Competition alters seasonal resource selection and promotes use of invasive shrubs by an imperiled native cottontail

Abstract Many ecosystems face multiple invaders, and interactions among invasive and native species may complicate conservation efforts for imperiled species. Examination of fine‐scale resource selection can be used to detect patterns in habitat selection resulting from species interactions and assess the value of specific resources, including invasive plants, to wildlife. We used animal location data with mixed‐effects resource selection models to examine seasonal competitive interactions and species‐specific selection for forage and cover resources by an imperiled native lagomorph, the New England cottontail Sylvilagus transitionalis and its nonnative competitor, the eastern cottontail S. floridanus in the eastern Hudson Valley, NY. We found evidence that resource selection by New England cottontails depended on the relative prevalence of eastern cottontails to New England cottontails. Where eastern cottontails were less prevalent New England cottontail selected for resources characteristic of early successional shrublands. Where eastern cottontails were more prevalent, New England cottontails selected for resources characteristic of later successional shrublands. New England cottontail use of certain invasive shrubs depended on the prevalence of eastern cottontails relative to New England cottontails, suggesting response to invasive plants is confounded by interactions with a nonnative competitor. Our results further emphasize the need for conservation efforts to consider invasive management within the ecosystem context. We demonstrate the utility of resource selection studies to assist in this regard by exploring competitive interactions in the absence of removal studies, while simultaneously assessing the impact of habitat components such as invasive vegetation on species of conservation concern. Synthesis and applications Resource selection studies can be directly applied to inform ongoing species conservation where multiple invaders are present or where species interactions influence resource selection. Fine‐scale assessments of resource selection, similar to those presented here, can be used to selectively manage habitat to benefit desired species within the ecosystem context.


| INTRODUC TI ON
Invasive species are among the top drivers of biodiversity loss and one of the primary challenges to conservation (Pressey, Cabeza, Watts, Cowling, & Wilson, 2007;Wilcove, Rothstein, Dubow, Phillips, & Losos, 1998). Competitive displacement of native species by nonnative competitors and habitat alterations resulting from the proliferation of invasive plants have both led to population declines or extinctions of native species (reviewed in Mooney & Cleland, 2001;Gurevitch & Padilla, 2004;Harris, 2009). In systems with multiple invaders, complex interspecific interactions may hinder or negate the efficacy of traditional management techniques (reviewed in Zavaleta, Hobbs, & Mooney, 2001;Glen et al., 2013;Ballari, Kuebbing, & Nuñez, 2016).
Resource selection studies provide a method to simultaneously assess the value of specific habitat components, including invasive plants, to wildlife and identify niche partitioning or displacement resulting from competitive interactions, without the use of removal (Douglas, Marsh, & Minckley, 1994;Schroeder et al., 2013;Wauters, Gurnell, Martinoli, & Tosi, 2002;Westhoff & Rabeni, 2013). Such studies commonly assume fitness is correlated with fundamental resources provided by habitat, such as food and cover, and the management of these resources might improve not only habitat quality but also fitness within populations (Thomas & Taylor, 2006). Thus, resource selection has become a common tool to identify and prioritize key ecosystem components to conserve when managing declining species (Cole, Jones, & Harris, 2005;Russo, Jones, & Migliozzi, 2002). Where competition is a concern for conservation, such as when one species is declining or nonnative competitors are present, studies of resource selection can help to identify differences in resource use between competing species that can be exploited by managers to benefit a desired competitor, while discouraging use by the undesirable species (Cole et al., 2005;Kenward & Holm, 1993).
Although eradication of invasives is the favored choice for conservation, some invasives act as facilitators to native species by providing or supplementing limited resources, potentially attracting individuals to invasive-dominated areas (reviewed in Zavaleta et al., 2001;Rodriguez, 2006). In some instances, invasive species have replaced or augmented native resources, becoming necessary for the persistence of endangered species (Van Riel, Jordaens, Martins, & Backeljau, 2000;Zavaleta et al., 2001), and their removal could be detrimental to recovery efforts if appropriate native species are not also restored. Quantifying resource use can illuminate the effect of these invasives on native species to inform invasive management and native species conservation (DeGrandchamp, Garvey, & Colombo, 2008;Recio, Mathieu, Virgós, & Seddon, 2014).
A clear understanding of fine-scale resource selection can inform guidelines to alter competitive interactions and improve habitat at a scale relevant to management. Our goal was to assess whether the prevalence of nonnative eastern cottontails Sylvilagus floridanus influenced fine-scale selection of resources that provide forage and cover to the imperiled New England cottontail S. transitionalis within shrublands in eastern New York. As temporal variation in resource availability can alter resource selection patterns, and guidelines developed from resource use over single or pooled seasons may miss seasonally critical habitat components (McCall, Pilfold, Derocher, & Lunn, 2016;Stewart, Bowyer, Kie, Cimon, & Johnson, 2002), we also examined variability in selection by season. We hypothesized that if competitive interactions altered resource accessibility to New England cottontails, we would observe a difference in resource selection by New England cottontails between areas where competitors are more prevalent and where they are less prevalent, and that these differences may vary seasonally as a result of changes in resource availability. We simultaneously assessed selection for abundant invasive plant species, which are of uncertain value for New England cottontails, but are selected for food and cover by eastern cottontails (Morgan & Gates, 1983;Sweetman, 1949). We sought to identify seasonal patterns of resource use that might inform management for the benefit of the imperiled New England cottontail without enhancing populations of the nonnative eastern cottontail.
These methods consider ecosystem context to inform management and conservation at a scale relevant to the site-level habitat management for New England cottontails and can be applied more generally to studies of resource selection between competing species and for species within altered habitats.

