
when not singing, we employed a rope drag technique in
which a 25-m-long rope connecting two observers was
dragged across vegetation [15]. This technique was designed
to increase the detectability of grassland birds during the
nonbreeding season and increase observer efficiency and bird
detectability over fixed-distance point counts [22]. We identi-
fied birds visually when perched or in flight when flushed
from within and aurally by species-specific calls. Individuals
that could not be assigned to species were recorded as ?un-
known? or identified to the genus or family level (e.g., Ammo-
dramus spp.). These observations were only used in estimates
of community-wide abundance.
In order to obtain representative samples of bird assemb-
lages in fields differing in area without pseudo-replicating
[reviewed in 23], we varied the number of transects sampled
per field, aggregating information at the patch-scale prior to
any analysis. The smallest patches contained a single tran-
sect, while the number of transects surveyed per patch
increased with patch size up to a maximum of six. We
oriented and surveyed transects in a linear series that ran
through the geographic center of each with no transect
ending closer than 50 m from a patch edge. We calculated
median values of species richness per transect (hereafter
species density) and abundance at the patch scale, combin-
ing information from all site visits. Many grassland bird
species in this region are ?area-sensitive,? preferring to
settle and breed in larger patches of habitat [24]. Conse-
quently, in addition to the well-understood species-area
effect in which more species are found in larger patches
[24], species richness per unit area (i.e., ?species density?)
increases with patch size [24]. We used species density as a
metric to test the hypothesis that avian communities are also
area-sensitive during the spring migratory period.
To estimate patch-scale species richness, we used area
searches to survey portions of each field not covered by
transects. To maintain observer effort proportional to the
size of each patch, observers walked at a regular pace
through each field in a systematic pattern such that one
observer passed within 75 m of every point in a field exactly
once. Species detected during strip-transect surveys, includ-
ing those detected at a distance of >50 m, were pooled with
detections from area searches to provide an estimate of total
bird species richness within each patch.
Within-Patch Habitat Structure
During the second site visit, we characterized vegetation
structure of perennial plantings within each transect to deter-
mine how microhabitat gradients may affect spatial distribu-
tions of birds. We randomly selected five nonoverlapping
sampling points within each transect at which we recorded
vertical density of vegetation and canopy coverage. Vertical
density [an index of biomass, 25] was quantified bymeasuring
the minimum height of visual obstruction from 4 m in each
cardinal direction from a Robel pole at a height of 1 m [26].
We estimated canopy coverage on the basis of nonoverlapping
percentages of forbs and grass using a Daubenmire quadrat
viewed from 1.5 m directly above [27]. We assigned cover
estimates an index number corresponding to a range of vege-
tation coverage (100?5 %, 205?25 %, 3025?50 %, 4050?
75 %, 5075?95 %, and 6095?100 %) and computed mean
values of microhabitat variables at the patch scale. Variation in
Fig. 1 Map of the study region
in the southern peninsula of
Michigan. Locations of 12
mixed grass?forb (filled circles)
and 12 switchgrass (open circles)
study sites are indicated
Bioenerg. Res. (2013) 6:311?320 313
Author's personal copy
vegetation density within each patch was used to calculate an
index of habitat heterogeneity for that patch relevant to grass-
land birds [28]:
Vegetation heterogeneity index0
Pn
i?1
Max xi"Min xi? ?
Pn
i?1
xi
Where x is the minimum height of visual obstruction on a
Robel pole at of four subsamples at each sampling unit, n is
the number of sampling units in each study field, of which
there are n0# of transects?5i sampling units per field.
Patch- and Landscape-Scale Variables
Grassland birds commonly avoid settling in habitat patches
more extensively bordered by forest during the breeding sea-
son [reviewed in 23] and during autumn migration [15]. We
estimated the percent forest composition within a 1.5-km
radius surrounding study sites using ArcGIS 9.3 [29] and the
2009 Cropland Data Layer (56 m resolution) [30]. The accu-
racy of land-use categories was verified during site visits.
Arthropod Richness and Biomass
The distribution of migratory birds among and within patches
is commonly linked to arthropod food availability [17]. We
estimated the relative availability of terrestrial arthropod prey
among stopover sites via sweep net samples of above-ground
vegetation near the geographic center of each field during the
second site visit each year. Two within-patch sweep sample
transects were taken, beginning at a distance of 50 m in
opposite directions from the field center on a north?south axis
and consisting of 50 sweeps taken while slowly moving to-
ward the plot center. The two within-field samples were com-
bined and sealed in plastic bags and transferred to 90% ethanol
solution for storage. We identified individuals to the family
level. We estimated individual mass using published length-
regression estimates [31, 32] and computed total arthropod
biomass at the patch level. Finally, we estimated patch-scale
estimates of arthropod family richness using the Chao 2 as-
ymptotic richness estimators in the program EstimateS [33].
