Download chapter ii nest-site selection and reproductive

Texas Tech University, Quinn C. Emmering, May 2014
CHAPTER I
INTRODUCTION
Breeding habitat selection is firmly connected to fitness as conditions within
the breeding territory (e.g., microclimate, food abundance, and predation risk) can
strongly affect reproductive success (Cody 1985, Jaenike & Holt 1991, Martin 1998).
Therefore, selection should favor strategies that allow birds to assess spatial
heterogeneity in resources to bias selection toward higher quality sites to potentially
increase reproductive success (Lima 2009, Forsman et al. 2013). Heterogeneity in
breeding-site quality can scale up to the population level as demonstrated in sitedependent population regulation models (e.g., Rodenhouse & Holmes 1997, McPeek
et al. 2001) and empirical work (e.g., Rodenhouse et al. 2003, Zajac et al. 2008, Kelly
et al., in review). Site-dependent models, and habitat selection models in general,
invoke several key assumptions including the existence of (1) spatial heterogeneity in
primary resources, (2) biased settlement in sites with greater resources, and (3) the
ability of organisms to assess spatial heterogeneity in resources, either directly or
indirectly.
An important resource contributing to breeding-site quality for songbirds likely
includes the availability of spatial refugia from predators (or predator-poor space), as
the bulk of nest mortality is the result of predation. By several accounts, 80% of nest
failures are on average attributed to predation (Ricklefs 1969, Martin 1993, Remes et
al. 2012). Furthermore, nest predation has been identified as one of the possible
1
Texas Tech University, Quinn C. Emmering, May 2014
causes for declines of some Neotropical migratory birds in North America (Thompson
2007). This suggests birds likely possess behavioral traits to assess spatial variation in
predator activity to subsequently make adaptive breeding decisions.
Spatial and temporal variation in predator activity can emerge through several
behavioral mechanisms such as the underlying distribution of important resources and
the predator’s own risk of predation. Regardless of the mechanism, spatial
heterogeneity in predator activity can create small-scale spatial refugia for breeding
birds particularly when predation is incidental. Incidental predation being undirected
and occurring when generalist predators while foraging on relatively common primary
prey randomly encounter and consume scarce alternative prey (Vickery et al. 1992,
Yanes & Suarez 1996, Schmidt et al. 2001a, Vigallon & Marzluff 2005).
Observational (e.g., Spaans et al. 1998, Sergio et al. 2003, Forstmeier and
Weiss 2004, Roos & Pärt 2004, Morton 2005, Schmidt et al. 2006) and experimental
studies (e.g., Eggers et al. 2006, Fontaine & Martin 2006a, Fontaine & Martin 2006b,
Hua et al. 2013) have demonstrated that breeding birds select sites of lower predation
risk (further reviewed in Lima 2009). Most studies have investigated nest predator
avoidance at scales exceeding the territory level and quantify nest predation risk using
coarse measures of predator abundance. Examinations of nest predator avoidance at
smaller spatial scales (e.g., nest-site choice within a breeding pair’s territory)
combined with finer-grained measures of predator space use are surprisingly scarce
(Lima 2009). Yet, small-scale refugia may be an important metric for assessing nest
predation risk as it is likely a key component of breeding-site quality. If so,
2
Texas Tech University, Quinn C. Emmering, May 2014
determining the underlying factors that create spatial heterogeneity in local predator
abundance would be important information for enhancing and managing breeding bird
habitat.
Furthermore, spatial avoidance of nest predators suggests that birds are able to
assess spatial heterogeneity in local predator abundance. Nesting birds should be
receptive to direct cues of risk (e.g., predator presence) as well as exploit
communication systems (e.g., vocalizations; Peake 2005) that can indirectly reveal
areas of concentrated predator activity. But few investigators have experimentally
isolated the sources of information used (e.g., Eggers et al. 2006, Forsman & Martin
2009, Mönkkönen et al. 2009, Forsman et al. 2013). Identifying the assessment
mechanisms used by birds to evaluate local predator abundance could lend further
empirical support for site-dependent regulation of populations (Rodenhouse & Holmes
1997, McPeek et al. 2001). In addition, differences in information use between
sympatrically breeding birds may emerge from dissimilar sensory systems, natural
history constraints (e.g., ground vs. shrub nests) or ecological tradeoffs (e.g., safety vs.
microclimate; Eggers et al. 2006). Such interspecific differences could potentially
affect community structure (Fletcher 2008) by causing competitors to increasingly
overlapping in space or through dissimilar habitat use. Lastly, identifying the sources
of information birds use to avoid nest predators may help predict how particular
anthropogenic disturbances may disrupt information acquisition. For example,
anthropogenic noise may reduce the ability of birds to detect and use acoustic
3
Texas Tech University, Quinn C. Emmering, May 2014
information provided by heterospecifics (e.g., alarm calls) or intercept predator
communication for predator avoidance.
The study system
Ongoing research (1998 – present) at the Cary Institute of Ecosystem Studies
in southeastern New York, has monitored rodent (white-footed mice, Peromyscus
leucopus and eastern chipmunks, Tamias striatus) and songbird populations to
primarily understand the relationship between variation in rodent abundance/space use
and nest predation. During this time frame, annual variation in summer rodent
abundance has spanned over two-orders of magnitude in response to interannual
variation in synchronous oak masting events occurring in the prior autumn (Ostfeld et
al. 1996, Jones et al. 1998, Schmidt & Ostfeld 2008). These fluctuations are
ecologically relevant as rodents are important predators of songbird nests (Reitsma et
al. 1990, Jędrzejewska & Jędrzejewski 1998, Williams & Wood 2002, Cain et al. 2003,
Forstmeier & Weiss 2004, King & DeGraaf 2006, Clotfelter et al. 2007, Schmidt et al.
2008, Cox et al. 2012, Haché et al. 2014).
Studies have focused on two sympatric, territorial songbirds, veeries (Catharus
fuscescens) and ovenbirds (Seiurus aurocapilla). Veeries are small (28 g) thrushes
building open-cup nests predominately in shrubs (≤ 1.5 m in height) and infrequently
directly on the ground. In contrast, ovenbirds are obligate, ground-nesting warblers
(20 g) constructing dome-shaped nests possessing a side entrance. Ovenbirds and
veeries, like many Holarctic songbirds, are susceptible to nest predation from rodent
4
Texas Tech University, Quinn C. Emmering, May 2014
generalist predators that prey upon eggs, nestlings and even fledgling young. Previous
research from our study site has verified that rodents are important predators of some
songbird nests as annual nest daily-mortality rate in veeries is an increasing function
of rodent density (Schmidt 2003, Schmidt & Ostfeld 2008, Kelly et al., in review).
However, annual nest daily-mortality rate ovenbirds this is not affected by rodent
abundance (Kelly et al., in review).
Spatial heterogeneity in predation risk to incidental prey may be generated by
the proportion of space used by generalist predators, where space use may largely be
determined by their assessment of the profitability of foraging patches (Brown 1988).
For example, mice may direct the bulk of their foraging efforts in patches with a
greater abundance of primary prey (e.g., arthropods or tree seeds) and/or with low
predation risk. Resulting unused space by generalist predators can create spatial
refugia for incidental prey to increase the probability of survival (Schmidt et al. 2006,
Schauber et al. 2009, Tyler 2011). Furthermore, several experiments suggest that
veery nest success is also related to smaller-scale mouse activity near the nest site
(Schmidt et al. 2001a, 2001b, 2003b, 2006, Schmidt 2006).
General summary of research and hypotheses
My dissertation examines the relationship between veery and ovenbird nest
success, nest-site selection and spatial heterogeneity in small-scale rodent activity. All
research took place in eastern deciduous forest on the property of the Cary Institute of
Ecosystem Studies in the Hudson Valley of New York. To investigate variation in
5
Texas Tech University, Quinn C. Emmering, May 2014
rodent activity I conducted two studies (Chapters II and III). The first study (Chapter
II), utilized a long-term dataset of rodent captures collected from six 2.25 ha trapping
grids using a spatially-explicit approach to examine spatial patterns in rodent activity.
In the second study (Chapter III), I used track plates to measure the spatial activity of
nest predators (rodents and mesopredators) on a shorter time scale. The final data
chapter describes a playback experiment to determine if songbirds eavesdrop on
chipmunk vocalizations to assess their spatial heterogeneity to make nest-site
decisions. Based on the earlier work described above, I formulated three general
predictions: (1) To lower the probability of nest predation ovenbirds and veeries
would place nests in locations of relatively lower rodent activity than other available
sites. (2) Veery and ovenbird nest success would be negatively related to small-scale
rodent activity surrounding nest sites. (3) Lastly, in response to playbacks simulating
“hotspots” of chipmunk activity, ground-nesting songbirds should place nests
significantly further from plots broadcasting chipmunk calls compared to control plots
that broadcasted frog calls or were silent.
6
Texas Tech University, Quinn C. Emmering, May 2014
CHAPTER II
NEST-SITE SELECTION AND REPRODUCTIVE SUCCESS OF
VEERIES AND OVENBIRDS IN RESPONSE TO SPATIAL
HETEROGENEITY IN RODENT ACTIVITY
Introduction
Spatial and temporal variation in predator activity creates enemy-free space for
prey to take refuge, increasing the probability of survival for adults or offspring.
Spatial heterogeneity in predator space use can be generated through a variety of nonmutually exclusive mechanisms that emerge as a result of underlying heterogeneity in
primary resources (Roos & Pärt 2004), the predators’ own risk of predation
(Forstmeier & Weiss 2004, Schmidt 2006, Dutra et al. 2011, Malo et al. 2012),
competition (Durant 1998), habitat complexity (Bowman 1980, Lecomte et al. 2008,
Gorini et al. 2012) and the energetic state of the animal (Brown et al. 1992). The
availability of spatial refugia from predation may be especially important for less
mobile prey that are restricted to a central location and are thus limited in the types of
behavioral responses they can use to avoid predators.
The eggs and young of songbirds are an example of vulnerable, sessile prey
that are anchored to a central location for a month or more. Indeed, studies have
consistently shown that nest failure is primarily the result of predation as, on average,
80% of nest failures are attributed to predation (Ricklefs 1969, Martin 1993, Remeš et
7
Texas Tech University, Quinn C. Emmering, May 2014
al. 2012). In addition, nest predation is thought to be principally incidental (Vickery et
al. 1992, Yanes & Suarez 1996, Schmidt et al. 2001a, Schmidt et al. 2004a, Vigallon
& Marzluff 2005), in that it is largely undirected and occurs when generalist predators,
while foraging for primary prey, serendipitously encounter and consume scarce
secondary prey (Vickery et al. 1992, Schmidt et al. 2001a, McKinnon et al. 2013).
Therefore, selection should favor strategies that allow birds to proactively evaluate
spatial heterogeneity in nest-predation risk to select breeding sites that lower the
probability of reproductive failure (Lima 2009).
Examples of nest predator spatial avoidance include red-backed shrikes
(Lanius collurio) that shift their breeding territories in response to annual variation in
the spatial distribution of magpie (Pica pica) and hooded crow (Corvus corone)
territories, both regular nest predators of shrikes (Roos & Pärt 2003). Similarly,
territoriality in eagle owls (Bubo bubo) creates relatively predator-free zones within
the interstitial spaces between adjacent owl territories where black kites (Milvus
migrans) choose to nest (Sergio et al. 2003). Large-scale predator removals (Fontaine
& Martin 2006a) and playback experiments (Hua et al. 2013) have found lower
occurrence or abundance of several breeding songbird species compared to control
areas. The importance of selecting breeding sites of lower predation risk is
highlighted in cases where birds potentially tradeoff food availability for safety. For
instance, Morton (2005) found ground-nesting ovenbirds (Seiurus aurocapilla)
consistently selected territories along forest edges as chipmunks were nearly absent
from these sites. Moreover, he determined that these sites had significantly lower food
8
Texas Tech University, Quinn C. Emmering, May 2014
availability for ovenbirds than the forest interior, where chipmunks were in abundance.
Likewise, the dusky warbler (Phylloscopus fuscatus), a shrub-nesting songbird,
responded to increasing chipmunk abundance by building nests higher and in more
isolated shrubs surrounded by open tundra further removed from larger closed shrub
stands (Forstmeier & Weiss 2004) despite the extra energy demands expended by
flying from isolated patches to forage. Chipmunks avoided open tundra presumably to
reduce predation risk associated with their own predators, diurnal short-eared owls
(Asio flammeus) and foxes (Forstmeier & Weiss 2004).
Within a breeding bird’s territory, small-scale heterogeneity in nest-predation
risk can be generated by generalist predators evaluating of the costs and benefits of
foraging patches. For generalist predators such as rodents, spatial variation in the
profitability of food patches balanced by the costs of predation risk may cause some
patches to become periodically unoccupied (Brown 1988). Thus generalist predator
activity becomes concentrated in low risk, high reward patches whereas unprofitable
space goes unused, potentially creating important refugia for songbirds to place their
nests. Schmidt & Ostfeld (2003b) provided evidence for such a mechanism. On six
2.25 ha rodent trapping grids they recorded similar white-footed mouse (Peromyscus
leucopus) densities (31 – 49 mice); however, within trapping grids the proportion of
space used by mice varied considerably (10 – 70%; Schmidt & Ostfeld 2003b). This
suggests the presence of unused space by mice may still persist even when their
densities are high. Moreover, they found that the daily mortality rate of artificial
songbird nests was an increasing function of the proportion of space used, but not the
9
Texas Tech University, Quinn C. Emmering, May 2014
number of individual mice caught within the same area around artificial nest sites
(Schmidt & Ostfeld 2003b). Further work by Schmidt et al. (2006) examined if smallscale mouse activity (sub-territory level) affected veery (Catharus fuscescens) nest
success and nest-site selection using nests that were found over five years on six
trapping grids (see above). Their study revealed (1) significant spatial heterogeneity
in mouse activity, some trap stations caught few or no mice (i.e., coldspots) while
captures at other stations were > 2 SD above the mean (i.e., hotspots); (2) annual veery
nest-predation rates showed both a significant positive relationship to mouse
abundance and a negative relationship to mouse-free space. (2) Veery nest success
was significantly related to mouse activity within a “nest neighborhood” (i.e., 9 trap
stations within a 30 × 30 m area centered on a nest). (4) Lastly, 75% of nests were
associated with sites of below-average mouse activity suggesting that veeries are
capable of assessing small-scale mouse activity (Schmidt et al. 2006). The crux of this
research suggests that predator density may be a misleading indicator of nest-predation
risk and that predator space use may poorly correlate with, or lag behind, changes in
density (Schmidt & Ostfeld 2003, Schmidt & Schauber 2007). Furthermore, smallscale predator activity surrounding nest sites might provide better estimates of nestpredation risk than larger-scale abundance estimates. This could partly explain why
some studies find a lack of congruence between predator abundance and nest success
Surprisingly though, few studies have attempted to examine avoidance of nest
predators at smaller spatial scales such as nest placement made within a breeding
pair’s territory (Lima 2009). As described above, research has primarily focused on
10
Texas Tech University, Quinn C. Emmering, May 2014
adjustments that birds make when selecting territories or larger-scale habitat decisions
(Roos & Pärt 2003, Sergio et al. 2003, Fontaine & Martin 2006a, Hua et al. 2013). In
addition, nest-predation risk is usually quantified by collecting coarse abundance
measurements of nest predators (Schmidt & Schauber 2007). However, methods such
as transect surveys or point counts (e.g., Forstmeier & Weiss 2004, Morton 2005) are
limited in their ability to quantify small-scale space use of nest predators. Quantifying
small-scale activity predator activity may be an important metric for assessing nestpredation risk as predator-poor space is likely a key component of breeding-site
quality. My objective was to quantify small-scale spatial activity of two common
predators of songbird nests, white-footed mice and eastern chipmunks (Tamias
striatus). And furthermore, to determine if nest success and nest-site selection of
ovenbirds and veeries were related to small-scale spatial heterogeneity in rodent
activity. Where “spatial activity”, hereafter, refers to metrics that quantify the
aggregate amount a specific location is used by mice or chipmunks. For example, a
particular location may record an equal value as a result of a single mouse visiting five
times or from five different mice visiting the location once. Hence, spatial activity
attempts to quantify space use intensity, rather than the number of individuals, at
spatially-referenced locations.
To accomplish these objectives, I expanded on the work of Schmidt et al.
(2006). I now examine 15 years of trapping data which captures a greater breadth of
rodent population fluctuations and incorporates a more spatially explicit approach
using randomization procedures to determine regions of high and low mouse and
11
Texas Tech University, Quinn C. Emmering, May 2014
chipmunk activity. I used Spatial Analysis by Distance Indices (SADIE; Perry et al.
1995, 1999). SADIE measures and detects nonrandomness in the overall spatial
pattern and calculates local indices for identifying clusters of sample locations as
being arranged randomly or consisting of local neighborhoods of similarly sized
counts in the form of ‘patches’ or ‘gaps’ (Perry et al. 1999). Such spatially-explicit
methods may provide improved estimates regarding the spatial pattern of rodent
activity. Furthermore, I include a heterospecific comparison of sympatrically breeding
veeries and ovenbirds whose nesting ecology differs markedly (e.g., shrub vs. ground
nests), however their nests are both accessible to mice and chipmunks.
