Does urbanization ameliorate the effect of endoparasite infection in kangaroo rats?

Abstract Urban development can fragment and degrade remnant habitat. Such habitat alterations can have profound impacts on wildlife, including effects on population density, parasite infection status, parasite prevalence, and body condition. We investigated the influence of urbanization on populations of Merriam's kangaroo rat (Dipodomys merriami) and their parasites. We predicted that urban development would lead to reduced abundance, increased parasite prevalence in urban populations, increased probability of parasite infection for individual animals, and decreased body condition of kangaroo rats in urban versus wildland areas. We live trapped kangaroo rats at 5 urban and 5 wildland sites in and around Las Cruces, NM, USA from 2013 to 2015, collected fecal samples from 209 kangaroo rats, and detected endoparasites using fecal flotation and molecular barcoding. Seven parasite species were detected, although only two parasitic worms, Mastophorus dipodomis and Pterygodermatites dipodomis, occurred frequently enough to allow for statistical analysis. We found no effects of urbanization on population density or probability of parasite infection. However, wildland animals infected with P. dipodomis had lower body condition scores than infected animals in urban areas or uninfected animals in either habitat. Our results suggest that urban environments may buffer Merriam's kangaroo rats from the detrimental impacts to body condition that P. dipodomis infections can cause.

& Zandbergen, 2011; Vickers et al., 2015). Negative impacts of urbanization are sometimes sublethal and difficult to detect, particularly when wildlife populations persist in an area rather than experiencing large declines in population size or local extinction (Birnie-Gauvin et al., 2016;Giraudeau et al., 2014;Valcarcel & Fernández-Juricic, 2009;Zanette et al., 2011). In some cases, wildlife respond to disturbances from vehicles, humans, and domestic animals as a perceived risk or as a perceived competitor, spending time and energy responding to these disturbances instead of foraging (Shier et al., 2012;Valcarcel & Fernández-Juricic, 2009;Zanette et al., 2011). This decrease in foraging and increase in energy expenditure can lead to reduced food provisioning for young and reduced reproduction (Bonnington et al., 2013;Zanette et al., 2011). For example, an experimental study showed that blackbirds (Turdus merula) exposed to a domestic cat model exhibited decreased care for young (Bonnington et al., 2013).
However, the effects of urbanization on wildlife are not necessarily negative. Urban areas also offer potential benefits such as readily available urban food sources (e.g., garbage, compost piles, ornamental/fruit trees, bird feeders, and pet food), and urban denning and roosting opportunities (e.g. urban planted trees, gardens, and basements; Becker et al., 2015;Oro et al., 2013). These anthropogenic resources provide opportunities for wildlife, and may increase urban wildlife populations as compared to wildland habitats. For example, raccoons and foxes can have higher population densities in urban areas, and foxes had decreased mortality in urban versus wildlands; this has been associated with anthropogenic food sources (Oro et al., 2013;Prange et al., 2003;Recio et al., 2015;Riley et al., 1998).
Both urbanization and parasite infection may affect body condition of wild animals, and responses to urbanization can be complex (Murray et al., 2019). Evidence from a variety of species indicates that urbanization can lead to decreased body condition (Hellgren & Polnaszek, 2011;Lomas et al., 2015;Murray et al., 2019;Ware et al., 2015). In addition, anthropogenic food sources are sometimes of relatively low nutritional quality, which may place wildlife in a nutrient-deficient state and influence maintenance and reproductive capability (Birnie-Gauvin et al., 2016;Oro et al., 2013;Plummer et al., 2013). Further, wildlife that are infected with parasites can have decreased body condition as compared to non-infected animals (Debeffe et al., 2016;Stien et al., 2002;Vandegrift et al., 2008). Some parasite-infected wildlife experience decreased reproduction (Altizer et al., 2003;Gooderham & Schulte-Hostedde, 2011;Hudson, 1986;Vandegrift & Hudson, 2009;Watson, 2013), which can lead to population declines. Importantly, urbanization and parasite infection may have interactive effects on wildlife (Murray et al., 2019): a variety of parasites and disease-causing agents have been detected in animals living in urban and suburban environments, including viruses, bacteria, and endoparasites (Adam et al., 2016;Clinton et al., 2010;Gordon et al., 2016;Korpe et al., 2016;Sibley et al., 2009). Some of these disease-causing organisms are transmittable to humans (i.e., zoonotic) and/or livestock and domestic animals.
Due to the increased growth of urban development, it is important to understand how wildlife are impacted by expanding urban areas. These expanding urban areas can impact wildlife abundance, parasite prevalence, and body condition; urban environments may also facilitate the interaction of parasite infection and other potential stressors, exacerbating their impacts on wildlife. We investigated the effects of urbanization on population density, parasite presence and prevalence, and body condition in Merriam's kangaroo rats (Dipodomys merriami). Kangaroo rats, which are granivorous and largely solitary, are a highly suitable study group in which to examine the influence of urban development on wildlife ecology and disease. These rodents are widespread throughout the western United States and are found in both wildland and urban environments (DaVanon et al., 2016;Germaine et al., 2001). A variety of both endoparasites and ectoparasites have been documented in Merriam's kangaroo rat (Decker et al., 2001;Ford et al., 2004;Holdenried & Quan, 1956;Iturbe-Morgado et al., 2017;King & Babero, 1974;Martínez-Salazar et al., 2016;Stout & Duszynski, 1983), and kangaroo rats may be involved in the enzootic maintenance of zoonotic parasites and diseases (Antolin et al., 2002;Decker et al., 2001;Ford et al., 2004;Holdenried & Quan, 1956;King & Babero, 1974).
D. merriami has also been identified as a potential hyper-reservoir (a species that carries two or more zoonoses) for zoonotic diseases (Han et al., 2015). We predicted that kangaroo rats would have a lower population density, increased parasite infection, increased parasite prevalence, and decreased body condition in urban versus wildland habitats.

