3720
|
SUR ACI et Al.
(e.g., recreation, hunting; Kays et al., 2017; Naidoo & Burton, 2020)
and human footprint on the landscape (e.g., habitat fragmentation,
development; Smith et al., 2019; Suraci et al., 2020; Venter et al.,
2016)— have different and often opposing effects on mammals,
likely because these two disturbance types represent distinct sets
of filters that interact differently with species traits. For instance,
mammal body size and trophic position may determine whether the
immediate presence of humans induces fear responses that result
in reduced habitat use and suppressed activity (Clinchy et al., 2016;
Ordiz et al., 2019; Suraci, Clinchy, et al., 2019) or whether human
presence leads to indirect benefits through relaxed predation/
competition (Berger, 2007; Muhly et al., 2011). Species traits may
similarly determine mammal responses to human footprint. Species
with large space requirements may be more negatively impacted
by habitat loss and fragmentation (Crooks et al., 2017; Ripple et al.,
2014), whereas those with higher dietary flexibility may benefit from
increased resource availability in modified landscapes (Bateman &
Fleming, 2012; Newsome & Van Eeden, 2017). Across disturbance
types, suites of traits may be strongly related to both the likelihood
that a species will occur in areas of high human influence (Aronson
et al., 2016; Evans et al., 2011; Santini et al., 2019), as well as the
intensity with which a species uses such areas when present (e.g.,
the number of individuals present and/or the frequency with which
a site is visited; Lewis et al., 2015; Moll et al., 2018; Suraci, Clinchy,
et al., 2019), potentially allowing ecologists to predict shifts in mam-
mal community structure and species interactions with increasing
disturbance intensity.
However, variation among populations may obscure the link be-
tween species- level traits and measured responses to human distur-
bance. Within a given mammal species, populations frequently vary
in the intensity or directionality of their response to a given distur-
bance type depending on local conditions, including habitat produc-
tivity and exposure to anthropogenic mortality (Belote et al., 2020;
Kays et al., 2017; Moreno- Rueda & Pizarro, 2009; Sévêque et al.,
2020). Indeed, studies of recreation impacts in protected areas com-
monly report contrasting responses to human presence by different
populations of the same species (Bateman & Fleming, 2017; Patten
& Burger, 2018; Reed & Merenlender, 2008; Reilly et al., 2017), and
use of developed areas may also vary among populations based on
trade- offs between anthropogenic threat and resource availability
(Bateman & Fleming, 2012; Carlos et al., 2009). Therefore, elucidat-
ing general patterns in mammal responses to human disturbance re-
quires explicitly accounting for variation among populations as well
as across species.
Here, we examine the link between mammal species traits and
responses to human disturbance at the continental scale, hypoth-
esizing that species with particular combinations of trait values
are more negatively impacted by human influence. Specifically, we
hypothesized that larger, more carnivorous species and those with
slower life history strategies (i.e., longer maturation periods, slower
reproductive rates) are more negatively affected by both human
presence and human footprint, given that these species are typically
more likely to come into conflict with humans (Oriol- Cotterill et al.,
2015; Ripple et al., 2014) and may experience higher rates of an-
thropogenic mortality (Darimont et al., 2015; Hill et al., 2020). To
test our hypotheses, we compiled camera trap data for 24 medium-
to- large ungulate and carnivore species from 61 study areas across
North America (Figure 1a), which collectively represent a substantial
proportion of the North American range for all mammal species in
our analysis. Each camera trapping project deployed cameras across
gradients of both human presence (Figure 1b) and human footprint
(Figure 1c), covering a broad range of both disturbance types, from
undeveloped, remote landscapes to well used parks and urban cen-
ters. Our analysis addresses two objectives. We first quantify mam-
mal species responses to human disturbance across North America,
incorporating variation among populations of the same species to
determine the net effect of human presence and human footprint
on habitat use and activity levels for each species. We then model
mammal responses to anthropogenic disturbance as a function of
species ecological and life history traits to discern the mechanistic
drivers of human influence on mammal communities.
2 | MATERIALS AND METHODS
2.1 | Camera trapping projects and species
We compiled data from 61 camera trapping studies (here after,
“projects”) from across the continental United States, Canada, and
Mexico, representing 3212 unique camera locations sampled for a
total of 454,252 trap days. Details of each camera trapping project
are presented in Table S1. Projects were conducted between 2007
and 2019, ranged in spatial extent between 0.4 and 61,506 km
2
(
± SD = 3473.1 ± 9834.9), deployed camera traps at three to 487
unique camera sites (
± SD = 52.6 ± 87.6) and operated for between
63 and 106,480 trap days (
± SD = 7446.7 ± 17,488.5). Although
the specific locations across North America sampled in this study
were driven by the availability of existing camera trap data sets,
we endeavored to cover a large and representative proportion of
the continent and to focus on areas with overlapping mammal spe-
cies composition. We focused our analyses on 24 medium- to- large
mammal species in the orders Artiodactyla and Carnivora that were
reliably identifiable from camera trap images and which repre-
sented three trophic guilds: herbivores, omnivores, and carnivores
(Table S2). We only included those species that were detected by
at least three camera trapping projects and with a total of at least
100 independent detections to ensure convergence of occupancy
models (see below). Due to data limitations, we treated eastern and
western spotted skunks (Spilogale putorius and Spilogale gracilis) as
a single species. We considered different camera trapping projects
to approximate separate populations of each focal species, while
acknowledging that there may be some overlap among adjacent
projects.
We used the geographic location of each camera site to stan-
dardize the spacing between sites by (i) treating groups of camera
sites within 10 m of each other as a single site and (ii) subsampling