Based on earlier work I formulated two central hypotheses. First, to lower the
probability of nest predation, I predicted ovenbirds and veeries would place nests in
areas of lower rodent activity. Second, I predicted that veery and ovenbird nest
success would be more strongly predicted by small-scale rodent activity surrounding
their nest sites than larger-scale measures of rodent abundance. Lastly, I also expected
heterospecific differences in response to the two rodent nest predators. Ovenbirds and
veeries both place their nests on or near the ground (veeries ≤ 1.5 m in shrubs;
Heckscher 2004, Bevier et al. 2005) so presumably both species’ nests are similar in
their accessibility to ground-foraging rodents. However, heterospecific differences in
nest-site microhabitats and/or architecture (Martin 1995), how parent birds manage
risk (Martin et al. 2000, Eggers et al. 2008) could influence their respective predation
risk and consequently which predator they avoid. Based on earlier research from our
study site (Emmering & Schmidt 2011; Chapter 4), I predicted that ovenbirds would
12
Texas Tech University, Quinn C. Emmering, May 2014
nest in areas of lower chipmunk activity. Emmering & Schmidt (2011) demonstrated
that veeries and ovenbirds placed nests significantly further from playbacks
broadcasting chipmunk vocalizations than sites broadcasting frog calls or silent
controls. Ovenbirds placed nests > 20 m from playbacks of chipmunks whereas the
veery response was half the distance of ovenbirds, placing nests only 10 m further
from chipmunk playbacks. Playback results suggest that ovenbirds may particularly
use chipmunk vocalizations to assess chipmunk activity in space and subsequently are
able to avoid concentrations of chipmunks to reduce nest-predation risk.
Methods
Field research was conducted on the property of the Cary Institute of
Ecosystem Studies (41º 50´ N, 73º 45´ W) located in Dutchess County, Millbrook,
New York, USA from 1998 – 2012. The property consists of ~325 ha of continuous
eastern deciduous forest with a oak-dominated canopy (Quercus rubra and Q. prinus).
Common understory trees and shrubs include oak and maple (Acer spp.) saplings,
witch hazel (Hamamelis virginiana) as well as nonnative honeysuckle (Lonciera spp.)
and Japanese barberry (Berberis thunbergii).
Trapping protocol and rodent enumeration
I used a 15-year dataset (1998 – 2012) of mouse and chipmunk captures
collected annually (May – August) from six 2.25 ha permanent trapping grids (GC,
GE, TE, TC, HC, and HE). Grids consisted of an 11 × 11 array (12 × 10 for one grid)
13
Texas Tech University, Quinn C. Emmering, May 2014
of 121 trap stations equally spaced 15 m apart. Each trap station consisted of two
Sherman live traps (7.6 × 8.9 × 30 cm) placed side by side with a cover board on top
for protection from sun and precipitation. The trapping schedule rotated between grids
such that one pair of grids was trapped every 2 – 3 weeks between trapping sessions.
Traps were baited daily with oats between 16:00 – 18:00 hrs. and checked the
following two mornings between 07:30 – 11:30 hrs. On cold nights (≤ 10C),
sunflower seed and raw cotton were added to traps in addition to the oats to provide a
fattier food source and warm bedding material. Captured animals were given a unique
numbered ear tag, measured for standard morphometrics and examined for
reproductive status and age then released where captured (for more details, see Jones
et al. 1998). The above methods were approved by an institutional animal care and
use committee.
Mouse and chipmunk abundance were estimated using different estimators as
the probability of capturing mice is higher than chipmunks. Mouse abundance was
estimated for each grid’s trapping week using minimum number known alive (MNA).
For chipmunk abundance, a Jolly-Seber (JS) heterogeneous survival model was used
(Pledger et al. 2010). Both chipmunk and mouse abundance estimates from, or
interpolated for, the approximate midpoint of the songbird breeding season (June 15)
were used for all subsequent analyses.
To quantify rodent activity total captures (not individuals) per trap station of
mice, chipmunks and combined (mice + chipmunks) were enumerated for each grid ×
year combination. Total captures were summed over the migratory songbird breeding
14
Texas Tech University, Quinn C. Emmering, May 2014
season from approximately the first week May – 1 August. I reasoned that total
captures, not the number of individuals captured in a trap, likely provides a better
proxy for rodent activity. To explain, a single mouse captured several times probably
poses a greater threat to a nesting songbird than a single individual caught only once.
During this timeframe a minimum of three trapping sessions (or six trapping nights
total) for each grid × year combination occurred with a single exception (GE in 2012)
that had only 2 trapping sessions. Otherwise, out of 90 total grid × year combinations
< 88% had 6 – 8 trapping nights/year; a few grid × year combinations had ≤ 10 (n=9)
trapping nights.
Nest data
From May – July of each year, nest searching took place on all trapping grids
but only nests found within 15 m of trapping grids were used in the following analyses.
From 2005 – 2012, I recorded nest locations using a GPS unit and determined each
nest’s nearest trap station using ArcGIS 9.1; prior to 2005, the nearest trap station was
recorded in the field for each nest. Active nests were monitored at minimum every
three days until young fledged or the nest was depredated. Only a single nest per
breeding pair/territory/year was used for all analyses due to the lack of independence
between associated renests. Single nesting attempts regardless of fate (abandoned,
unknown, depredated or fledged) were used to investigate nest-site choice. To
examine the relationship between rodent activity and reproductive success, nests were
classified as either fledged or failed. Nests were considered fledged if they
15
Texas Tech University, Quinn C. Emmering, May 2014
successfully fledged at least one nestling. Completed nests that became inactive
before the earliest possible fledging date, whether due to abandonment or from clear
signs of depredation (e.g., broken eggs, destroyed nest), were considered failed.
Spatial Analysis by Distance Indices (SADIE)
Grid-level (large scale) and small-scale rodent activity (i.e., rodent captures)
within nest neighborhoods was analyzed using Spatial Analysis by Distance Indices
(SADIE). SADIE is a set of nonparametric methods developed by Perry et al. (1995,
1999) specifically designed to examine count data collected at spatially-referenced
sample locations. SADIE allows an investigator to (1) test spatially-referenced count
data for nonrandom spatial patterns, (2) detect clusters of “patches” and “gaps”
composed of neighboring sample locations with relatively high or low counts,
respectively, and (3) assess the amount to which each sample location contributes to
the overall spatial pattern (Perry et al. 1995, Perry et al. 1999). Furthermore, output
can be illustrated in corresponding mapped plots depicting local clustering (e.g.,
Figure 2.1; Perry et al. 1995, Perry et al. 1999). SADIE techniques are advantageous
in that they deliberately distinguish between density and clustering, discounting
isolated large (or small) values that can heavily influence geostatistical models (Perry
et al. 1999).
SADIE cluster analysis uses a transportation algorithm to find the optimal way
to minimize the distance that sample locations with above-average counts (donor
units) would need to distribute and partition their “surplus” to those locations with
16
Texas Tech University, Quinn C. Emmering, May 2014
below-average counts (receiver units), thereby equalizing counts across a sample grid
into a regular arrangement. This total distance is known as the “distance to regularity”,
D. Generally, larger values of D indicate greater spatial aggregation of counts.
Permutations of observed counts (≈6000) among the sample locations are used to
create a sampling distribution of corresponding values of D. An aggregation index, Ia,
is estimated by dividing the observed value of D by the mean of the sampling
distribution. This procedure was carried out for mouse and chipmunk captures to
assess the overall pattern of rodent activity on trapping grids.
Secondly, we used SADIE to quantify the individual contribution of each trap
station to the overall spatial pattern by estimating local cluster indices () for captures
of mice, chipmunks and combined (mice + chipmunks; Perry et al. 1999). Cluster
indices, in part, are calculated using the same transportation algorithm as D to
minimize the distance to regularity and partition outflows to below-average sample
locations. By convention, donor units (i) are positive and receiver units (j) are
negative. The core of the cluster index value is Ƴ , the average outflow distance from
donor units weighted by the strength of the individual flows to receiver units. Thus Ƴi
is formulated by:
Ƴi = ∑j dij vij/∑j vij
where i is a donor unit (at position xi, yi), j is a receiver unit (at position xj, yj); dij and
vij are the distance and surplus amount from unit i to unit j, respectively. Additionally,
 is standardized by the terms oƳ, iƳ and cƳ. In brief, cƳ is the mean value of Ƴ when
the count at sample unit i is followed around to all the locations it landed for each
17
Texas Tech University, Quinn C. Emmering, May 2014
iteration of the randomization, iƳ is similar to cƳ except that it is the mean value of Ƴ
with respect to location x, y. Lastly, oƳ is the average value of cƳ over all sample units,
thus it is the same value for all sample units and counts. Finally, the cluster index
value for donor units (above-average counts), denoted by i, is calculated by:
i = Ƴi (oƳ/iƳ cƳ)
Likewise, receiver units (below-average counts), j, are similarly formulated.
Values of i > 1.0 constitute members of patches and values of j < –1.0
constitute members of gaps (hereafter “hotspots” and “coldspots” of rodent activity,
respectively). By convention, values of j ≤ –1.5 and i ≥ 1.5 are considered to
represent significant clustering of low and high counts, respectively, where a unit was
half as much again than the value expected by chance. A trap station assigned a value
of j = –2.0 would indicate that captures at that location were twice that expected by
chance. While somewhat arbitrary, the values of j ≤ –1.5 and i ≥ 1.5 when plotted
with their corresponding randomized distributions approximate or exceed their 5th and
95th centiles, respectively (Perry et al. 1999, Winder et al. 2001). Cluster index values
enclosed between contours ±1 are regions where counts are approximately the sample
mean ( ≈ 0.0) or indicate random placement relative to neighboring units (Perry et al.
1999). As with D described earlier, formal randomization tests allow the estimation of
p-values associated with j and i to evaluate whether observed regions of coldspots
and hotspots are significant.
The unique properties of the SADIE approach remove two potential biases
associated with the trapping data. First, the number of trapping nights varied annually
18
Texas Tech University, Quinn C. Emmering, May 2014
and between grids (Schmidt et al. 2006 resolved this by calculating captures per night
instead of using raw counts). The terms described above that standardize Ƴ correct for
such differences (Perry et al. 1995), making grids and years directly comparable. For
example, if two grids had the same relative spatial arrangement of counts but one had
double the number of captures, the index values would still be equal. Also, trapping
grids could experience an edge effect whereby traps located on grid boundaries
compete less for rodent captures than interior traps (Schmidt et al. 2006) potentially
causing edge traps to have higher captures. However, in SADIE, index values of
relatively large, isolated counts are discounted so that its cluster index value would be
substantially lower than an equivalent value found amid a neighborhood of high
counts (Perry et al. 1999). Likewise, the same holds true for isolated low counts.
Therefore, discounting relatively large values reduces a potential edge effect.
Analyses of nest-site choice and nest success
Cluster indices associated with trap stations provide a local measure of the
magnitude of rodent activity relative to other locations on a trapping grid. Hence, the
index acted as a proxy for nest-predation risk with the assumption that risk is an
increasing function of local rodent activity near a nest site.
Cluster indices of mice, chipmunks, and combined rodents were examined at
three spatial scales around nest sites – the focal trap nearest the nest and two “nest
neighborhoods”. Five-trap nest neighborhoods consisted of the focal trap plus its four
nearest neighboring traps 15 m away, approximating a 225 m2 area (Figure 2.2).
19
Texas Tech University, Quinn C. Emmering, May 2014
Nine-trap nest neighborhoods extended the five-trap neighborhood to include four
additional traps at 45˚ and ≈ 21 m from the focal trap, approximating a 900 m2 area
(Figure 2.2). Nests whose nearest focal trap was along a grid boundary had only three
traps and six traps associated with its two neighborhood scales. For each nest, mean
cluster index values of nest neighborhoods were used for all subsequent analyses.
Randomization procedures (R version 3.0.0, The R Foundation for Statistical
Computing) were used to determine if veeries and ovenbirds disproportionately placed
nests in areas of lower rodent activity relative to the grid-specific environment in
which they nested than would be predicted by chance. Nest locations were
bootstrapped among trap stations, but restricted to the respective grid × year
combinations in which they were observed, to create a sampling distribution of mean
neighborhood cluster indices under the null hypothesis that nests were placed at
random with respect to the spatial activity of mice and chipmunks. This procedure
was carried out for each pairwise combination of rodent metric (mice, chipmunks and
combined) and nest-site scale (focal, five- and nine-trap neighborhoods) for a total of
nine tests per songbird species. Bonferroni corrections were used to adjust alpha
levels for multiple comparisons to a value of 0.005 (0.05/9 tests per songbird species).
All randomization tests consisted of 1000 iterations. I used all veery and ovenbird
nests found within 15 m of trapping grids regardless of nest fate.
Logistic regression (JMP® Pro ver. 10.0.2, SAS Institute Inc., Cary, NC) was
used to examine the influence of (1) grid-level rodent abundance, (2) rodent activity at
three scales around the nest site and (3) predator species (mice, chipmunks or
20
Texas Tech University, Quinn C. Emmering, May 2014
combined) on ovenbird and veery nest failure. Akaike’s Information Criterion
corrected for small sample size (AICc) was used to evaluate candidate models
(Anderson 2008). Separate model sets were run separately for veeries and ovenbirds,
each consisting of 21 candidate models. No interaction terms were included in the
models. Mouse and chipmunk abundance (at grid-level) and activity metrics were also
not included together in models as the combined metrics (mice + chipmunk) allowed
investigation of the combined effect of the two rodent species. Likewise, models did
not include more than one nest neighborhood scale (e.g., focal trap and 9-trap
neighborhoods).
Model strength was assessed based on multiple selection criteria including
values of AICc, ∆AICc, Akaike weights () and evidence ratios. Individual variables
were evaluated using 95% confidence intervals of estimates not overlapping with zero
and by calculating importance values (Burnham & Anderson 2002) for single
variables by summing  across all models which included the respective variable. All
nests categorized as fledged (veery, n=41; ovenbird, n=23) or failed (veery, n=47;
ovenbird, n=28) were used to examine nest failure.
Results
From 1998 – 2012, we found a total of 848 veery and 554 ovenbird nests
across the study site; out of this total 88 veery and 60 ovenbird nests were within ≤ 15
m of the six trapping grids (Table 2.1). Over the 15-year study, there were a total of
90 grid × year combinations in which nests could have been found of which 52 and 40
21
Texas Tech University, Quinn C. Emmering, May 2014
grid × year combinations had veery and ovenbird nests, respectively (68 grid × year
combinations having either species’ nests). A maximum of four nests for each species
were found in any given grid × year combination.
Large-scale patterns of rodent activity
Indices of aggregation (Ia) of mouse and chipmunk captures (counts) varied
among years and grids (Figure 2.3). Mean Ia values (of the six trapping grids)
indicated counts were frequently aggregated into clusters as values were consistently
greater than 1.0 over the 15-year study period (Figure 2.3). Mean Ia values ranged
from 1.06 – 1.36 for mice and 1.15 – 1.47 for chipmunks (Figure 2.3).
Nest choice, nest success and small-scale rodent activity
Randomization tests showed that for all nest neighborhoods and rodent metrics
examined, veery and ovenbird nests were not placed in areas with significantly lower
(all p ≥ 0.027, α = 0.005) mean cluster indices than would be predicted by chance
(Table 2.2).
The top ranked model investigating veery nest predation included rodent
activity (mice + chipmunks) within the 9-trap nest neighborhood and rodent
abundance (Table 2.3). The second highest model (∆AICc = 0.32) included only
rodent abundance, having a similar Akaike weight as the highest ranked model (wi =
0.34 and 0.29, respectively). Two other models also had strong support (∆AICc ≤ 2.0)
which also included metrics of combined rodent activity and abundance (Table 2.3).
22
Texas Tech University, Quinn C. Emmering, May 2014
Lastly, the model containing mouse abundance only was modestly supported (∆AICc =
4.45, likelihood = 0.11). Rodent activity in 9-trap neighborhoods was included in two
out of the top four models; however, in both cases its confidence intervals straddled
zero (Table 2.4a). Rodent abundance was found in the top three models (Table 2.4a)
and having much greater importance values (0.910) than the mouse abundance (0.083)
variable. All other candidate models (R=16) in the veery model set were weakly
supported (evidence ratios ≥ 16.15, model likelihoods ≤ 0.062,  ≤ 0.021).
For ovenbirds, the model consisting of chipmunk abundance and activity in the
9-trap nest neighborhood obtained the highest rank (Table 2.3), followed closely by
two models – rodent activity within the 5-trap nest neighborhood plus rodent
abundance (∆AICc = 0.59) and for the second, rodent abundance only (∆AICc = 0.91).
Overall, the top six models had ∆AICc values < 2.0 and the top 15 obtained ∆AICc
values ≤ 4.0 with evidence ratios ≤ 8.0 although Akaike weights were low for most (
< 0.10). Only two parameter estimates had confidence intervals that did not overlap
zero – chipmunk activity in the 9-trap nest neighborhood and mouse abundance (Table
2.4b), with chipmunk activity having a modestly higher importance value (0.233) than
mouse abundance (0.138).
