Urban Ecosyst (2017) 20:435–448

DOI 10.1007/s11252-016-0605-6

Cavity nesting birds along an urban-wildland gradient: is human

facilitation structuring the bird community?
Jorge A. Tomasevic 1 & John M. Marzluff 1

Published online: 6 October 2016

# Springer Science+Business Media New York 2016

Abstract Urbanization brings dramatic and sudden changes cavity-nesting bird community within the most urbanized
to ecological conditions affecting natural communities. sites. Flexible nest site selection and human facilitation pro-
Cavity-nesting birds, both primary and secondary (hereafter vide new opportunities for native cavity-nesting birds in a
PCN and SCN, respectively), may be limited in this novel rapidly changing world.
environment because of reduced abundance of nesting sites
(e.g. snags and cavities) and competition for cavities with non- Keywords Stress gradient hypothesis . Nest box .
native species. But humans can also directly and indirectly Cavity-nesting birds . Non-native species
provide nest sites (e.g., nest boxes, crevices on houses), espe-
cially for SCN species, potentially partially compensating for
negative effects. We investigated whether and how PCNs and Introduction
humans facilitated the cavity-nesting bird community along a
gradient of urbanization. To do so, we estimated the abun- For decades, the study of species interactions in community
dance of cavity-nesting species between 1998 and 2010 at and population ecology has been mainly approached from the
135, 1-km2 sites that differed in the degree of urbanization point of view of negative interactions, notably competition
(0–100 % forest cover). Also, we found 367 nests on a subset and predation (Butterfield 2009). However, recent consider-
of 31 sites. PCNs (n = 67 nests) nested mostly on snags ations have shown that facilitation (i.e., any positive interaction
(98.5 %), while native SCNs (n = 141) used both natural between two species where at least one of them is benefited and
(71.63 %) and anthropogenic (28.37 %) cavities. Non-native neither is harmed, Hacker and Gaines 1997; Stachowicz 2001),
SCNs (n = 159 nests) used mostly anthropogenic cavities can structure ecological communities and regulate population
(98.11 %). PCN abundance facilitated native SCN abundance dynamics (Stachowicz 2001; Bruno et al. 2003; Brooker et al.
on sites with more than 12 % forest cover at 1-km2 scale, but 2008), especially under stressful conditions (Bertness and
not at less forested sites. There, native SCNs nested primarily Callaway 1994).
(59 %) in anthropogenic cavities. Human facilitation allowed The urban ecosystem is a stressful environment for native
native SCNs to successfully use and reproduce where snags species that require native vegetation or undisturbed settings.
were scarce, changing the composition and structure of the As urbanization expands globally, novel ecosystems have
emerged (Kowarik 2011) and native bird communities have
Electronic supplementary material The online version of this article
been affected by habitat loss and degradation, but also by
(doi:10.1007/s11252-016-0605-6) contains supplementary material, changes in the abundance of food, disease transmission, pre-
which is available to authorized users. dation and other interactions with native and non-native spe-
cies (Blair 1996; Marzluff 2001; Kowarik 2011; Aronson
* Jorge A. Tomasevic et al. 2014). As such, the influence of facilitation to animals
[email protected] within urban ecosystems may be investigated by considering
the stress-gradient hypothesis (SGH, Bertness and Callaway
1
School of Environmental and Forest Sciences, University of 1994), which posits that the importance of facilitative interac-
Washington, Box 352100, Seattle, WA 98195-2100, USA tions is greatest at high levels of stress and it is less relevant
436 Urban Ecosyst (2017) 20:435–448

when stress levels are lower and other biotic interactions, such facilitation by humans. In accordance with the stress-gradient
as competition, are more important. This shift in the form of hypothesis, we expect that the effects of human facilitation
species interactions has been documented for plants with dif- (if any) on cavity nesting birds will be strongest in sites
ferent life histories and origins (native or non-native) in a wide with medium to high levels of disturbance (urban cores and
variety of ecosystems and climates (He et al. 2013), yet testing suburban areas) and that humans may play a role structuring
of this framework in animal ecology is still lacking (Barrio the community by favoring some species over others.
et al. 2013). In fact, it seems that the common conception
among animal ecologists is that competition is higher in stress-
ful than in less stressful conditions (Barrio et al. 2013). Methods
Urbanization is a multidimensional stress as biotic and abi-
otic dimensions are affected at the same time, normally on a Study sites and focal species
cline between urban cores and the surrounding natural areas
(Blair 1996; Marzluff 2001). Under these circumstances, it has We studied 5 primary cavity nesting bird species (e.g. wood-
been shown that different species respond differently to these peckers, hereafter PCN) and 7 secondary cavity nesting bird
novel habitats (Marzluff et al. 2015). Urban cores usually species (hereafter SCN) between 1998 and 2010 along an
support high levels of avian biomass concentrated in few na- urban-wildland gradient in the greater Seattle area (Table 1).
tive and non-native species (Beissinger and Osborne 1982; This area comprises a mosaic of urban, suburban and forested
Blair 1996; Melles et al. 2003; Chace and Walsh 2006; land cover from the Puget Sound to the Cascade Mountain
Chapman and Reich 2007; Møller 2009; MacGregor-Fors foothills (Fig. 1). We used forest cover (%, at a 1 km2 scale)
et al. 2012; Aronson et al. 2014; Shanahan et al. 2014; Sol estimated based on classified 30 m resolution land cover data
et al. 2014). Native species, on the other hand, may be more based on 2007 Landsat TM and ETM satellite imagery to
prevalent in middle to low levels of urbanization, where more define our urban-wildland gradient (Alberti et al. 2006). This
vegetation (native and non-native) is available, reaching the dataset included 14 land cover categories, two of which where
highest richness at middle levels of urbanization (ca. 30–70 % Bforest^ (coniferous forest, >80 % coniferous trees, and deci-
forest in the landscape, Hansen et al. 2005; Marzluff 2005; duous and mixed forest, 10–80 % deciduous or mixed forest,
Tratalos et al. 2007; Blair and Johnson 2008; Pennington and see Hepinstall et al. 2008 for more details). We chose forest
Blair 2012; Shanahan et al. 2014). cover as our independent variable because most of the species
Although cities are stressful for many avian species, we studied live in the forest (so it is a direct quantification of
human supplements may facilitate the presence of others their habitat) and sites have less forest normally because they
(Faeth et al. 2005). People benefit birds by providing habitat have been converted to some level of urban use in our study
in their backyards using landscaping, bird feeders, nest area (Hepinstall et al. 2008). We used imagery from 2007
boxes, and water features, among other practices (Davies because changes to forest cover after this point were minimal,
et al. 2009; Clucas et al. 2011; Clucas and Marzluff 2011; including sites that were developed during the early years of
Marzluff 2014), increasing bird richness (Robb et al. 2008; our study. All bird species are native to this area, except for
Clucas and Marzluff 2015) and population size (Fuller et al. European Starling (Sturnus vulgaris) and House Sparrow
2008). (Passer domesticus), both SCNs. These 12 species represent
The main objective of our work was to understand facilita- the full assemblage of diurnal land bird cavity nesters and all
tive interactions between humans and a particular group of the woodpecker species present in the area. We did not include
birds (cavity-nesting species) on a gradient of habitat distur- cavity nesting waterfowl or owls for logistical reasons.
bance, and therefore stress, which we consider to be correlated
with the degree of urbanization. Cavity-nesting birds may be Cavity-nesting bird community characterization
particularly sensitive to urbanization because they are normal-
ly limited by the availability of dead trees for nesting (Newton We described the cavity-nesting bird community using species
1998) and habitat (e.g. forest cover). These key resources are presence, richness, and relative abundance of all the species
usually lost or significantly reduced in urban areas (Blewett mentioned. To do so, we estimated these parameters at 135,
and Marzluff 2005; Donnelly and Marzluff 2006; Davis et al. 1 km2 square plots in sites with different levels of urbanization
2014; LaMontagne et al. 2015) affecting occurrence and abun- (0–100 % forest). On each site, we counted all birds seen and
dance of cavity-nesting species and other species that may heard within 4–8 50-m-fixed-radius points for 10 min (Ralph
interact with them as part of the ecological community (e.g. et al. 1993). We accounted for this variation in effort by having
nest web, Martin and Eadie 1999). We studied the response of similar number of point counts for most sites (mean ± SE,
cavity-nesting species richness, abundance, community struc- 7.05 ± 0.18 points), but also by visiting the sites multiple times
ture, nesting patterns and nesting success along a gradient of during the breeding season (once a month during the breeding
urbanization, paying particular attention to direct and indirect season, i.e. April–August) and over multiple years, resulting
Urban Ecosyst (2017) 20:435–448 437

