Influence of food subsidies on the foraging ecology of a synanthropic species in protected areas

,


IntroductIon
Human activities are now considered to be the predominant drivers of global change (Ellis et al. 2010) and consequently have widespread and often negative impacts on ecosystems (Vitousek et al. 1997, Kareiva et al. 2007).In addition to direct impacts through habitat loss and fragmentation, ecosystems are altered indirectly by the widespread availability of food subsidies to wildlife that result from human activities.
There is growing evidence that human-derived food subsidies to wildlife have individual-and population-level impacts, with subsequent shifts in community structure and trophic cascades (Newsome et al. 2014).Certain species, referred to collectively as synanthropes (Johnston 2001), benefit from associations with human activities (Hansen et al. 2005), and have increased in abundance as a result of the availability of predictable human foods (Kristan and Boarman 2003, Withey and Marzluff 2008, Webb et al. 2011, Newsome v www.esajournals.org WEST ET AL. et al. 2014).In turn, elevated populations of synanthropes threaten a number of less abundant species through predation, competition, and hybridization (Simberloff and Stiling 1996, Jerzak 2001, Vigallon and Marzluff 2005, Marzluff and Neatherlin 2006).Impacts to sensitive species can extend beyond human-dominated areas, as "spillover predation" can occur when synanthropes move into less altered landscapes and prey on sensitive species (Marzluff et al. 1994, Rand andLouda 2006).As a result, the conservation benefit of protected areas can diminish as human activity increases within and around these areas (Radeloff et al. 2012, Wood et al. 2014).Consequently, understanding how individuals respond to spatial and temporal patterns in the availability of food subsidies can inform the conservation of sensitive species occurring in protected areas.
One particularly challenging synanthropic species problem involves nearly global increases in the abundance and distribution of species in the family Corvidae (ravens, crows, magpies, and jays) which have occurred in response to urbanization and human-induced changes in food resources.Corvids threaten a number of sensitive native species through predation on nests and juveniles (Masselink 2001, Webb et al. 2011) and are particularly difficult to manage because they are widely and abundantly distributed (Boarman 2003) and can learn to avoid control measures (Cornell et al. 2011, Peterson andColwell 2014).While several landscape-scale studies have investigated the effects of humanderived food subsidies on corvid abundance and distribution in urban areas and along urbanrural gradients (Vigallon and Marzluff 2005, Marzluff and Neatherlin 2006, Marzluff et al. 2007, Withey and Marzluff 2008), relatively little research has focused on the response of corvid populations to human-derived food resources inside protected areas (Scarpignato andGeorge 2013, Walker andMarzluff 2015).In addition, most research to date has addressed populationlevel effects of human-derived resource subsidies on synanthropic species, while relatively few studies have investigated how changes in food subsidies alter aspects of individual behavior such as foraging patterns and space use (Boutin 1990, Schoech and Bowman 2001, Scarpignato and George 2013).
Here, we sought to understand how point sources of human foods within protected areas influenced the foraging ecology of Steller's jays (Cyanocitta stelleri) by combining stable isotopebased diet assessments with fine-scale information on space use.Steller's jays have increased in abundance in western North America over the past several decades, presumably because of the expansion of human activities and associated food resources within their coniferous forest habitat (Peery and Henry 2010, Sauer et al. 2014, Wood et al. 2014).Within parks, Steller's jay densities are particularly high around point sources of human foods such as campgrounds.Steller's jays are considered to be important nest predators of marbled murrelets (Brachyramphus marmoratus), a threatened seabird that often nests in remnant stands of old-growth forests occurring in state and federal parks.Indeed, corvid predation, including predation by jays, contributes to low murrelet reproductive rates in these areas, likely because park campgrounds with high jay densities often occur within key murrelet nesting areas (Peery et al. 2004, Peery andHenry 2010).A range of management actions have been implemented in parks to reduce the impact of Steller's jays on marbled murrelets (Peery andHenry 2010, Gabriel andGolightly 2014), but all are hindered by a dearth of information about jay ecology and the patterns and processes by which human foods subsidize jays.Thus, our specific objectives were to (1) compare the diet of individual jays using subsidized (campgrounds) and unsubsidized (forests) habitats, (2) determine whether the consumption of human foods varies by age and sex and in response to park visitation, (3) relate jay space use to the consumption of human foods, and (4) determine the spatial scale at which human foods in campgrounds subsidize jays.For the fourth objective, we estimated how far jays typically traveled between forest and campground habitats to consume human foods, our logic being that such movements could potentially benefit jay populations at spatial scales that exceed campground footprints.