Discussion
Spatial pattern analysis of total summer captures revealed consistent spatial
heterogeneity of mouse and chipmunk activity over the 15 year study period (Figure
2.3). Hence, in most years and sites spatial heterogeneity in rodent activity is likely
23
Texas Tech University, Quinn C. Emmering, May 2014
available to provide leastwise predator-poor space for nesting songbirds to lower the
probability of nest predation.
Veeries and ovenbirds did not place nests in areas of relatively lower rodent
activity than expected by chance (Table 2.2). This was true for all three spatial scales
examined around their nests including the focal trap nearest the nest and the 5- and 9trap nest neighborhoods, areas approximating 225 m2 and 900 m2, respectively, around
nest sites (Figure 2.2). However, results indicated that the two species differ in the
factors that influence their reproductive success. Ovenbird nest success was best
described by chipmunk activity within a 30 × 30 m area (9-trap neighborhood; Figure
2.2) around their nests (Tables 2.3 and 2.4b). Chipmunk abundance was also included
in the top model, but importance values and confidence intervals suggest its relative
influence was not as substantial as chipmunk activity (Table 2.4b). Mouse abundance
also seems to be an important factor in describing ovenbird nest success, as it was the
only other variable whose confidence intervals did not overlap with zero and its
importance value was only modestly lower than chipmunk activity in 9-trap
neighborhoods (Table 2.4b). In contrast, rodent abundance more strongly described
veery nest success than spatial activity as four of the top five models included rodent
abundance (Table 2.3). Additionally, total rodent abundance and mouse abundance
were the only two variables with confidences intervals that did not overlap zero.
However, rodent activity at the focal trap, 5-trap and 9-trap neighborhood scales were
all included in the top four veery models indicating that space use by rodents at one or
more of these scales may also have some influence on veery nest success.
24
Texas Tech University, Quinn C. Emmering, May 2014
These results partly corroborate an earlier playback experiment (Emmering &
Schmidt 2011; Chapter 4 of dissertation) which provides evidence that ovenbirds
eavesdrop on chipmunk calls to avoid nesting near hotspots of chipmunk activity. We
had originally predicted that both ovenbirds and veeries would respond strongly to
experimental playbacks of chipmunk vocalizations as shrub-nesting veeries are also
vulnerable to chipmunk nest predation (Emmering & Schmidt 2011). Interestingly
though, the magnitude of the veery response was only half that of ovenbirds. In the
current study, while ovenbirds did not choose nest sites with lower chipmunk activity
regression models did indicate that ovenbird nest success was related to chipmunk
activity near their nests. Veeries also did not avoid areas of greater chipmunk or
mouse activity on trapping grids. Instead, combined rodent abundance (mice +
chipmunks), rather than local activity, appears to be a more important factor
determining their nest success. There are several non-mutually exclusive hypotheses
for the lack of a relationship between rodent activity and responses of nesting
songbirds. For example, veeries and ovenbirds may be behaviorally constrained to
nest within certain microhabitats (e.g., mesic areas) or nest substrates (e.g., shrubs)
limiting their ability to select nest sites of lower predator activity. Alternatively, birds
may show plasticity in other behavioral traits to reduce predation risk. For example,
veeries may nest within denser vegetation for greater concealment (Marzluff 1988;
Eggers et al. 2006) or reduce the number of feeding visits to young (Martin et al.
2000). Also, unlike obligate ground-nesting ovenbirds, veeries are not constrained to
adjust the height of their nests.
25
Texas Tech University, Quinn C. Emmering, May 2014
Over the 3-month songbird breeding season trapping grids were not
continuously operated but, instead sampled usually 3 – 4 times (6 – 8 nights) per
season and total captures were summed for this duration. Summing total summer
captures to assess spatial activity, in particular, may be a less efficient method for
estimating space use by rodents. Spatial activity of mice can vary considerably over a
twelve-week period at our study site (Schauber et al. 2009) suggesting that predation
risk is also not temporally homogenous across sites. Quantifying each trapping
session individually could provide a more accurate (or current) estimate of rodent
activity that is likely more relevant for prospecting birds when making nest-site
decisions and for investigating nest success. My study and Schmidt et al. (2006) used
total captures (including repeated captures of individuals), not the number of
individuals captured in a trap. The rationale behind this was that individual rodents
caught several times in the same trap are likely a greater threat to nests than a single
individual caught only once. The assumption being that individuals caught less are
not as active or exploratory and consequently use less space than individuals caught
frequently. This reasoning may require further evaluation as this assumption may be
inaccurate.
Predator density, spatial activity, or any single proxy of predation risk may
poorly predict site occupancy and/or reproductive success of prey (Schmidt & Ostfeld
2003, Schmidt & Schauber 2007) as the behavioral strategies used to reduce risk likely
vary among coexisting prey. Prey that use enemy-free space can potentially decrease
mortality to themselves and their offspring irrespective of overall predator densities in
26
Texas Tech University, Quinn C. Emmering, May 2014
the surrounding environment (Schmidt et al. 2006, Schmidt & Schauber 2007). Nest
success of veeries and ovenbirds is at least partly described by measures of both
rodent density and spatial activity (Table 2.3). But the predictive strength of the
metrics differed for the two sympatric songbirds (Table 2.4). Veery nest success was
more strongly described by the combined, grid-level abundance of mice and
chipmunks whereas small-scale chipmunk activity was the better predictor of ovenbird
nest success. Analysis of either measure alone might cause one to conclude that
rodents are not important predators of some songbird nests and therefore fail to
uncover important ecological relationships. These results highlight the need for
careful selection of metrics to examine prey responses to predation risk. This could
partly explain why some studies find a lack of congruence between predator
abundance and nest success (Chalfoun & Schmidt 2012, Kelly et al., in review).
In conclusion, these findings provide evidence that small-scale spatial variation
in chipmunk activity, and not abundance, was the better predictor of ovenbird nest
success. This relationship corroborates research demonstrating ovenbirds avoiding
perceived concentrations of chipmunks as found in an experimental playback study
(Emmering & Schmidt 2011). However, the observed differences between veeries and
ovenbirds in their response to spatial variation in rodent activity suggest that they
differ in vulnerability to rodent nest predators or adopt different behavioral strategies
for managing predation risk. Spatial heterogeneity in rodent activity as recorded at
our study site could provide breeding ovenbirds and other ground-nesting songbirds
the means to cope with large pulses of rodent abundance (Ostfeld et al. 1996). Future
27
Texas Tech University, Quinn C. Emmering, May 2014
research into the underlying mechanisms that produce heterogeneity in rodent space
use (e.g., vegetation structure) could lend greater understanding on how changes in
forest communities may affect ecological interactions.
28
Texas Tech University, Quinn C. Emmering, May 2014
Figure 2.1 Example of a trapping grid depicting local cluster indices of mice captured in 2012. SADIE
analysis indicated plot counts were clustered overall (Ia = 1.38), and for both hotspots (v𝑖 = 1.30) and
coldspots (v𝑗 = -1.39) of mouse activity.
29
Texas Tech University, Quinn C. Emmering, May 2014
Figure 2.2 Diagram of an 11 × 11 trapping grid illustrating three
scales at which rodent activity was examine: focal, five and nine
trap neighborhoods. Adjacent traps at 90˚ and 45˚ are spaced 15
and 21 m, respectively.
30
Texas Tech University, Quinn C. Emmering, May 2014
(a)
(b)
Figure 2.3 Mean (± s.e.m.) index of aggregation (Ia) of the six trapping grids for (a) mouse and (b)
chipmunk captures. Ia > 1.0 indicates an aggregation of counts; a value of Ia ≤ 1.0 or less indicates a
random or regular aggregation of counts, respectively.
31
Texas Tech University, Quinn C. Emmering, May 2014
Table 2.1 Sample sizes of veery and ovenbird nests from 1998 – 2012 found on
each rodent trapping grid: GC, GE, HC, HE, TC and TE.
32
Texas Tech University, Quinn C. Emmering, May 2014
Table 2.2 Results of nest-site selection analysis from restricted bootstrapped
tests of cluster index values conducted at three spatial scales for (a) veeries
and (b) ovenbirds. Nest sample sizes = n, standard deviation = SD; p-values are
estimated from the left tail of bootstrapped sampling distributions. Alpha
levels were adjusted using Bonferroni corrections for multiple comparisons to
α = 0.005
33
Texas Tech University, Quinn C. Emmering, May 2014
Table 2.3 Summary of logistic regression analyses examining the relationship
between nest failure, rodent activity and abundance in veeries and ovenbirds,
where PL = mice, TS = chipmunks; R = mice + chipmunks, abund = abundance,
focal = nearest trap station, 5-trap and 9-trap = metrics of activity in nest
neighborhoods, Log(L) = Log-likelihood function, K = the number of parameters,
wi = Akaikie weights and ML = model likelihood.
34
Texas Tech University, Quinn C. Emmering, May 2014
Table 2.4 Parameter estimates (β), standard errors (SE) and lower and upper
95% confidence intervals for parameters in the top ranked models for (a) veery
and (b) ovenbird from Table 2.3. Asterisks indicate confidence intervals not
overlapping zero.
35
Texas Tech University, Quinn C. Emmering, May 2014
CHAPTER III
EXAMINING SPATIAL ACTIVITY OF MAMMALIAN
PREDATORS AT SONGBIRD NEST SITES USING TRACK
PLATES
Introduction
Spatial heterogeneity in predator activity creates refugia for prey to occupy and
lower their risk of predation through reduced predator efficiency (Gause 1934,
Huffaker 1958, Hilborn 1975, Hastings 1977). Spatial refugia are especially important
for sessile, relatively immobile prey and vulnerable life history stages, that are not able
to easily flee from predators. Instead, immobile prey may rely more on finding
enemy-free space (Blaustein et al. 2004, Resetarits & Wilbur 1989, Walls 1995,
Schmidt et al. 2006, Schauber et al. 2009).
The eggs and young of songbirds provide a clear example of vulnerable, sessile
prey anchored to a central location for a month or more. On average, predation
accounts for 80% of nest failures in songbirds (Ricklefs 1969, Martin 1993, Remes et
al. 2012), hence nest-site selection is closely linked to fitness and population
demography (Martin 1998). Further, nest predation is thought to be principally
incidental (Vickery et al. 1992, Yanes & Suarez 1996, Schmidt et al. 2001a, Schmidt
et al. 2004a, Vigallon & Marzluff 2005, Schmidt et al. 2006), in that it is largely
undirected and occurs when generalist predators, while foraging for primary prey,
serendipitously encounter and consume less common secondary prey (Vickery et al.
36
Texas Tech University, Quinn C. Emmering, May 2014
1992, Schmidt et al. 2001a, McKinnon et al. 2013). If nest predation is largely
incidental, spatial refugia, or perhaps more appropriately “predator-poor” space, may
be created through predator behavior (Schmidt & Ostfeld 2003a, Schmidt & Schauber
2007). Examples include the economics of foraging where generalist predator activity
is concentrated in low risk, high reward patches. Less profitable patches are mostly
ignored, resulting in space that is periodically unoccupied by potential nest predators
(Brown 1988, Schmidt et al. 2001a, Schmidt & Ostfeld 2003a). Also, predator
territoriality or central-place foraging which spatially restricts or “anchors” predators
to core activity centers, such as the nests of raptors (Sergio et al. 2003) or mammalian
burrows and larders (Bowers 1995), leaves interstitial spaces between neighboring
territories relatively predator-poor for prey to take refuge (Mech 1977, Bowers 1995,
Sergio et al. 2003).
Localized measures of predator activity focused around nest sites might better
predict the probability of nest success instead of coarser measures of predator
abundance collected at large scales, yet such studies are uncommon (Lima
2009). After all, incidental, ground-foraging predators are more likely to detect and
locate nests as their proximity to the nest increases (Carthey et al. 2011) as the cues
associated with nests (e.g., nestling vocalizations and scent) become more
evident. Rodents, for instance, primarily search for prey via olfaction (Slotnick 2001)
and mice more successfully depredate visually cryptic prey where prey cues are
patchy and tightly associated with prey location (Carthey et al. 2011). Taken together,
incidental encounters of nests and localization of prey cues by generalist predators
37
Texas Tech University, Quinn C. Emmering, May 2014
likely happens, depending upon the species, within close proximity to the nest.
Therefore, nest vulnerability should be strongly correlated with local heterogeneity in
predator activity (Schmidt et al. 2006)
Several studies have demonstrated that birds proactively assess predation risk
and subsequently adjust habitat selection and/or reproductive investment (Spaans et al.
1998, Sergio et al. 2003, Forstmeier and Weiss 2004, Roos & Pärt 2004, Morton 2005,
Eggers et al. 2006, Fontaine & Martin 2006a, Fontaine & Martin 2006b, Zanette et al.
2011, further reviewed in Lima 2009). Birds may assess risk based on a single species
of predator that disproportionately depredates nests due to its numerical dominance
(e.g., Schmidt et al. 2001a, Forstmeier & Weiss 2004). Alternatively, prey may
simultaneously monitor and evaluate risk based on the combined activity of several
potential predators to make optimal breeding decisions (Fontaine & Martin 2006a,
Fontaine & Martin 2006b, Zanette et al. 2011). Which predators birds monitor
probably depends on a variety of factors including relative predator abundances,
previous predation by a specific species (Jackson et al. 1989, Lima 2009), and nest
type-location (e.g., open cup-shrub, enclosed dome-ground; Fontaine et al. 2007), to
name just a few. Determining which species are significant predators of nests within
an ecosystem has clear benefits for wildlife-management strategies and understanding
avian population demography (Benson et al. 2010, Ellis-Feleg et al. 2012).
Research from our study site has demonstrated a strong relationship between
veery (Catharus fuscescens) nest success and mouse activity surrounding nest sites
(~15 m radius), and furthermore, that veeries may preferentially select sites of below38
Texas Tech University, Quinn C. Emmering, May 2014
average mouse activity (Schmidt et al. 2006). Chipmunks, however, might only
weakly affect the reproductive success of open-cup nesters like veeries (Schmidt et al.
2001b). But the opposite may be true for ovenbirds. Emmering and Schmidt (2011;
Chapter 4 of this dissertation) recently found that ovenbirds and veeries placed nests
significantly further from playbacks broadcasting chipmunk vocalizations than sites
broadcasting frog calls or silent controls. However, the magnitude of the ovenbird
distance response was twice that of the veery response as ovenbirds placed nests 20 m
further from chipmunk playbacks than controls. Although other results have found
that combined rodent density (mice + chipmunks) had little effect on ovenbird daily
mortality (Schmidt & Ostfeld 2003a, Kelly et al., in review). These results suggest
that veeries may be less susceptible to nest predation by chipmunks but for ovenbirds
the opposite may be true, though this remains somewhat unclear. Analysis of rodent
activity within the nest neighborhood of ovenbirds has yet to be examined as earlier
research efforts focused more on veeries, and as a consequence, insufficient samples
of nests were found to examine these relationships (Schmidt et al. 2006).
Study objectives were to determine (1) which specific ground-foraging species
are important nest predators, (2) if predator activity is related to nest fate, (3) do nest
sites have lower predator activity than nearby random sites, and lastly (4) if predator
species and activity differ between two coexisting, ground-nesting passerines,
ovenbirds and veeries. To accomplish this, I present a new approach to quantify
predation risk surrounding nest sites using track plates (Connors et al. 2005). Track
plates are a highly portable and economical alternative to expensive video systems
39
Texas Tech University, Quinn C. Emmering, May 2014
(Ribic et al. 2012). By recording an animal’s tracks, plates can record the activity of
multiple individuals and species in space, including larger mesopredators. Further,
predator species can be identified based on their species-specific footprints.
Two main predictions were formed based on previous research. First, to
reduce the risk of nest predation ovenbirds and veeries would construct nests in
locations with lower predator activity than randomly chosen sites within their
territories (≤ 50 m). And second, nest sites that successfully fledged young would
have less nest-predator activity than nest sites that were depredated. Lastly, I
conducted a heterospecific comparison to explore if predator activity surrounding
ovenbird and veery nest sites differed. Ovenbirds and veeries both place their nests on
or near the ground (veeries ≤ 1.5 m in shrubs; Heckscher 2004, Bevier et al. 2005) so
presumably both species’ nests are similar in their accessibility to ground-foraging
predators. However, heterospecific differences in nest-site microhabitats preferences
(Martin 1995), construction and how parent birds manage risk at the nest site (Martin
et al. 2000, Eggers et al. 2008) could influence their respective predation risk and
where they place their nests. Based on evidence from our study site (described above),
I predicted ovenbird nests sites would have fewer chipmunks than nest sites of veeries.
Methods
The study was conducted on the property of the Cary Institute of Ecosystem
Studies (41º 50´ N, 73º 45´ W) located in Dutchess County, Millbrook, New York,
USA from 2006 – 2008 during the songbird breeding season (May – August). The
40
Texas Tech University, Quinn C. Emmering, May 2014
property consists of ~325 ha of continuous eastern deciduous forest with a oakdominated canopy (Quercus rubra and Q. prinus). Common understory trees and
shrubs include oak and maple (Acer spp.) saplings, witch hazel (Hamamelis
virginiana) as well as nonnative honeysuckle (Lonicera spp.) and Japanese barberry
(Berberis thunbergii).