Statistical Analysis
We used a multi-model inference approach to determine the
relative importance and effect size of seven environmental
variables and feedstock type (Table 2) in explaining varia-
tion in (1) the community-wide species richness and (2)
species density, (3) total bird abundance, and (4) abundance
of grassland specialist/obligate bird species [sensu 34], (see
Table 1). Because birds may weigh microhabitat structure,
patch size, and land cover composition simultaneously in
making settlement decisions, we consider potential interac-
tions between feedstock and patch size, and feedstock and
forest cover in predicting avian diversity and abundance
metrics. We log transformed patch size prior to analysis
because species richness generally increases with patch size
in an asymptotic and nonlinear fashion [35]. Vegetation
density was square root transformed to achieve normality.
Because avian community variables followed Poisson
distributions typical of count data, and were overdispersed,
we modeled arthropod community variables using either
negative binomial or quasi-Poisson regression with a log
link function [36]. We began by entering all independent
variables and the above-mentioned interaction terms into a
full generalized linear model. Next, all possible subsets of
the that model were analyzed using the multimodel infer-
ence package, MuMin in R v. 2.11.1 [37]. We used MuMin
to estimate model coefficients and bias-corrected quasi-
AICc values, an adaptation of Akaike?s Information Criteri-
on (AIC) that accounts for potential overdispersion in count
data and contains a small sample size adjustment [38]. We
used differences between the QAICc-best model and other
candidate models (?QAICc) to calculate Akaike weights
(?+) for each candidate. We then summed the weights of
ranked models to construct a 90% confidence set of candidate
models. Akaike weights were then recalculated for each mod-
el in the 90 % confidence set and used to calculate model-
averaged parameter estimates and summed variable weights
for each variable in the global model. Finally, Akaike weights
for classes of variables were summed to assess the relative
importance of different characteristics associated with study
sites in explaining avian community metrics (Table 3). A full
discussion of the information-theoretic approach to model/
variable selection used here can be found in [36].
Because study fields in closer physical proximity may
have similar characterstics, we computed the Moran?s I test
for residual spatial autocorrelation (60 km distance band) for
the best models using the spatial dependence package,
spdep in R [39]. We found no evidence of autocorrelation
(0.08>I>0.03; P>0.32 for all tests).
Results
Bird and Arthropod Community Composition
We identified 97.4 % of the 301 individuals detected within
transects to the species level. In total, we detected 35 bird
species, with equal total species richness in both feedstocks
(n020) and similar obligate species richness in each
314 Bioenerg. Res. (2013) 6:311?320
Author's personal copy
Table 1 Bird species (n036) detected in during spring migration within mixed grass?forb and switchgrass fields (n012, each) in southern
Michigan
Species Switchgrass Mixed grass?forb
American Goldfinch (Spinus tristus) X X
American Kestrel (Falco sparverius) X
American Robin (Turdus migratorius) X X
Barn Swallow (Hirundo rustica) X X
Brown-headed Cowbird (Molothrus ater) X X
Bobolink (Dolichonyx oryzivorus)a X X
Clay-colored Sparrow (Spizella pallida) X X
Chipping Sparrow (Spizella passerina) X X
Common Grackle (Quiscalus quiscula) X X
Common Yellowthroat (Geothlypis trichas) X X
Eastern Bluebird (Sialia sialis) X X
Eastern Kingbird (Tyrannus tyrannus) X X
Eastern Meadowlark (Sturnell magna)a X X
European Starling (Sturnus vulgaris) X
Field Sparrow (Spizella pusilla) X X
Grasshopper Sparrow (Ammodramus savannarum)a, b X X
Henslow's Sparrow (Ammodramus henslowii)a,c,d X
Horned Lark (Eremophila alpestris) X
Killdeer (Charadrius vociferus) X
Mallard (Anas platyrhynchos) X X
Mourning Dove (Zenaida macroura) X
Northern Flicker (Colaptes auratus) X X
Northern Harrier (Circus cyanus)a,b X
Ring-necked Pheasant (Phasianus cholchicus) X X
Red-tailed Hawk (Buteo jamaicensis) X X
Red-winged Blackbird (Agalaius phoeniceus) X X
Savannah Sparrow (Passerculus sandwichensis)a X X
Sedge Wren (Cistothorus platensis)a X X
Song Sparrow (Melospiza melodia) X X