Study areas
Our study was conducted in four parks in California containing remnant old-growth forest  1).Paired sampling was conducted at campground and noncampground (forest) sites in each park.Forest sites were selected based on accessibility, the presence of territory-holding jays, and were located 1 to 2 km from campgrounds, as previous studies suggest that Steller's jays do not travel more than 1 km to access human food (Marzluff and Neatherlin 2006).Stable isotope assessments of jay diets were conducted for all four parks, whereas radiotelemetry work was limited to jays sampled in Big Basin.

Captures and sampling
Steller's jays were captured during the breeding season (April to August) using live traps (model 1045; Havahart, Lititz, Pennsylvania,   (Pyle et al. 1987).We define AHY jays as those known to have hatched before the calendar year of banding and HY jays as those known to have hatched during the calendar year of banding.The sex of all individuals was determined through observations of sexspecific vocalizations, tail length measurements (Pyle et al. 1987) and later confirmed using polymerase chain reaction methods of extracted DNA from blood samples (Griffiths et al. 1998).Blood samples used for molecular sexing and stable isotope analysis were collected by puncture of the brachial vein using a sterile 26-gauge needle and transferred into capillary tubes (150 μL).

Stable isotope analysis
We used red blood cells (hereafter, blood) to characterize the level of human-derived foods consumed by jays as they reflect diet items incorporated up to approximately 40 d prior to sampling (Hobson andClark 1992, Bearhop et al. 2002).Stable isotopes are a useful tool for reconstructing consumer diets (Hobson and Clark 1992, Caut et al. 2008, Ben-David and Flaherty 2012) as the isotopic ratios of nitrogen ( 15 N/ 14 N, noted δ 15 N) and carbon ( 13 C/ 12 C, noted δ 13 C) in consumer tissues predictably reflect those in their diets (DeNiro and Epstein 1978).Human foods are often derived from or processed using corn and cane sugar, which are enriched in δ 13 C (Jahren et al. 2006).This δ 13 C signal is traceable in the wildlife that consume human-derived foods (Newsome et al. 2010) and is thus useful for determining Steller's jays' reliance on human food subsidies.
Blood samples collected for stable isotope analyses were centrifuged to separate red blood cells from sera and were frozen at −20°C until dried for analysis.We sampled jay diet sources (invertebrates, berries, acorns, conifer seeds, and human-derived foods) at capture sites in each park (northern California: 2011, Butano park: 2012and 2013, and Big Basin: 2011-2013) to estimate the proportional contribution of isotopically distinct items to jay diets (listed in Appendix S1).Potential diet sources were sampled at jay capture sites approximately once every 2 weeks throughout each field season to account for temporal variation in the isotopic signatures of diet items.We also collected representative human food samples at campgrounds when we observed jays foraging in unattended campsites and when we observed campers providing food directly to jays.All invertebrate samples were classified to order.We did not include marbled murrelets eggs or nestlings as a potential dietary source in stable isotope analyses of Steller's jay diet as only a very small fraction of jays were likely to have consumed murrelet eggs given the much greater abundance of jays than murrelet nests in our sampling areas (see Appendix S2).In addition, mixing models tend to attribute some proportion of all dietary sources considered to a consumer's diet, even rare or nonexistent ones, in this case potentially overestimating the importance of murrelets in jay diets and thus underestimating more frequently consumed food sources (Parnell et al. 2008, Phillips et al. 2014).
Blood samples were dried for 48 h at 56°C and homogenized with a spatula.Diet samples were dried for 72 h at 56°C and homogenized in a ball mill.For 13 C and 15 N analysis, samples were weighed, placed in tin capsules, and were sent to the Stable Isotope Facility at the University of Wyoming to be analyzed with a Costech 4010 elemental analyzer attached to a Thermo Finigan Delta Plus XP Continuous Flow Isotope Ratio Mass Spectrometer.Results are provided as per mil (parts per thousand [‰]) ratios relative to the international standards of Pee Dee Belemnite (PDB; δ 13 C) and atmospheric nitrogen (AIR; δ 15 N) with calibrated internal laboratory standards.P-values >0.05), and consequently diet groups from each year were combined for each park.Human-derived foods were indistinguishable isotopically among the four parks.Invertebrates and mast differed among Big Basin Redwoods State Park, Butano State Park, and the two combined northern California parks, but were indistinguishable between Jedediah Smith Redwoods State Park and Prairie Creek Redwoods State Park.Thus, K nearest neighbor randomization tests and mixing models were conducted by pooling data for northern California sites, but the two central California parks were treated separately.