| Study system and species
The shrubland obligate New England cottontail is a lagomorph endemic to New England and eastern New York, USA ( Figure 1). The New England cottontail has experienced a range-wide population decline concurrent with regional losses of successional shrublands F I G U R E 1 A New England cottontail (Sylvilagus transitionalis) in native-dominated ericaceous shrubland site with an oak (Quercus spp.) canopy. Photograph was taken by remotely triggered trail camera (i.e., dense woody shrub communities associated with early successional forest regeneration) over the 20th century resulting from reforestation and human development (Litvaitis, 1993). The species persists within five isolated populations covering less than 14% of its historic range (Fenderson, Kovach, Litvaitis, & Litvaitis, 2011;Litvaitis et al., 2006). Efforts to recover New England cottontails are largely dependent on creating and maintaining a network of suitable shrubland patches (Fuller & Tur, 2012). However, restoration of shrublands to benefit New England cottontails has been complicated by competitive interactions with the nonnative eastern cottontail, the establishment of invasive plants within successional shrublands, and a poor understanding of seasonal resource needs of New England cottontails (Litvaitis et al., 2008).
Native to extreme southern New York and west of the Hudson River, NY, USA (Nelson, 1909), eastern cottontails were introduced east of the Hudson River for hunting in the early 20th century (Foster, Motzkin, Bernardos, & Cardoza, 2002;Probert & Litvaitis, 1996). They have since become widely established, frequently cooccurring with New England cottontails at the patch scale (Probert & Litvaitis, 1996). Avoidance and antagonistic interactions between New England and eastern cottontails have been recorded in captive trials, but neither species dominated the interactions, suggesting competitive displacement of resident New England cottontails by eastern cottontails was unlikely to occur through antagonistic interactions (Probert & Litvaitis, 1996). However, the eastern cottontail is more general in its habitat requirements and may be able to colonize shrublands at an earlier successional stage than is suitable for New England cottontail occupancy (Probert & Litvaitis, 1996). This system of "prior rights" would confer a colonization advantage to eastern cottontails and limit later colonization or may result in displacement of New England cottontails in co-occupied successional shrublands (Probert & Litvaitis, 1996).
Similarly, displacement from successional shrublands into coniferous forest and ericaceous shrublands has been hypothesized as a driver of the range-wide decline in Appalachian cottontails (Russell, Moorman, & Guynn, 1999).
There is also concern that common invasive shrubs alter habitat quality for New England cottontails, but their effect on fitness of cottontails is not well understood (Warren, Litvaitis, & Keirstead, 2016). Successional shrublands may be particularly prone to invasion by exotic plant species (Johnson, Litvaitis, Lee, & Frey, 2006), and within Northeastern shrublands, invasive Japanese barberry Berberis thunbergii and multiflora rose Rosa multiflora are particularly pervasive, causing changes in vegetation structure, and reducing native plant diversity (Silander & Klepeis, 1999;Yurkonis, Meiners, & Wachholder, 2005). Multiflora rose provides forage to cottontail rabbits, and eastern cottontails are known to consume Japanese barberry during winter (Dalke & Sime, 1941;Sweetman, 1949).
However, the presence of Japanese barberry reduces the biomass of other plant species (Silander & Klepeis, 1999) and its value as a food plant to New England cottontails is unknown. Further, these invasives are associated with higher tick burdens on New England cottontails, which may negatively impact fitness (Mello, 2018).
Although New England cottontail body condition and survival are sensitive to resource availability (Smith & Litvaitis, 2000;Villafuerte, Litvaitis, & Smith, 1997), seasonal differences in the resource needs of New England cottontails are poorly understood. Prior studies of New England cottontail habitat have been primarily constrained to data collected during winter (Barbour & Litvaitis, 1993;Buffum, McGreevy, Gottfried, Sullivan, & Husband, 2015;Villafuerte et al., 1997), which is a limiting period for survival but does not consider differences in resource availability between summer and winter or variation in the resource needs of juveniles or reproducing individuals during the late spring to summer reproductive season. shrub understory, and late successional shrublands were characterized by high canopy closure, low forb and grass cover, and moderate-to-high shrub densities. Persistent shrublands included high canopy closure ericaceous (i.e., mountain laurel, Kalmia latifolia and blueberry Vaccinium spp.) shrublands, and forested wetlands with a dense understory often consisting of sweet pepperbush (Clethra alnifolia) and swamp azalea (Rhododendron viscosum) (Appendix S1, Supporting Information). The two most abundant invasive shrubs within the study region were Japanese barberry, which has a broad range of light tolerances (Silander & Klepeis, 1999) and is common within mid and late successional shrublands, and multiflora rose, which occurred most frequently within early and mid-successional shrublands (Appendix S1, Supporting Information). Several common native shrubs such as Rubus spp., dogwood (Cornus spp.), and

