Diet of Florida coyotes in a protected wildland and suburban habitat Melissa M. Grigione & Prabir Burman & Sarah Clavio & Steve J. Harper & Denara Manning & Ronald J. Sarno Published online: 19 March 2011 # Springer Science+Business Media, LLC 2011 Abstract Coyotes ( Canis latrans ) arrived to Florida (USA) in the 1960s and are currently found throughout most of the state. The purpose of this study was to determine if the diet of Florida coyotes differed between suburban and wildland habitat types or across seasons. Fresh coyote fecal samples were collected from wildland and suburban habitats in Pinellas County, Florida (USA; 27°54 ′ N, 82°41 ′ W) from May 2005 to March 2007. Diet items in the 49 wildland and 71 suburban samples were identified to the lowest possible taxonomic level by gross morphological characteristics and medullary configurations of dorsal guard hairs. A Poisson regression was utilized to determine the main and interactive effects of habitat and season on the number of different food items per sample. Coyotes in the wildland habitat had greater diet diversity than suburban coyotes. In addition, anthropo- genic waste was recovered over twice as often from coyote fecal samples collected in the suburban habitat. In the wildland habitat, vegetative matter (96%), Insecta (53%), and Rodentia (45%) were recovered most often, as opposed to berries (56%) and Lagomorpha (32%) in the suburban habitat. In both habitats, vegetative matter, berries, and Lagomorpha were recovered most often from coyote fecal samples, whereas Odocoileus virginianus , Lagomorpha , and berries varied the most between wet and dry seasons. This study suggests Urban Ecosyst (2011) 14:655 – 663 DOI 10.1007/s11252-011-0159-6 M. M. Grigione ( * ) Department of Biology, Pace University, 861 Bedford Road, Pleasantville, NY 10570, USA e-mail: mmgrigione@pace.edu P. Burman Department of Statistics, University of California-Davis, Davis, CA 95616, USA S. Clavio : D. Manning Department of Environmental Science and Policy, University of South Florida, Tampa, FL 33647, USA S. J. Harper Pinellas County Environmental Management Environmental Lands, 3620 Fletch Haven Drive, Tarpon Springs, FL 34688, USA R. J. Sarno Department of Biology, Hofstra University, Hempstead, NY 11549-1000, USA that as urbanization increases, diet diversity for the coyote will likely decrease and consumption of anthropogenic items will likely increase. As a result of this, human – coyote conflicts may become more common — particularly in counties, like Pinellas, that are undergoing urbanization. Keywords Canis latrans . Carnivores . Coyotes . Diet . Florida . Urban ecology Introduction Coyotes are opportunistic foragers whose diet varies geographically, as well as between rural and urban habitats (Stratman and Pelton 1997; Quinn 1997; Fedriani et al. 2001; Grinder and Krausman 2001; Lingle et al. 2005; Prugh 2005; Azevedo et al. 2006). For example, Fedriani et al. (2001) reported that anthropogenic foods comprised 25% (dry season) and 14% (wet season) of coyote diets in urban sites in California, whereas anthropogenic foods accounted for only 3% (dry season) and trace amounts (wet season) of coyote diet in adjacent rural areas. Since 1960, human population growth in Florida has increased by nearly 3% a year (Barnett and Dobshinsky 2007). By 2020, an estimated 20 million people will live in Florida, and by 2060 this number will rise to approximately 39 million. As a consequence of this growth, conversion of wildlands for housing and other development are predicted to require nearly 2,832,800 ha of additional land (Barnett and Dobshinsky 2007). Coyotes arrived in Northwestern Florida in 1970. From 1981 to 1988, the number of counties occupied by coyotes rose from 18 to 48 (Florida is comprised of 67 counties). Their numbers appear to be increasing statewide (McCown and Scheick 2007). Given the ability of coyotes to expand their range and thrive in human-dominated landscapes, we assessed coyote diet between suburban and wildland habitats in Pinellas County. With a human population size of approximately 19,000,000, Pinellas is the 6th most populated county in the state (USCB 2009). In Pinellas, we investigated how diet composition differed between coyotes residing in suburban and wildland/protected habitats and whether