Statistical analysis of stable isotope data
We estimated the proportional contribution of isotopically distinct diet groups to jay diets using a Bayesian stable isotope mixing model, Stable Isotope Analysis in R (SIAR) 4.2 (Parnell et al. 2008) within R 3.1.0(R Development Core Team 2014).SIAR models were conducted separately for each park in central California, but jay samples from the two northern California parks were combined due to small sample sizes.Initially, we ran SIAR models considering three biologically distinct groups: human-derived foods, mast (fruit from trees and shrubs, including acorns, pine seeds, and berries), and invertebrates (P < 0.05 for all groups based on K randomization tests; Rosing et al. 1998;Fig. 2;Appendix S1).Given the variability in δ 15 N values across our invertebrate samples and to test the robustness of results from our mixing models, we also estimated jay diets with SIAR models using finer diet groupings.Specifically, K randomization tests revealed five isotopically distinct groups within each park (P < 0.05 for all groups): three invertebrate groups (group 1: Araneae, Coleoptera, and Hemiptera; group 2: nonpredatory Diptera and Hymenoptera; group 3: predatory Diptera and Hymenoptera), human-derived foods, and a fifth group that included mast and invertebrates in the orders Lepidoptera and Orthoptera.Mast and Lepidoptera/Orthoptera diet sources were not grouped in the second set of SIAR models-and thus treated as a fifth and sixth group-as they represent biologically relevant diet sources and because the elemental concentration of these diet sources varies significantly.Models were parameterized with uniform priors using default values for MCMC estimation (chain length = 2 × 10 5 , burning = 5 × 10 4 , thinning = 15); consequently, the distributions of source contributions are based on 10 5 estimations and are presented as Bayesian 95% credible intervals, which represent each prey group's likely proportional contribution to Steller's jay diet (Parnell et al. 2010).The median values of proportional diet estimates were considered the most likely level of contribution to the diet, recognizing that solutions can occur anywhere within the credible interval (Jackson et al. 2009, Parnell et al. 2010).In addition to requiring isotope estimates for consumers and potential diet items, SIAR models require estimates of elemental concentration (wt% C, wt% N) and trophic discrimination (which is the isotopic difference between source and consumer as food is incorporated into the tissues).We used the following isotopic discrimination factors provided in the literature based on an experimental study that estimated discrimination factors for a passerine diet similar to that of wild Steller's jays: +1.8‰ (SD = 0.66) for δ 15 N, and +1.5‰ (SD = 0.12) for δ 13 C (Pearson et al. 2003).We used the average measured elemental concentrations for each diet group (Appendix S1).
To characterize factors influencing patterns in the consumption of human-derived foods, we used raw isotope values from blood as a proxy for jay diet, with increased enrichment of  , 2011-2013 (n = 197).Diet sources show mean with error bars depicting the standard deviation.Diet sources were adjusted using a discrimination factor of 1.5 ± 0.12 and 1.8 ± 0.66 for δ 13 C and δ 15 N, respectively.δ 13 C indicating greater consumption of humanderived foods (Phillips et al. 2005, Newsome et al. 2010).We used an analysis of covariance (ANCOVA) framework where δ 13 C was the response variable and Julian date of each sample (20 d prior to sampling date; our estimate of the half-life of blood turnover for Steller's jays) was used as a continuous predictor variable, while year, park, sex, and age class were used as categorical predictor variables.To understand the relationship between the consumption of human-derived foods and changes in human visitation (a proxy for availability of humanderived foods), we conducted a linear regression with δ 13 C derived from blood and mean number of occupied campsites as our response and predictor variables, respectively.Here, we used the number of occupied campsites as an index of human presence, which we estimated by averaging the number of occupied campsites between 17 and 23 d prior to each blood sampling date (using 20 d prior to each sampling date as the median).We estimated the half-life of stable isotopes (blood turnover) in Steller's jay blood (20 d) using known half-life data for red blood cells in American crows (Corvus brachyrhynchos), which is approximately 30 d (Hobson and Clark 1992) and the allometric scaling relationship between time and body mass, M 1/4 , where M is mass (West et al. 1997).This analysis of associations between δ 13 C and human visitation was restricted to Big Basin due to smaller sample sizes and a limited range in the number of occupied campgrounds at the other parks.