| Study area
Viburnum spp. that are palatable to New England cottontails (Pringle, 1960) were at highest densities within early successional and persistent shrublands, and typically had low densities in late successional shrublands.
Removal experiments to study competition between the cottontail species within our system were not possible given the similarity in appearance between New England and eastern cottontails, and state regulations concerning holding time of wild-caught animals.
However, sites differed in both the presence and the abundance of eastern cottontails relative to New England cottontails, permitting us to stratify resource selection analyses for New England cottontails based on the prevalence of eastern cottontails.

| Location data acquisition
Cottontails were captured in single-door box traps baited with apple following methods in Ryan, Gavard, Cheeseman, Cohen, and Whipps (2016). To minimize bias associated with differences in resource use, traps were placed along transects spanning entire sites at rabbit sign or every 25 m in the absence of sign. We confirmed species identity for all cottontails using two restriction digests, performed on an amplified target section of mitochondrial DNA extracted from collected ear tissue as described by Ryan et al. (2016) following modified protocols outlined by Litvaitis and Litvaitis (1996), Litvaitis, Litvaitis, Lee, and Kocher (1997), and Kovach, Litvaitis, and Litvaitis (2003).
All cottontails over 800 g were affixed with a 24-g radio collar with a zip tie closure (Advanced Telemetry Systems, Isanti, MN). Juvenile cottontails weighing less than 800 g were affixed with a 1.1-g glueon radio transmitter (Advanced Telemetry Systems) with a wingedmesh attachment following methods in Estes-Zumpf and Rachlow (2007). Effort was made to recapture juveniles once they weighed 800 g to replace glue-on transmitters with radio collars.
Triangulation and homing were used to locate cottontails 2-3 times weekly, year-round (Cheeseman, 2017). All locations for each individual were obtained >24 hr apart to help ensure independence of observations. We located cottontails during both active periods (2 hr before sunset to 2 hr after sunrise) and resting periods (2 hr after sunrise to 2 hr after sunset) (