there was an influence of season on diet. We predicted that coyote diet would be more diverse in the protected wildland habitat and that coyotes would consume more anthropogenically-generated items in suburban areas. Information obtained from this developed county could provide insight into how future urbanization within the state could impact the diet of the area ’ s top predators. Methods Study sites Study sites were classified as “ Wildland (protected) ” and “ Suburban ” based on their percentages of natural habitat. These classifications correspond to definitions estab- lished by Marzluff et al. (2001). Brooker Creek Preserve (BCP) — defined as an area containing unsettled land that may occasionally include dwellings — is approximately 86% natural. In contrast, our suburban areas are approximately 27% natural and are typified by moderate to high density single-family housing with lot sizes of 0.1 to 1.0 ha (Marzluff et al. 2001). 656 Urban Ecosyst (2011) 14:655 – 663 Wildland habitat BCP (27°54 ′ N, 82°41 ′ W) is a 3,350-ha protected area that is actively managed for natural resource protection. Located in the northeast corner of Pinellas County, BCP abuts densely populated residential areas (Fig. 1). The study site consists of extensive pine flatwoods and freshwater swamps. Fauna include white-tailed deer ( Odocoileus virginianus ), raccoons ( Procyon lotor ), armadillo ( Dasypus novemcinctus ), bobcat ( Lynx rufus ), otter ( Lutra canadensis ), wild turkey ( Meleagris gallopavo ), red-shouldered hawks ( Buteo lineatus ), wood storks ( Mycteria americana ), bobwhite quail ( Colinus virginianus ), and gopher tortoise ( Gopherus polyphemus ). Suburban habitat Three suburban sites in Pinellas County were sampled, comprising 433-ha (Fig. 1). These sites were primarily vegetated, such as habitats associated with power line rights-of-way, bike trails, and inactive railroads. All three areas were surrounded by dense suburban development. The wet season for our study was defined as June 1 through October 31 and the dry season as November 1 through May 31 (Chen and Gerber 1990). Field methods Coyote feces were collected between May 2005 and March 2007 from foot paths, power line rights-of-way, and bike trails in BCP and suburban areas. Routes were traversed on foot, bicycle, and All Terrain Vehicles (ATV) three times a week. Each fecal sample was measured (length and diameter) and photographed in the field. Species identification was Fig. 1 Map of suburban ( left panel ) and wildland ( right panel ) study areas in Pinellas County, Florida. Study areas are outlined in yellow Urban Ecosyst (2011) 14:655 – 663 657 determined by feces dimensions and accompanying tracks. Locations of coyote scat were recorded with a handheld GPS unit. Each sample was placed in a paper bag. In order to avoid accumulating domestic dog feces, only fecal samples containing hair and bone fragments and/or accompanied by coyote tracks were collected (Wooding et al. 1984). Similarly, in order to avoid collecting bobcat feces, only samples in excess of 2.5 cm in diameter were collected (Gompper et al. 2003). Laboratory methods Food items in feces were identified to the lowest possible taxonomic category based on bone, teeth, nails, and hair that were recovered from each sample. After removing approximately 4 g of fecal material from each sample for parasite examination as part of a parallel study, the remaining material was oven-dried at 60 – 80°C for at least 48 h to kill any latent parasites (Wagner and Hill 1994). After desiccation, each sample was individually placed in the top of a combination of wire mesh sieves (Stratman and Pelton 1997) and washed thoroughly with a garden spray hose attached to a sink faucet. Remaining food items were transferred to paper plates (one for each sieve) and air dried overnight. Remains were separated into the following categories: hair, bones, teeth, nails, feathers, reptile fragments, insect fragments, vegetative matter, berries, anthropogenic waste (e.g., trash, rope, plastic wrappers), and unknown. Dorsal guard hair of prey was then separated based on gross morphological characteristics (i.e., color, color bands, and color band locations) and slides were made for prey identification. Hair was identified only according to gross morphological characteristics