Table 1 Number of nests of cavity nesting species found on the urban-wildland gradient of the Greater Seattle area between 1998 and 2010

Species Type of substrate

Natural Anthropogenic Total nests

Snag Tree Other Total House/ building Nest box Other Total

Primary Cavity Nesting Species (PCN)

Downy woodpecker (Picoides pubescens) 5 0 0 5 0 0 0 0 5
Hairy woodpecker (Picoides villosus) 23 1 0 24 0 0 0 0 24
Pileated woodpecker (Dryocopus pileatus) 3 0 0 3 0 0 0 0 3
Red-breasted sapsucker (Sphyrapicus ruber) 10 7 0 17 0 0 0 0 17
Northern flicker (Colaptes auratus) 17 0 0 17 1 0 0 1 18
Total PCN 58 8 0 66 1 0 0 1 67
Secondary Cavity Nesting Species (SCN)
Native species
Black-capped chickadee (Poecile atricapillus) 36 6 0 42 0 6 0 6 48
Chestnut-backed chickadee (Poecile rufescens) 38 9 0 47 0 1 0 1 48
Bewick’s wren (Thryomanes bewickii) 2 0 4 6 1 2 3 6 12
Red-breasted nuthatch (Sitta canadensis) 6 0 0 6 0 0 0 0 6
Violet-green swallow* (Tachycineta thalassina) 0 0 0 0 27 0 0 27 27
Total native SCN species 82 15 4 101 28 9 3 40 141
Exotic Species
European starling (Sturnus vulgaris) 3 0 0 3 116 0 1 117 120
House sparrow (Passer domesticus) 0 0 0 0 30 9 0 39 39
Total exotic SCN species 3 0 0 3 146 9 1 156 159
Total SCN 85 15 4 104 174 18 4 196 300
Total nests 143 23 4 170 175 18 4 197 367

*Migratory species

on an average of 726.15 ± 147.14 point counts per year. The We estimated relative ocurrence and relative abundance
point-count stations at least 150 m apart to reduce the proba- within areas of small and fixed radius because land cover
bility of double-counting individuals on two plots. We (and within developments is not homogeneous and distance to au-
field assistants) counted birds between 05:00 and 12:00, ral detections (the majority of cases in our heavily forested and
avoiding days when weather conditions would interfere with built landscapes) may be difficult to estimate accurately
the detectability of the birds (rain, heavy winds, etc.). All (Hutto 2016). Because we used naïve estimates, our determi-
participants on this study where trained on a consistent man- nation of relative abundance in forests may be biased low
ner on how to detect and record the species during point relative to distance-corrected estimates. Marzluff et al.
counts to reduce observers’ bias. In sites with built and forest- (2015) reported that the overall detectability was higher on
ed portions, we allocated more points to the built portion (up the urban portion than the forested portion of our study sites.
to 6 out of 8) to account for higher variability of habitat and Therefore, we exercise caution in interpreting species re-
the bird community (Donnelly and Marzluff 2004; Donnelly sponses where abundance is lowest in forest and highest in
and Marzluff 2006). However, because not all sites have both more developed sites. This situation is only expected for
forest and built portions, the total allocation of point-counts urban-associated species such as the European Starling and
was closer to a 3:7 ratio (304 point counts in the forest vs. 726 House Sparrow, however, bias in their detection is minimal
on the built matrix). This allocation allowed us to more fully because of the gregarious and conspicuous nature.
describe the diversity of birds found at each site. We then
averaged the relative abundance, expressed as number of in- Nesting patterns and nesting success
dividuals per count, to the site level. We used the average
relative abundance per year to account for differences in effort Because cavity-nesting birds can be limited by the availability
between sites. of nesting sites (Newton 1998), our first approach to study
438 Urban Ecosyst (2017) 20:435–448

Fig. 1 Study area in Seattle, WA.

Land cover types grouped as
urban/suburban (in different
degrees of urbanization), forest
(including coniferous, deciduous
or mixed) and other. Dots do not
represent size of the sites

facilitation was to assess whether humans had any influence failure was recorded when possible. Because different sites
on this potentially limiting factor. We recorded nests in a sub- were followed for different number of years, we used nests
set of 31 sites that were subject to a more intensive monitoring per year (average nests) as our dependent variable.
scheme that included spot-mapping, mist-netting, and nest Our second approach to study facilitation was to assess the
searching (Marzluff et al. 2015). We sampled these sites for role of PCNs on the community. PCN species are providers of
different time spans, ranging from 1 to 10 years (mean ± SE, natural cavities that can be used as nesting sites by SCN spe-
6.06 ± 0.48 years). These sites represent a gradient of urban- cies when PCNs abandon them (Martin and Eadie 1999), so
ization between urban cores and wild lands ranging from 0 to we tested for relationships between the abundance of PCN and
99.8 % forest remaining (mean ± SE, 42.61 ± 4.93 %, Fig. 1). SCN species, as an indicator of facilitation where nesting sites
We used two nest-searching schemes: dedicated to cavity may be the limiting factor.
nesting birds (2000–2002, see Blewett and Marzluff 2005)
and opportunistic (1998–1999, 2003–2010). The dedicated Statistical analyses
search scheme was more intensive and focused on both urban
and forested portions of the study plots. The opportunistic We used logistic regression to determine the relationship be-
search scheme was mostly focused on the forested portion of tween forest cover and the presence of each individual species.
the study sites, but still recording nests in the urban portion We also grouped species by guild (PCN, native SCN, non-
where detections were easier as well. Field crews searched for native SCN, all SCN) and looked for changes in richness pat-
nests approximately 2–3 h/site/week during the breeding sea- terns related to forest cover using Poisson log-linear regression,
son (April – August). They recorded all nests found, plus the linear regression, and non-linear regression. We related other
date when they were found, the species they belonged to, and continuous data (e.g. relative abundance) to forest cover with
the substrate in which the nest was built. As for substrate, they standard linear or non-linear regression as deemed appropriate
recorded the species and type (e.g. tree, snag, stump, house, based on visual inspection of the data and quantitative model
pole, etc.), and we later assigned its origin (natural or anthro- selection using the Akaike Information Criterion (AIC; Akaike
pogenic). We visited each nest found every three days to de- 1973, Wagenmakers and Farrell 2004). We calculated AIC
termine its fate. We considered a nest successful if at least one values using package AICcmodavg (version 2.0–4, Mazerolle
chick was able to fledge. If the nest attempt failed, the cause of 2016) and we calculated Delta AIC and AIC weights using
Urban Ecosyst (2017) 20:435–448 439