Nest-predation risk was assumed to be an increasing function of predator
spatial activity. Conceptually, the degree of predator activity recorded in an area may
be similar where either an individual mouse exhibited high spatial activity or where
multiple mice were only moderately active (Schauber et al. 2009). Either case could
result in an equal number of track plates being tracked. In addition, we also assumed
nest predation is more commonly the result of undirected, incidental predation
(described above; Vickery et al. 1992, Yanes & Suarez 1996, Schmidt et al. 2004a).
Incidental predators were thus not expected to direct their foraging activities or shift
their territories to areas containing a nest.
Track plates were used to quantify nest-predator activity. Several track plates
placed in an area can be used to record the presence/absence of specific grounddwelling species and can therefore be used to assess their spatial activity. Track plates
are advantageous compared to live traps as traps may be less effective for estimating
small-scale activity of rodents. Traps restrict an animal’s movements for extended
periods of time and can affect the behavior of individuals by attracting them with bait
(i.e., “trap happy”) or making them “trap shy” after initial capture. Furthermore, track
plates (1) are highly portable, (2) allow predator identity to species by recording
41
Texas Tech University, Quinn C. Emmering, May 2014
species-specific footprints (3) record the activity of multiple individuals and predator
species, including mesopredators too large for traps and (4) are economical compared
to video-monitoring systems that have been used in other nest predation studies (e.g.,
Ribic et al. 2012). Track plates have also been successfully used at our study site by
other researchers to estimate rodent space use in investigations on the effects of mouse
activity on gypsy moth pupa survival (Connors et al. 2005, Schauber et al. 2009) and
the effects on space use by immunochallenged mice (Schwanz et al. 2011).
Individual track plates consisted of a 14  22 cm acetate sheet (fastened to
aluminum sheeting with two paper clips for support), which was coated with a waterresistance solution of graphite, alcohol and mineral oil that provides a dry surface for
recording footprints (Connors et al. 2005, Figures 3.1 – 3.3). However, a few nests (<
5) had only 8 track plates due to obstacles within 15 m of the nest such as standing
water or roads. Spacing between a track plate’s nearest plate was ~7.6 m and ~11.8 m
for 5-m and 15-m plates, respectively.
Arrays of track plates were setup around nest sites and at randomly chosen
control sites where no nests were found. Track plates were evenly distributed as an
array of 12 plates encircling nest and paired control sites with four plates placed at 5 m
and eight plates placed at 15 m from the nest (Figure 3.4). Control sites were placed
50 m away from each nest using a random compass bearing chosen a priori and
centered on an area or nesting substrate typical of ovenbirds or veeries (e.g., multistemmed shrub). To ensure other conspecific nests were not near focal control sites
we intensively searched for nests in the area during track plate set up but also
42
Texas Tech University, Quinn C. Emmering, May 2014
throughout the field season. If another conspecific’s nest was found, a new random
compass bearing was selected for the control site. Some paired control sites may not
have fallen within the same territory as its paired nest site. For veeries, mean territory
size based on radio-tracking males at our study site was 1.80 ha (≈ 75 m radius;
Belinsky, K.L. unpublished data). Assuming veery territories are roughly circular and
nests are centrally located, control sites placed 50 m away may still fall within the
territory of the paired nest site. Ovenbird territories, though never measured at our
study site, are typically smaller and range from 0.13 – 0.38 ha (Smith & Shugart 1987)
which could have a radius of ≈ 20 – 35 m. So for ovenbirds, paired control sites were
likely outside of the territory in which the nest was located.
Track plates without acetate sheets (aluminum backing only) were set out at
least three days prior to data collection to allow animals to habituate to the plates.
Acetate sheets were then added to the aluminum backing and inspected every other
day for six consecutive days for a total of three checks (or 36 total plate-checks/nest).
Individual plates were scored as “tracked” or “not-tracked” for each of the three
checks and for each species which included white-footed mouse (Peromyscus
leucopus), eastern chipmunk (Tamias striatus), gray squirrel (Sciurus niger), red
squirrel (Tamiasciurus hudsonicus), raccoon (Procyon lotor), opossum (Didelphis
virginiana) or other less common mesopredators such as foxes (e.g., Vulpes vulpes;
Connors et al. 2005). Acetate sheets were replaced when ≥ 25% of the sheet was
removed by tracks or rain. Significant rainfall events would usually remove the
graphite coating from a large number of acetate sheets in which case track data was
43
Texas Tech University, Quinn C. Emmering, May 2014
not recorded on the scheduled check date. Instead acetate sheets were replaced to be
checked again two days later. Occasionally, individual track plates were scored as
“unreadable” due to rain, plates being flipped or acetate sheets accidently affixed
upside down.
Over the three years of the study, track plates were placed around 48 ovenbird
and 59 veery nests (Table 3.2). I compared nest predator activity between: (1) paired
nest and control sites, (2) fledged and depredated nests and (3) ovenbird and veery
nest sites (hereafter, heterospecific analysis). Nests were considered fledged if they
successfully fledged at least one nestling. Only nests confirmed as depredated or
fledged were used in the fate analysis. For the heterospecific analysis, I pooled all
nests, regardless of their fate, but excluded any nests that only had eight track plates;
however, nests with only eight plates were used in the paired-site and fate analysis.
Seven metrics were examined based on different predator assemblages (Table
3.1). The first metric summed the total number of predator species recorded per plate
to assess predator richness around a nest site in which all ground-dwelling predators
were included (listed above). For instance, if a single plate had tracks from both mice
and chipmunks it was scored a ‘2’, if it had tracks from three nest predator species it
was scored a ‘3’, and so on. This metric was subdivided into three additional metrics
composed of smaller predator guilds (Table 3.1). The final three metrics only
estimated presence/absence of mice, chipmunks and all predator species by calculating
the proportion of plates that were scored as tracked or not tracked, ‘1’ or ‘0’,
44
Texas Tech University, Quinn C. Emmering, May 2014
respectively. All metrics were calculated based on 36 plate checks (12 plates × 3
check days).
Statistical analysis
Summed track plate scores over the three checks were used to calculate mean
metric values for each site (nest, control) and used for all subsequent analysis. Mean
scores were used instead of summed scores per site as the number of plates
occasionally varied per site (see methods). Control site values were not used in the
fate or heterospecific analyses.
All statistical analyses were performed using R version 3.0.0 (The R
Foundation for Statistical Computing). To test if paired nest and control sites differed
significantly in nest predator activity, paired t-tests were conducted for each metric
(Table 3.1). The paired t-test was well-suited for this experimental design not only
because of the paired association between nest and control sites but additionally it
does not assume groups are normally distributed with equal variances (Zar 1999).
Rather the paired t-test assumes that the differences between groups are normally
distributed (Zar 1999). Visual inspection of normal Q-Q plots and Shapiro-Wilk
normality tests confirmed that control and nest site differences for veeries were
normally distributed with the exception of the mesopredator metric, which I thereafter
excluded from the site analysis. For the ovenbird dataset, site differences for three
metrics (mice + chipmunks, mice and presence/absence) were not normally distributed
so they were normalized using arcsine-square root transformations. For the six
45
Texas Tech University, Quinn C. Emmering, May 2014
comparisons, a Bonferroni correction was used to adjust an alpha level of 0.05 to
0.008.
For the fate and heterospecific analyses, permutations were used to generate
sampling distributions of the test statistic under the null hypothesis that groups (i.e.,
dead vs. fledged or ovenbird vs. veery) were no different than would be expected if
mean metric values were randomly assigned to different groups. The absolute
difference between group means, DIFsim, was used as the test statistic to contrast with
the observed absolute difference between groups (DIFobs). P-values were estimated
by comparing DIFobs with the tails of the bootstrapped distribution by dividing all
values of |DIFsim| ≥ |DIFobs| by the number of bootstrapped iterations (e.g., Figure 3.5).
Alpha levels were adjusted for the seven comparisons using a Bonferroni correction to
adjust alpha to 0.007.
Bootstrapping of metric values was handled differently for the fate and
heterospecific analysis. In the fate analysis, values were bootstrapped (1000
iterations) between groups (depredated and fledged) but constrained within years to
obtain a mean group difference for each respective year that were then averaged to
compute a single value of DIFsim. The heterospecific analysis did not restrict
randomizations but instead shuffled metric values between groups and years (100,000
iterations). Lastly, only nests from 2006 and 2007 were used for the heterospecific
analysis as track plates were not placed around veery nests in 2008.
46
Texas Tech University, Quinn C. Emmering, May 2014
Results
Nest and control sites
Nest-predator activity was consistently higher at ovenbird nests than their
paired control sites (Table 3.3), but all metrics were nonsignificant (t48 ≥ -2.230, p ≥
0.030, α = 0.008; Table 3.3). Likewise, predator activity also did not differ
significantly between veery nests and their paired control sites for all of the metrics
examined (-1.575 ≤ t58 ≤ -0.346, p ≥ 0.113; Table 3.3).
Nest fate
Predator activity metrics did not differ significantly between depredated and
fledged ovenbird nests (all metrics, p ≥ 0.030, α = 0.007; Table 3.4). Depredated
veery nests had, with the exception of the mesopredator metric, consistently higher
predator activity than fledged nests, however, these differences did not differ
significantly (all metrics, p ≥ 0.252; Table 3.4).
Heterospecific comparison
The proportion of track plates recording nest predators (i.e., presence/absence
metric) and chipmunks was 10% greater around veery nests than ovenbird nests (Table
3.5) and this difference was significant for both comparisons (p < 0.0001, α = 0.007).
However, the five remaining metrics were not significantly different between veery
and ovenbird nests (p ≥ 0.011, α = 0.007).
47
Texas Tech University, Quinn C. Emmering, May 2014
Discussion
Analysis of track plate results found no relationship between predator activity
and nest success, for both songbird species (Table 3.4). Results also suggest that
veeries and ovenbirds do not select nest sites with lower nest predator activity
compared to unused sites near or within their territories. Though not significant,
results showed just the opposite from our predictions – nest predator activity was
higher around nest sites. Ovenbird nest sites had higher total predator activity than
random sites and this difference was likely driven by rodents as only the all predator
and all rodent metrics were higher (Table 3.3).
For veeries, track plate results were similar to findings that analyzed mouse
and chipmunk captures from permanent trapping grids over the course of the songbird
breeding season (Chapter 2). Both studies failed to find a relationship between veery
nest-site selection, nest success and small-scale rodent activity (Tables 2.2 – 2.4).
Likewise, analyses of trapping data found ovenbirds also did not place nests in areas
of lower rodent activity (Table 2.2), but differently ovenbird nest success was related
to chipmunk activity within a 30 × 30 m area around their nests (Tables 2.3 – 2.4).
Track plates and trapping grids differed markedly in the temporal scale at which
sampling occurred (7 days vs. 3 months, respectively) which may partly explain why
study results differed.
At least at the spatial scale and metrics we examined, our findings suggest that
veeries and ovenbirds do not select nest locations based on the activity of several
predator species to reduce the risk of nest predation or that there is a relationship with
48
Texas Tech University, Quinn C. Emmering, May 2014
local activity and reproductive success. Several non-mutually exclusive hypotheses
could explain our results. One alternative is that the species we examined (Table 3.1)
are not important predators of veery or ovenbird nests. Certainly, some of the
predators investigated may not be as significant as others, but there is good evidence
from our study site (including unpublished video evidence) and other systems that
mice (Reitsma et al. 1990, Williams and Wood 2002, Cain et al. 2003, Clotfelter et al.
2007, Schmidt and Ostfeld 2008, Cox et al. 2012), chipmunks (Cain et al. 2003, King
and DeGraaf 2006, Schmidt et al. 2008, Haché et al. 2014), squirrels (Reitsma et al.
1990, Cain et al. 2003, Haché et al. 2014) and mesopredators (Crooks & Soulé 1999,
Schmidt 2003, Rodewald & Kearns 2011, Ellis-Feleg et al. 2012) all regularly
depredate nests.
Spatial refugia may be less effective at preventing predation from highly
mobile avian nest predators like birds of prey which could partly explain the lack of a
relationship between mammal activity and nest success. In years of low rodent
abundance, birds of prey (e.g., barred owls, Strix varia, broad-winged hawks, Buteo
platypterus) may behaviorally respond by switching to alternative prey including
songbird nests and fledglings (Jędrzejewska & Jędrzejewski 1998, Selås 2001,
Schmidt & Ostfeld 2003, Schmidt et al. 2008). Over the 3-year study duration, rodent
densities (mice and chipmunks) fluctuated dramatically which included a 15-year high
in 2007 of 83.4 animals ha-1 to 22.5 animals ha-1 in 2008, nearly a four-fold decrease.
Crash years, as seen in 2008, could result in higher nest predation by avian predators
to compensate for now sparse rodent prey (Schmidt & Ostfeld 2003, Schmidt et al.
49
Texas Tech University, Quinn C. Emmering, May 2014
2008). Whether nesting songbirds are able to assess and respond appropriately to such
changes in the predator community remains unresolved.
Another explanation for observed similarities in predator activity between sites
could be that predator activity varied little at the scale of bird territories. If the scale at
which predator activity is spatially autocorrelated exceeds that of the territory scale
then territory-level heterogeneity would be low, implying that refugia may not exist.
Leastwise, prospecting birds might be limited in the number of spatial refugia
available for nesting or in their ability to perceive minor variation in activity.
Research by Schauber et al. (2009) at our study site suggests this might be the case.
They found that the spatial scale of autocorrelation in mouse activity was frequently
larger than the scale at which we placed paired track plate arrays or songbird
territories (range 44 – 1000 m; Schauber et al. 2009). Whether the activity of other
nest predators is spatially autocorrelated at similar scales is unknown.
These results also raise the possibility as to whether the distribution of groundforaging birds and generalist predators are actually spatially associated. White-footed
mouse and chipmunk diets during the spring and summer months consist largely of
arthropod prey (Hamilton 1941, Batzli 1977, Synder 1982, Lackey 1985, Wolff et al.
1985) as are the diets of veeries and ovenbirds especially during the breeding season.
Food abundance is considered one of the principle determinants of territory quality for
breeding songbirds (Rodenhouse et al. 1997) and is particularly well documented for
ovenbirds (e.g., Smith & Shugart 1987, Burke & Nol 1998). Therefore groundforaging rodents and songbirds may have similar site preferences based on shared prey
50
Texas Tech University, Quinn C. Emmering, May 2014
causing their densities to be spatially correlated. In such case, settling birds might
demand what Brown & Kotler (2004) describe as “hazardous duty pay” where settling
birds would require greater food resources to compensate for the increased risk of nest
predation. Alternatively, birds may simply prioritize safety ahead of food abundance
biasing their selection of territories towards those with fewer predators. Durrant
(1998) discovered a similar case where cheetahs were more often found in sites of
lower food abundance to avoid their competitors and also intra-guild predators –
hyenas and lions. Or parent birds may manage risk using other behavioral strategies
including reduced parental activity at the nest (Martin & Ghalambor 1999, Martin et al.
2000), defensive strategies (Caro 2005, Ellison & Ribic 2012, Gloag et al. 2013) or
nest-site characteristics (Martin 1998, Marzluff 1988, Eggers et al. 2006, Stevens 2013,
Lovell et al. 2013).
Predicted heterospecific differences in predator activity were partly supported.
Predator activity was greater at veery compared to ovenbird nest sites and this
difference appeared to be chiefly caused by chipmunks (Table 3.5), which is
consistent with results from an experimental playback study conducted at our study
site (Emmering & Schmidt 2011; Chapter 4 of dissertation). In response to playbacks
of chipmunk vocalizations that simulated “hotspots” of chipmunk activity, ovenbirds
and veeries placed nests significantly further from plots broadcasting chipmunk calls
compared to control plots that broadcasted frog calls or were silent. However,
ovenbirds nested at twice the distance of veeries (20 m vs. 10 m, respectively) from
chipmunk playbacks. This differential aversion to perceived chipmunk activity could
51
Texas Tech University, Quinn C. Emmering, May 2014
partly explain the large difference in chipmunk activity recorded around their
respective nest sites and together suggests that ovenbird nests may be more vulnerable
to predation by chipmunks. In addition, microhabitat nest-site preferences could
explain the greater chipmunk activity around veery nest sites. Veeries frequently nest
in shrubs such as barberry and honeysuckle both of which can form larger patches
which may provide food (i.e., berries) or protective cover from predators for
chipmunks (e.g., Forstmeier & Weiss 2004) or other nest predators (Dutra et al. 2011).
Ovenbirds differ in that they are obligate ground-nesters, building dome-shaped nests
that are frequently placed in locations with sparse to no cover (Van Horn & Donovan
1994). Since veeries experience greater predation from mice (Schmidt et al. 2001a,
Schmidt & Ostfeld 2003a), we predicted mouse activity would be greater at their nests.