| Vegetation data acquisition
The abundance and density of shrub cover, height of cover, available herbaceous forage, and amount of tree canopy are thought to impact habitat quality for New England cottontails (Barbour & Litvaitis, 1993;Buffum et al., 2015;Warren et al., 2016); therefore, our sampling efforts sought to characterize these variables. We sampled vegetation in two seasons, defined by the availability of resources to cottontails: the leaf-off season (November-April) and leaf-on season (May-October). Data were collected at the centroid of every cell (hereafter "plot," n = 1,191) of a 50-m grid spanning entire sites. During the leaf-off season, the amount of canopy closure by branches and evergreen vegetation, hereafter "persistent canopy closure," was measured using a spherical densitometer held at a height of 1 m. To assess stem density by species, we counted the number of stems <7.5 cm dbh for all woody plant species, hereafter "shrubs," at a height of 0.5 m within 10 × 1-m 2 belt transects at each plot in the leaf-off season (Barbour & Litvaitis, 1993). During the leaf-on season we assessed leaf-on canopy closure, hereafter "seasonal canopy closure" as well as vegetation height and cover.
Seasonal canopy closure was estimated with a spherical densitometer as above. We estimated shrub height within a gridded 1-m 2 quadrat at each centroid to the nearest 5 cm (Matenaar, Bazelet, & Hochkirch, 2015;Warren et al., 2016). As cover provided by forbs was seasonally variable, we separately estimated the height of forbs within the plot using the same methodology (Matenaar et al., 2015).
The proportion cover of herbaceous graminoid (e.g., grasses, sedges, and rushes), shrub, and forb vegetation within each gridded 1-m 2 quadrat was visually estimated to the nearest 1%. To examine cottontail selection of particular vegetation types, for each plot we measured stem density of multiflora rose, Japanese barberry, and pooled palatable stems. Stem density and shrub height did not vary seasonally, so we included them in analyses for both seasons. Data were imported into ArcMap 10.4.1 (ESRI, Redlands, CA), and vegetation metrics were resampled to 10-m resolution across sites using bilinear interpolation (Bonnot, Millspaugh, & Rumble, 2009;Stabach, Laporte, & Olupot, 2009;Vellend, Bjorkman, & McConchie, 2008).

| Data analysis
We sampled available resources within a buffer around each used location equal in diameter to the mean movement distance of cottontail species between successive locations (Northrup, Hooten, Anderson, & Wittemyer, 2013), creating one sampling unit for each used point. We removed non-habitat (i.e., roads, lakes, and mature forest) from the sampling unit prior to random point generation. Northrup et al. (2013) suggested that the sampling accuracy of point process models could be improved if multiple available points were paired to each used location; we therefore randomly selected up to 20 "unused" locations within each sampling unit, constrained to be far enough apart to not fall in the same pixel as other points. We Hereafter, we refer to these designations as "less prevalent" where eastern cottontails made up one in six or fewer known living cottontails and "more prevalent" where eastern cottontails comprised greater than one in six known living cottontails. We also assessed resource selection by eastern cottontails where they were more prevalent than one in six cottontails, for comparison. Our sample size for eastern cottontails was not sufficient to assess their resource selection at sites where they were less prevalent. As this cutoff could be biased if a species-specific response to trap effort existed, we assessed trapping bias using a Pearson's correlation of the ratio of Eastern to New England cottontails known to be alive from trapping and telemetry efforts to the same ratio obtained from noninvasive genetic pellet sampling for a concurrent parasitological study. Pellet survey data were available from 17 identical site and year combinations of 2014 and 2015.
We modeled resource use relative to availability separately for each species in each season as a function of seasonal tree canopy closure (leaf-on only), persistent tree canopy closure, forb cover (leaf-on only), shrub cover, graminoid cover (leaf-on only), forb height (leaf-on only), shrub height, and stem density of Japanese barberry, multiflora rose, and pooled native palatable stems (Appendix S4, Supporting Information). Stem densities were rescaled to stems/0.1 m 2 for analysis. Models were fit using mixedeffects conditional logistic regression in a Bayesian framework in rjags using the jags function within the jagsUI wrapper (Kellner, 2016;Plummer, 2003) in program R v. 3.

(R Foundation for
Statistical Computing, Vienna, Austria). We ran 3 chains in parallel for 500,000 iterations using flat normal priors for all parameters, and a burn-in length of 100,000. We considered models converged if the R-hat statistic was <1.1 (Gelman & Rubin, 1992). We inferred support for variables if the 95% Bayesian credible interval for the relevant regression coefficient did not overlap zero (Kéry, 2010).
Because cottontails inhabit early-and mid-successional stages, we tributions. If, for a particular random effect, σ i is high relative to β i, and if SE σ i is low relative to σ i , then there is evidence for the importance of including that random effect in the model (McCall et al., 2016), so we used these ratios as criteria for keeping particular random effects in our model. Comparisons between seasons, species, and eastern cottontail prevalence categories were made by examining the degree of overlap in prediction intervals for each variable. We interpreted selection or avoidance for a range of values where prediction intervals did not overlap 0.5 and a difference in selection or avoidance of particular resources where the prediction intervals overlapped by less than 25% (Cumming & Finch, 2005).

| Mapping predictions
We mapped predicted resource use for each value relative to the mean shrubland condition to facilitate comparison across sites. We extracted site-specific resource values for each 10-m raster cell and calculated predicted use and 95% prediction intervals for each cell using study-level average resource values as the available reference for each plot. Predicted resource use and 95% prediction intervals were then mapped in ArcMap 10.4.1.