Las Cruces encompasses several urban parks and open spaces
with natural vegetation and is surrounded by undeveloped desert (Bureau of Land Management lands). The study area is part of the Chihuahuan desert ecoregion and the climate is characterized as arid or semi-arid, with peak rainfall occurring during summer monsoons with smaller secondary rain events during the winter months.
Mean annual temperature is 14.70°C and mean annual precipitation is 245.1 mm (Havstad et al., 2006

| Site characteristics
We established sites that were similar in native vegetation in urban (n = 5 within Las Cruces city limits) and wildland environments (n = 5 on federal and state properties surrounding Las Cruces; Table 1), and that were large enough to accommodate a 1 ha trapping grid.
Wildland sites were located ≥500 m from paved roads, and all study sites were at least 1 km from each other ( Figure 1). Eight of the sites were characterized in Hurtado and Mabry (2017), with two additional sites added here (DACC and SS). Methods for characterizing vegetation at study sites follow Hurtado and Mabry (2017). Line-point intercept transects were used to quantify vegetation cover at all sites.
Six 50-m line-point intercepts were randomly dispersed across each site (total of 300 m surveyed per site). To compare the percent cover by grasses, shrubs, forbs, bare ground, and rocks between urban and wildland sites, we used Wilcoxon rank sums tests. An urban index for the study area was created using Landsat Thematic Mapper (LTM) imagery at the 30-m spatial resolution (http://www.ngdc. noaa.gov/metadata). To quantify the degree of urban development surrounding each site, we measured the proportion of pixels within 500 m of the center of each study site that represented impervious surface, which tends to be materials associated with urbanization, such as roads, cement, and buildings, and serves as a proxy for urbanization ( Figure 1; see also Hurtado and Mabry (2017) for details).
To verify the urbanization index, we counted all the buildings at a site (housing units and commercial units) within a 1-km buffer. All processing was conducted in ArcGIS 10.1 (ESRI). There was a strong correlation between the number of buildings at a site and the urban index (Pearson's correlation: r 2 = .93, t 8 = 7.15, p < .01), indicating that the urban index we created was a good measure of urbanization.
A Wilcoxon rank sums test was used to compare the urban index between urban and wildland sites.