This prediction was not supported as mouse activity was very similar around the two
species nest sites (Table 3.5).
52
Texas Tech University, Quinn C. Emmering, May 2014
Table 3.1 Description of track plate metrics. For each metric, mean counts were calculated from 36 total plate checks (12
plates × 3 checks) for each nest. Some species, like red fox (Vulpes vulpes), were very uncommon and thus were grouped
into a single mesopredator guild; unknown mesopredator species were also included within the mesopredator guild.
Metric
Description
Species included
Summation of all nest predator species that tracked a
All predators
plate, e.g., if a plate was tracked by a mouse, chipmunk
and raccoon, the plate was scored a '3'
White-footed mouse, eastern
chipmunk, gray squirrel, red squirrel,
raccoon, opossum and other
mesopredators*
White-footed mouse, eastern
All rodents
Summation of all rodent species that tracked a plate
chipmunk, gray squirrel and red
squirrel
Mice and
chipmunks
Mesopredators
Mice
Chipmunks
Summation of mice and chipmunks
Summation of mesopredators
Presence/absence of a track left by Peromyscus spp.
only
Presence/absence of a track left by Tamias striatus only
53
White-footed mouse and eastern
chipmunk
Raccoon, opossum, fox, coyote and
mustelids
White-footed mouse only
Eastern chipmunk only
Texas Tech University, Quinn C. Emmering, May 2014
White-footed mouse, eastern
Presence/absence
Presence/absence, scored as ‘1’ or ‘0’ respectively, of a
chipmunk, gray squirrel, red squirrel,
track left by any nest predator under consideration
raccoon, opossum and other
mesopredators
54
Texas Tech University, Quinn C. Emmering, May 2014
Figure 3.1 Image illustrating the application of
Figure 3.2 Image demonstrating the water resistance of a
water-resistant solution to acetate sheet.
track plate coated with graphite solution.
55
Texas Tech University, Quinn C. Emmering, May 2014
(a)
(b)
(c)
Figure 3.3 Examples of footprints left on track plates by one or more of the following: (a) mice, (b) raccoons and
(c) chipmunks. Track plates measured 14  22 cm while the area covered by graphite was slightly less by 1-2 cm.
56
Texas Tech University, Quinn C. Emmering, May 2014
TRACK PLATE ARRAY
10m
5m
nest
15m
unk tracks
Figure 3.4 Illustration of track plate array encircling a nest. Control sites
were arranged similarly. Plates were placed at 5 m and 15 m from the
Arrays –nest
Atornest
& random
control
controlsites
site center.
Plate dimensions
were approximately 14  22
cm and
hence are not
to scale
in figure.
sites ~50m
distance
from
nest.
57
Texas Tech University, Quinn C. Emmering, May 2014
All rodent species
Figure 3.5 Histogram displaying the bootstrapped sampling distribution of the
absolute difference between depredated and fledged veery nests (DIFsim) from
1000 bootstrapped iterations. The dotted red line represents the observed mean
difference, DIFobs. All values of DIFsim ≥ the value of DIFobs constitute the
estimated p-value of 0.621 (i.e.,, 621/1000 iterations).
58
Texas Tech University, Quinn C. Emmering, May 2014
Table 3.2 Sample sizes of nests by year and reproductive fate for each
analysis conducted. Note each nest was also paired with second track plate
array (i.e., control site) ~50 m away.
59
Texas Tech University, Quinn C. Emmering, May 2014
Table 3.3 Results of nest-site analysis from paired t tests to compare means of nest and control sites. Nests were
pooled from all three years of the study to report observed means of each metric. Alpha levels were adjusted using
Bonferroni corrections for multiple comparisons to α = 0.008. Refer to Table 3.1 for detailed descriptions of each
60
Texas Tech University, Quinn C. Emmering, May 2014
Table 3.4 Results of fate analysis from bootstrapped tests of the mean
difference between dead and fledged nests. Nests were pooled from all three
years of the study to calculate observed means of each metric. DIFobs is the
observed difference of mean dead and fledged nests. Alpha levels were
adjusted using Bonferroni corrections for multiple comparisons to α = 0.007.
61
Texas Tech University, Quinn C. Emmering, May 2014
Table 3.5 Results of heterospecific analysis from bootstrapped tests of the
mean difference between veery and ovenbird nests. Means are of all nests
for 2006 and 2007 combined. DIFobs is the observed difference of the mean
between veery and ovenbird. Alpha levels were adjusted using Bonferroni
corrections for multiple comparisons to α = 0.007. Refer to Table 3.1 for
62
Texas Tech University, Quinn C. Emmering, May 2014
CHAPTER IV
NESTING SONGBIRDS ASSESS SPATIAL HETEROGENEITY
OF PREDATORY CHIPMUNKS BY EAVESDROPPING ON
THEIR VOCALIZATIONS
Introduction
Environmental heterogeneity creates uncertainty for organisms that limits their
ability to select behavioral strategies appropriate for the current (or anticipated future)
conditions. Information reduces uncertainty (Dall et al. 2004); hence acquiring
information is critical for optimal decision-making. Information on predation risk is
particularly important for organisms selecting breeding sites, as safety from predators
is an important component of site quality (Martin 1995; Blaustein et al. 2004;
Forstmeier & Weiss 2004; Rodenhouse et al. 2003). Thus organisms should attend to
information which enables them to select sites with a low probability of predation.
A rapidly growing body of evidence has documented that birds use a multitude
of information sources when selecting breeding habitats and territories (Viitala et al.
1995; Part & Doligez 2003; Safran 2004; Betts et al. 2008; Hromada et al. 2008;
Forsman & Martin 2009; Lima 2009). Sources include pre-breeding cues such as the
presence of conspecifics (Fletcher 2007) and post-breeding cues from prior years, such
as personal reproductive success (Haas 1998) or the presence of fledglings (Betts et al.
2008). Conspecific presence may be useful for finding appropriate habitat patches,
63
Texas Tech University, Quinn C. Emmering, May 2014
but less so for estimating small-scale spatial heterogeneity necessary for selecting
individual territories or nest-sites (e.g., Morton 2005). Likewise, performance-based
cues, such as personal or conspecific reproductive success, can be useful as integrated
measures of territory quality; however, they require high temporal correlation between
years to accurately forecast future success (Safran 2004). Thus in environments with
low annual predictability, for example, in pulsed-resource systems (McShea 2000;
Ostfeld & Keesing 2000; Clotfelter et al. 2007; Schmidt & Ostfeld 2008), birds should
place a premium on using current information, when possible, to locate refugia from
nest predators. However, species differences in life histories, sensory ecology, and/or
habitat constraints may result in interspecific differences in information use.
Although previous studies have documented that birds make breeding habitat
adjustments in response to predation risk (e.g., Marzluff 1988; Sergio, Marchesi &
Pedrini 2003; Forstmeier & Weiss 2004; Fontaine & Martin 2006; Schmidt, Ostfeld &
Smyth 2006; Peluc et al. 2008; Lima 2009), only a handful have identified the
source(s) of information that they use (e.g., Eggers et al. 2006; Forsman & Martin
2009; Mönkkönen et al. 2009). Prey should be sensitive to direct cues of risk (i.e.,,
predator activity) given by predators themselves. However, experimental
manipulation of individual cues are uncommon, and while predator removals can
demonstrate the prey’s behavioral response they do not determine which individual
cues are used to assess to risk. We suggest prey may frequently exploit the acoustic
signals of predators acquired through ‘eavesdropping’, the process where unintended
receivers intercept the signals of others to acquire information (Peake 2005).
64
Texas Tech University, Quinn C. Emmering, May 2014
Communication is often clearly audible and in the public domain thereby reducing the
time and acquisition costs of directly assessing predators to facilitate avoidance.
Our work on ground-nesting passerines has previously demonstrated (1) rodent
nest predators, including white-footed mice (Peromyscus leucopus) and eastern
chipmunks (Tamias striatus), have significant spatial heterogeneity in activity
(Schmidt, Ostfeld & Smyth 2006; Goodwin et al. 2005); (2) there is a strong negative
correlation between veery (Catharus fuscescens) fledging success and rodent activity
surrounding nest sites (Schmidt, Ostfeld & Smyth 2006; Schmidt & Schauber 2007);
and (3) veeries build nests at sites with below average rodent activity (Schmidt,
Ostfeld & Smyth 2006), suggesting that they acquire and use information associated
with rodent activity to select nest sites with a lower risk of nest predation. Because
annual fluctuations in rodent abundance can be of two orders of magnitude (Ostfeld,
Jones & Wolff 1996; Jones et al. 1998; Schmidt & Ostfeld 2008), the predictability of
spatial refugia for nesting passerines based on prior success may be limited. This
suggests birds may place a premium on information available during the current
breeding period to assess spatial heterogeneity in nest-predation risk. Chipmunks are
vociferous throughout the breeding season, and their frequent calling bouts with
neighboring conspecifics may provide a proximate cue of predation risk for nesting
passerines to assess and avoid “hotspots” of chipmunk activity. As chipmunks will
depredate eggs, nestlings and even fledgling birds (Schmidt, Rush & Ostfeld 2008),
nest-site and territorial adjustments in response to acoustic cues emitted by chipmunks
may substantially lower the risk of nest and juvenile predation. Such behavior may be
65
Texas Tech University, Quinn C. Emmering, May 2014
paramount for coping with high densities of nest predators that occur in the wake of
masting seed crops (McShea 2000; Schmidt, Ostfeld & Smyth 2006, Schmidt &
Ostfeld 2008).
To test whether passerines eavesdrop on chipmunks as a means to assess local
chipmunk activity and consequently avoid areas of higher perceived nest predation
risk, we conducted an experimental playback study to create spatial heterogeneity in
perceived risk. We used three types of playbacks at the beginning of the breeding
season: (1) chipmunk vocalizations (increased nest predation risk), (2) gray tree frog
calls (Hyla versicolor; low-risk control) and (3) no playback (silent control). We
examined two potential responses to experimental playbacks: (1) nest-site placement
in two ground-nesting passerines, veeries and ovenbirds (Seiurus aurocapilla) and (2)
bird activity in two guilds of nesting birds, ground and canopy. We predicted veeries
and ovenbirds would nest further away from chipmunk playback plots versus control
plots, and the presence or activity of ground-nesters would be reduced near chipmunk
playback plots relative to control plots, whereas canopy-nesters would show no
change in activity.
Methods
General procedures
Our experiment was conducted in oak-dominated, eastern deciduous forest on
the property of the Cary Institute of Ecosystem Studies located in Dutchess County,
Millbrook, New York, USA. In spring 2006 we established 32 experimental plots: 16
66
Texas Tech University, Quinn C. Emmering, May 2014
broadcasting chipmunk vocalizations and 16 silent control plots. Each playback plot
consisted of 3 CD/speaker stations (T100-CD Trutech, USA; 40-1441 speakers, Radio
Shack, Fort Worth, TX, USA) housed in small plastic containers (13  17  34 cm) to
protect the equipment from the elements. Playback stations were equally spaced 30 m
apart forming an equilateral triangle and faced inwards. Control plots were similarly
arranged; however no playback equipment was present and vertices were marked with
a single small flag. All plots (playback and control) were separated by at least 200 m,
twice the maximum distance at which we measured nest locations, to avoid
pseudoreplication of nests found on experimental plots. Playbacks began May 7 prior
to territory establishment (the first nest under construction was found on May 12 and
May 17 in 2006 and 2007, respectively) of focal migrant songbirds through peak
settlement (after June 15) and ran daily for the approximate six week experiment in
each study year. Each speaker intermittently broadcasted chipmunk vocalizations [8085 dB SPL (re. 20µPa) at 5 m] for ~12 min/hr beginning in early morning when fresh
batteries were added (~0630-0900) until ~1600 when batteries were depleted.
Playbacks were halted on days of heavy rain (2-3 days/year).
Chipmunk and frog playbacks were produced from recordings of multiple
individuals at the site using a Marantz PMD-670 field recorder (Middlesex, UK)
connected to a Sennheiser ME-62 microphone (Buckinghamshire, UK) and Telinga
parabolic disc (Tobo, Sweden). Chipmunk vocalizations included ‘chips’, ‘chucks’
and ‘trills’ (Dunford 1970; Elliot 1978) from ~12 wild individuals (plus 8 additional
individuals added in 2007). Recordings were compiled and edited using Raven Pro
67
Texas Tech University, Quinn C. Emmering, May 2014
1.3 (Cornell Laboratory of Ornithology, Ithaca, NY, USA) to remove background
noise and untargeted vocalizations. Final chipmunk and frog playbacks thus contained
several call types from multiple individuals.
We amended the experimental design in 2007 to include a non-silent,
procedural control using gray tree frog calls. Gray tree frogs are common at the site
and unlikely to elicit either heterospecific attraction or repulsion, which is a risk if
using avian vocalizations (Fletcher 2008). To accommodate the frog treatment in
2007 we added 8 plots increasing the total to 40. In addition, we rotated treatment
types between years. The 16 silent control plots used in 2006 were switched to
chipmunk playbacks, whereas 12 of the 2006 chipmunk plots played frog calls. The
remaining 4 chipmunk plots used in 2006 and the 8 new plots were used as silent
controls in 2007. In 2007 we also added a regular rotation (every 3 days) of 3 unique
chipmunk exemplars each playing a different assortment of chipmunk calls.
Otherwise, all other previously described methods used in 2006 were continued in
2007.
Nest-site selection
We searched for nests within a 100 m radius of all stations. Search efforts
focused only on veeries and ovenbirds whose nests are vulnerable to chipmunk
predation. Ovenbirds nest exclusively on the ground and veeries directly on or close
to the ground in shrubs (≤ 1 m). After nests became inactive, we measured the
distance of each nest to the nearest playback or silent station (i.e.,, flagged vertex)
68
Texas Tech University, Quinn C. Emmering, May 2014
with a measuring tape. For analyses, we included only those nests that had been built
≤ 2 days following termination of the playbacks.
Point counts
To examine if adult birds had reduced activity in areas perceived as risky (i.e.,,
chipmunk playback plots), we conducted 3 replicate, 5-min point counts (Hutto,
Pletschet, & Hendricks 1986) from the center of all study plots. We recorded all
migrant species that bred on site grouping them into two nesting guilds: a groundshrub-nesting (hereafter, ground-nesting guild) and a canopy nesting guild. While
canopy nests may still be vulnerable to predation by chipmunks, they are less likely to
be encountered by chipmunks that bias the majority of their activities on the ground;
hence canopy-nesters were considered at relatively lower risk than ground-nesters.
The ground-nesting guild included ground-shrub-nesting veery and obligate groundnesting ovenbird, worm-eating warbler (Helmitheros vermivorus), and black-andwhite warbler (Mniotilta varia). The canopy-nesting guild consisted of yellow-billed
cuckoo (Coccyzus americanus), black-billed cuckoo (Coccyzus erythropthalmus),
blue-gray gnatcatcher (Polioptila caerulea), eastern wood-pewee (Contopus virens),
great-crested flycatcher (Myiarchus crinitus), red-eyed vireo (Vireo olivaceus),
yellow-throated vireo (Vireo flavifrons), black-throated green warbler (Dendroica
virens), scarlet tanager (Piranga olivacea), rose-breasted grosbeak (Pheucticus
ludovicianus), and Baltimore oriole (Icterus galbula). The first count was conducted 7
days (2006) and 14 days (2007) after starting the experiment; thereafter each replicate
69
Texas Tech University, Quinn C. Emmering, May 2014
point count was separated by 7-10 days over the course of the experiment. Counts
were performed between 0500-0900 on days without inclement weather (rain, fog, or
steady breeze). After arrival to the plot, counts were delayed 5 minutes before
initiating to allow birds to resume normal activities. During the count we recorded all
birds detected by sight or sound within 50 m from the center of plots. Each individual
detected was placed within one of 5 distance categories: 0-10 m, 10-20 m, 20-30 m,
30-40 m, and 40-50 m. If, however, an estimated distance was difficult to assign we
divided the sighting (assigned ½) between the two adjacent categories. Playbacks
were turned on after the completion of the point count.
Assessment of mammalian nest-predator activity using track plates
To determine if chipmunks or other mammalian nest predators [gray squirrels
(Sciurus niger), white-footed mice and raccoons (Procyon lotor)] responded to our
playbacks, we analyzed track plates that recorded predator activity in bird territories
from a concurrent study (Chapter 3 of dissertation). Track plates consisted of a 14 
22 cm acetate sheet (fastened to aluminum sheeting for support) coated with a waterresistance graphite/alcohol/oil solution to record activity via footprints (Connors et al.
2005). Track plates were distributed as an array of 12 plates (4 at 5m, 8 at 15m)
encircling veery and ovenbird nests on chipmunk plots (n=29 nests) and control plots
(n=34 nests). Arrays of aluminum backing were first set out at least 3 days prior to
data collection to allow animals to habituate to the plates. Acetate sheets were then
added to aluminum backing and inspected every other day for 6 consecutive days for a
70
Texas Tech University, Quinn C. Emmering, May 2014
total of 3 checks. Plates were scored as “tracked” or “not-tracked” by individual
species (see Connors et al. 2005). We calculated the proportion of 36 plates (3 checks
 12 plates) tracked by each species over the 6-day period for each array of plates.