Swamp Sparrow (Melospiza georgiana) X X
Tree Swallow (Tachycineta bicolor) X X
Turkey Vulture (Cathartes aura) X X
Unknown sparrow spp. X
Spizella sparrow spp. X
Vesper Sparrow (Pooecetes gramineus)a X X
White-crowned Sparrow (Zonotrichia leucophrys) X
Wilson's Snipe (Gallinago delicata) X X
Yellow Warbler (Dendroica petechia) X
Totals 33 (5) 33 (6)
Species totals in parentheses represent obligate grassland species on April 1?May 4, 2010. Species names in bold and species totals in parentheses
represent those for which published research indicates they breed exclusively within grassland habitats
a Obligate grassland species
bMichigan species of concern
cMichigan endangered species
d Species appearing on the Audubon watchlist (National Audubon Society 2007) [58]
Bioenerg. Res. (2013) 6:311?320 315
Author's personal copy
planting type (switchgrass05, mixed grass?forb06, Ta-
ble 1). Several species of high state and national conserva-
tion status (e.g., Grasshopper Sparrow, Henslow?s Sparrow,
and Northern Harrier) were detected. We identified 222
arthropods from 101 families. Terrestrial arthropod biomass
was greater in mixed grass?forb plantings, but richness was
similar between feedstocks (Table 2).
Vegetation Structure, Patch Size, and Land Cover
Study fields varied in size from 2.0 to 70.8 ha. Microhabitat
heterogeneity, vegetation density, percent forest cover sur-
rounding study fields, and the size of study fields did not
differ between switchgrass and mixed grass plantings
(Table 2). On average, the percentage of ground cover in
Table 2 Comparisons of seven explanatory variables and avian community metrics between mixed grass?forb (n012) and switchgrass fields (n0
12) in southern Michigan
Variable Mixed grass?forb Switchgrass t22 P
Within-patch
Microhabitat heterogeneity index (0?2) 1.03 (0.13) 1.00 (0.17) 0.16 0.55
Mean minimum height of Robel pole intercept (cm) 20.0 (2.3) 41.1 (7.1) 2.83 0.01
Mean ground cover index broad-leaved plants (1?6) 1.5 (0.2) 1.3 (0.2) 0.75 0.46
Arthropod biomass (mg/sample) 2.43 (0.80) 0.67 (0.27) 2.07 0.05
Arthropod richness (# families) 11.4 (3.1) 5.1 (1.8) 1.70 0.10
Patch and landscape scale
Patch size (ha) 16.4 (5.4) 7.6 (2.7) 1.45 0.16
Surrounding land cover in forest (1.5-km radius) (%) 31.0 (0.1) 31.0 (0.1) 0.15 0.98
Avian community metrics
Species richness 8.6 (0.7) 6.9 (1.1) 1.27 0.22
Species density (species/ha) 6.6 (1.6) 12.2 (2.0) 1.60 0.13
Community-wide abundance 3.5 (0.8) 0.5 (0.2) 1.55 0.13
Obligate grassland species abundance 1.1 (0.4) 0.5 (0.2) 1.32 0.20
Mean values and standard errors (SE) are given for parameter estimates along with critical and P values of independent samples t tests
Table 3 Parameter estimates (??unconditional SE) from the QAICc-
best model predicting the (1) species richness, (2) species density, (3)
abundance of all birds, and (4) abundance of obligate grassland bird
species exploiting switchgrass and mixed grass?forb plantings as
spring migratory stopover habitat in southern Michigan
Parameter Species richness Species density Total abundance Obligate abundance
? (SE) ?+ ? (SE) ?+ ? (SE) ?+ ? (SE) ?+
Intercept 1.97 (0.42)*** 1.11 (0.50)*** 1.46 (0.45)*** 0.07 (0.96)
Feedstocka 0.21 2.00 (0.28)* 0.44 0.13 0.27
Habitat heterogeneity 0.18 0.17 0.14 0.15
Forb cover 0.44 0.16 0.14 0.29
Arthropod biomass (mg/sample) 16.80 (24.8)** 0.38 0.15 0.14 0.12
Arthropod richness (families/sample) 0.37 0.14 0.13 0.14
Vegetation density (ht. of pole intcpt, cm) 0.13 0.23 0.13 0.13
Log patch size (ha) 0.25 0.28 0.15 0.14
% forest cover (1.5-km radius) ?0.87 (0.64)*** 0.76 ?1.31 (0.91)* 0.81 0.14 0.43
Feedstock?Log patch size 0.03 0.04 0.13 0.03
Feedstock?% forest cover 0.00 0.02 0.13 0.02
Variable weights (0???1) quantify relative support for each variable across the entire set of models, which contained all possible combinations of
predictors. Variable weights obtained by summing ? for all models in the set that included a given variable; weights approximate the likelihood a
given variable will be included in the model in repeated runs of the experiment. Blank spaces indicated a parameter was not included in the QAICc-
best model
*0.01?P?0.05; **0.001?P>0.01; ***P<0.001 (significance codes for likelihood ratio ?2; critical values)
aMixed grass?forb was the reference category
316 Bioenerg. Res. (2013) 6:311?320
Author's personal copy
broad-leaved plants was low, averaging between 5 and 25 %
in both planting types.