and medullary configurations (Wilkins et al. 1982). To aid in identification, hair, bones, teeth, and nails were compared to specimens housed at the Florida Museum of Natural History (Thornton et al. 2004). Statistical analyses We defined diet diversity as the number of different food items per fecal sample. Seasonal variation was categorized into “ wet ” and “ dry. ” The percentage of coyote fecal samples containing each item was determined. A Poisson regression with the total number of different food items/sample as the response variable was utilized to determine the relationship between diet diversity and habitat, season, and H*S interaction. In order to determine if presence or absence of anthropogenic waste in coyote diets was related to habitat (H), season (S), or an H*S interaction, we used a logistic regression. In this analysis, habitat and season were indicator variables. H had values 0 (wildland) and 1 (suburban), as did S (0 [dry] and 1 [wet]). Results The occurrence of food items varied by habitat type (Fig. 2). The Poisson regression indicated that coyotes in the wildland habitat had greater diet diversity than suburban coyotes ( P < 0.001), with average numbers of different diet items per sample of 3.18 and 1.69, respectively. We assume that Y was distributed as Poisson( 1 ) where ln( 1 )=)= β 0 + β 1 H+ β 2 S+ β 3 HS. We investigated this model and all its submodels, and applied the well known model selection criteria such as AIC and BIC [see Dobson and Barnett (2008)] in order to select the most parsimonious model; both methods of model selection yielded the same model in which the number of diet items was habitat dependent. 658 Urban Ecosyst (2011) 14:655 – 663 Effect of season and the interaction between habitat and season were not statistically significant ( P =0.5178). The best fitted model had ln ð b 1 Þ ¼ b " 0 þ b " 1 H with b " 0 ¼ 1 : 1580 , b " 1 ¼ : 6332 , SE ð b " 1 Þ ¼ 0 : 1214 . Therefore, the Z-statistic for testing H 0 : β 1 =0 against H 1 : β 1 ≠ 0 has Z ¼ b " 1 = SE ð b " 1 Þ ¼ 5 : 2150 , and the P -value was approximately equal to 1.8×10 − 7 Therefore, habitat as a factor was highly significant. This model indicated the mean number of diet items was approximately 1.9 times higher in the wildland than the suburban habitat. In the wildland habitat, 8% of the samples contained anthropogenic waste compared to 18% of samples from suburban areas. The logistic regression model took the form: logit p ð Þ ¼ log p = 1 p ð Þ ð Þ ¼ " 0 þ " 1 H þ " 2 S þ " 3 HS , where p was the probability that anthropogenic waste is present. We tested this model and all its submodels using AIC and BIC model selection criteria. BIC criterion led to the null model. In other words, the BIC criterion selected the model in which the presence of anthropogenic waste did not depend on habitat or season. Based on the AIC criterion, the best model was logit ð b p Þ ¼ b " 0 þ b " 1 H , where b " 0 ¼ 2 : 4204 , b " 1 ¼ 0 : 9249 , SE ð b " 1 Þ ¼ 0 : 6053 . For this model the Z-statistic for testing H 0 : β 1 =0 against H 1 : β 1 ≠ 0 is Z ¼ b " 1 = SE ð b " 1 Þ ¼ 1 : 53 , p =0.13. The AIC criterion resulted in a model in which the probability of presence of anthropogenic waste depended only on habitat (AIC = 99.308), but this dependence was rather weak. Therefore, it is reasonable to conclude that the data did not provide strong evidence that the presence/absence of anthropogenic waste depended on either habitat or season (AIC = 101.950). In the wildland habitat, Insecta and berries were found most frequently in the samples collected during the wet season whereas Lagomorpha , Rodentia , and Cervidae appeared more frequently during the dry season (Fig. 3a). In the suburban habitat, Didelphidae , Procyonidae , Felidae , Insecta , Aves , and Testudines were only recovered during the wet season (Fig. 3b). During the wet season, 68% of the fecal samples collected from the suburban habitats contained berries, but this number declined to 22% during the dry season (Table 1). Conversely, during the wet season only 21% of the samples collected from the suburban habitat contained Sylvilagus spp., but during the dry season this number rose to 67% (Table 1). Fig. 2 Coyote diet in wildland/protected and suburban habitats, Pinellas County, Florida (May 2005 – March 2007) Urban