Track plate data were analyzed using ANOVA with the arcsine-transformed
proportion of plates tracked as the dependent variable and year and playback treatment
(chipmunk, control) as independent variables. We found no difference in the
proportion of track plates by treatment marked by chipmunks (F1,60 = 0.232, p =
0.632), mice (F1,60 = 0.004, p = 0.951), squirrels (F1,60 = 1.425, p = 0.237), or raccoons
(F1,60 = 0.416, p = 0.521), suggesting that movements of chipmunks and other
potential nest predators were unaffected by playbacks. Detections of other mammals
(e.g., opossums, Didelphis virginiana) were too infrequent to analyze.
Statistical analysis
Nest distance data were analyzed using ANOVA (SYSTAT 12.02.00, Chicago,
IL, USA) with distance as the dependent variable; independent variables included
playback treatment, year, and species (veery, ovenbird) and all 2- and 3-way
interaction terms. For point counts, we estimated the mean number of birds recorded
for the 3 replicate counts on each experimental plot. We hypothesized that birds were
more likely to reduce their activity nearer than farther from chipmunk playback
stations. Shifting activities (or territories) only a short distance away, for example 2030 m, may not seem substantial, however given the scale of chipmunk territories
(0.03-0.40 ha; Snyder 1982) this distance would likely be sufficient to lower predation
71
Texas Tech University, Quinn C. Emmering, May 2014
risk. For analysis, we used a hierarchical approach first using Akaike’s information
criterion (AICc, corrected for small sample sizes) to determine if bird presence/activity
based on point count detections changed over distance between treatments, and
furthermore, what distance (comparing 10 m increments) was most parsimonious with
our data. AICc scores were based on mixed effects general linear models (GLM) with
year, distance (near vs. far), treatment and their interactions as fixed effects, and site
(i.e.,, plot) as a random effect. For each nesting guild we compared five models. Four
models uniquely categorized near and far distances (0-10 vs. 10-50 m, 0-20 vs. 20-50
m, etc.), and a fifth model had no distance term (0-50 m). We identified the most
parsimonious model by the lowest AICc score, and used that model to assess statistical
significance of the fixed and random effects specified above.
Prior to analyses we examined the 2007 data to corroborate that procedural
activities did not affect bird responses and pooled the two control (silent and frog)
treatments. To do this, we compared nest distance and point count data across the two
controls using two-tailed t-tests. Nest distances were nearly identical between the two
control treatments (difference in mean distances between frog and silent treatments
was 0.4 m and 3.7 m for veery and ovenbirds, respectively; p > 0.80 in both
comparisons). Point count data on the two controls were analysed for all distance
categories (0-10, 0-20 m, etc.) and the two nesting guilds (ground, canopy); all
analyses were non-significant (t22 ≥ -1.967, p ≥ 0.062). Thus, we concluded
procedural effects were not present justifying combining frog and silent controls into a
single control group in all subsequent analyses.
72
Texas Tech University, Quinn C. Emmering, May 2014
Results
Nest-site selection
During the two-year study we found 70 veery (n=27 for chipmunk, n=43 for
control) and 32 ovenbird (n=15 for chipmunk, n=17 for control) nests on experimental
plots. On average, birds placed nests at significantly greater distances (playback
treatment: F1,94 = 7.347, p = 0.008) from chipmunk playbacks than controls (frog +
silent; Fig. 1). The species  treatment effect was not statistically significant (F1,94 =
2.179, p = 0.143). However, ovenbirds responded more strongly, on average, to
chipmunk playbacks than veeries in both absolute mean distances (57.8 ± 4.6 vs. 37.8
± 6.1 m for ovenbird and veery, respectively) and the increased distance relative to the
control (20.0 m vs. 9.4 m, for ovenbird and veery, respectively; Fig. 1). All other
effects and interactions were not significant.
Point counts
For the ground-nesting guild, the top-ranked model included two distance
categories at 0-30 m (near) and 30-50 m (far); Table 1. All lower ranked models had
very little support (AICc ≥ 9.90, model weight () < 0.01; Table 1). GLM analysis
of the top model indicated significant year (F1,100 = 7.6, p = 0.007) and distance (F1,100
= 38.4, p < 0.001) effects. Most relevant, the total number of ground-nesting birds
detected did not differ by treatment (main treatment effect: F1,100 = 0.05, p = 0.82), but
fewer ground nesters were detected near chipmunk plot centers relative to control
73
Texas Tech University, Quinn C. Emmering, May 2014
plots, while more ground nesters were detected far from chipmunk plot centers relative
to control plots (treatment × distance interaction: F1,100 = 12.6, p = 0.001; Fig. 2a).
For the canopy-nesting guild, the top-ranked model included two distance
categories at 0-30 m (near) and 30-50 m (far); Table 1. All lower ranked models had
very little support (AICc ≥ 32.1; Table 1). GLM analysis of the top model indicated
a distance effect (F1,100 = 21.0, p < 0.001) with fewer birds recorded within 0-20 m
versus 20-50 m; all other main effects (F1,100 < 2.6, p > 0.10) and interaction (F1,100 <
3.14, p > 0.08) were not significant (Fig. 2b).
Discussion
Our results indicate that two ground-nesting species, ovenbirds and veeries,
eavesdrop on chipmunk vocalizations and use this cue to adjust nest-site placement,
and furthermore, to alter territory settlement and/or reduce conspicuous behaviors (e.g.,
calling frequency). Relative to controls, nests of ovenbirds were, on average, 20 m
farther from chipmunk playback stations than control stations (Fig. 1). Chipmunk
home ranges vary in size from 0.03-0.40 ha (i.e.,, radius between 9.8-35.7 m for
circular home ranges; Dunford 1970; Snyder 1982) with greater activity usually
focused in the middle of territories at “activity centers” around burrow entrances
(Elliot 1978; Bowers 1995). Hence, a shift of 20 m from areas with high perceived
chipmunk activity is likely to place the nest outside of a chipmunk’s home range, or at
very least, farther from risky activity centers. In addition, newly fledged songbirds are
weak fliers remaining relatively close to the nest and within the natal territory for
74
Texas Tech University, Quinn C. Emmering, May 2014
several days post-fledging (Rush & Stutchbury 2008; Vitz & Rodewald 2010).
Therefore, constructing nests away from areas of greater chipmunk activity could
additionally function to reduce fledgling mortality as chipmunks are also important
predators of fledgling birds at our study site (Schmidt, Rush & Ostfeld 2007).
Veeries too nested, on average, farther away from chipmunk playbacks relative
to controls, however the strength of the response was less than half of what we
observed with ovenbirds, and may be less biologically significant. There are several,
non-mutually exclusive hypotheses for the weaker response observed in veeries. First,
mice may be more important nest predators than chipmunks based on experimental
evidence (e.g., Schmidt et al. 2001). Second, at our study site veeries show a strong
preference for mesic habitat surrounding small forested wetlands (Schmidt et al. 2005).
Thus habitat preferences may behaviorally constrain veeries (more so than ovenbirds)
from making significant nest-site or territory adjustments (unlike Betts et al. 2008).
This hypothesis remains untested, and sites containing both mesic and xeric habitats
were used for the study. Lastly, veeries may also use alternative sources of
information, such as returning to territories and nest sites where they successfully
raised a brood in the past (Greenwood & Harvey 1982; Switzer 1993; Schmidt 2001;
Hoover 2003). Repeated use of territories and nest sites by veeries is high at our site,
for example, individual shrubs have been repeatedly used as nest substrates 3-5 times
over a 10 year period (K.A. Schmidt, unpublished data). However, without a colorbanded population we lacked the ability to quantify fidelity of individual birds. While
veeries may not respond like ovenbirds by constructing nests further away from
75
Texas Tech University, Quinn C. Emmering, May 2014
chipmunk playbacks this does not necessarily mean they do not eavesdrop on
chipmunk calls to make nest-site decisions. Veeries may, alternatively, show
plasticity in other features at the nest-site. For example, veeries may place nests in
denser vegetation for greater concealment (Marzluff 1988; Eggers et al. 2006), or
conversely, nest in less dense vegetation that would expose chipmunks to their own
predators (as in Forstmeier & Weiss 2004). Also, unlike obligate ground-nesting
ovenbirds, veeries are not constrained to adjust the height of their nests.
The number of birds detected during point counts corroborated the nest
placement data. Ground-nesting species, which included ovenbirds and veeries, were
less frequently detected near (0-30 m) chipmunk playback plots relative to controls
(Fig. 2a). This pattern could be the result of two non-mutually exclusive responses.
One explanation is that there was a relative shift in the distribution of ground-nesting
species from plot centers to the periphery, as predicted if birds adjusted territorial
margins to avoid areas perceived as risky. Alternatively or concomitantly, adult birds
may have displayed less overt behaviors (e.g., singing frequency; further reviewed in
Lima 2009) in response to chipmunk playbacks lowering the detectability of
individuals. For example, Fontaine and Martin (2006) found when they removed
predators from experimental plots that singing activity of several passerine species
significantly increased relative to plots with intact predator communities. We found a
similar behavioral response to the perceived presence of nest predators, but uniquely
we demonstrated a differential response between two nesting guilds based on the
foraging niche of the predator as ground-nesting species responded to playbacks of
76
Texas Tech University, Quinn C. Emmering, May 2014
chipmunks while canopy-nesting species with relatively safer and higher nest sites did
not.
We used track plates to monitor the spatial activity of chipmunks and other
potential nest predators across treatments, and were able to rule out the alternative
hypothesis that birds responded to differences in the activity of rodents and other
predator species and not strictly to our experimental playbacks. We did not examine
changes in the frequency of natural chipmunk vocalizations in response to
experimental playbacks. However, if wild chipmunks increased their vocalizations in
chipmunk plots it would only reinforce our experimental protocol, whereas if
increased vocalizations occurred off plots it would diminish differences between
treatments and impede our ability to demonstrate significant effects. Thus, either
scenario strengthens our findings that chipmunk vocalizations are an important
influence on avian settlement. Lastly, it is possible chipmunk predators (e.g., raptors)
may have been attracted to chipmunk playbacks, and it was these and not chipmunks
per se that ovenbirds and veeries responded to. However, raptors were recorded on <
5 point counts during the 2-year study making this an unlikely scenario.
Our results show quantitative differences in information use between veeries
and ovenbirds, extending the small number of interspecific comparisons made to date
(see Nocera, Forbes & Giraldeau 2006 and Parejo et al. 2007 vs. Doligez, Danchin &
Clobert 2002; also Coolen et al. 2003 for an example in social foraging). Differences
between species may arise due to their sensory abilities, natural history constraints
(e.g., nest type; discussed above), or particular ecological tradeoffs (e.g., safety vs.
77
Texas Tech University, Quinn C. Emmering, May 2014
optimal microclimate; Eggers et al. 2006). These interspecific differences can
potentially affect community structure (Fletcher 2008), especially when the
information has potentially high fitness benefits as is the case when assessing spatial
heterogeneity in predation risk. For instance, interspecific differences in assessment
or use of information may result in dissimilar habitat use (i.e.,, near versus far from
predator activity) or overlap of interspecific competitors. Furthermore, temporally
variable environments, such as our pulsed-resource system (Schmidt & Ostfeld 2008),
may confer a competitive advantage to those species using proximate cues providing
current information on predation risk. Hence, ovenbirds may be better equipped to
cope with interannual fluctuations of predator abundance and, in turn, have lower
predation rates compared to veeries which do not appear to use chipmunk
vocalizations to make nest-site decisions. Indeed, at our study site chipmunk activity
measured around ovenbird nest sites as well as predation rates are lower than recorded
at veery nests (Q.C. Emmering, unpublished data), but whether differences in nest
predation rates are due to chipmunks remains untested.
Past studies have demonstrated birds assess predator activity and subsequently
make adaptive breeding decisions (Jędrzejewska & Jędrzejewski 1998; Spaans et al.
1998; Larsen 2000; Forstmeier & Weiss 2004; Roos & Pärt 2004; Fontaine & Martin
2006; Peluc et al. 2008) but rarely have assessment mechanisms been identified. Our
results are consistent with the few studies that have also manipulated predator cues.
For instance, Eggers et al. (2006) demonstrated Siberian jays (Perissoreus infaustus)
more than doubled the distance between nest sites used in previous years and nested in
78
Texas Tech University, Quinn C. Emmering, May 2014
significantly denser vegetation in response to experimental playbacks of corvid nest
predators. By confining least weasels (Mustela nivalis) with nest boxes to add
olfactory and visual (weasel hair) cues of nest predation risk, Mönkkönen et al. (2009)
found that pied flycatchers (Ficedula hypoleuca) avoided risky nest boxes exposed to
weasels significantly more so than control boxes lacking weasel cues. Lastly,
Forsman & Martin (2009) recently demonstrated that songbird hosts of the parasitic
brown-headed cowbird (Molothrus ater) eavesdrop on cowbird vocalizations to find
‘brood parasite-free’ space.
Spatial heterogeneity in predation risk creates spatial refugia that can be
potentially exploited by prey and consequently lower predator efficiency (Mech 1977;
Fontaine & Martin 2006; Schmidt, Ostfeld & Smyth 2006; Schauber et al. 2009).
Models of predator-prey dynamics indicate this reduced efficiency has important
consequences for prey persistence and the stability of prey and predator populations
(Huffaker 1958; Hilborn 1975; Lewis & Murray 1993; Goodwin et al. 2005; Schauber
et al. 2007). In addition, our results may have implications for models of sitedependent regulation in heterogeneous landscapes (Rodenhouse, Sherry & Holmes
1997). Site-dependent regulation produces density dependence through a simple
mechanism of heterogeneity in site (i.e.,, territory) quality where individuals fill up
available sites according to their rank order of quality (Rodenhouse, Sherry & Holmes
1997). Spatial heterogeneity in local predator abundance is one of several
mechanisms that can create spatial heterogeneity in territory quality (Rodenhouse et al.
2003; Schmidt, Ostfeld & Smyth 2006). If such differences in local predator
79
Texas Tech University, Quinn C. Emmering, May 2014
abundance can be detected via acoustic or other cues, the mechanism that we
document here may partly underlie some of the empirical support for site-dependent
regulation, especially in avian populations (e.g., Rodenhouse et al. 2003; Sillett,
Rodenhouse & Holmes 2004; Zajac, Solarz & Bielanski 2008).
Our study, and a growing list of others (e.g., Blaustein et al. 2004; Eggers et al.
2006; Mönkkönen et al. 2009), suggests that information on spatial distribution of
predator activity may be readily available through the communication of predators
themselves. Constrained in many cases to give vocal, chemical or other social cues to
signal territoriality or to attract mates, predator communication is often publicly
available to eavesdropping prey. Thus, heterospecific eavesdropping may be a
common feature of predator-prey interactions that allows prey to locate spatial refugia
and provide greater stability to predator-prey dynamics. In turn, there may be
selective pressures on the evolution of predator communication to avoid informing
their prey (e.g., Deecke, Ford & Slater 2005) or vice versa (Bernal et al. 2007).
Combining the work of theorists and animal behaviorists is thus bringing to bear how
communication and networks of eavesdroppers may have larger ecological
consequences on predator-prey interactions than previously suspected.
80
Texas Tech University, Quinn C. Emmering, May 2014
Mean distance to playback station (m)
70
Chipmunk
Control
60
50
40
30
20
10
0
Veery
Ovenbird
Figure 4.1 Mean (± SE) distance of nests to nearest chipmunk
playback (gray bars) and control (frog + silent; white bars)
stations for veeries and ovenbirds.
81
Texas Tech University, Quinn C. Emmering, May 2014
2.5
(a)
chipmunk
2.0
control
1.5
Mean number of birds counted
1.0
0.5
0.0
0-30
2.5
30-50
(b)
2.0
1.5
1.0
0.5
0.0
0-30
30-50
Distance from plot centre (m)
Figure 4.2 Mean (± SE) number of birds in the ground-nesting (a) and canopynesting (b) guilds detected during point counts at chipmunk playback (gray bars)
and control plots (frog + silent; white bars). Counts were separated into near and
far from plot centers based on model selection procedures applied separately for
each nesting guild (see Results). The treatment × distance category was significant
only for the ground-nesting guild.
82
Texas Tech University, Quinn C. Emmering, May 2014
Table 4.1 Model comparison using the information theoretic index (AICc) to test for
differences in the number of (a) ground-nesting and (b) canopy-nesting birds detected
near and far from playback stations during point counts. We used a sliding cutoff (in
10 m increments) to produce 5 models: 4 models with a near-far cutoff occurring at
10, 20, 30, and 40 m, and a fifth model (0-50 m) without a near-far categorization.
Data were analyzed as a mixed effects general linear model (GLM) with the fixed
effects of year (yr), playback treatment (tr), and distance (dist), random effects of
playback station (PB), and interaction effects of treatment  year and treatment 
distance. K indicates the number of parameters for individual models. In both
analyses, the AIC weight of the top model was > 0.99. Results of the GLM for the top
ranked model based on AIC was used to examine statistical significance (see Results).