package qpcR (version 1.4–0, Spiess 2014). We compared dif- (McCune and Grace 2002). In our case, we calculated differ-
ferences on means using paired t-test and we used Chi-squared ences in community structure with Bray-Curtis distance,
tests to test for differences on proportions on nest success be- which accounts for differences in abundance between species
tween different substrates or guilds (Zar 1999). taking species identity into consideration (McCune and Grace
We looked for changes in trend or thresholds in nesting 2002). We used the function metaMDS with the same para-
patterns related to the urban-wildland gradient using a seg- meters for all ordinations (autotransform = FALSE, distance =
mented linear regression analysis (package segmented v0.2– BBray^, tol = 1e-5, trymax =50). We tried uni-, bi- and tridi-
9.4 for R, (Muggeo 2008) on the average number of nests mensional NMDS and selected the dimension that yielded the
(analyzing natural and anthropogenic nests separately) per highest reduction in stress when comparing to the previous
guild. We compared this approach to other models, however dimension.
our main objective was not to determine the best model of the We then tested for relationships between the SCN commu-
full relationship, but rather to determine where any significant nity and forest cover using PERMANOVA (McCune and
threshold in use of resources occurred (e.g., nest substrate), Grace 2002) and the function adonis(), with Bray-Curtis dis-
which may provide evidence of anthropogenic facilitation by tance and 1000 permutations for each run (package vegan for
direct or indirect provision of nest sites. For this analysis, we R). After determining significant relationships between forest
considered each guild (PCN, native SCN, non-native SCN) cover and community structure, we plotted that relationships
individually. In the case of native SCN, we divided our in ordination space with the function ordisurf() (available on
dataset, analyzing Violet-green Swallow (Tachycineta package vegan). We calculated the residuals of this relation-
thalassina) separately from the other species. We took this ship by extracting the predicted forest cover from the gener-
approach because this species has significantly different hab- alized additive model (GAM) used by ordisurf() and the ob-
itat requirements from the others, using open areas for forag- served values for each site. We used these residuals to also
ing, and its gregarious nature results in a patchy distribution interpret the potential role of human facilitation on different
on the landscape (Brown et al. 2011). Whenever we found a levels of urbanization, as we compared the predicted forest
significant change in trend using the segmented regression cover based on the community structure with the actual
analysis (which yields a point estimate and a 95 % CI for amount of forest cover of the site.
the break point), we also ran a two-sided Davies’ test (package We also tested for differences in community structure
segmented v0.2–9.4 for R, Muggeo 2008) to refine the esti- above and below the threshold we found in nesting patterns
mate. We used the results of the Davies’ test as a conservative by native SCN on anthropogenic substrates using Multi-
approach to inform other analyses that built upon these results. response Permutational Procedure (MRPP, McCune and
Based on the shift in use of nesting substrates at higher Grace 2002, available on package vegan). MRPP is a nonpara-
levels of urbanization that we found for native SCNs, we metric technique used to test for differences among groups
further divided our dataset into two groups (above and below making no distributional assumptions about the data (such as
the break point). We calculated relative abundance, relative multivariate normality and homogeneity of variances). We
occurrence and dominance of each species for each group. used Bray-Curtis distance (McCune and Grace 2002) with
We then studied differences in community structure and the 1000 permutations.
effect of urbanization on each group using a multivariate ap- We also looked for evidence of natural facilitation (i.e. by
proach (McCune and Grace 2002). PCNs). To do so, we tested for a relationship between the
As suggested for community structure data (McCune and abundance of PCN and SCN using Log-linear regression
Grace 2002), we checked for rare species. We did not need to (Zar 1999). Our expectation was that PCN abundance should
delete rare species, however, as all of the species we included influence SCN abundance in places where nesting sites could
were present in more than 5 % of the sites. In order to equalize be a limiting factor and PCNs are the main provider of such
the contribution of each site regardless of their total abundance resource (Martin and Eadie 1999).
of birds, we used relativization by species maxima and by site We conducted all analyses on R v3.1.0 (BSpring Dance^)
total for the SCN community data (package vegan v 2.2–1 for for Mac (R Core Team 2014) using RCommander v 2.1–7
R, Oksanen et al. 2013). This centers the focus on the structure (Fox 2005) and Rstudio v 0.98.1049 (RStudio Team 2014).
of the community rather than on the abundance. We used Non-
Metric Multidimensional Scaling (NMDS) to visualize the
differences in community structure between sites (package Results
vegan v 2.2–1 for R, Oksanen et al. 2013). NMDS is a useful
ordination technique that makes no assumptions about the Patterns of species presence and richness
form of the relationships among variables. It also relieves
the Bzero-truncation^ problem that other ordination techniques Native and non-native cavity nesters responded to the gradient
have and it can be constructed with any distance measure of urbanization in distinct ways. After evaluating different
440 Urban Ecosyst (2017) 20:435–448

models using AIC (Table 2), we found that PCNs richness was Species relative abundance
positively associated with forest cover, which was best de-
scribed with a log-linear relationship (Poisson log-linear, Relative abundance of different cavity nesting groups showed
z = 4.77, df = 131, p < 0.0001, Fig. 2a). All five species were different responses to the reduction in forest cover. As expected,
present on sites ranging from 16 to 100 % cover, but their the average relative abundance of PCN species was lower
presence declined rapidly when forest cover was less than (mean ± SE, 0.17 ± 0.01 ind/count/year) than the abundance
16 %. As a group, richness of native SCNs peaked at 37 % of SCN species (mean ± SE, 3.63 ± 0.17 ind/count/year) across
forest and declined with more or less forest cover (Quadratic all sites. PCN’s relative abundance was not significantly corre-
regression, F2,130 = 3.71, p = 0.027, Fig. 2b; Table 2 shows lated with the percentage remaining forest (Linear regression,
alternative models evaluated). Non-native SCNs’ richness Adj, R2 = −0.005, F1, 126 = 0.40, p = 0.53, Fig. 2d). However,
was negatively related to forest cover (Linear regression, some PCN species tended to be more abundant on either ex-
Adj. R2 = 0.30, F1,131 = 58.71, p < 0.0001, Fig. 2c; see treme of the urbanization gradient, while other species peaked
Table 2 for alternative models contrasted). Most of the sites at intermediate levels of urbanization. The most abundant PCN
with less than 40 % forest cover contained native and non- species was the Northern Flicker (mean ± SE, 0.11 ± 0.01 in-
native species, but non-native species were absent on sites dividuals/count/year), which was negatively related to forest
with 90 % or more forest cover. The results for presence at cover (Table 3, Suppl1c). Hairy Woodpecker (Picoides villosus)
the guild level held for nearly all the individual species abundance increased with increasing forest (Table 3, Suppl1b),
(Table 3). Rather than presence of all species decreasing while Downy (Picoides pubescens, Suppl1a) and Pileated
as forest cover increased, native species were typically Woodpecker (Dryocopus pileatus, Suppl1d) abundance peaked
positively associated forest cover and non-native species at sites with intermediate (ca. 40 %) forest cover.
were negatively associated with it (Table 3). Exceptions Although the overall average relative abundance was not
included species known to exploit open areas, or edges statistically different between native and non-native SCNs
between forest and developed lands, such as Black-capped (1.82 vs. 1.81 ind/count/year, two-tailed paired t-test,
Chickadee (Poecile atricapillus) and Northern Flicker t127 = 0.06, p = 0.95), their distributions along the gradient
(Colaptes auratus) that yielded no significant linear relationship of urbanization were substantially different. Native SCN
with forest cover (Table 3). species showed a non-linear unimodal trend peaking in