Avian Community Metrics
Avian species richness was primarily affected by forest
cover and arthropod biomass (Table 3). The QAICc-
best model (?01.52) predicted a significant increase in
richness in patches with higher arthropod biomass and in
landscapes with lower forest cover (Fig. 2a). Higher bird
species richness was also associated with greater terrestrial
arthropod biomass. Forest cover received strong support
across the entire model set (?+00.76) indicating this variable
was consistently included in the most predictive models.
Support for arthropod biomass and other variables was gen-
erally lower (?+?0.44; Table 3), suggesting they were less
likely to be included in the model in repeated runs of the
experiment.
Species density was affected by feedstock type and
forest cover. The QAICc-best model (?01.17) predicted
roughly 100 % increase in species density in switch-
grass fields compared mixed grass?forb plantings
(Tables 2 and 3). Patches with an increasing percentage
of forest cover surrounding them exhibited decreasing
species density (Fig. 2b). These effects were moderately
to strongly supported (percent forest cover, ?+00.81;
feedstock, ?+00.64) compared to other predictors
(?+?0.23; Table 3).
Global models predicting total bird abundance and that
of obligate grassland species fit the data reasonably well
(total abundance, ?02.43; obligate abundance, ?01.14),
but QAICc-best models for each analyses contained only
the intercept. Support for the importance of predictor
variables and interactions were poor for all variables
across both model sets and for both abundance measures
(?+?0.33; Table 3). Support for models including inter-
actions between feedstock type and either patch size or
forest cover were extremely poor (?+?0.13; Table 3)
across models sets for all for avian community metrics
we examined.
Discussion
This study represents one of the first to assess the relative
biodiversity value of alternative next-generation agroenergy
feedstocks in North America empirically [also see 12, 15,
40], and is the first to investigate the habitat associations of
grassland birds during spring migration. Most studies inves-
tigating how birds select migratory stopover habitat have
focused efforts at a single small spatial scale [20], and
research has taken place almost exclusively within forested
ecosystems. We simultaneously compared the influence of
factors operating at different spatial scales [see also 20, 41],
focusing our efforts on the entire migratory bird community
and on grassland specialists to understand better how factors
intrinsic and extrinsic to agroenergy plantings may more
broadly affect the ability of agricultural landscapes to act
as migratory stopover habitat. We found some support for
the contention that switchgrass and mixed grass?forb feed-
stocks differ in their suitability as en route habitat during
spring. However, habitat composition at the landscape scale
appeared to be more broadly important than differences
among feedstocks in attracting diverse bird assemblages to
agroenergy feedstocks during spring migratory movements.
We detected bird species of state and continental conserva-
tion concern in fields planted with both feedstock types. The
total number of bird species detected in switchgrass and mixed
grass?forb plantings was identical, and the number of obligate
grassland bird species detected was similar. Results from our
multi-model inference analysis indicate that switchgrass fields
supported greater species density (species richness per unit
area), but that bird species richness in switchgrass fields was
similar to that of mixed grass?forb plantings.
Vegetation heterogeneity in switchgrass and mixed
grass?forb plantings was similar, but switchgrass fields
had greater average vegetation density than mixed grass?
forb fields. This difference in average vegetation density
between feedstock types, however, was not linked to the
elevated species density we identified in switchgrass fields.
These results contrast with bird?microhabitat relationships
identified during the fall migratory period in this study
Fig. 2 Partial regressions of
bird species richness vs. percent
forest cover surrounding
plantings (1.5-km radius) (a),
and bird species density (species
richness/hectare) vs. percent
forest cover surrounding
plantings (1.5-km radius) (b).