Radiotelemetry
To characterize individual space use, we fitted a subsample of jays captured in Big Basin in 2011, 2012, and 2013 with backpack-mounted radiotransmitters (model A1050; Advanced Telemetry Systems, Isanti, Minnesota, USA; Buehler et al. 1995).Transmitters had an expected lifespan of 12 weeks and were affixed using 2.5-mm Teflon ribbon (Rappole and Tipton 1991).We tracked jays from 1 May to 15 August, a period that largely coincides with known Steller's jay nest initiation dates in coastal California (Brown 1964).We used a handheld telemetry receiver (model R410; Advanced Telemetry Systems) to relocate focal individuals through daily groundbased telemetry surveys, which were conducted between 07:00 and 19:00 h by walking trails and driving roads in search of signals emanating from radio-marked individuals, allowing at least 2 h between location determinations taken on the same day (Swihart and Slade 1985).When a signal was heard, we used homing techniques (White and Garrott 2012) and triangulation to locate the individual's location (recorded using GPS units; Garmin 72h, Garmin, Olathe, Kansas, USA).We also opportunistically collected nocturnal roost locations (21:00-04:00 h) for each individual and included these locations in estimates of home range size.
We estimated jay core area and home range size as the area (ha) within 50% and 95% probability contours of fixed kernel density estimates, respectively (Seaman et al. 1999, Kernohan et al. 2001).We used least squares cross-validation for bandwidth (i.e., smoothing parameter) estimation (Silverman 1986, Gitzen et al. 2006) in the Geospatial Modeling Environment for ArcGIS (Beyer 2012).To characterize the use of campgrounds by radio-marked jays in Big Basin, we constructed a geospatial layer representing campground (by walking campground boundaries with a handheld GPS) and noncampground (all remaining forested land cover) areas with ArcInfo GIS software (ESRI 2014) and estimated the proportion of each individual's home range that overlapped campground areas.
We combined stable isotope analyses of blood sampled from radio-marked jays with core and home range data from each individual to relate jay space use and foraging behavior.We determined whether the size of individual jay's home range was correlated with δ 13 C enrichment (i.e., the consumption of humanderived foods) using partial correlation analysis to control for the effect of Julian date.Our expectation was that home range size and δ 13 C enrichment would be negatively related given that space use often contracts with increasing food availability, including in synanthropic species (Boutin 1990, Shochat et al. 2004).We also tested whether the proportion of each jay's home range that overlapped campgrounds was correlated with δ 13 C in the diet, again using partial correlation while controlling for Julian date.Analyses were conducted separately for males and females because of differences in home range size between sexes.