| RE SULTS
We collared 80 New England cottontails and 68 eastern cottontails and collected a total of 5,375 locations to be used in resource selection analyses (Table 1). The number of New England cottontails monitored per site ranged from 1 to 11 (x̅ = 5.2) and the number of eastern cottontails ranged from 1 to 14 (x̅ = 5.7; Appendix S5, Supporting Information). Available resources did not differ between eastern cottontail "more prevalent" and "less prevalent" sites in any consistent manner among categories of shrubland (Appendix S6, Supporting Information). The proportion of eastern cottontails within five sites changed between years; we accordingly changed their designation for our analyses (Table 1).
We detected no species-specific bias in trapping, based on correlation between species composition in our captured population and in pellet samples (r = 0.642, n = 17 site x year combinations, p = 0.005). Mean movement distance between successive locations did not differ between cottontail species and equaled 75 m (Cheeseman, 2017). As such, we defined available locations as those within 75 m of each used coordinate. To create vegetation data layers, we sampled vegetation in 1,191 plots in both leaf-off and leaf-on seasons.
For all models, σ i was high relative to β i and all SE σ I were low relative to σ i for the random effect of individual in our models for New England cottontails (Appendix 7, Supporting Information) and eastern cottontails (Appendix 8, Supporting Information).
However, all SE σ i were high relative to σ i for the random effects of site and year, such that we had little evidence for their importance as a source of variation in resource use. Thus, we proceeded to make inferences with models including only the random effect of individual. As where eastern cottontails were less prevalent, New England cottontails also selected for moderate values of proportion shrub cover in these areas (Figure 3b).

| Resource selection
Where eastern cottontails were less prevalent in the leaf-on season, New England cottontails did not select for shrub cover, but did use densities of multiflora rose characteristic of early-and midsuccessional shrublands (Figure 3e), and avoided native palatable stems (Figure 3f). In these areas, New England cottontails displayed strong selection of seasonal canopy starting at low proportion closure (Figure 4a), selected for tall forbs (Figure 4b), and low-to-intermediate proportion graminoid cover (Figure 4c), all characteristic of early successional shrublands. However, where eastern cottontails were more prevalent in the leaf-on season, New England cottontails selected for the high values of persistent canopy closure (Figure 3a) typified by mid-and late successional forests, and moderate-to-high values of shrub cover (Figure 3b).  were more prevalent in the leaf-off season, New England cottontail resource selection was more characteristic of interior late successional shrublands than mid-successional shrublands and forest edges ( Figure 5). During the leaf-on season, where eastern cottontails were less prevalent, New England cottontail resource selection was characteristic of early-to mid-successional shrublands ( Figure 5). In contrast, where eastern cottontails were more prevalent, New England cottontail relative probability of use during the leaf-on season was characterized by resources consistent with mid-to late successional shrublands ( Figure 5). Shifts in resource use by New England cottontails, in sites where the eastern cottontail prevalence designation shifted between years, were also evident (Appendix 11, Supporting Information).
Eastern cottontail use was characteristic of early successional shrublands in both seasons, with only slight shifts in use toward resources characteristic of older successional shrublands during the leaf-off season ( Figure 5, 95% prediction intervals reported in Appendices S8 and S9, Supporting Information).