Parameter estimates are based on
model-averaged values
Bioenerg. Res. (2013) 6:311?320 317
Author's personal copy
system in which vegetation structural heterogeneity was the
single most important within-patch factor shaping both bird
community composition and abundance [15].
Seasonal differences in habitat structure may explain
these apparently conflicting results. For example, habitat
heterogeneity is commonly associated with species richness
of fall migrants in forested systems [e.g., 18, 19, 42?44],
and may be particularly important in attracting species with
a range of habitat structural preferences to grasslands during
fall when within-field heterogeneity is relatively low (fall0
0.35, SD00.18; spring01.0, SD00.53) [this study, and 15
with comparable methodology]. Though seasonal changes
in habitat use at the species and community level have been
noted in temperate forested systems [42, 45?47], they have
not been explicitly linked to variation in microhabitat struc-
ture. This lack of a consistent between-season relationship
between bird diversity and feedstock vegetation character-
istics might result from the cumulative effects of variable
species-specific responses to habitat structure [48]. Alterna-
tively, it could reflect the region- or habitat-specific depen-
dence of habitat structure in mediating (1) availability of or
accessibility to food [49], (2) resource competition within or
between species [50], or (3) risk of predation [51].
When selecting breeding habitat, many grassland special-
ists avoid habitat patches within highly forested landscapes
[49, 52], and this forest avoidance may be a mechanism
shaping area sensitivity in breeding grassland specialist
birds [23]. In addition, the richness of en route migratory
bird assemblages has been known to increase with patch
size in forested [19, 50] and grassland habitats [15]. In this
study, forest cover (1.5-km radius scale) was the single most
important factor we identified in predicting reduced species
richness and species density in both feedstock types, inde-
pendently of patch size. This indicates that a significant
portion of the avian species assemblage that exploits feed-
stock plantings exhibits forest avoidance behavior in select-
ing patches of stopover habitat during spring migration, and
that patch size per se is relatively unimportant.
This result is somewhat surprising given that these same
bird community metrics were not linked to forest avoidance
in Robertson et al.?s [15] study conducted in this system
during the fall migratory period. In that study, some indi-
vidual birds detected in mixed grass?forb and switchgrass
plantings were likely remaining in their natal habitat patches
and staging for migration. Consequently, the inclusion of
post-breeding individuals in models of fall migratory bird
assemblages may have confused the abundance of individ-
uals associated with particular breeding habitats with the
attractiveness of these patches for migrants. Even so, this
does not negate the potential importance of mixed grass?
forb and switchgrass habitats for post-breeding individuals
staging for migration. It does potentially confound factors
shaping the distribution of these individuals from that of
those stopping over during the migratory journey. In con-
trast, the present study was conducted during a time period
when the only non-migrant bird species present should be a
low abundance of a few overwintering resident species
(mostly Song Sparrows and American Goldfinches) [53].
Collectively, and consistent with the results of studies of this
system during other time periods [12, 15], our results imply
that cultivation of perennial-based agroenergy crops in less
forested landscapes will attract a greater diversity of migratory
bird species. Moreover, it appears that en route grassland
specialist species will benefit more from this crop placement
strategy during the fall [15] than spring migratory period.
Spatial variation in the biomass or diversity of arthropod
prey is commonly associated with the distribution of migrat-
ing birds [reviewed in [18, 20, 48]. This is not surprising given
the energetic demands of migration and the need to rapidly
refuel [54].We found weak evidence that bird species richness
was positively related to arthropod food availability, paralle-
ling a similar relationship during the fall migratory period in
this system [15]. We captured 40 times fewer individual
arthropods per patch than Robertson et al. [15] did during
the fall migratory period in this system, representing a mar-
ginal mean biomass roughly six times smaller (mixed grass?
forb comparison only, spring02.4 mg, SD02.7; fall0
15.1 mg, SD012.0). Yet, variation in arthropod biomass dur-
ing the fall migratory period was not linked to distributions of
migrants [15]. Again, potential relationships may have been
obscured by the combining observations of post-breeding and
actively migrating individuals in that study. Alternatively,
birds may more closely track the distributions of arthropods
during spring when that food source is rarer. Seeds, too, may
represent an important caloric resource important to many
grassland migrants whose distribution could influence settle-
ment decisions during spring migration, but that we did not
examine in this study.