Patterns of Steller's jay space use
We radio-marked a total of 62 Steller's jays in Big Basin, 50 in campgrounds and 12 at forest sites (campground: n = 38 males and n = 12 females; forest: n = 8 males and n = 4 females).We obtained an average of 34.9 locations per individual (range: 31-40) and tracked individuals for an average of 87 d (range: 62-124 d).We identified four general patterns of space use by radiomarked jays: (1) individuals captured in campgrounds with at least 50% of their core areas inside campgrounds ("campground jays"; n = 35), (2) individuals captured in campgrounds with less than 50% of their core areas inside campgrounds that made extensive use of adjacent surrounding forests ("periphery jays"; n = 15), (3) individuals captured in forests that made exclusive use of forests and never visited campgrounds ("forest jays"; n = 7), and (4) individuals captured in forests that used both campgrounds and forests ("intermediate jays"; n = 5; Fig. 3).Campground jays were regularly detected within campground boundaries as 68% of diurnal locations (n = 718/1057) and 86% of nighttime  roosting locations (n = 120/140) occurred within campground boundaries (Fig. 4a).Periphery jays were detected less frequently in campgrounds with 47% of diurnal (n = 217/462) and 30% of roosting (n = 18/60) locations occurring within campground boundaries (Fig. 4b).Intermediate jays were detected within campground boundaries less frequently than the other two groups with 27% of diurnal locations (n = 54/197) and 10% of roosting locations (n = 2/20) occurring within campground boundaries (Fig. 4c).The frequency distribution of distances from campgrounds was relatively uniform for intermediate jays from 0.4 to 2.0 km and approximately 8% of their locations occurred 1.8 to 2.0 km from campgrounds (Fig. 4c).

Patterns in Steller's jay diet inferred from stable isotopes
We analyzed 295 individual Steller's jay blood samples for δ 13 C and δ 15 N stable isotope ratios (Jedediah Smith: n = 20; Prairie Creek: n = 25; Big Basin: n = 197; Butano: n = 53).Both δ 13 C and δ 15 N isotope signatures of Steller's jay blood samples fell within the mixing space of the three diet groups considered (Fig. 2).δ 13 C signatures were much greater for humanderived foods than mast and invertebrates, and jays enriched in δ 13 C could thus reasonably be assumed to consume a greater proportion of human foods.
Human-derived foods made up a significant proportion of the diet of Steller's jays using campgrounds, and campground jays consumed more human-derived foods than forest jays based on three-source SIAR models (Fig. 5).Indeed, human foods constituted from 35% (95% CI: 28-42%; Butano) to 53% (95% CI: 48-57%; Big Basin) of the diet of campground jays, but only 14% (95% CI: 10-46%; northern parks) to 25% (95% CI: 13-36%; Big Basin) of the diet of forest jays.SIAR models also suggested that both intermediate and periphery jays consumed more human foods than forest jays in Big Basin and that intermediate jays consumed more human foods than forest jays in the northern parks (Fig. 5).Intermediate jays were identified in northern parks based on space-use data in Goldenberg et al. (2016).Mast generally represented a larger proportion of the "natural" dietary component in all four groups of jays, with invertebrates making up a smaller portion of jay diets (Fig. 5).Note that the apparent difference in the relative importance of mast and invertebrates between SIAR models (Fig. 5) and the bivariate isotopic plot (Fig. 2) was likely due to greater nitrogen concentration in invertebrates that led to similar estimates of δ 15 N in jays and invertebrates (Fig. 2), despite the greater importance of mast as a diet source (elemental concentration was incorporated in SIAR models).Campground jays appeared to shift their diets toward human-derived foods by consuming fewer invertebrates (Fig. 5).Results from SIAR mixing models using six diet sources yielded similar results as the three-source model, supporting the finding that human-derived foods made up a significant proportion of the diet of jays sampled in campgrounds (Appendix S3).