Ecosyst (2011) 14:655 – 663 659 Discussion Coyote diet in the wildland habitat was more diverse than that in the suburban area. The number of possible food items in a nature preserve such as BCP, though not measured in the present study, is likely greater than in an suburban area. Of the diet items consumed in both habitat types, Sylvilagus spp. varied the least between suburban (32%) and BCP (29%), whereas vegetative matter varied the most (96% protected, 25% suburban). Eight percent of fecal samples collected within BCP contained anthropogenic waste. While the origin of this cannot be determined, we assume that at least some of it was obtained from surrounding residential areas. Although not statistically significant, anthropogenic waste was recovered over twice as often from coyote fecal samples collected in the suburban habitat (18%). Fig. 3 a Seasonal variation in coyote diet in the wildland habitat. Pinellas County, Florida (May 2005 – March 2007). b Seasonal variation in coyote diet in the suburban habitat. Pinellas County, Florida (May 2005 – March 2007) 660 Urban Ecosyst (2011) 14:655 – 663 Anthropogenic waste was also found in only half as many samples collected during the wet season (15%) as opposed to the dry season (28%) from the suburban habitat. This could possibly be related to food scarcity. The results of the present study for seasonal variation of anthropogenic waste in suburban coyote fecal samples are very similar to those of Fedriani et al. (2001) where during the wet season only 14% of coyote fecal samples contained anthropogenic waste, but during the dry season this number rose to 25% in an adjacent suburban area. Combining data from both habitats indicates that most fecal samples contained vegetative matter (54%) and berries (46%). While Stratman and Pelton (1997) found that fruit accounted for 80% of coyote diet in northwestern Florida, our study documented berries accounting for only 46% of coyote diet. Rodentia and Insecta were recovered from 29% and 24% of the feces, respectively. Of the Rodentia consumed, Sigmodon hispidus was recovered most often (15%). Only 7% of fecal samples contained Sherman ’ s fox squirrel ( S. niger shermanii ), a Species of Special Concern (FWC 2004). Felidae was recovered from 2% of combined samples — 1% from each study area. These data reveal less consumption of Felidae than other studies, such as 13% from coyotes in Kentucky (Crossett and Elliott 1991). The accumulation of anthropogenic waste has resulted in the proliferation of red foxes ( Vulpes vulpes ), Ring-billed ( Larus delawarensis ) and California gulls ( Larus californicus ) Norway rats ( Rattus norvegicus ), and raccoons ( Procyon lotor ) worldwide (Wilson 2005). Table 1 Diet items consumed by coyotes in wildland (BCP) and suburban habitats, Pinellas County, Florida (May 2005 – March 2007) Diet item Wet Season Dry season BCP Urban BCP Urban n (%) n (%) n (%) n (%) Didephidaie 0 (0) 2 (4) 0 (0) 0 (0) Didelphia virginiana 0 (0) 2 (4) 0 (0) 0 (0) Lagomorpha 6 (22) 11 (21) 8 (36) 12 (67) Sylvilagus spp. 6 (22) 11 (21) 8 (36) 12 (67) Rodentia 11 (41) 11 (21) 11 (50) 2 (11) Sciurus carolinensis 1 (4) 5 (9) 0 (0) 2 (11) Sciurus niger shermanii 2 (7) 4 (8) 2 (9) 0 (0) Sigmodon hispidus 8 (30) 2 (4) 8 (36) 0 (0) Gemoys pinetis 0 (0) 0 (0) 1 (5) 0 (0) Cervidae 8 (30) 0 (0) 8 (36) 0 (0) Odocoileus virginianus (fawn) 8 (30) 0 (0) 0 (0) 0 (0) Odocoileus virginianus (adult) 0 (0) 0 (0) 8 (36) 0 (0) Procyonidae 0 (0) 1 (2) 0 (0) 0 (0) Procyon lotor 0 (0) 1 (2) 0 (0) 0 (0) Felidae 1 (4) 1 (2) 0 (0) 0 (0) Insecta 17 (63) 3 (6) 9 (41) 0 (0) Vegetation 25 (93) 15 (28) 22 (100) 3 (17) Berries 12 (44) 36 (68) 3 (14) 4 (22) Aves 4 (15) 5 (9) 2 (9) 0 (0) Testudines 1 (4) 1 (2) 1 (5) 0 (0) Anthropogenic Waste 2 (7) 8 (15) 2 (9) 5 (28) Urban Ecosyst (2011) 14:655 – 663 661 The proliferation of these species has resulted in damage to infrastructure, disease spread, and increasing conflict with humans. With regard to coyotes, as opportunistic feeders, it would be prudent for wildlife officials to closely monitor the possible interaction between anthropogenic waste consumption and human – coyote conflict. This study suggests that as wild areas become reduced as the result of urbanization diet diversity for the coyote will likely decrease and consumption of