| Abundance and density estimates
We estimated population size of kangaroo rats in urban and wildland areas in 2013 using closed population Huggins p and c models in Program MARK (White & Burnham, 1999), using Akaike's Information Criterion corrected for small sample size (AICc) to determine the most parsimonious model of the influence of urbanization on kangaroo rat abundance. We used the Huggins model because we assumed that the kangaroo rat populations at our study sites approximated closed populations over our relatively short trap periods (3 nights). We tested for differences in capture (p) and recapture (c) probabilities between urban and wildland sites. Two sites, 1 each from urban (AH) and wildland (TT), were dropped from the analysis due to low captures (only 3 individuals were captured at either site in 2013). The model averaging function in Program MARK was used to estimate population size, and population density was calculated by dividing the estimated abundance in urban or wildland by 4 ha, the total area trapped in each habitat type.  (Cacciò et al., 2002).

| Parasite presence
The forward primer for the first reaction was Gia7 (5′-AAGCCCGA CGACCTCACCCGCAGTGC-3′) and the reverse primer was Gia759

| Parasite prevalence
Parasite prevalence (number infected/number tested) was determined for all 10 sites (Jovani & Tella, 2006). The number of individuals tested from each site ranged from 8 to 42, with mean ±1 SE = 20.10 ± 3.08. A sample size of 10-20 decreases uncertainty in estimates of prevalence, without losing data to low sample cut-offs (Jovani & Tella, 2006). Prevalence of infection by habitat (urban vs. wildland) was compared using a Wilcoxon signed rank test.

| Relationship between individual infection status and habitat
We determined if habitat affected individual infection status using binomial generalized linear mixed models (GLMMs) implemented in the R package lmerTest (Kuznetsova et al., 2017). Infection was scored as presence/absence of each parasite species for each individual kangaroo rat. We ran separate binomial GLMMs for infection with Pterygodermatites dipodomis and Mastophorus dipodomis, with habitat as a fixed factor and site and year as random factors.

| Site characteristics
There was no difference between urban and wildland sites in any measured environmental variables other than the urban index

| Abundance and population density
We found equal support for the null model of no effect of habitat on either capture (p) or recapture (c) probability and a model that included different values for p and c within habitats (Table 3). An

| Parasite presence
Seven endoparasite species were detected in kangaroo rats via fecal flotation and molecular barcoding (

| Relationship between individual infection status and habitat
We did not detect effects of habitat on an individual animal's probability of infection with either parasite. Overall, Merriam's kangaroo rats living in wildland habitats appeared to have a slightly lower rate of infection with P. dipodomis as compared to urban animals (

| D ISCUSS I ON
We expected to find that urbanization would negatively affect Merriam's kangaroo rats, and predicted that populations in urban parks would have lower population density, higher parasite prevalence, a higher probability of parasite infection for individuals, and lower body condition than in populations in undeveloped desert habitats. Instead, we found no effect of urbanization on any variable examined, other than an interaction between urbanization and in-  Vandegrift & Hudson, 2009;Vandegrift et al., 2008).
The negative effects of infection may be due to chronic immune stress, which may reduce body condition and decrease reproduction (Brooks & Mateo, 2013;Vandegrift et al., 2008). Further, P. peromysci infection can alter behavior and increase the likelihood of depredation for Peromyscus (Luong et al., 2011); however, we did not investigate the effects of infection on behavior in D. merriami due to sample size limitations. We observed no effect of M. dipodomis infection or habitat on kangaroo rat body condition. Other researchers have found differences in Mastophorus infection by habitat type, but similar to our results, they also found no difference in body condition by habitat (Lafferty et al., 2010).