| D ISCUSS I ON
We provide evidence for displacement of a native species by an introduced species from otherwise selected resources. Moreover, the use of low-to-moderate densities of the common invasive shrub Japanese barberry was higher where the introduced competitor was more prevalent than where it was less prevalent. Competitive displacement by nonnative species has resulted in negative demographic effects and has led to population declines and extinctions (reviewed in Cleland, 2001 andHarris, 2009). For example, competition with gray squirrels Sciurus carolinensis over food resources has led to widespread declines in Eurasian red squirrel S. vulgaris where the two are now sympatric (Bertolino, Montezemolo, Preatoni, Wauters, & Martinoli, 2014). In this study, we observed evidence that New England cottontails were displaced into later successional-stage shrublands where eastern cottontails are prevalent. Of particular concern is displacement of New England cottontails into Japanese barberry and tall shrubs in the leaf-off season where eastern cottontails were more prevalent. Although tall shrubs and Japanese barberry may provide additional and novel escape cover, habitat quality and forage availability influence overwinter survival of New England cottontails (Smith & Litvaitis, 2000;Villafuerte et al., 1997) and these resources may not provide adequate forage (Warren et al., 2016). Furthermore, high tick burdens have been linked to population crashes in cottontails (Smith & Cheatum, 1944), and for New England cottontails, higher tick burdens have been observed where sites are dominated by invasive vegetation, including Japanese barberry (Mello, 2018 Lees & Bell, 2008;Zavaleta et al., 2001). Where competition with a nonnative species is a conservation concern and eradication is undesirable or infeasible, managing habitat to improve survival or recruitment of native species is an option (Cole et al., 2005). In European forests, management to promote hazel Corylus avellana over oak Quercus robur within deciduous forests has been recommended to alter competitive interactions in favor of the native Eurasian red squirrel where it is being displaced by invasive gray squirrels (Kenward & Holm, 1993). Here, we present evidence that New England cottontails use shrublands with high canopy closure but that eastern cottontails avoid these areas. This is similar to the competitive interactions hypothesized to occur between Appalachian and eastern cottontails, where Appalachian cottontails are displaced from successional shrublands into mature forest and ericaceous shrublands (Russell et al., 1999). We suggest that where competition with eastern cottontails is a concern and eastern cottontail removal is deemed infeasible, managing for large patches of dense, high canopy closure successional shrublands, ericaceous shrublands with intermixed gap phase processes, or shrub-covered forested wetlands would allow for New England cottontails use, without encouraging use by eastern cottontails. Adaptive management of habitat based on our hypotheses and analysis of the fitness consequences of resource selection by New England cottontails in our system are ongoing.
Antagonistic interactions between resident New England cottontails and colonizing eastern cottontails that could explain altered F I G U R E 5 Example of (a) site-level cover types and (b) predicted resource selection in the leaf-off season by New England cottontails where eastern cottontails were less prevalent, (c) New England cottontails where eastern cottontails were more prevalent, (d) eastern cottontails where they were more prevalent, and (e) predicted resource selection in the leaf-on season by New England cottontails where eastern cottontails were less prevalent, (f) New England cottontails where eastern cottontails were more prevalent and (g) eastern cottontails where they were more prevalent.

Predicted Selection
High Low resource use by New England cottontails in the presence of competing eastern cottontails have not been observed (Probert & Litvaitis, 1996). However, the ability for generalist eastern cottontails to colonize early successional shrublands before they are suitable for New England cottontail occupancy may limit colonization by New England cottontails (Probert & Litvaitis, 1996) and could partially explain the pattern of resource use observed here. It is also probable that a similar form of scramble competition occurs within sites at the scale of resource selection, where eastern cottontails use areas at an earlier successional stage than is suitable for New England cottontails, and are able to hold them as they mature into suitability. Moreover, New England cottontail resource selection shifted seasonally regardless of eastern cottontail prevalence, from dense, high canopy shrublands in the leaf-off season to less dense, open-canopy shrublands in the leafon season, likely in response to seasonal availability of forage and apparent cover provided by tall forbs in the leaf-on season. We propose that seasonal shifts away from resources associated with early successional shrublands result in a small-scale form of competitive release in these areas benefiting eastern cottontails. Individual niche expansion resulting from competitive release is documented (Bodey, McDonald, & Bearhop, 2009;Bolnick et al., 2010), and such behavioral responses by resident eastern cottontails to seasonal competitive release within early successional shrublands provide a mechanism for competitive displacement of New England cottontails that is not reliant on patterns of colonization or displacement through antagonistic interactions.

| CON CLUS IONS
Using conditional resource selection functions, we observed that resource selection, including selection for common invasive plants, by the imperiled native New England cottontail varied by the relative prevalence of a nonnative competitor. Our results suggest seasonal competitive release as a mechanism that facilitates competitive displacement of resident New England cottontails by its nonnative competitor. We caution that current management strategies may benefit the nonnative competitor over the imperiled target species. Our results suggest silvicultural approaches such as clearcutting or mowing to create successional shrublands may favor the nonnative competitor. Instead, we suggest seed tree cuts, shelter wood cuts, or selective thinning to create canopy gaps could be used to adaptively manage sites to promote the native species where its nonnative competitor is present. Treatments

AUTH O R S' CO NTR I B UTI O N S
AEC and JBC conceived of the ideas and designed methodology with input from CMW and SJR. AEC collected the data. Genetic work was supervised by CMW. AEC and JBC analyzed the data. AEC and JBC led manuscript preparation. All authors contributed substantially to drafts and gave final approval for publication.

DATA ACCE SS I B I LIT Y
Data available from the Dryad Digital Repository: https://doi. org/10.5061/dryad.72qb1k2.