83
Texas Tech University, Quinn C. Emmering, May 2014
CHAPTER V
CONCLUSIONS
Research summary
Heterogeneity in predator activity creates predator-free space for prey to
occupy and increase their chance of survival and/or successfully reproducing. For
nesting songbirds, predator-free space can act as spatial refugia for safeguarding
defenseless eggs and young from incidental nest predators. Since predation is the
primary cause of nest failure in songbirds, nest-site selection is closely linked to
fitness and population demography; hence, the presence and use of spatial refugia may
have far-reaching consequences for songbirds.
Information reliably associated with nest failure, such as predator activity, can
be used to adjust breeding decisions leading to higher reproductive success. Songbirds
likely possess behavioral traits that allow them to assess, either directly or indirectly,
cues of predator activity. Communication between predators is often conspicuous and
available in the public domain, including eavesdropping prey that can reduce time and
acquisition costs associated with directly assessing predator activity. Previous
research has demonstrated that birds often make breeding-site adjustments based on
predation risk. Less common are investigations of the adjustments birds make in
response to small-scale predator activity and experiments identifying the sources of
information birds use to assess this heterogeneity.
84
Texas Tech University, Quinn C. Emmering, May 2014
The objective of my research was to determine if breeding veeries (Catharus
fuscescens) and ovenbirds (Seiurus aurocapilla) utilize “coldspots” of predator
activity as nesting refugia to increase reproductive success. I quantified small-scale
predator activity surrounding nest sites and estimated predator density (mice,
chipmunks) at a larger scale (grid-level). Predator activity was estimated at similar
spatial scales but over dissimilar temporal scales using two different techniques.
Track plates estimated predator activity (rodents and mesopredators) surrounding nests
over the course of a week. Whereas, capture data from permanent trapping grids
estimated “hotspots” and “coldspots” of mouse and chipmunk activity from total
summer captures (May – July). Additionally, I conducted a playback experiment to
explore whether breeding songbirds eavesdrop on the calls of a common nest predator,
the eastern chipmunk (Tamias striatus), to assess spatial heterogeneity in chipmunk
activity to make breeding site decisions. Control plots consisted of playbacks of a
non-predator (frog calls) and plots with no playbacks. In response to playbacks I
compared the activity of ground- and canopy-nesting guilds as they likely differ in
their susceptibility to chipmunk predation. Lastly, I recorded veery and ovenbird nest
locations relative to experimental treatment and control plots.
Results summary
As earlier research has found (Schmidt et al. 2006, Schauber et al. 2009,
Schartel 2011), mouse and chipmunk activity at our study site were spatially
heterogeneous across years and trapping grids. The presence of such spatial
85
Texas Tech University, Quinn C. Emmering, May 2014
heterogeneity may be sufficient to create unoccupied space that can be used as refugia
for songbirds to nest within to successfully raise young. However, analyses of rodent
capture and track plate data, found that local predator activity was only weakly, or
unrelated, to veery nest success (Tables 2.3, 2.4, 3.4). Instead, total rodent abundance
(mice + chipmunks) had substantially greater support for describing veery nest success
(Table 2.4). Veeries also did not place nests in sites of lower predator activity (Table
3.3). If indeed predator activity is only weakly related to veery nest success, it’s not
unexpected that veeries would show no preference for sites based on variation in nest
predator activity.
A somewhat different story emerged with ovenbirds. Similar to veeries,
analyses of both rodent capture and track plate data showed that ovenbirds did not
place nests in areas of lower mouse or chipmunk activity (Tables 2.2 & 3.3). However,
information theoretic analyses indicated ovenbird nest predation was best described by
chipmunk activity within a 30 × 30 m area surrounding their nests (Table 2.3 – 2.4).
Mouse abundance, and to a lesser extent total rodent activity and abundance, also may
be important predictors of ovenbird nest predation. Although, model interpretations
warrant some caution as model weights were low (Table 2.3).
The playback experiment (Emmering & Schmidt 2011) corroborated the
observed heterospecific differences as discussed above. In response to experimental
playbacks, veeries and ovenbirds placed nests significantly further from plots
broadcasting chipmunk vocalizations compared to control plots (frog, no playbacks).
However, ovenbirds responded more strongly than veeries, placing nests 20 m further
86
Texas Tech University, Quinn C. Emmering, May 2014
from chipmunk playbacks compared to controls, whereas veeries responded at half the
magnitude, nesting only 9.4 m further from chipmunk playbacks (Figure 4.1). In
comparisons of two nesting guilds, ground-nesters (including veeries and ovenbirds)
significantly reduced their activity within plot centers suggesting a shift in breeding
territories or reduction of overt behaviors (e.g., singing; Figure 4.2). Canopy-nesting
species, less vulnerable to predation to ground-based chipmunks, showed no
differential response to playbacks (Figure 4.2).
My results contrast with those of Schmidt et al. (2006). Although abundance
was quantified at two different scales (grid versus study site), both studies found that
rodent abundance was related to veery nest success as did recent work by Kelly et al.
(in review). Dissimilarly, Schmidt et al. (2006) found that veery nest predation was
significantly related to mouse activity as quantified by two metrics, neighborhood
mouse activity and mouse-free space (Schmidt et al. 2006). An important distinction
between Schmidt et al.’s (2006) metrics and SADIE techniques, is that cluster indices
quantify spatial aggregations of trap station counts relative to one another in that the
magnitude of the index at a trap station is independent of the count there (refer to
Figure 2.1; Perry et al. 1999, Winder et al. 2001). This spatially-explicit approach
should provide a more reliable estimate of spatial patterns of rodent activity for
examining relationships with nest predation.
87
Texas Tech University, Quinn C. Emmering, May 2014
Population implications
Choice of breeding sites is closely tied to fitness and population demography
as conditions within the breeding territory, particularly predation risk, can strongly
affect reproductive success (Cody 1985, Jaenike & Holt 1991, Martin 1998). The
availability of predator-free space as potential nesting refugia could be an essential
resource for birds to successfully reproduce and hence influence their territory
settlement decisions.
Observed heterospecific differences in response to spatial variation in rodent
activity could result in meaningful consequences for veery and ovenbird populations.
For veeries, local rodent activity around nest sites appears weakly or unrelated to nest
success. If this is indeed the case, it’s not unexpected that veeries also showed no
preference for selecting sites based on variation in rodent activity. A lack of a
relationship between local rodent activity and veery nest predation suggests finding
and nesting within spatial refugia may be an ineffective strategy for veeries to increase
reproductive success. Unlike veeries, predator activity, particular of chipmunks, is a
good predictor of ovenbird nest success. Moreover, ovenbirds appear to eavesdrop on
chipmunk vocalizations to assess heterogeneity in their activity and may subsequently
select safer nest sites. This difference may allow ovenbird populations to be more
resistant to the large temporal fluctuations of generalist predators that characterize
pulsed-resource systems like our oak-dominated study site (Schmidt & Ostfeld 2008,
see also McShea 2000, Clotfelter et al. 2007, Emmering & Schmidt 2011). Recent
analysis of ovenbird nest survival from our study site by Kelly et al. (in review)
88
Texas Tech University, Quinn C. Emmering, May 2014
supports this hypothesis. Analyzing a 13-year dataset they found no relationship
between ovenbird nest daily survival and rodent abundance; veery nest survival on the
other hand was strongly related to rodent abundance (Kelly et al., in review).
89
Texas Tech University, Quinn C. Emmering, May 2014
LITERATURE CITED
Anderson, D.R. 2008. Model based inference in the life sciences: a primer on the
evidence. New York: Springer
Batzli, G. O. 1977. Population Dynamics of the White-footed Mouse in Floodplain
and Upland Forests. American Midland Naturalist 97:18-32.
Benson, T. J., J. D. Brown, and J. C. Bednarz. 2010. Identifying predators clarifies
predictors of nest success in a temperate passerine. Journal of Animal Ecology 79:225234.
Betts, M.G., Hadley, A.S., Rodenhouse, N. & Nocera, J.J. (2008) Social information
trumps vegetation structure in breeding-site selection by a migrant songbird.
Proceedings of the Royal Society B-Biological Sciences, 275, 2257-2263.
Bernal, X.E., Page, R.A., Rand, A.S. & Ryan, M.J. (2007) Natural history miscellany Cues for eavesdroppers: Do frog calls indicate prey density and quality? American
Naturalist, 169, 409-415.
Bevier, L.R., Poole, A.F. & Moskoff, W. (2005) Veery (Catharus fuscescens), The
Birds of North America Online (A. Poole, Ed.). Ithaca: Cornell Lab of Ornithology
Blaustein, L., Kiflawi, M., Eitam, A., Mangel, M. & Cohen, J.E. (2004) Oviposition
habitat selection in response to risk of predation in temporary pools: mode of detection
and consistency across experimental venue. Oecologia, 138, 300-305.
Bowers, M.A. (1995) Use of space and habitats by the eastern chipmunk, Tamias
striatus. Journal of Mammalogy, 76, 12-21.
Bowman, G. B. and L. D. Harris. 1980. Effect of Spatial Heterogeneity on GroundNest Depredation. The Journal of Wildlife Management 44:806-813.
Brown, J. S. 1988. Patch use as an indicator of habitat preference, predation risk, and
competition. Behavioral Ecology and Sociobiology 22:37-47.
Brown, J. S., R. A. Morgan, and B. D. Dow. 1992. Patch Use Under Predation Risk 2.
A Test With Fox Squirrels, Sciurus niger. Annales Zoologici Fennici 29:311-318.
Brown, J. S. and B. P. Kotler. 2004. Hazardous duty pay and the foraging cost of
predation. Ecology Letters 7:999-1014.
90
Texas Tech University, Quinn C. Emmering, May 2014
Burke, D. M. and E. Nol. 1998. Influence of food abundance, nest-site habitat, and
forest fragmentation on breeding ovenbirds. Auk 115:96-104.
Burnham KP, Anderson DR (1998) Model selection inference: a practical informationtheoretic approach. New York: Springer
Cain, J. W., K. S. Smallwood, M. L. Morrison, and H. L. Loffland. 2006. Influence of
Mammal Activity on Nesting Success of Passerines. Journal of Wildlife Management
70:522-531.
Carthey, A. J. R., J. P. Bytheway, and P. B. Banks. 2011. Negotiating a noisy,
information-rich environment in search of cryptic prey: olfactory predators need
patchiness in prey cues. Journal of Animal Ecology 80:742-752.
Chalfoun, A. D. and K. A. Schmidt. 2012. Adaptive breeding-habitat selection: is it
for the birds? Auk 129:589-599.
Clotfelter, E.D., Pedersen, A.B., Cranford, J.A., Ram, N., Snajdr, E.A., Nolan, V. &
Ketterson, E.D. (2007) Acorn mast drives long-term dynamics of rodent and songbird
populations. Oecologia, 154, 493-503.
Cody, M. L., ed. 1985. Habitat Selection in Birds. New York: Academic Press.
Connors, M.J., Schauber, E.M., Forbes, A., Jones, C.G., Goodwin, B.J. & Ostfeld, R.S.
(2005) Use of track plates to quantify predation risk at small spatial scales. Journal of
Mammalogy, 86, 991-996.
Coolen, I., van Bergen, Y., Day, R.L. & Laland, K.N. (2003) Species difference in
adaptive use of public information in sticklebacks. Proceedings of the Royal Society
of London Series B-Biological Sciences, 270, 2413-2419.
Cox, W. A., R. T. Frank, III, and J. Faaborg. 2012. Species and temporal factors affect
predator-specific rates of nest predation for forest songbirds in the Midwest. The Auk
129:147-155.
Crooks, K. R. and M. E. Soule. 1999. Mesopredator release and avifaunal extinctions
in a fragmented system. Nature 400:563-566.
Dall, S.R.X., Giraldeau, L.A., Olsson, O., McNamara, J.M. & Stephens, D.W. (2005)
Information and its use by animals in evolutionary ecology. Trends in Ecology
&
Evolution, 20, 187-193.
Deecke, V.B., Ford, J.K.B. & Slater, P.J.B. (2005) The vocal behaviour of mammaleating killer whales: communicating with costly calls. Animal Behaviour, 69, 395-405.
91
Texas Tech University, Quinn C. Emmering, May 2014
Doligez, B., Danchin, E. & Clobert, J. (2002) Public information and breeding habitat
selection in a wild bird population. Science, 297, 1168-1170.
Dunford, C. (1970) Behavioral aspects of spatial organization in the chipmunk,
Tamias striatus. Behaviour, 36, 215-230.
Dutra, H., K. Barnett, J. Reinhardt, R. Marquis, and J. Orrock. 2011. Invasive plant
species alters consumer behavior by providing refuge from predation. Oecologia:1-9.
Eggers, S., Griesser, M., Nystrand, M. & Ekman, J. (2006) Predation risk induces
changes in nest-site selection and clutch size in the Siberian jay. Proceedings of the
Royal Society B-Biological Sciences, 273, 701-706.
Ellis-Felege, S. N., M. J. Conroy, W. E. Palmer, and J. P. Carroll. 2012. Predator
reduction results in compensatory shifts in losses of avian ground nests. Journal of
Applied Ecology 49:661-669.
Emmering, Q. C. and K. A. Schmidt. 2011. Nesting songbirds assess spatial
heterogeneity of predatory chipmunks by eavesdropping on their vocalizations.
Journal of Animal Ecology 80:1305-1312.
Ellison, K.S. & Ribic, C.A. 2012. Nest Defense. Video Surveillance of Nesting Birds
(ed. Ribic, C.A., Thompson, F.R., and Pietz, P.J.). Berkeley: University of California
Press. Studies in Avian Biology.
Elliott, L. (1978) Social behavior and foraging ecology of the eastern chipmunk
(Tamias striatus) in the Adirondack Mountains. Smithsonian Contributions to Zoology,
no. 265.
Fletcher, R.J. (2007) Species interactions and population density mediate the use of
social cues for habitat selection. Journal of Animal Ecology, 76, 598-606.
Fletcher, R.J. (2008) Social information and community dynamics: Nontarget effects
from simulating social cues for management. Ecological Applications, 18, 1764-1773.
Fontaine, J.J. & Martin, T.E. 2006a. Habitat selection responses of parents to offspring
predation risk: An experimental test. American Naturalist, 168, 811-818.
Fontaine, J. J. and T. E. Martin. 2006b. Parent birds assess nest predation risk and
adjust their reproductive strategies. Ecology Letters 9:428-434.
Forsman, J.T., Martin, T.E. (2009) Habitat selection for parasite-free space by hosts of
parasitic cowbirds. Oikos, 118, 464-470.
92
Texas Tech University, Quinn C. Emmering, May 2014
Forsman, J. T., M. Mönkkönen, E. Korpimäki, and R. L. Thomson. 2013. Mammalian
nest predator feces as a cue in avian habitat selection decisions. Behavioral Ecology
24:262-266.
Forstmeier, W. & Weiss, I. (2004) Adaptive plasticity in nest-site selection in response
to changing predation risk. Oikos, 104, 487-499.
Gause, G. F. 1934. The struggle for existence. Williams and Wilkins, Baltimore,
Maryland, USA.
Gloag, R., V. D. Fiorini, J. C. Reboreda, and A. Kacelnik. 2013. The wages of
violence: mobbing by mockingbirds as a frontline defence against brood-parasitic
cowbirds. Animal Behaviour 86:1023-1029.
Goodwin, B.J., Jones, C.G., Schauber, E.M. & Ostfeld, R.S. (2005) Limited dispersal
and heterogeneous predation risk synergistically enhance persistence of rare prey.
Ecology, 86, 3139-3148.
Gorini, L., J. D. C. Linnell, R. May, M. Panzacchi, L. Boitani, M. Odden, and E. B.
Nilsen. 2012. Habitat heterogeneity and mammalian predator-prey interactions.
Mammal Review 42:55-77.
Greenwood, P.J. & Harvey, P.H. (1982) The natal and breeding dispersal of birds.
Annual Review of Ecology and Systematics, 13, 1-21.
Haas, C.A. (1998) Effects of prior nesting success on site fidelity and breeding
dispersal: An experimental approach. Auk, 115, 929-936.
Haché, S., E. M. Bayne, and M.-A. Villard. 2014. Postharvest regeneration, sciurid
abundance, and postfledging survival and movements in an Ovenbird population. The
Condor 116:102-112.
Hamilton, W. J., Jr. 1941. The Food of Small Forest Mammals in Eastern United
States. Journal of Mammalogy 22:250-263.
Hastings, A. 1977. Spatial heterogeneity and the stability of predator-prey systems.
Theoretical Population Biology 12:37-48.
Heckscher, C. M. 2004. Veery nest sites in Mid-Atlantic Piedmont forest: Vegetative
physiognomy and use of alien shrubs. American Midland Naturalist 151:326-337.
Hilborn, R. (1975) The effect of spatial heterogeneity on the persistence of predatorprey interactions. Theoretical Population Biology, 8, 346-355.
93
Texas Tech University, Quinn C. Emmering, May 2014
Hoover, J.P. (2003) Decision rules for site fidelity in a migratory bird, the
prothonotary warbler. Ecology, 84, 416-430.