TA B L E 3 Results of Huggins p and c models in
Urban areas tend to maintain higher levels of plant productivity than surrounding wildland areas (e.g., active watering and nutrient inputs), and urban areas may moderate environmental fluctuations (e.g., heat islands, water runoff) as compared to surrounding wildlands (Faeth et al., 2005;Zhao et al., 2016). An increase in vegetation in urban areas may help kangaroo rats infected with P. dipodomis cope with parasite infection. Urbanization can influence processes at multiple ecological levels; for example, increased plant primary production could translate into effects on herbivores, predators, and parasites at higher trophic levels. We found no differences in percent cover by different functional groups of vegetation (grass, forbs, and shrubs) between urban and wildland sites, similar to the results of another study conducted in similar habitats in the same re- Note: Parasite, parasite detected; habitat, habitat type in which kangaroo rats were captured, number tested, number of kangaroo rats tested for the presence of that endoparasite, number infected, the number of kangaroo rats that were infected with that parasite, prevalence, prevalence of each parasite in the urban or wild populations of kangaroo rats. In some cases, only genus is cited, and this is due to the dearth of information on the parasites detected. We found no effect of urbanization on population density of Merriam's kangaroo rat. Further, our estimate of population density (~14 individuals/ha) was almost twice that reported by Lightfoot et al. (2012;7.35/ha). It is possible that these differences are due to differences in sampling. Lightfoot et al. (2012) included 11 years of population data, whereas we have only one year (2013), which had higher than average population density. However, estimated population density in this study was within the range of densities recorded over 11 years (Lightfoot et al., 2012). One reason that we may not have found an effect of urbanization on population density is that recently urbanized areas have more native vegetation as compared to older urbanized areas, which have a higher density of buildings and more isolated habitat fragments (Bolger et al., 1997). The urban sites included in this study were all within areas that were developed within the past 20 years (Hurtado & Mabry, 2019), so these urban parks with native vegetation may be similar enough to wildlands in environmental attributes that Merriam's kangaroo rats can persist.
The trapping periods used for population estimation were too short to allow us to estimate survival in urban versus wildland habitats.
We tested over 200 animals from both urban and wildland sites for parasite presence, and found that urbanization was not associated with parasite infection of individuals or population-level parasite prevalence. Although we had a large sample size of individuals, those individuals comprised just five populations from each habitat type for analyses of parasite prevalence; it is possible that increased replication may have detected differences in prevalence. However, a recent meta-analysis (Werner & Nunn, 2020) found no difference between urban and rural environments in parasite prevalence in rodent hosts, suggesting that the typical expectation that urbanization will lead to an increase in parasitism may not hold for this taxonomic group. Further, Werner and Nunn (2020) (Luong & Hudson, 2012). These insects are then presumably consumed by Merriam's kangaroo rats (Decker et al., 2001). Differences in insect diversity and abundance in urban versus non-urban areas have been documented (Bolger et al., 2000;Faeth et al., 2005;McIntyre, 2000) and attributed to pollution, alteration to water resources, and an increase in insect predators (Faeth et al., 2005;McIntyre, 2000).
However, we did not collect data on insect diversity or abundance; Finally, in some systems, bolder and more aggressive animals have been shown to have increased infections and/or be involved in a higher number of transmissions (Dizney & Dearing, 2013;Natoli et al., 2005). However, in a previous study conducted with animals from the same sites, we found that there was no difference in either boldness or aggression between urban and wildland kangaroo rats (Hurtado & Mabry, 2017), suggesting that these behaviors were not likely to influence individual-level infection rates. A relatively small sample size of animals that were both infected with parasites and included in behavior trials precluded us from examining the relationship between parasite infection and behavior.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are available at Dryad: https://doi.org/10.5061/dryad.8pk0p 2nns.