Table 2 Alternative models

relating richness, abundance and Model ΔAIC Relative Likelihood wi
nesting pattern to % forest cover
remaining. AIC are relevant to Guild richness versus forest cover
particular test presented. ΔAIC is PCN
relative to the best (first listed) Poisson Log-linear 0 1 0.9999
model among the candidates
Linear 21.63 2.00E-05 2.00e-05
SCN
Quadratic 0 1 0.923
Linear 4.97 8.32E-02 7.68E-02
Poisson Log-linear 88.33 6.59E-20 6.08E-20
Non-Native SCN
Linear 0 1 1
Poisson Log-linear 91.62 1.27E-20 1.27E-20
Guild relative abundance versus forest cover
SCN
Quadratic 0 1 0.997
Linear 11.81 0.002 0.002
Average number of nests on anthropogenic substrates
Native Secondary Cavity Nesting species (except Violet-green Swallow)
Exponential decay 0 1 0.952
Segmented regression 5.97 5.04E-02 0.048
Multinomial regression (3rd degree) 34.47 3.27E-08 3.11E-08
Quadratic regression 52.71 3.58E-12 3.41E-12
Linear regression 67.64 2.05E-15 1.95E-15
Urban Ecosyst (2017) 20:435–448 441

Fig. 2 Species richness and 5 0.7

average relative abundance a 0.6
d
4
(respectively) by guild along the 0.5
gradient of urbanization (% forest 3 0.4
cover) between 1999 and 2010:
primary cavity-nesters (a and d), 2 0.3
native secondary-cavity nesters (b 0.2
and e), and non-native secondary- 1
0.1
cavity nesters (c and f). p<0.0001 p = 0.53
0 0.0
Regression line and significance
0 20 40 60 80 100 0 20 40 60 80 100
presented. See BResults^ for more

Relative abundance (ind/point/year)

5 5
details
b e

Species richness
4 4

3 3

2 2

1 1
p=0.027 p < 0.001
0 0
0 20 40 60 80 100 0 20 40 60 80 100
5
c 8 f
4
6
3
4
2

1 2

p<0.001 p < 0.0001

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Forest cover(%) Forest cover(%)

abundance at ca. 36 % forest cover (Quadratic regression, (n = 24), Red-breasted Sapsuckers (Sphyrapicus ruber, n = 18)
Adj, R2 = 0.095, F2, 125 = 7.68, p < 0.001, Fig. 2e, see and Northern Flickers (n = 17). Most of the native SCN nests
Table 2 for alternative models). The abundance of Violet- belonged to Chestnut-backed Chickadees (n = 49) and Black-
green Swallows, which peaks at ca. 40 % forest cover, capped Chickadees (n = 49), while non-native species’ nests
appeared to drive this pattern (Suppl 2e), but we found that were mostly European Starling (n = 120).
the overall trend was robust even when excluding such PCN and SCN species differed in their use of natural and
species (Quadratic regression, Adj, R 2 = 0.044, F 2, anthropogenic nest substrates. PCNs relied almost exclusively
125 = 3.94, p = 0.022). On the other hand, non-native on natural substrates (98.51 %, Table 1), most of them snags
SCN decreased linearly as forest cover increased (Linear (86.57 %, Table 1). The only PCN species that we document-
regression, Adj, R2 = 0.27, F1, 126 = 47.15, p < 0.0001, ed using anthropogenic substrates was the Northern Flicker.
Fig. 2f). We also found variability in the relative abundance We found no PCN nests on sites with less than 6 % forest
patterns among different SCN species (See Suppl2 and remaining.
Suppl3). While some native species were more abundant The proportionate use of natural and anthropogenic sub-
as forest cover increased (e.g. Chestnut-backed Chickadee, strates was significantly different between native and non-
Poecile rufescens, Table 3, Suppl2c); others decreased (e.g. native SCNs (X238 = 79.64, p < 0.001). Native SCNs used
Black-capped Chickadee, Table 3, Suppl2a). The abun- mostly natural substrates (71.63 %). Most natural nests oc-
dance of both non-native species decreased with increase curred on snags (58.16 %) and occurred along the entire extent
in forest cover (Table 3, Suppl3). of the gradient of urbanization (Fig. 3a) suggesting little influ-
ence of forest cover (Linear regression, F 1, 28 = 0.21,
Nesting patterns p = 0.65). We did not find nests in natural substrates if the
remaining forest cover was less than 9 % (Fig. 3a).
We found a total of 367 nests of both PCNs (n = 67 of 5 native The number of nests on anthropogenic sites by native
species) and SCNs (n = 300; 141 of 5 native species, and 159 of SCNs (except for Violet-green Swallow) decreased exponen-
2 non-native species) in natural and anthropogenic substrates tially with increases of forest cover (Exponential decay model,
(Table 1). Most of the PCN nests were of Hairy Woodpeckers residual std. error = 0.06, t28 = 9.336, p < 0.0001), and we
442 Urban Ecosyst (2017) 20:435–448

Table 3 Relationship between forest cover and presence and abundance of cavity-nesting bird species studied

Species Presence vs. Forest cover Abundance vs. Forest cover

Reg. model Rel. Z (df = 131) P Reg. model Rel. Adj. R2 F or z P

(df = 127)

Primary Cavity Nesting Species (PCN)

Downy woodpecker (Picoides pubescens) Logit + 2.411 0.0159 Linear NS
Hairy woodpecker (Picoides villosus) Logit + 5.048 <0.0001 Linear + 0.22 36.65 <0.0001
Pileated woodpecker (Dryocopus pileatus) Logit + 3.068 0.0022 Linear NS
Red-breasted sapsucker (Sphyrapicus ruber) Logit + 3.29 <0.0001 Linear (+) 0.02 3.837 0.0523
Northern flicker (Colaptes auratus) Logit NS Linear - 0.07 11 0.0012
Secondary Cavity Nesting Species (SCN)
Native species
Black-capped chickadee (Poecile atricapillus) Logit NS Linear - 0.11 16.61 <0.0001
Chestnut-backed chickadee (Poecile rufescens) Logit + 2.143 0.0321 Linear + 0.09 13.88 <0.0003
Bewick’s wren (Thryomanes bewickii) Logit - 3.864 <0.001 Linear - 0.02 4.144 0.0446
Red-breasted nuthatch (Sitta canadensis) Logit + 2.605 0.0092 Linear NS
Violet-green swallow* (Tachycineta thalassina) Logit - 2.122 0.0338 Linear NS
Exotic Species
European starling (Sturnus vulgaris) Logit - 4.075 <0.0001 Linear - 0.22 37.39 <0.0001
House sparrow (Passer domesticus) Logit - 5.515 <0.0001 Poisson - 5.883 <0.0001