Hromada, M., Antczak, M., Valone, T.J. & Tryjanowski, P. (2008) Settling decisions
and heterospecific social information use in shrikes. PLoS ONE, 3, e3930.
Hua, F., R. J. Fletcher, K. E. Sieving, and R. M. Dorazio. 2013. Too risky to settle:
avian community structure changes in response to perceived predation risk on adults
and offspring. Proceedings of the Royal Society B: Biological Sciences 280.
Huffaker, C.B. (1958) Experimental studies on predation: dispersion factor and
predator–prey oscillations. Hilgardia, 27, 343-383.
Hutto, R.L., Pletschet, S.M. & Hendricks, P. (1986) A fixed-radius point count
method for nonbreeding and breeding-season use. Auk, 103, 593-602.
Jackson, W.M., Rohwer, S. & Nolan, V., Jr. (1989). Within season breeding dispersal
in prairie warblers and other passerines. Condor 91, 233–241.
Jaenike, J. and R. D. Holt. 1991. Genetic-variation for habitat preference - evidence
and explanations. American Naturalist 137:S67-S90.
Jędrzejewski, B. & Jędrzejewski, W. (1998) Predation in vertebrate communities: the
Bialowieza primeval forest as a case study. Springer-Verlag, New York.
Jones, C.G., Ostfeld, R.S., Richard, M.P., Schauber, E.M. & Wolff, J.O. (1998) Chain
reactions linking acorns to gypsy moth outbreaks and Lyme disease risk. Science, 279,
1023-1026.
Kelly, J.K., Schmidt, K.A., & Ostfeld, R.S. Coexisting ground-nesting passerines are
not alike: rodents and conspecific abundance differentially affect nest survival in the
ovenbird (Seirus aurocapilla) and the veery (Catharus fuscescens), in review
King, D. I. and R. M. DeGraaf. 2006. Predators at bird nests in a northern hardwood
forest in New Hampshire. Journal of Field Ornithology 77:239-243.
Lackey, J., D. Huckaby, and B. Ormiston. 1985. Peromyscus leucopus. Mammalian
Species 247:1-10.
Larsen, T. (2000) Influence of rodent density on nesting associations involving the
bar-tailed godwit, Limosa lapponica. Ibis, 142, 476-481.
94
Texas Tech University, Quinn C. Emmering, May 2014
Lecomte, N., V. Careau, G. Gauthier, and J.-F. Giroux. 2008. Predator behaviour and
predation risk in the heterogeneous Arctic environment. Journal of Animal Ecology
77:439-447.
Lewis, M.A. & Murray, J.D. (1993) Modelling territoriality and wolf deer interactions.
Nature, 366, 738-740.
Lima, S.L. (2009) Predators and the breeding bird: behavioral and reproductive
flexibility under the risk of predation. Biological Reviews, 84, 485-513.
Lovell, P. G., G. D. Ruxton, K. V. Langridge, and K. A. Spencer. 2013. Egg-Laying
Substrate Selection for Optimal Camouflage by Quail. Current Biology 23:260-264.
Martin, T. E. 1993. Nest Predation and Nest Sites - New Perspectives on Old Patterns.
Bioscience 43:523-532.
Martin, T.E. (1995) Avian life-history evolution in relation to nest sites, nest predation,
and food. Ecological Monographs, 65, 101-127.
Martin, T.E. 1998. Are microhabitat preferences of coexisting species under selection
and adaptive? Ecology 79: 656-670.
Martin, T. E., J. Scott, and C. Menge. 2000. Nest predation increases with parental
activity: separating nest site and parental activity effects. Proceedings of the Royal
Society of London Series B-Biological Sciences 267:2287-2293.
Marzluff, J.M. (1988) Do pinyon jays alter nest placement based on prior experience.
Animal Behaviour, 36, 1-10.
McKinnon, L., D. Berteaux, G. Gauthier, and J. Bêty. 2013. Predator-mediated
interactions between preferred, alternative and incidental prey in the arctic tundra.
Oikos 122:1042-1048.
McPeek, M. A., N. L. Rodenhouse, R. T. Holmes, and T. W. Sherry. 2001. A general
model of site-dependent population regulation: population-level regulation without
individual-level interactions. Oikos 94:417-424.
McShea, W.J. (2000) The influence of acorn crops on annual variation in rodent and
bird populations. Ecology, 81, 228-238.
Mech, L.D. (1977) Wolf-pack buffer zones as prey reservoirs. Science, 198, 320-321.
Monkkonen, M., Forsman, J.T., Kananoja, T. & Ylonen, H. (2009) Indirect cues of
nest predation risk and avian reproductive decisions. Biology Letters, 5, 176-178.
95
Texas Tech University, Quinn C. Emmering, May 2014
Morton, E.S. (2005) Predation and variation in breeding habitat use in the ovenbird,
with special reference to breeding habitat selection in northwestern Pennsylvania.
Wilson Bulletin, 117, 327-335.
Nocera, J.J., Forbes, G.J. & Giraldeau, L.A. (2006) Inadvertent social information in
breeding site selection of natal dispersing birds. Proceedings of the Royal Society BBiological Sciences, 273, 349-355.
Ostfeld, R.S., Jones, C.G. & Wolff, J.O. (1996) Of mice and mast. Bioscience, 46,
323-330.
Ostfeld, R.S. & Keesing, F. (2000) Pulsed resources and community dynamics of
consumers in terrestrial ecosystems. Trends in Ecology & Evolution, 15, 232-237.
Parejo, D., White, J., Clobert, J., Dreiss, A. & Danchin, E. (2007) Blue tits use
fledgling quantity and quality as public information in breeding site choice. Ecology,
88, 2373-2382.
Part, T. & Doligez, B. (2003) Gathering public information for habitat selection:
prospecting birds cue on parental activity. Proceedings of the Royal Society of London
Series B-Biological Sciences, 270, 1809-1813.
Peake, T. (2005) Eavesdropping in communication networks. Animal Communication
Networks (ed. P. McGregor), pp. 13-37. Cambridge University Press.
Peluc, S.I., Sillett, T.S., Rotenberry, J.T. & Ghalambor, C.K. (2008) Adaptive
phenotypic plasticity in an island songbird exposed to a novel predation risk.
Behavioral Ecology, 19, 830-835.
Perry, J. N. 1995. Spatial-analysis by distance indexes. Journal of Animal Ecology
64:303-314.
Perry, J. N., L. Winder, J. M. Holland, and R. D. Alston. 1999. Red-blue plots for
detecting clusters in count data. Ecology Letters 2:106-113.
Pledger, S., K. H. Pollock, and J. L. Norris. 2010. Open Capture–Recapture Models
with Heterogeneity: II. Jolly–Seber Model. Biometrics 66:883-890.
Porneluzi, Paul, M. A. Van Horn and T.M. Donovan. 2011. Ovenbird (Seiurus
aurocapilla), The Birds of North America Online (A. Poole, Ed.). Ithaca: Cornell Lab
of Ornithology; Retrieved from the Birds of North America
96
Texas Tech University, Quinn C. Emmering, May 2014
R Core Team (2013). R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. URL http://www.Rproject.org/
Reitsma, L. R., R. T. Holmes, and T. W. Sherry. 1990. Effects of Removal of Red
Squirrels, Tamiasciurus hudsonicus, and Eastern Chipmunks, Tamias striatus, on Nest
Predation in a Northern Hardwood Forest - an Artificial Nest Experiment. Oikos
57:375-380.
Remeš, V., B. Matysioková, and A. Cockburn. 2012. Long-term and large-scale
analyses of nest predation patterns in Australian songbirds and a global comparison of
nest predation rates. Journal of Avian Biology 43:435-444.
Resetarits, W. J. and H. M. Wilbur. 1989. Choice of oviposition site by Hyla
chrysoscelis - role of predators and competitors. Ecology 70:220-228.
Rodewald, A. D. and L. J. Kearns. 2011. Shifts in dominant nest predators along a
rural-to-urban landscape gradient. Condor 113:899-906.
Ribic, C.A., Thompson, F.R., and Pietz, P.J., eds. Video Surveillance of Nesting Birds.
Berkeley: University of California Press. 2012. Studies in Avian Biology.
Ricklefs, Robert E. 1969. An Analysis of Nesting Mortality in Birds. Smithsonian
Contributions to Zoology 9:1-48.
Rodenhouse, N.L., Sherry, T.W., & Holmes, R.T. (1997) Site-dependent regulation of
population size: A new synthesis. Ecology, 78, 2025-2042.
Rodenhouse, N.L., Sillett, T.S., Doran, P.J. & Holmes, R.T. (2003) Multiple densitydependence mechanisms regulate a migratory bird population during the breeding
season. Proceedings of the Royal Society of London Series B-Biological Sciences, 270,
2105-2110.
Roos, S. & Pärt, T. (2004) Nest predators affect spatial dynamics of breeding redbacked shrikes (Lanius collurio). Journal of Animal Ecology, 73, 117-127.
Rush, S.A. & Stutchbury, B.J.M. (2008) Survival of fledgling Hooded Warblers
(Wilsonia citrina) in small and large forest fragments. Auk, 125, 183-191.
Safran, R.J. (2004) Adaptive site selection rules and variation in group size of barn
swallows: Individual decisions predict population patterns. American Naturalist, 164,
121-131.
97
Texas Tech University, Quinn C. Emmering, May 2014
Schartel, T. E. 2011. Quantifying and manipulating spatiotemporal trends in rodent
space use and consumption rates on incidentally encountered prey. Master's. Southern
Illinois University, Carbondale, Illinois.
Schauber, E.M., Goodwin, B.J., Jones, C.G. & Ostfeld, R.S. (2007) Spatial selection
and inheritance: Applying evolutionary concepts to population dynamics in
heterogeneous space. Ecology, 88, 1112-1118.
Schauber, E.M., Connors, M.J., Goodwin, B.J., Jones, C.G. & Ostfeld, R.S. (2009)
Quantifying a dynamic risk landscape: heterogeneous predator activity and
implications for prey persistence. Ecology, 90, 240-251.
Schmidt, K.A., Goheen, J.R., Naumann, R., Ostfeld, R.S., Schauber, E.M. &
Berkowitz, A. (2001a) Experimental removal of strong and weak predators: Mice and
chipmunks preying on songbird nests. Ecology, 82, 2927-2936.
Schmidt, K. A., J. R. Goheen, and R. Naumann. (2001b) Incidental nest predation in
songbirds: Behavioral indicators detect ecological scales and processes. Ecology
82:2937-2947.
Schmidt, K. A. 2003a. Linking frequencies of acorn masting in temperate forests to
long-term population growth rates in a songbird: the veery (Catharus fuscescens).
Oikos 103:548-558.
Schmidt, K. A. 2003b. Nest predation and population declines in Illinois songbirds: a
case for mesopredator effects. Conservation Biology 17:1141-1150.
Schmidt, K. A. and R. S. Ostfeld. 2003a. Songbird populations in fluctuating
environments: Predator responses to pulsed resources. Ecology 84:406-415.
Schmidt, K. A. and R. S. Ostfeld. 2003b. Mice in space: Space use predicts the
interaction between mice and songbirds. Ecology 84:3276-3283.
Schmidt, K.A. 2004a. Site fidelity in temporally correlated environments enhances
population persistence. Ecology Letters, 7, 176-184.
Schmidt, K. A. 2004b. Incidental predation, enemy-free space and the coexistence of
incidental prey. Oikos 106:335-343.
Schmidt, K.A., Nelis, L.C., Briggs, N. & Ostfeld, R.S. (2005) Invasive shrubs and
songbird nesting success: Effects of climate variability and predator abundance.
Ecological Applications, 15, 258-265.
98
Texas Tech University, Quinn C. Emmering, May 2014
Schmidt, K.A., Ostfeld, R.S. & Smyth, K.N. (2006) Spatial heterogeneity in predator
activity, nest survivorship, and nest-site selection in two forest thrushes. Oecologia,
148, 22-29.
Schmidt, K.A. & Schauber, E.M. (2007) Behavioral indicators of predator space use:
Studying species interactions through the behaviour of predators. Israel Journal of
Ecology & Evolution, 53, 389-406.
Schmidt, K.A. & Ostfeld, R.S. (2008) Numerical and behavioral effects within a
pulse-driven system: Consequences for shared prey. Ecology, 89, 635-646.
Schmidt, K.A., Rush, S.A. & Ostfeld, R.S. (2008) Wood thrush nest success and postfledging survival across a temporal pulse of small mammal abundance in an oak forest.
Journal of Animal Ecology, 77, 830-837.
Schwanz, L. E., D. Brisson, M. Gomes-Solecki, and R. S. Ostfeld. 2011. Linking
disease and community ecology through behavioural indicators: immunochallenge of
white-footed mice and its ecological impacts. Journal of Animal Ecology 80:204-214.
Selas, V. 2001. Predation on reptiles and birds by the common buzzard, Buteo buteo,
in relation to changes in its main prey, voles. Canadian Journal of Zoology-Revue
Canadienne De Zoologie 79:2086-2093.
Sergio, F., Marchesi, L. & Pedrini, P. (2003) Spatial refugia and the coexistence of a
diurnal raptor with its intraguild owl predator. Journal of Animal Ecology, 72, 232245.
Sillett, T.S., Rodenhouse, N.L & Holmes, R.T. (2004) Experimentally reducing
neighbor density affects reproduction and behavior of a migratory songbird. Ecology,
85, 2467-2477.
Slotnick, B. (2001) Animal cognition and the rat olfactory system. Trends in Cognitive
Sciences, 5, 216–222.
Smith, T. M. and H. H. Shugart. 1987. Territory Size Variation in the Ovenbird: The
Role of Habitat Structure. Ecology 68:695-704.
Snyder, D.P. (1982) Tamias striatus. Mammalian Species, 168, 1-8.
Spaans, B., Blijleven, H.J., Popov, I.U., Rykhlikova, M.E. & Ebbinge, B.S. (1998)
Dark-bellied brent geese, Branta bernicla bernicla, forego breeding when Arctic
Foxes, Alopex lagopus, are present during nest initiation. Ardea, 86, 11-20.
99
Texas Tech University, Quinn C. Emmering, May 2014
Stevens, M. 2013. Evolutionary Ecology: Knowing How to Hide Your Eggs. Current
Biology 23:R106-R108.
Switzer, P.V. (1993) Site fidelity in predictable and unpredictable habitats.
Evolutionary Ecology, 7, 533-555.
Thompson, F. R. 2007. Factors affecting nest predation on forest songbirds in North
America. Ibis 149:98-109.
Van Horn, M. A. and T.M. Donovan. 1994. Ovenbird (Seiurus aurocapilla), The Birds
of North America Online (A. Poole, Ed.). Ithaca: Cornell Lab of Ornithology
Vickery, P. D., M. L. Hunter, and J. V. Wells. 1992. Evidence of incidental nest
predation and its effects on nests of threatened grassland birds. Oikos 63:281-288.
Vigallon, S. M. and J. M. Marzluff. 2005. Is nest predation by Steller's Jays
(Cyanocitta stelleri) incidental or the result of a specialized search strategy? Auk
122:36-49.
Viitala, J., Korpimaki, E., Palokangas, P. & Koivula, M. (1995) Attraction of kestrels
to vole scent marks visible in ultraviolet-light. Nature, 373, 425-427.
Vitz, A.C. & Rodewald, A.D. (2010) Movements of fledgling Ovenbirds (Seiurus
aurocapilla) and Worm-eating Warblers (Helmitheros vermivorum) within and
beyond the natal home range. Auk, 127, 364-371.
Walls, S. C. 1995. Differential vulnerability to predation and refuge use in competing
larval salamanders. Oecologia 101:86-93.
Williams, G. E. and P. B. Wood. 2002. Are traditional methods of determining nest
predators and nest fates reliable? An experiment with wood thrushes (Hylocichla
mustelina) using miniature video cameras. Auk 119:1126-1132.
Winder, L., C. J. Alexander, J. M. Holland, C. Woolley, and J. N. Perry. 2001.
Modelling the dynamic spatio-temporal response of predators to transient prey patches
in the field. Ecology Letters 4:568-576.
Wolff, J. O., R. D. Dueser, and K. S. Berry. 1985. Food Habits of Sympatric
Peromyscus leucopus and Peromyscus maniculatus. Journal of Mammalogy 66:795798.
Yanes, M. and F. Suarez. 1996. Incidental nest predation and lark conservation in an
Iberian semiarid shrub steppe. Conservation Biology 10:881-887.
100
Texas Tech University, Quinn C. Emmering, May 2014
Zajac T., Solarz W. & Bielanski, W. (2008) Site-dependent population dynamics: the
influence of spatial habitat heterogeneity on individual fitness in the sedge warbler,
Acrocephalus schoenobaenus. Journal of Avian Biology, 39, 206-214.
Zanette, L. Y., A. F. White, M. C. Allen, and M. Clinchy. 2011. Perceived Predation
Risk Reduces the Number of Offspring Songbirds Produce per Year. Science
334:1398-1401.
Zar, J.H. 1999. Biostatistical Analysis. Upper Saddle River: Prentice Hall. 663 p.
101