anthropogenic items will likely increase. Consequently human – coyote conflicts might increase, as well. Trends like this will be particularly important in areas like Pinellas County where the rapid disappearance of natural habitats changing to urbanized lands is widespread. Acknowledgements We wish to thank Jay Jones, Catherine Hughes, and Michelle Dachsteiner for their assistance in the field and laboratory. We express appreciation to Dr. Bruce Rinker and the staff at Brooker Creek Preserve for providing access to the Preserve, use of an ATV, and for being an integral part of this study. We are grateful for the assistance of Pinellas County Animal Control, specifically Kenny Mitchell, Dr. Welch Agnew, and Richard Stahl. Lastly we would like to express our sincere thanks to Pinellas County for funding this research. References Azevedo FC, Lester V, Gorsuch W, Lariviere S, Wirsing AJ, Murray DL (2006) Dietary breadth and overlap among five sympatric prairie carnivores. J Zool 269:127 – 135 Barnett J, Dobshinsky A (2007) An alternative future: Florida in the 21st Century 2020 2040 2060. University of Pennsylvania: Philadelphia Chen E, Gerber JF (1990) Climate. In: Myers RL, Ewel JJ (eds) Ecosystems in Florida. University of Central Florida Press, Orlando, pp 11 – 34 Crossett RL, Elliott CL (1991) Winter food habits of red foxes and coyotes in central Kentucky. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 45:97 – 103 Dobson A, Barnett A (2008) An introduction to generalized linear models. Chapman & Hall: Boca Raton (FL) Fedriani JM, Fuller TK, Sauvajot RM (2001) Does availability of anthropogenic food enhance densities of omnivorous mammals? An example with coyotes in southern California. Ecography 24:325 – 331 Florida Fish and Wildlife Conservation Commission (FWC) (2004) Florida ’ s endangered species, threatened species, and species of special concern. http://myfwc.com/imperiledspecies/pdf/Endangered- ThreatenedSpecial-Concern-2004.pdf Gompper ME, Goodman RM, Kays RW, Ray JC, Fiorello CV, Wade SE (2003) A survey of the parasites of coyotes ( Canis latrans ) in New York based on fecal analysis. J Wildl Dis 39:712 – 717 Grinder M, Krausman PR (2001) Home range, habitat use, and nocturnal activity of coyotes in an urban environment. J Wildl Manage 65:887 – 898 Lingle S, Pellis SM, Wilson WF (2005) Interspecific variation in antipredator behaviour leads to differential vulnerability of mule deer and white-tailed deer fawns early in life. J Anim Ecol 74:1140 – 1149 Marzluff JM, Bowman R, Donnelly R (2001) A historical perspective on urban bird research: trends, terms, and approaches. In: Marzluff JM, Bowman R, Donnelly R (eds) Avian ecology and conservation in an urbanizing world. Kluwer, Norwell, pp 2 – 17 McCown W, Scheick B (2007) The coyote in Florida. Fish and Wildlife Research Institute. Florida Fish and Wildlife Conservation Commission (myfwc.com). Tallahassee, Florida Prugh LR (2005) Coyote prey selection and community stability during a decline in food supply. Oikos 110:253 – 264 Quinn T (1997) Coyote ( Canis latrans ) food habits in three urban habitat types of western Washington. Northwest Sci 71:1 – 5 Stratman MR, Pelton MR (1997) Food habits of coyotes in Northwestern Florida. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 51:269 – 275 Thornton DH, Sunquist ME, Main MB (2004) Ecological separation within newly sympatric populations of coyotes and bobcats in South-Central Florida. J Mammal 85:973 – 982 United States Census Bureau (USCB) (2009) http://quickfacts.census.gov/qfd/states 662 Urban Ecosyst (2011) 14:655 – 663 Wagner GD, Hill EP (1994) Evaluation of southeastern coyote diets during the wild turkey reproductive season. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 48:173 – 181 Wilkins L, Langworthy M, Rathbun CD, Sullivan R (1982) Identification of the dorsal guard hairs of some of Florida mammals. Technical Report prepared for the Florida Game and Fresh Water Fish Commission. University of Florida, Gainesville, Florida Wilson DE (2005) Mammal species of the world: a taxonomic and geographic reference. Johns Hopkins University Press, Baltimore Wooding JB, Hill EP, Sumner PW (1984) Coyote food habits in Mississippi and Alabama. Proc Annu Conf Southeast Assoc Fish Wildl Agencies 38:182 – 188 Urban Ecosyst (2011) 14:655 – 663 663