Relationship between diet and space use of jays in campgrounds
We observed an inverse relationship between δ 13 C enrichment and home range size for male campground and periphery jays after controlling for Julian date (partial r = −0.39,P = 0.01, n = 38; Fig. 7a); thus, males consuming a relatively high proportion of human foods tended to use smaller areas.The proportion of male home ranges that overlapped campgrounds was also positively associated with δ 13 C enrichment (partial r = 0.47, P = 0.002; n = 38; Fig. 7b); in other words, males that primarily used campgrounds tended to have a greater proportion of human foods in their diet.

dIscussIon
Our study was unique in that we combined isotope-based dietary assessments with detailed information on individual space use to provide an understanding of how human-derived foods in protected areas subsidize a synanthropic species.The enrichment of δ 13 C in human-derived foods allowed us to demonstrate that these foods comprise a greater proportion of the diet of campground than forest jays, presumably because of heavy human visitation and associated food subsidies in campgrounds.Indeed, human-derived foods constituted more than half the diet of campground jays and thus were more important proportionally than both mast and insects combined, in the most heavily visited park (Big Basin with >685,000 visitors in 2013).While the consumption of human-derived foods by campground jays was less pronounced in the least visited park (Butano with 57,034 visitors in 2013), human-derived foods nevertheless constituted 35% of jay diets and could be an important subsidy at this site given their high caloric content.Thus, concordance in results among multiple parks suggests that the subsidization of jays is a regional phenomenon.
Steller's jays in campgrounds exhibited an apparent functional response by consuming an increasing proportion of human-derived foods as park visitation increased.Park visitation peaks in July and August and coincides with the jay fledgling period such that food subsidies are likely most prevalent during the energetically expensive nestling and fledgling provisioning stages.Of note, breeding jays in Big Basin appeared to preferentially feed their juveniles humanderived foods in campgrounds as HY individuals were more enriched in δ 13 C than AHY individuals.Collectively, these results suggest that food subsidies may increase jay reproductive output in campgrounds, as has been observed in numerous other wildlife studies (Rodewald et al. 2011, Webb et al. 2011, Oro et al. 2013, Newsome et al. 2014), and potentially provide a source of young birds that disperse into murrelet nesting areas further from parks.Additional research, however, is needed to understand the potential fitness consequences of human foods in campgrounds, as well as possible larger-scale population effects.
In Big Basin, jays roosting, and presumably breeding, outside of campgrounds regularly traveled to campgrounds to exploit human foods.These individuals could be classified into two categories: jays roosting and breeding in forests at the periphery of campgrounds that regularly made short-distance movements to forage in campgrounds (i.e., periphery jays) and jays Fig. 7. Carbon isotope ratios (δ 13 C) of blood samples from radio-tagged male (filled) and female (unfilled) Steller's jays captured in Big Basin campgrounds (including "periphery" individuals) with respect to size of home range (a) and proportion of core area overlapping campgrounds (b).Solid and dashed lines represent the partial correlation (controlling for Julian date) between variables for males and females, respectively.that roosted in interior forests and traveled longer distances (1-2 km) to forage in campgrounds (i.e., intermediate jays).Jays in both groups were enriched for δ 13 C and appeared to consume more human-derived foods than forest jays.Periphery jays may have been subordinate to individuals relegated to roosting and breeding in areas adjacent to campgrounds with limited access to human foods given their intermediate level of δ 13 C enrichment (compared with forest and campground jays).Indeed, pairs of breeding jays exhibit aggressiveness and intolerance over conspecifics in areas immediately surrounding their nests (Brown 1963(Brown , 1964(Brown , 1969)), suggesting that periphery jays may encounter a "landscape" of aggressiveness when attempting to exploit human-derived foods in campgrounds containing high jay densities, recognizing that we do not have observations of social interactions among periphery and campground jays to support this hypothesis.Intermediate jays included two individuals that made periodic longer-distance foraging trips from forests to campgrounds and three individuals that shifted their use of space from forests to campgrounds over the course of the breeding season.Moreover, the fact that five of 12 jays captured and radio-marked in forests were observed foraging in campgrounds (i.e., were intermediate jays) and 37-45% of their diet was comprised of human foods suggests that campgrounds subsidized approximately 42% of the local jay population, recognizing that our modest sample size of jays captured in forests resulted in considerable sampling uncertainty.Moreover, movements by intermediate jays suggests that the "ecological footprint" of campgrounds extended approximately 2 km beyond their spatial footprint, at least in terms of the subsidies they provided to jays during the breeding season.Indeed, our finding that individual jays detected 2 km from campgrounds were subsidized by human foods differs from previous finding that Steller's jays did not travel more than 1 km to access human foods (Marzluff and Neatherlin 2006).Our results, then, provide additional evidence that campgrounds subsidize Steller's jays beyond their boundaries and potentially indicate that the quality of oldgrowth habitat surrounding campgrounds may be reduced for nesting marbled murrelets by these subsidies.