found a strong break on the slope of this relationship nests were placed on houses and buildings (Table 1). Non-
(Segmented linear regression, Adj. R2 = 0.92, F3,26 = 112.5, native SCNs nested along most of the gradient (3–75 % forest
p < 0.0001, Fig. 3b, see Table 2 for alternative models evaluated) cover remaining) although we found 95 % of these nests on
at 9.45 % forest cover (95%CI: 2.87, 16.04, Fig. 3b). This is, sites with 50 % forest remaining or less (Fig. 3c). We found a
when forest cover was less that 9.45 %, the average number of tendency for nests of this group to decline with increasing
nests decreased much faster than when forest cover was great forest cover (Linear regression, F1, 28 = 3.541, p = 0.07), how-
than 9.45 %. A more conservative estimate gauging the change ever we found no significant break point for this relationship
in slope of the relationship between use of anthropogenic cavi- (Two-sided Davies test, p > 0.85).
ties and forest cover yielded a significant break point at 12.78 %
forest cover (Two-sided Davies test, p < 0.0001), which we used
for subsequent analyses. And although the exponential decay Nesting success
model was more parsimonious than the segmented regression
as shown with the AIC analysis (Table 2), the segmented model We determined the fate of 58/67 (86.6 %) PCN nests and 296/
helped us answer the question of whether there was a shift in 300 (98.7 %) SCN nests. A large proportion of PCN nests were
trend on this relationship and objectively determine where that successful (n = 48, 71.64 %). The most common cause of failure
threshold was. This cannot be achieved with the exponential in natural nests was eviction by European Starlings (n = 3),
decay model given the gradual nature of the decay. We found followed by abandonment (n = 2) and unknown (n = 2).
no significant linear relationship between average number of Other causes of failure include predation (n = 1) and snag falling
Violet-green Swallow nests and forest cover (Linear regression, (n = 1). The only PCN nest we recorded in anthropogenic sub-
F1, 28 = 0.96, p = 0.34). Most of the anthropogenic substrates strates failed also due to interference from European Starlings
where native SCNs (all species combined) nested were houses, (Northern Flicker nest). When looking at SCNs, 266 of the nests
buildings or other anthropogenic structures. Only 6.38 % of were successful (39 native and 146 non-native nests in anthro-
the nests of native SCN were placed in nest boxes. All pogenic substrates and 80 native and 1 non-native nest in natu-
native SCN species, except for Red-breasted Nuthatch ral substrates, 88.67 %). In this case, nest success was signifi-
(Sitta canadensis), used anthropogenic substrates when forest cantly higher in anthropogenic structures than in natural sub-
cover was less than 13 %. strates (94.4 % vs. 81 %, X 21 = 13.03, p = 0.0003), a trend that
Non-native SCNs placed almost all their nests on anthro- we also observed when looking at native SCNs in isolation
pogenic substrates (98.11 %, Table 1) with very little use of (97.5 % vs. 79.2 %, X 21 = 5.6, p = 0.018). The most frequent
natural substrates (1.89 %, Table 1). Most (91.82 %) of the cause of failure for SCN in natural nests was predation (n = 6),
Urban Ecosyst (2017) 20:435–448 443

1.5
a p < 0.001). Sites on the extremes of the gradient of forest cover
were also represented at either end of the ordination space
1.0
along the x-axis (NMDS, Fig. 4), which reflects their dissimi-
larities in the structure and composition of their SCN bird com-
munity. The nature of the relationship between the SCN com-
0.5 munity and forest cover changed above 40 %. Below 40 %
forest cover there is a non-linear relationship between forest
cover and community structure (Fig. 4), however above this
0.0 threshold there is a clear linear relationship with forest cover as
Average number of nests (nests/year)

0 20 40 60 80 100 indicated by the parallel isoclines in Fig. 4. In fact, below 40 %

1.0 forest cover there is wide variation in the composition and
b structure of the community, reflected by the wide dispersion
0.8 of sites along the y-axis of ordination space (Fig. 4).
We found a significant difference in the SCN community
0.6
structure and composition above and below the threshold where
0.4 native SCNs significantly shift to nest predominantly in anthro-
pogenic substrates (i.e. 12 % forest. MRPP, Within-group
0.2 chance-corrected agreement A = 0.0677, p = 0.00099, Bray-
Curtis distance). In fact, the relative importance of each species
0.0
in the community changed between sites below and above this
0 20 40 60 80 100
threshold (Fig. 5). For example, European Starlings dominated
10 c the community on sites below 12 % followed by House
Sparrow and all the native species decreasing their dominance
8 in a linear fashion, while sites above 12 % of forest remaining
6
were dominated by Chestnut-backed Chickadees, Violet-green
Swallows, European Starlings and Black-capped Chickadees to
4 a similar degree. Examining the location of sites in ordination
space and their relationship with forest cover, we found that
2
61.4 % of the sites below 12 % forest where predicted to have
0 more than 12 % forest, which means that the structure and
0 20 40 60 80 100 composition of these sites was more similar to sites with more
Forest cover(%) forest than what they actually had. On the other hand, only
Fig. 3 Relationship (using segmented regression analysis) between 5.95 % of the sites above 12 % forest were predicted to have
average number of nests of cavity nesting birds per site and number of less forest than that, which suggests that the community of
years surveyed and percentage of forest cover remaining along a gradient
of urbanization between 1999 and 2010. a Native secondary cavity those sites have deteriorated to be similar to sites with less
nesting species using natural substrates, (b) native secondary cavity forest. In fact, those individual sites had higher abundance of
nesting species using anthropogenic substrates, (c) non-native non-native SCN than native SCNs (average ± SE native/non-
secondary cavity nesting species using anthropogenic substrates. native abundance ratio of those sites was 0.35 ± 0.068, while
Inverted triangle represents break point or threshold. See BResults^ for
more details the rest of the sites above 12 % forest had an average native/
exotic abundance ratio of 9.78 ± 2.302). Sites below 12 % had
followed by unknown cause (n = 5) and abandonment (n = 4). an average native/non-native ratio of 2.06 ± 0.871.
The remaining 6 nests failed by causes that were observed only Finally, we found a significant positive relationship between
once (e.g. construction, logging, etc.). We could not determine the abundance of PCN species and the abundance of native
the cause of failure for most failed nests in anthropogenic sub- SCN species on sites above 12 % forest (Log-linear regression,
strates (n = 9). When looking at nests in anthropogenic sub- Adj. R2 = 0.1387, F1, 82 = 14.37, p < 0.0003). However,
strates, native SCN species had similar nest success as non- this relationship did not hold for sites below 12 % forest
(Log-linear regression, Adj. R2 = 0.042, F1, 42 = 2.88, p = 0.10).
native species did (97.5 % vs. 93.6 %, X 21 = 0.92, p = 0.34).

Effects on SCN community structure Discussion

SCN community structure and composition were significantly Facilitation was prevalent in our system. As expected, wood-
related to forest cover (PERMANOVA, pseudo-F1, 126 = 29.11, peckers (PCNs) facilitated native SCN species (Martin and
444 Urban Ecosyst (2017) 20:435–448

Fig. 4 Non-metric
multidimensional scaling
representation of the secondary
cavity-nesting bird community
along a gradient of urbanization
between 1999 and 2010. Open
circles represent sites that have
12 % forest cover or higher. Black
circles represent sites with less
than 12 % forest cover. Isoclines
represent relationship between
community structure and
composition and forest cover