We draw inferences from bird?habitat relationships based
on extant variation in within-patch habitat structure, yet
most of the switchgrass and mixed grass?forb fields we
studied were not actively managed for agrofuel production.
Moreover, we have not investigated how the migratory
birds? assemblages associated with perennially based feed-
stocks compare with those of contemporary agrofuel crops
like corn, or if birds use corn fields in spring at all. In
addition to the implementation of more intensive cultivation
methods, industrialization of perennially based agroenergy
crops will likely include chemical applications and the se-
lection of high-biomass genotypes that have potential to
reduce plant species diversity [55] and structural heteroge-
neity, especially in monocultural systems. Consequently, the
bird?habitat relationships we have identified can best be
considered as useful in informing best farming practices
and suggesting management tools in the development of
sustainable agroenergy production systems.
318 Bioenerg. Res. (2013) 6:311?320
Author's personal copy
More research is needed in order to properly evaluate the
full potential of perennially based crops to act as en route
habitat for grassland birds in the USA. Because most fields
we studied were not harvested for biomass during the pre-
vious fall, over-winter matting of standing grass and forbs
may cause delayed spring vegetation growth compared to
pastures and other adjacent habitat. Moreover, if spring
migrant use of perennial crops is limited by a lack of food
availability or vegetation structure as it is in annual crops
[56], the timing of fall harvest may have a profound impact on
spring migrant use of fields. For this reason, harvest strategies
(e.g., partial or staggered harvest) that have the effect of
increasing spring re-vegetation rate or enhancing within- or
between-field structural diversity during the migratory period
could be a management tool for increasing avian biodiversity
at one or more spatial scales [15, 57] and in both fall and
spring. Collectively, the impact of harvesting perennial agro-
energy feedstocks on migratory grassland birds remains un-
clear and is an important focus for future research.
We conclude that candidate next-generation perennial
agroenergy feedstocks have potential to act as spring migra-
tory stopover habitat for a significant assemblage of habitat
generalist and grassland specialist birds, especially if pro-
ducers can be encouraged to locate plantings within rela-
tively unforested regions. While our results indicate some
differences in the species richness of birds stopping over in
mixed grass?forb and switchgrass plantings, the largely
similar en route land bird assemblages in these two planting
types suggest no clear advantage of one feedstock vs. the
other in attracting a diversity or abundance of migrant birds.
In the end, perennial feedstocks may represent an important
conservation opportunity for grassland birds, but the realiza-
tion of any such potential appears highly dependent upon the
development and implementation of crop management strate-
gies that enhance the suitability of fields to grassland birds and
whether such management is economically sustainable.
Acknowledgments This work was funded by the Migratory Bird
Center of the Smithsonian Conservation Biology Institute, US Fish
and Wildlife Service (grant # 30181AG045), the DOE Great Lakes
Bioenergy Research Center (DOE BER Office of Science DE-FC02-
07ER64494), DOE OBP Office of Energy Efficiency and Renewable
Energy (DE-AC05-76RL01830), and the U.S. National Science Foun-
dation LTER program.
References
1. European Union (2009) Directive 2009/28/EC of the European
Parliament and of the Council of 23 April, 2009 on the promotion
of the use of energy from renewable sources and amending and
subsequently repealing directives 2001/77/EC and 2003/30/EC
2. US Congress (2007) Energy Independence and Security Act of
2007. Renewable fuels directive. http://www.govtrack.us/congress/
billtext.xpd?bill0h110-6. Accessed 26 May 2009
3. West T, Dunphy-Guzman K, Sun A, Malczynski L, Reichmuth D,
Larson R et al (2009) Feasibility, economics, and environmental
impact of producing 90 billion gallons of ethanol per year by 2030.
Sandia National Laboratories, Livermore
4. Green RE, Cornell SJ, Scharlemann JPW, Balmford A (2005)
Farming and the fate of wild nature. Science 307:550?555
5. Millennium Ecosystem Assessment (2005) Ecosystems and human
well-being: biodiversity synthesis. World Resources Institute,
Washington, DC
6. Fargione JE, Cooper TR, Flaspohler DJ, Hill J, Lehman C, McCoy
T et al (2009) Bioenergy and wildlife: threats and opportunities for
grassland conservation. Bioscience 59:767?777
7. Meehan TD, Hurlbert AH, Gratton C (2010) Bird communities in
future bioenergy landscapes of the UpperMidwest. PNAS 107:18533?
