Discussion
Host movement is a missing component in predicting vector-borne disease
emergence, spread, and transmission (Hartemink et al. 2015;
Dougherty et al. 2018) and has mainly been explored through
mathematical models (Hartfield et al. 2011; Sumner et al.2017; Tardy et al. 2021). However, individual movement is often
not accounted for when modeling the transmission process (Fofana &
Hurford 2017). Our results demonstrate individual resource selection of
development intensity is modified by deer’s sex, tendency to utilize
habitat at the human interface, and their location in the urban matrix.
This finding suggests modeling individual-scale movement is essential
when assessing hosts’ distribution of ticks and tick-borne pathogens
into residential areas.
We observed seasonal sex-based movement differences important for vector
dispersal in urban deer. Males moved faster and selected developed
habitats more frequently than females, that instead maintained smaller
HRs with more forested habitat than developed landcover. This suggests
male deer may disproportionately distribute ticks outside of natural
areas while females maintain ‘source’ tick populations in urban green
spaces through providing bloodmeals and short-distance dispersal
opportunities. Further, because male movement was slowest through urban
areas during the pre-breeding and females during the breeding season,
male movements could drive the distribution of feeding nymphs, while
females could propagate feeding adult ticks to human interfaces.
Our analysis identified predictable responses in deer use of specific
landcover classes at the individual scale, highlighting the potential to
understand deer response under different urban development scenarios.
Deer use of wetland and herbaceous habitats, low-intensity development,
and highly vegetated residential areas scaled alongside the availability
of these habitats. In contrast, highly developed public and residential
landcovers were used less than the amount available, indicating a
threshold may exist for urban deer’s tolerance of impervious surfaces
and high human activity. This finding is consistent with studies that
found deer use declines with increasing housing density (Urbanek &
Nielsen 2013). However, larger HRs were associated with increased use of
highly developed areas, suggesting development may be more utilized as
natural resources become limited. Finally, deer used developed
landcovers more at night compared to daytime hours. This diel activity
pattern supports global meta-analyses showing increased wildlife
nocturnality in response to human activity (Gaynor et al. 2018)
and provides evidence of anthropogenic-driven alteration of deer space
use.
Our results indicate urban deer have an immediate response to
landscape-dependent fragmentation and strongly avoid development at the
spatial scale of 100m when establishing their HRs. In our study, deer
avoided the urban environment and depended on natural habitats when
establishing their HR but were more tolerant of urban features within
the HR, using those resources according to their availability or, for
some individuals, selecting them. Our findings on the spatial scale of
second order selection to human impact may differ from other work that
observed second order selection over a larger distance (Nagy-Reiset al. 2019) because SI’s urban landscape presents extreme
spatial heterogeneity over short distances (Band et al. 2005),
forcing deer to respond to the surrounding environment more imminently
than in more natural environments.
We identified block level characteristics important for deer resource
use that were associated with tick presence – suggesting residential
attributes may increase or decrease the likelihood of tick introduction
and survival in ecotones that directly interface with humans. This
finding provides evidence to LIF influencing deer’s perception and use
of the landscape as observed through their movement. Although our models
showed deer overall preferred forested habitat, there was high variation
in use of non-forested landcover types and stronger avoidance as
development intensity of non-forested habitats increased. While
residential block-1 was selected for by few individuals, 30% of models
using fine thematic landcover resulted in a neutral response, indicating
that some deer may not use residential landcover more or less than
expected by chance and that there is not strong aversion of this
habitat. In comparison, over 90% of models showed deer select against
block-2 and only 3% of models resulted in a neutral response towards
block-2. This contrast between the two residential landcover types
indicates that deer prioritize more accessible concentrated resources in
block-1.
Critically, our findings provide further evidence to support the need
for a landscape lens of tick-borne disease (Diuk-Wasser et al.2021). We observed individual movement responses that can directly
impact the risk landscape for urban tick-borne disease. Deer that
maintained smaller HRs occupied more natural habitats where they may
amplify vectors if habitat is suitable for I. scapularis survival
(e.g. deciduous forest). The diversity and percentage of more developed
landcover types within HRs increased with HR size. Individuals with
larger HRs may functionally connect selected habitat types resulting in
conduits of movement. The juxtaposition of these patterns of deer space
use may jointly contribute to the amplification and dispersal of ticks,
two components at the core of increasing tick-borne disease incidence
and spread. As observed in models examining the role of landscape
connectivity for deer on tick-borne disease risk (VanAcker et al.2019; Tardy et al. 2021), higher functional connectivity for deer
can enhance the spread of ticks between isolated habitat patches. Home
ranges of male (30-1049 ha; average: 170 ha) and female (24-79 ha;
average: 50 ha) deer far exceed the spatial extent of most forest
patches sampled in studies that laid the foundation for the dilution
effect theory (0.3 to 19 ha) (Allan et al. 2003; LoGiudiceet al. 2008), indicating that patches where nymphal tick density
and infection prevalence were estimated and treated independent from one
another were likely functionally connected through deer (and potentially
other host) movement. Thus, we recommend using the scale of the animal’s
space use to examine the spatial unit of influence that wildlife hosts
have on tick-borne disease dynamics (Bolzoni et al. 2012). This
study addresses this by leveraging tools and analytical approaches from
movement and disease ecology to reconcile the hierarchical structure of
resource selection with variation in spatial behaviors exhibited by
individual animals, rarely attempted before in an urban setting where
the outcome of habitat selection impacts zoonotic hazard. With increased
attention on translating movement mechanisms to spatial epidemiological
modeling (Manlove et al. 2022), we hope this work provides a
foundation to formalize integrating movement and epidemiological
datasets.
The individual based hierarchical approach employed in this study
increased our ability to identify movement behaviors that would have
otherwise been missed with a single-scale mean-population approach.
Examining the response of deer to development across spatial scales
provided insight into how urban deer differ from those in more natural
landscapes in their response to human activity during second-order
selection. By examining space use in a hierarchical manner, we gained a
nuanced understanding of how deer both avoid and exploit anthropogenic
development and resources in human-dominated environments, effectively
shedding light on how ecological relationships emerge at the
human-wildlife interface altering the state of zoonotic hazards. The
individual-based modeling framework allowed us to see consistency in
movement behaviors across individuals (ie. patterns across sexes) and
important movement anomalies (ie. high variation in HR size and use of
high intensity development). Finally, translating deer’s observed
movement behaviors to an area with unobserved space-use through
simulation revealed how resource selection can determine an animal’s use
of the human interface, modulating transmission risk. We observed how
simulations based on deer that selected highly forested, connected
residential blocks resulted in high dispersal probability into the urban
matrix than the simulation based on the individual which showed neutral
selection to non-forested landcovers. Simulating future space use from
individual movement models advances our understanding of how
host-environment interactions through movement connects to the spatial
spread or concentration of vectors and pathogens across landscapes and
at human interfaces and improves the use of static risk-maps to display
infectious disease risk.