Eadie 1999) on sites with moderate to low levels of urbaniza- present on sites where the may not have been without hu-
tion (specifically >12 % forest cover where snag abundance man facilitation.
also increases; Blewett and Marzluff 2005), as suggested by Our finding that PCN and SCN diversity did not change in
the strong and positive correlation between the abundance of a linear fashion along a gradient of urbanization, contrasts
both groups. In addition, as urbanization increases (reducing with previous findings that cavity nesting bird species richness
forest cover to levels below 12 %), key nesting resources significantly increases with forest cover (DeGraaf and
required by cavity nesting species are commonly lost Wentworth 1986; Tilghman 1987; Pidgeon et al. 2007). It is
(Blewett and Marzluff 2005; Harper et al. 2005; Davis et al. important to note that these studies analyzed PCN and SCN
2014; LaMontagne et al. 2015) and we found that in these species together, which may obscure the guild-specific trends
more urban sites facilitation by humans enabled some native we report, especially because we found that people rarely
SCN species to overcome reductions in forest and natural provide direct nesting opportunities for PCN species (in con-
cavities, by switching to nest on anthropogenic substrates. trast with SCN species which can use nest boxes) and they
This anthropogenic facilitation reduced the dependence of were less likely to take advantage of indirect anthropogenic
SCNs on their natural facilitators in areas of moderate to ex- nesting opportunities than SCN species, with the exception of
treme urbanization, as expected by the stress-gradient hypoth- the Northern Flicker. This adaptable woodpecker was the most
esis (Bertness and Callaway 1994). This shift is especially abundant PCN, especially in high levels of urbanization where
important, as PCNs species started to disappear on sites with parks were present. This is not surprising as flickers are
less than 16 % forest cover, which combined with the lack of known to exploit open areas for foraging on the ground
snags could have jeopardized the persistence of native SCN (Moulton and Adams 1991; Elchuk and Wiebe 2003; Wiebe
species in urban settings. By taking advantage of the direct, and Moore 2008), and while a completely urbanized site
and especially indirect, nesting opportunities provided by would be unsuitable, it is favored in areas were both trees
humans (Clucas et al. 2011), the composition and structure and lawns are available (e.g. parks, Tilghman 1987;
of the secondary cavity-nesting bird community changed Morrison and Chapman 2005). Although this species seems
and diversity was maintained in and beyond highly urban- to have many tools to deal with the changes in the habitat due
ized areas, as shown by richness, abundance, nesting and to urbanization, it also faces other challenges, such as compe-
community structure data. The fact that ca. 60 % of the tition with European Starlings (Ingold 1994; Ingold 1996;
sites with less than 12 % forest had a SCN bird community Fisher and Wiebe 2006).
that was typical of sites greater forest cover (as seen on European Starlings dominated the SCN community on the
Fig. 4) is a strong indication that native species were most urban sites (less than 12 % forest remaining) and
Urban Ecosyst (2017) 20:435–448 445

Fig. 5 Relative dominance of

0.4
secondary cavity-nesting birds on
sites along a gradient of
urbanization between 1999 and
EUST
2010. Species (on alphabetical
order): BCCH: Black-capped
Chickadee, BEWR: Bewick’s
Wren, CBCH: Chestnut-backed 0.3
HOSP
Chickadee, EUSP: European
Starling, HOSP: House Sparrow,

Relative Dominance
RBNU: Red-breasted Nuthatch,
VGSW: Violet-green Swallow.
Scientific names presented on text
0.2 EUST
BCCH

CBCH

VGSW

BCCH
HOSP
BEWR
0.1
VGSW
RBNU

BEWR CBCH

RBNU
0.0 Below 12% Forest Above 12% Forest

1 2 3 4 5 6 7
Species Rank

declined in abundance swiftly as forest cover increased. They lotor) or raptors (e.g. Cooper’s Hawk, Accipiter cooperii and
caused nest failures among Northern Flickers (3 out of 8 fail- Barred Owl, Strix varia) may thrive in urbanized areas
ures) and Red-breasted Sapsuckers (2 out of 3 failures), but we (DeStefano and DeGraaf 2003; Chace and Walsh 2006;
found no evidence of competition with native SCN. Our sam- Rullman and Marzluff 2014), and even free-ranging domestic
ple size of nest failure is not large enough to make population- animals, like cats (Felis catus) may result in significantly high
wide inference on the consequences of European Starlings on rates of predation (Loss et al. 2013). So, although this Bsafe
these species, but the pattern we found aligns with findings nesting zone^ hypothesis has not been supported for open-
that Red-breasted Sapsuckers may be experiencing population nesters (Jokimäki et al. 2005), which may suffer greater nest
reductions due to European Starlings, while the effects on predation due to high densities of corvids and sciurids in urban
Northern Flickers and SCNs are less pronounced country- areas (Marzluff et al. 2007), nests in cavities are known to be
wide (Koenig 2003). On the other hand, House Sparrows safer than open-cup nests against predation (Martin and Li
did not threaten nesting success among the species we studied. 1992). And, while urban areas may harbor a wide assemblage
In fact, it seemed that both non-native SCN species were of predators (Haskell et al. 2001; Rullman and Marzluff
somehow constrained to areas with human presence, 2014), these hunters may not have developed a reliable search
which is not unexpected (McKinney 2002; Blair 2004), image for anthropogenic nests or those nests may be hard to
but the fact that even when these non-native species ven- access.
tured into forested settings they rarely nested in natural Cavity-nesting bird abundance (lumping PCN and SCN)
substrates, was surprising. has been documented to decrease from natural areas into ur-
Native SCNs had higher nesting success in anthropogenic ban areas (DeGraaf and Wentworth 1986; Tilghman 1987;
than in natural substrates suggesting lower predation and com- Blair and Johnson 2008). We found that the abundance of
petition pressure when nesting on these novel substrates. It has PCN and SCN species responded differently along the gradi-
been suggested that urban areas may have fewer native pred- ent, and analyzed them accordingly. SCN abundance is typi-
ators than natural areas (Adams 1994) potentially making cally reduced in urban areas compared to natural areas as some
them Bsafe zones^ for nesting (Gering and Blair 1999). sensitive species are lost (Blair 1996; Blair and Johnson 2008)
While it is often true that some natural predators are lost to and/or key resources are reduced (e.g. snags, Blewett and
urbanization (especially large species), other smaller general- Marzluff 2005; Harper et al. 2005; Blair and Johnson 2008;
ist species like coyotes (Canis latrans), raccoons (Procyon Davis et al. 2014; LaMontagne et al. 2015). In our case, we
446 Urban Ecosyst (2017) 20:435–448