18538
8. Fletcher RJ, Robertson BA, Evans J, Doran P, Alavalapati JRR,
Schemske D (2011) Biodiversity conservation in the era of bio-
fuels: risks and opportunities. Front Ecol Env 9:161?168
9. Tilman D, Hill J, Lehman C (2006) Carbon-negative biofuels from
low-input high-diversity grassland biomass. Science 314:1598?
1600
10. Sauer JR, Hines JE, Fallon J (2008) The North American breeding
bird survey, results and analysis 1966?2007. Version 5.15.2008.
USGS Patuxent Wildlife Research Center, Laurel
11. Scharlemann JPW (2008) Can bird research clarify the biodiversity
benefits and drawbacks of biofuels? Ibis 150:640?642
12. Robertson BA, Doran P, Robertson R, Loomis L, Schemske D
(2011) Perennial biomass feedstocks enhance avian diversity. GCB
Bioenergy 3:235?246
13. Askins RA, Chavez-Ramirez F, Dale BC, Haas CA, Herkert JA
(2007) Conservation of grassland birds in North America: under-
standing ecological processes in different regions. Ornithol
Monogr 124:1?46
14. Sillett TS, Holmes RT (2002) Variation in survivorship of a mi-
gratory songbird throughout its annual cycle. J Anim Ecol 71:296?
308
15. Robertson BA, Robertson JR, Loomis ER, Doran JP, Schemske
DW (2011) Avian use of perennial biomass feedstocks as post-
breeding and migratory stopover habitat. PLoS One 6(3):E16941
16. Sample DW, Mossman MJ (1997) Management of habitat for
grassland birds: a guide for Wisconsin. Wisconsin Department of
Natural Resources, Madison
17. Moore FR, Gauthreaux SAJ, Kerlinger P, Simons TR (1995) Habitat
requirements during migration: important link in conservation. In:
Martin TE, Finch DM (eds) Ecology and management of neotropical
migratory birds. Oxford University Press, Oxford, pp 121?144
18. Rodewald PG, Brittingham MC (2004) Stopover habitats of land-
birds during fall: use of edge-dominated and early-successional
forests. Auk 121:1040?1055
19. Martin TE (1980) Diversity and abundance of spring migratory
birds using habitat islands on the Great Plains. Condor 82:430?439
20. Buler JJ, Moore FR, Norris K (2007) A multi-scale examination of
stopover habitat use by birds. Ecology 88:1789?1802
21. Emlen JT (1977) Estimating breeding season bird densities from
transect counts. Auk 94:455?468
22. Fletcher RJ, Dhundale JA, Dean TF (2000) Estimating non-
breeding season bird abundance in prairies: a comparison of two
survey techniques. J Field Ornith 71:321?329
23. Ribic CA, Koford R, Herkert JR, Johnson DH, Niemuth ND,
Naugle DE et al (2009) Area sensitivity in North American grass-
land birds: patterns and processes. Auk 126:233?244
24. Herkert JR (1994) The effects of habitat fragmentation on Mid-
western grassland bird communities. Ecol Appl 4:461?471
25. Harmoney KR, Moore KJ, George R, Brummer EC, Russell JR
(1997) Determination of pasture biomass using four indirect meth-
ods. Agron J 89:665?672
Bioenerg. Res. (2013) 6:311?320 319
Author's personal copy
26. Robel RJ, Briggsa JN, Datytonan D, Hulbert DLC (1970) Rela-
tionships between visual obstruction measurements and weight of
grassland vegetation. J Range Manage 23:295?297
27. Daubenmire R (1959) A canopy-coverage method of vegetation
analysis. Northwest Sci 33:43?64
28. Weins JA (1974) Habitat heterogeneity and avian community
structure in North American grasslands. Am Midl Nat 91:195?213
29. ESRI (2008) ArcGIS. ESRI, Redlands
30. United States Department of Agriculture (2010) 2009 Cropland
data layer. National Agricultural Statistics Service, Washington,
D.C
31. Rogers LE, Hinds WT, Buschbom RL (1976) A general weight vs.
length relationship for insects. Ann Ent Soc Am 69:387?389
32. H?dar JA (1996) The use of regression equations for estimation of
arthropod biomass in ecological studies. Acta Ecol 17:421?433
33. Colwell RK (1997) EstimateS: statistical estimation of species
richness and shared species from samples. Version 5. User?s guide
and application. Available: http://viceroy.eeb.uconn.edu/estimates.