SIAR mixing models indicated that humanderived foods represented a nonnegligible component of the diet of forest jays that were never observed in campgrounds.Forest jays may have had access to human foods provided by visitors using extensive park trail networks to access forested areas containing jays sampled for isotopic analysis.Alternatively, some forest jays may have traveled to campgrounds (a behavior identified in intermediate jays) and consumed human foods prior to sampling, a pattern that would be reflected in the isotopic composition of sampled blood.Forest birds may also have made undetected movements to exploit food resources during radiotracking, although such behavior was readily detected in intermediate jays that either made regular foraging trips or permanently shifted their use to campgrounds.Finally, SIAR models may have overestimated the importance of human foods in the diet of Steller's jays, as dietary estimates can be sensitive to choice of discrimination factor (Kurle et al. 2013) and we adjusted prey isotopic signatures using factors developed from a surrogate species (Pearson et al. 2003).However, the estimated proportional contribution of human-derived foods in the diet of forest jays was lowest in the park with the fewest visitors (Butano), a pattern that would be expected if forest jays were in fact consuming some human foods.
Observed movement and diet patterns may also help explain previous observations of high Steller's jay densities in areas of parks frequented by humans in both central California and other regions (Suddjian 2009, Walker andMarzluff 2015).Among campground jays, individuals with home ranges mostly in campgrounds tended to rely more on human-derived foods and had relatively small home ranges, suggesting that human-derived foods allowed individuals to meet their dietary needs within relatively small areas.This finding mirrors a general tendency in birds, where individuals reduce their foraging areas when food is abundant and allows more individuals to occupy a given area (Boutin 1990, Shochat et al. 2004).Tolerance of conspecifics (home range overlap) also appeared to be high in males in campgrounds, despite the fact that this species is generally considered to hold distinct breeding territories (Oberski and Wilson 1991).However, territoriality in birds often weakens v www.esajournals.orgWEST ET AL.
when food is abundant and the incentive to exclude conspecifics from resources decreases, which could explain observed patterns in spacing among jays exploiting human subsidies (Ewald and Carpenter 1978, Hixon 1980, Schoener 1983).
Results from this study have several implications for marbled murrelet conservation as most remaining nesting habitat for this species in California occurs within protected areas receiving heavy recreational visitation (Baker et al. 2006).Of note, 65% of known murrelet nests in central California (Peery et al. 2004), most of which failed due to nest predation, occurred within the ecological footprint (i.e., 2 km) of Big Basin campgrounds, suggesting that point sources of human foods subsidize Steller's jays in important murrelet nesting areas.All four sampled parks have long-standing visitor education and trash management programs designed to reduce food subsidies to Steller's jays, and ultimately to reduce nest predation on marbled murrelets.Our finding that a high proportion of the diet of Steller's jays breeding in and around park campgrounds was comprised of human-derived foods indicates that these programs have not fully achieved their intended effect.While reducing food subsidies in areas receiving high levels of camping and recreation is challenging, changing human behavior for the purposes of species conservation through education and enforcement has been successful in some instances (Craighead et al. 1995, Hopkins et al. 2014).More recently, controlled taste aversion has been implemented to reduce Steller's jay predation on murrelet nests in many regional parks, a strategy that involves deploying noxious eggs that mimic murrelet eggs in appearance at landscape scales (Gabriel and Golightly 2014).Our results suggest that such a strategy would likely need to consider dense deployments of mimic eggs in campgrounds given that jays appear to aggregate in these areas due to the prevalence of humanderived foods.Alternatively, removal of Steller's jays from campgrounds could be an effective means of reducing jay densities in marbled murrelet habitat, but would likely need to be continued on a regular basis to account for jays likely to disperse into these areas from surrounding forests.Ultimately, moving campgrounds and associated human-derived food resources out of marbled murrelet nesting habitat (old-growth forests) may be the most effective long-term strategy for reducing subsidies to jays in these areas.
In conclusion, our work suggests that point sources of food subsidies in areas of high human activity can influence the behavior and foraging strategies of synanthropic species in less developed areas and over relatively large spatial scales (Brashares et al. 2001, Marzluff and Neatherlin 2006, Wood et al. 2014).Conservation efforts for rare species can thus benefit from identifying key locations on the landscape and periods in the life cycle where food subsidies benefit synanthropes, even for populations within protected areas.
v www.esajournals.orgWEST ET AL. nesting habitat for marbled murrelets: Big Basin Redwoods State Park, Butano State Park, Jedediah Smith Redwoods State Park, and Prairie Creek Redwoods State Park (Fig.