did not observe a linear reduction in the abundance of native because some native species that are sensitive to habitat deg-
SCN, but rather a non-linear trend. This contrasted with the radation, such as cavity-nesting birds, can use resources pro-
pattern we found for PCN, where the overall pattern was vided by humans (directly or indirectly). Although we only
mostly driven by the abundance of Northern Flicker, the most quantified a partial aspect of this facilitation (provision of
abundant in the group; and for non-native SCN, where both nesting sites), there is evidence that the presence and success
species had similar abundances, but their responses to urban- of native SCNs may also influenced by other forms of human
ization were the same. In the case of native SCN, most species facilitation, such as provision of food (Clucas and Marzluff
had similar levels of abundance (except for the Red-breasted 2012).
Nuthatch that had between half to a fifth of the average abun- The importance of positive interactions in ecological com-
dance of the other species), but had their peaks of abundance munities may profoundly affect our understanding of nature,
on different areas of the gradient. And thus, adding their abun- even changing our approach to how ecological communities
dances resulted on the hump-shaped pattern we observed. are structured, how species niches and important resources are
The hump-shaped pattern of diversity along an urbaniza- defined, and where species conservation efforts should be
tion gradient is not unexpected (Marzluff 2005). It could be focused (Bruno et al. 2003; Butterfield 2009). Our study
the result of several factors (Graham and Duda 2011), which adds to the increasing realization that humans not only
are not limited to: a) habitat heterogeneity found at medium destroy ecological function, they can also actively facilitate
levels of urbanization promotes the coexistence of it (Marzluff 2014). Broadening our facilitory role may
synanthropic species along with early successional (e.g. improve the conservation of biological diversity in an in-
Black-capped Chickadee and Bewick’s Wren, (Marzluff creasingly urban world.
2005) and forest specialists species (e.g. Chestnut-backed
Chickadee, Marzluff 2005), where factors such as develop-
Acknowledgments Tina (Rohila) Blewett, Heather Cornell, Roarke
ment age, exotic shrub cover, urban land cover, and forest Donnelly, Laura Farwell, Cara Ianni, David Oleyar, Stan Rullman,
aggregation that affect different species in different degrees Thomas Unfried, Kara Whittaker, Sean Williams, and John Withey
(Donnelly and Marzluff 2006) occur in different levels in (among many others) helped collect the data we present. Jon Bakker,
close proximity; b) dynamic disturbances at intermediate Carol Bogezi, Jack DeLap, Michael Heimbuch, Joshua Lawler, Loma
Pendergraft, Martin Raphael, Kaeli Swift, and Lauren Walker provided
levels of urbanization, which are characteristic of develop- invaluable comments to improve this manuscript. This research was
ment (Marzluff et al. 2015), also promote the coexistence of funded by U.S. National Science Foundation (DEB-9875041, IGERT-
competing species by changing habitat conditions and reduc- 0114351, BCS 0120024, and BCS 0508002). JAT was partially funded
ing the chance of competitive exclusion, as predicted by the by a Fulbright-Conicyt scholarship and by the School of Environmental
and Forest Sciences scholarship.
Intermediate Disturbance Hypothesis (Connell 1978); c) over-
lap of human and PCN facilitation may augment the number
of nesting sites or other resources, although we found no ev-
idence of additive effects on the provision of nest sites where References
we found the peak of abundance, other forms of human facil-
itation (e.g. bird feeders) may have a direct contribution to the Adams LW (1994) Urban wildlife habitats: a landscape perspective. U of
Minnesota Press, Minneapolis
abundance of birds (Fuller et al. 2008; Robb et al. 2008) and Akaike H (1973) Information theory and an extension of the maxi-
may have a larger influence in suburban areas in Seattle mum likelihood principle. In: Petrov BN, Caski F (eds)
(Clucas et al. 2011; Clucas and Marzluff 2012); and/or d) Proceedings of the Second International Symposium on Information
reduced predation pressure regulating the cavity nester Theory. Akademiai Kiado, Budapest, pp. 267–281
population. The first three alternatives seem reasonable, Alberti M, Marzluff JM, Waddell P, Handcock M (2006) Modeling inter-
actions among urban development, land-cover change, and bird di-
but the last seems least likely as our study sites harbor versity. NSF Final Report BE/CNH 120024
a rich community of avian predators (Marzluff et al. 2007; Aronson MF, La Sorte FA, Nilon CH et al (2014) A global analysis of the
Rullman and Marzluff 2014). impacts of urbanization on bird and plant diversity reveals key an-
thropogenic drivers. P Roy Soc Lond B Bio 281:20133330
Barrio IC, Hik DS, Bueno CG, Cahill JF (2013) Extending the stress-
Conservation implications
gradient hypothesis–is competition among animals less common in
harsh environments? Oikos 122:516–523
In cities and their suburbs, the combination of natural re- Beissinger SR, Osborne DR (1982) Effects of urbanization on avian
sources and human subsidies may produce unexpected out- community organization. Condor 84:75–83
comes from the conservation perspective. In fact, urban and Bertness MD, Callaway R (1994) Positive interactions in communities.
Trends Ecol Evol 9:191–193
suburban areas are rarely conceived of as a place to practice
Blair RB (1996) Land use and avian species diversity along an urban
species conservation despite their biological and social impor- gradient. Ecol Appl 6:506–519
tance (Marzluff 2002; Miller and Hobbs 2002). Our data sug- Blair R (2004) The effects of urban sprawl on birds at multiple levels of
gest that conservation can be successful in urban areas biological organization. Ecol Soc 9
Urban Ecosyst (2017) 20:435–448 447

Blair RB, Johnson EM (2008) Suburban habitats and their role for birds in Gering JC, Blair RB (1999) Predation on artificial bird nests along an
the urban–rural habitat network: points of local invasion and extinc- urban gradient: predatory risk or relaxation in urban environments?
tion? Landsc Ecol 23:1157–1169 Ecography 22:532–541
Blewett CM, Marzluff JM (2005) Effects of urban sprawl on snags and Graham JH, Duda JJ (2011) The humpbacked species richness-
the abundance and productivity of cavity-nesting birds. Condor 107: curve: a contingent rule for community ecology. Int J Ecol.
678–693 doi:10.1155/2011/868426
Brooker RW, Maestre FT, Callaway RM et al (2008) Facilitation in plant Hacker SD, Gaines SD (1997) Some implications of direct positive inter-
communities: the past, the present, and the future. J Ecol 96:18–34. actions for community species diversity. Ecology 78:1990–2003
doi:10.1111/j.1365-2745.2007.01295.x Hansen AJ, Knight RL, Marzluff JM et al (2005) Effects of exurban
Brown CR, Knott AM, Damrose EJ (2011) Violet-green Swallow development on biodiversity: patterns, mechanisms, and research
(Tachycineta thalassina). The Birds of North America Online. needs. Ecol Appl 15:1893–1905
doi:10.2173/bna.14 Harper MJ, McCarthy MA, van der Ree R (2005) The abundance of
Bruno JF, Stachowicz JJ, Bertness MD (2003) Inclusion of facilitation hollow-bearing trees in urban dry sclerophyll forest and the effect
into ecological theory. Trends Ecol Evol 18:119–125 of wind on hollow development. Biol Conserv 122:181–192
Butterfield B (2009) Effects of facilitation on community stability and
Haskell DG, Knupp AM, Schneider MC (2001) Nest predator abundance
dynamics: syntesis and future directions. J Ecol 97:1192–1201.
and urbanization. In: Marzluff JM, Bowman R, Donnelly R (eds)
doi:10.1111/j.1365-2745.2009.01569.x
Avian ecology and conservation in an urbanizing world. Kluwer
Chace JF, Walsh JJ (2006) Urban effects on native avifauna: a review.
Academic Publishers, Boston, pp. 243–258
Landsc Urban Plan 74:46–69
He Q, Bertness MD, Altieri AH (2013) Global shifts towards positive
Chapman KA, Reich PB (2007) Land use and habitat gradients determine
species interactions with increasing environmental stress. Ecol Lett
bird community diversity and abundance in suburban, rural and
16:695–706. doi:10.1111/ele.12080
reserve landscapes of Minnesota, USA. Biol Conserv 135:527–541
Clucas B, Marzluff JM (2011) Coupled relationships between humans Hepinstall JA, Alberti M, Marzluff JM (2008) Predicting land cover
and other organisms in urban areas. In: Breuste JH, Elmqvist T, change and avian community responses in rapidly urbanizing envi-
Guntenspergen G et al (eds) Urban ecology: patterns, processes, ronments. Landsc Ecol 23:1257–1276
and applications. Oxford University Press, Oxford, pp. 135–147 Hutto RL (2016) Should scientists be required to use a model-based
Clucas B, Marzluff JM (2012) Attitudes and actions toward birds in urban solution to adjust for possible distance- based detectability bias?
areas: human cultural differences influence bird behavior. Auk Ecol Appl 26:1287–1294
129:8–16 Ingold DJ (1994) Influence of nest-site competition between European
Clucas B, Marzluff JM (2015) A cross-continental look at the patterns of starlings and woodpeckers. Wilson Bull 110:227–241
avian species diversity and composition across an urbanisation gra- Ingold DJ (1996) Delayed nesting decreases reproductive success in
dient. Wildl Res. doi:10.1071/WR15007 northern flickers: implications for competition with European star-
Clucas B, Marzluff JM, Kübler S, Meffert P (2011) New directions in lings. J Field Ornithol 67:321–326
urban avian ecology: reciprocal connections between birds and Jokimäki J, Kaisanlahti-Jokimäki M-L, Sorace A et al (2005) Evaluation
humans in cities. In: Perspectives in Urban Ecology. Springer, pp of the Bsafe nesting zone^ hypothesis across an urban gradient: a
167–195 multi-scale study. Ecography 28:59–70
Connell JH (1978) Diversity in tropical rain forests and coral reefs. Koenig WD (2003) European starlings and their effect on native cavity-
Science 199:1302–1310 nesting birds. Conserv Biol 17:1134–1140
Davies ZG, Fuller RA, Loram A et al (2009) A national scale inventory of Kowarik I (2011) Novel urban ecosystems, biodiversity, and conserva-
resource provision for biodiversity within domestic gardens. Biol tion. Environ Pollut 159:1974–1983
Conserv 142:761–771 LaMontagne JM, Kilgour RJ, Anderson EC, Magle S (2015) Tree cavity
Davis A, Major RE, Taylor CE (2014) Distribution of tree-hollows availability across forest, park, and residential habitats in a highly
and hollow preferences by parrots in an urban landscape. Emu urban area. Urban Ecosyst 18:151–167
114:295–303 Loss SR, Will T, Marra PP (2013) The impact of free-ranging domes-
DeGraaf RM, Wentworth JM (1986) Avian guild structure and habitat tic cats on wildlife of the United States. Nat Commun 4:1396.
associations in suburban bird communities. Urban Ecol 9:399–412 doi:10.1038/ncomms2380
DeStefano S, DeGraaf RM (2003) Exploring the ecology of suburban
MacGregor-Fors I, Morales-Pérez L, Schondube JE (2012) From forests
wildlife. Front Ecol Environ 1:95–101
to cities: effects of urbanization on subtropical mountain bird com-
Donnelly R, Marzluff JM (2004) Importance of reserve size and land-
munities. Stud Avian Biol-Ser 45:33–48
scape context to urban bird conservation. Conserv Biol 18:733–745
Martin K, Eadie JM (1999) Nest webs: a community-wide approach to
Donnelly R, Marzluff JM (2006) Relative importance of habitat quantity,
the management and conservation of cavity-nesting forest birds.
structure, and spatial pattern to birds in urbanizing environments.
Forest Ecol Manag 115:243–257
Urban Ecosyst 9:99–117
Elchuk CL, Wiebe KL (2003) Ephemeral food resources and high con- Martin TE, Li P (1992) Life history traits of open-vs. cavity-nesting birds.
specific densities as factors explaining lack of feeding territories in Ecology 73:579–592
Northern Flickers (Colaptes auratus). Auk 120:187–193 Marzluff JM (2001) Worldwide urbanization and its effects on birds. In:
Faeth SH, Warren PS, Shochat E, Marussich WA (2005) Trophic dynam- Avian ecology and conservation in an urbanizing world. Springer,
ics in urban communities. Bioscience 55:399–407 pp 19–47
Fisher RJ, Wiebe KL (2006) Nest site attributes and temporal patterns of Marzluff JM (2002) Fringe conservation: a call to action. Conserv Biol
northern flicker nest loss: effects of predation and competition. 16:1175–1176
Oecologia 147:744–753 Marzluff J (2005) Island biogeography for an urbanizing world: how
Fox J (2005) Getting started with the R commander: a basic-statistics extinction and colonization may determine biological diversity in
graphical user interface to R. J Stat Softw 14:1–42 human-dominated landscapes. Urban Ecosyst 8:157–177
Fuller RA, Warren PH, Armsworth PR et al (2008) Garden bird feeding Marzluff JM (2014) Welcome to Subirdia: sharing our neighborhoods
predicts the structure of urban avian assemblages. Divers Distrib 14: with wrens, robins, woodpeckers and other wildlife. Yale
131–137 University Press
448 Urban Ecosyst (2017) 20:435–448