Accessed 27 July 2010
34. Vickery PD, Tubaro PL, Dasilva JMC, Peterjohn BG, Herkert JR,
Cavalcanti RB (1999) Conservation of grassland birds in theWestern
Hemisphere. In: Vickery PD, Herkert JR (eds) Ecology and conser-
vation of grassland birds of theWestern Hemisphere. Stud Avian Biol
19. Cooper Ornithological Society, Camarillo, pp 2?26
35. Rosenzweig ML (1995) Species diversity in space and time. Cam-
bridge University Press, Cambridge
36. Burnham KP, Anderson DR (2002) Model selection and multi-
model inference: a practical information-theoretic approach, 2nd
edn. Springer, New York
37. R Development Core Team (2010) R: A language and environment
for statistical computing. R Foundation for Statistical Computing.
Vienna. Version 2.12.1. ISBN 3-900051-12-7. http://www.
R-project.org. Accessed 9 March 2012
38. Legendre P, Vaudor A (1991) The R Package: Multidimensional
analysis, spatial analysis. Department of Biological Sciences, Uni-
versity of Montr?al, Montr?al
39. Bivand R, Altman M, Anselin L et al. (2010) Package ?spdep?, R
Foundation for Statistical Computing, Vienna, Austria. http://
www.r-project.org/. Accessed 15 Nov 2010
40. Gardiner MA, Tuell JK, Isaacs R, Gibbs J, Ascher JS, Landis DA
(2010) Implications of three model biofuel crops for beneficial
arthropods in agricultural landscapes. Bioenergy Res 3:6?9
41. Deppe JL, Rotenberry JT (2008) Scale-dependent habitat use by
fall migratory birds: vegetation structure, floristics, and geography.
Ecol Monogr 78:461?487
42. Hutto RL (1985) Seasonal changes in the habitat distribution of
transient insectivorous birds in south eastern Arizona: competition
mediated? Auk 102:120?132
43. Moore FR, Kerlinger P, Simons TR (1990) Stopover on a Gulf
coast barrier island by spring trans-Gulf migrants. Wilson Bull
102:487?500
44. Champlin TB, Kilgo JC, Moorman CE (2009) Food abundance
does not determine bird use of early-successional habitat. Ecology
90:1586?1594
45. Blake JG (1984) A seasonal analysis of bird communities in
southern Nevada. Southwest Nat 29:463?474
46. Dinsmore JJ, Kent TH, Koenig D, Peterson PC, Roosa DM (1984)
Iowa birds. Iowa State University Press, Ames
47. Farley GH, Ellis LM, Stuart JN, Scorr NJ (1994) Avian species
richness in different aged stands of riparian forest along the middle
Rio Grande, New Mexico. Cons Biol 8:1098?1108
48. Petit DR (2000) Habitat use by landbirds along Neartic?Neotrop-
ical migration routes: implications for conservation of stopover
habitats. Stud Avian Biol 20:15?33
49. Ribic CA, Sample DW (2001) Associations of grassland birds with
landscape factors in southern Wisconsin. AmMidl Nat 146:105?121
50. Moore FR, Yong W (1991) Evidence of food based competition
among passerine migrants during stopover. Behav Ecol Sociobiol
28:85?90
51. Cimprich DA, Woodrey MS, Moore FR (2005) Passerine migrants
respond to variation in predation risk during stopover. Anim Behav
69:1173?1197
52. Bakker KK, Naugle DE, Higgins KF (2002) Incorporating land-
scape attributes into models for migratory grassland bird conser-
vation. Cons Biol 16:1638?1646
53. Chartier AT, Ziarno J (2004) A birders guide to Michigan. American
Birding Association, Colorado Springs
54. Berthold P, Terrill SB (1991) Recent advances in studies of bird
migration. Ann Rev Ecol Syst 22:357?378
55. Adler PR, Sanderson MA, Weimer PJ, Vogel KP (2009) Plant
species composition and biofuel yields of conservation grasslands.
Ecol Appl 19:2202?2209
56. Galle AM, Linz GM, Homan JH, Bleier WJ (2009) Avian use of
harvested crop fields in North Dakota during spring. West N Am
Nat 69:491?500
57. Roth AM, Sample DW, Ribic CA, Paine L, Undersander DJ,
Bartelt GA (2005) Grassland bird response to harvesting switch-
grass as a biomass energy crop. Biomass Bioenergy 28:490?498
58. National Audubon Society (2007) Audubon watchlist. http://
web1.audubon.org/science/species/watchlist. Accessed 7 Jan 2009
320 Bioenerg. Res. (2013) 6:311?320
Author's personal copy