Fig. 1 .
Fig. 1.Location of four protected areas in California where Steller's jays were sampled for isotopic analyses (all parks) and radio-marked (Big Basin Redwoods State Park only).

δ
13 C and δ 15 N values for diet groups did not differ significantly among years within any of the four sampled parks based on ANOVA (all v www.esajournals.orgWEST ET AL.

Fig. 2
Fig. 2. δ 13 C and δ 15 N isotope ratios of individual Steller's jays based on blood samples (circles) and their potential diet sources used in Stable Isotope Analysis in R diet calculations, sampled during the breeding season in Big Basin Redwoods State Park, 2011-2013  (n = 197).Diet sources show mean with error bars depicting the standard deviation.Diet sources were adjusted using a discrimination factor of 1.5 ± 0.12 and 1.8 ± 0.66 for δ 13 C and δ 15 N, respectively.

Fig. 3 .
Fig. 3. Representative core areas and home ranges (50% and 95% utilization distributions, respectively) for male forest, campground, periphery, and intermediate Steller's jays radio-marked in Big Basin Redwoods State Park in 2011.Also shown in (a) is the estimated 2-km "ecological footprint" of campgrounds and its overlap with 11 of the 17 known marbled murrelet nests in central California.

Fig. 4 .
Fig. 4. Frequency distribution of distances that (a) campground, (b) periphery, and (c) intermediate Steller's jays were detected from campgrounds in Big Basin Redwoods State Park during the breeding season (April-August), 2011-2013.The different shades of gray represent individual intermediate jays.Only one shade was used to represent campground and periphery jays because of the relatively large numbers of individuals sampled.

acknowledgMents
This project benefited from the assistance of the following field and laboratory technicians: D. Becker, M. Bourdeaux, J. Clark, J. Foggia, H.Jones, K. Kerans, B. Kirby, S. Matasick, J. Piorkowski, S. Price, M. Schlothan, E. Reid-Wainscoat, A. Hannah, and G. O'Toole.We also thank Portia Halbert and California State Parks staff for logistical assistance in the field.Finally, we thank Bill Karasov, Jon Pauli, T. Luke George, and Laird Henkel for their assistance throughout this project.The authors have no conflict of interest to declare.Funding was provided by Save the Redwoods League (MSN150028, MSN160820), the College of Agriculture and Life Sciences (UW-Madison), the Department of Forest and Wildlife Ecology (UW-Madison), the University of Wisconsin-Madison Office of the Vice Chancellor Research and Graduate Education (MSN164905), the U.S. Fish and Wildlife Service, the California Depart ment of Fish and Wildlife, and Dan Emmett.
lIterature cIted