Marzluff JM, Withey JC, Whittaker KA et al (2007) Consequences of Pidgeon AM, Radeloff VC, Flather CH et al (2007) Associations of forest
habitat utilization by nest predators and breeding songbirds across bird species richness with housing and landscape patterns across the
multiple scales in an urbanizing landscape. Condor 109:516–534 USA. Ecol Appl 17:1989–2010
Marzluff JM, Clucas B, Oleyar MD, DeLap J (2015) The causal response R Core Team (2014) R: A language and environment for statistical com-
of avian communities to suburban development: a quasi-experimen- puting. R Foundation for Statistical Computing, Vienna
tal, longitudinal study. Urban Ecosyst:1–25. doi:10.1007/s11252- Ralph CJ, Geupel GR, Pyle P, et al (1993) Handbook of field
015-0483-3 methods for monitoring landbirds. USDA Forest Service/UNL
Mazerolle MJ (2016) AICcmodavg: Model selection and multimodel Faculty Publications 105
inference based on (Q)AIC(c). R package version 2.0–4. Robb GN, McDonald RA, Chamberlain DE, Bearhop S (2008) Food for
http://CRAN.R-project.org/package=AICcmodavg thought: supplementary feeding as a driver of ecological change in
McCune B, Grace JB (2002) Analysis of ecological communities. MjM avian populations. Front Ecol Environ 6:476–484
software design Gleneden Beach, OR RStudio Team (2014) RStudio: integrated development for R. RStudio,
McKinney ML (2002) Urbanization, biodiversity, and conservation. Inc., Boston. http://www.rstudio.com/
Bioscience 52:883–890 Rullman S, Marzluff JM (2014) Raptor presence along an urban-wildland
Melles S, Glenn S, Martin K (2003) Urban bird diversity and landscape gradient: influences of prey abundance and land cover. J Raptor Res
complexity: species-environment associations along a multiscale 48:257–272
habitat gradient. Conserv Ecol 7
Shanahan DF, Strohbach MW, Warren PS et al (2014) The challenges of
Miller JR, Hobbs RJ (2002) Conservation where people live and work. urban living. In: Gil D, Brumm H (eds) Avian urban ecology.
Conserv Biol 16:330–337 Oxford University Press, Oxford, pp. 3–20
Møller AP (2009) Successful city dwellers: a comparative study of the
Sol D, González-Lagos C, Moreira D et al (2014) Urbanisation tolerance
ecological characteristics of urban birds in the western Palearctic.
and the loss of avian diversity. Ecol Lett 17:942–950
Oecologia 159:849–858
Spiess A-N (2014) qpcR: modelling and analysis of real-time PCR data.
Morrison JL, Chapman WC (2005) Can urban parks provide habitat for
R package version 1.4-0. http://CRAN.R-project.org/package=qpcR
woodpeckers? Northeast Nat 12:253–262
Moulton CA, Adams LW (1991) Effects of urbanization on foraging Stachowicz JJ (2001) Mutualism, facilitation, and the structure of ecolog-
strategy of woodpeckers. In: Wildlife conservation in metropolitan ical communities. Bioscience 51:235–246
environments, NIUW Symp Ser, pp 67–73 Tilghman NG (1987) Characteristics of urban woodlands affecting breed-
Muggeo VM (2008) Segmented: an R package to fit regression models ing bird diversity and abundance. Landsc Urban Plan 14:481–495
with broken-line relationships. R News 8:20–25 Tratalos J, Fuller RA, Evans KL et al (2007) Bird densities are associated
Newton I (1998) Population limitation in birds. Academic Press, London with household densities. Glob Chang Biol 13:1685–1695
Oksanen J, Blanchet FG, Kindt R, et al (2013) Vegan: Community Wagenmakers E-J, Farrell S (2004) AIC model selection using Akaike
Ecology Package. R package version 2.0–7. http://CRAN.R- weights. Psychon B Rev 11:192–196. doi:10.3758/BF03206482
project.org/package=vegancia Wiebe KL, Moore WS (2008) Northern Flicker (Colaptes auratus). The
Pennington DN, Blair RB (2012) Using gradient analysis to uncover Birds of North America Online. doi:10.2173/bna.166
pattern and process in urban bird communities. Stud Avian Biol- Zar JH (1999) Biostatistical analysis, 4th edn. Prentice-Hall, Inc., Upper
Ser 45:9–31 Saddle River