Factors Determining Coyote ( Canis latrans ) Diets

Although studies have documented the potential for coyote (Canis latrans) food use to negatively affect wildlife populations and domesticated animals, they are often equivocal, possibly because most are of small spatial extent, and little is known of factors determining coyote diets. Our objectives were to quantify the diet and identify factors determining coyote food use, particularly game species and livestock, over a large spatial and temporal extent.

expanded its range throughout most of North America [1].Increasing numbers of coyotes in areas such as the southeastern United States [2] [3] are a concern as they may feed on a variety of food items, including important game species, livestock, and pets [4] [5].Coyotes have been implicated in the decline of numerous wildlife species, negatively affecting them through competition for resources (e.g., bobcats [Lynx rufus]; [6]), predation (e.g., white-tailed deer [Odocoileus virginianus; hereafter, deer]; [7]), and changes in community structure [8] [9] [10].
Although studies have documented the potential for coyote food use to negatively affect wildlife and domesticated animals, they are often equivocal, possibly because most are of small spatial extent, which can mask broader habitat and ecosystem effects [11].In addition, beyond a limited knowledge of the influence of season and animal sex and age, we lack an understanding of determinants of coyote food use [11].A better understanding of these determinants is needed to tailor management strategies and mitigate the negative effects of coyotes.
Our objectives were to quantify the diet and identify factors determining coyote food use, particularly game species and livestock, over a large spatial and temporal extent.We expected coyotes would be opportunistic and have a diverse diet, where one or few items did not dominate; exhibit age, sex, and size differences in diet, to reduce intraspecific competition for food, with older, more experienced animals consuming larger and more difficult to prey upon items, and coyote size positively related to prey size consumed; and have diets that vary with season and location, to take advantage of available food assemblages, such as deer and livestock calves.Finally, we wished to investigate if some combination of these factors might further affect coyote food use.

Materials and Methods
We opportunistically obtained coyote carcasses and gastrointestinal (GI) tracts from hunters, trappers, and as road-kills throughout Florida from December 2011 to February 2015.Road-kills were typically less than 2 days old at the time of collection.Coyote carcasses were sexed and weighed, and a canine from the lower jaw was removed, aged via cementum annuli, and used to assign coyote age class (Matson's Laboratory LLC, Milltown, MT; [20]; Table 1).We determined body mass classes arbitrarily (i.e., dividing the range of values into three equal parts), and following Giuliano et al. (1989), an ocular kidney fat index was used to assess coyote condition (Table 1).Collection date, method, and location (Table 1) were documented for all animals.To examine the effects of season on coyote food use, we partitioned collection dates into three relevant season  1).We removed gastrointestinal tracts from carcasses, rinsed contents with warm water, and separated materials using a 4-sieve kit (Hubbard #3076 Screen Four Sieve Kit).After air-drying for 24 hours, we sorted samples into components (e.g., hair, bone, plant material, insects, etc.).Items were identified micro-and macroscopically using attributes such as hair length, color, and scale patterns by comparison to reference collections [21] [24].We separated deer fawns and adults based on hair characteristics following Wilkins et al. (1982).Dietary items were grouped for analyses, with some items appearing in multiple groups.Because we were particularly interested in factors determining coyote use of game species and livestock, we separately examined these groups (Table 1).We counted any type of item found within a coyote GI tract only once, regardless of how many of the individual food item were in the GI tract.All dietary items (e.g., Virginia opossum [Didelphis virginiana]) and food groupings (e.g., small mammals) were expressed as percent (%) occurrence (i.e., number of coyotes with the food item or group/total number coyotes).
To understand determinants of coyote dietary habits, we employed logistic regression in an information-theoretic framework [25].For each food group (i.e., consumed or not consumed), we developed and evaluated the same set of 22 a priori, single-and multiple-variable candidate models (Table 1) based on the literature, prior knowledge and field experience, and study objectives.Individual models were limited to 3 predictor variables to reduce the likelihood of overfitting.We examined Akaike's Information Criterion with small-sample correction (AIC c ) values, AIC c differences (ΔAIC c ), Akaike weights (w i ), and model goodness of fit (−2 log-likelihood; when P ≤ 0.10, models were considered to fit) for models with different combinations of predictor variables, and considered models with ΔAIC c < 2 supported.Where multiple models were supported, we used model averaging to increase precision of inference and examine the relative contribution of each variable from all supported models [25].When 85% confidence intervals (CI) for variables within supported models overlapped with zero, we considered them to have a weak effect on the dependent variable and be uninformative [26].For brevity and clarity, we only present results of supported models.All analyses were conducted using R software [27].
All research and animal welfare protocols were reviewed and approved: University of Florida Animal Research Permit (003-11WEC) and Florida Fish and Wildlife Conservation Commission Research Permit (SPGS-11-68).

Results
Coyotes (n = 263) had a diverse diet, consuming vegetation, insects, mammals (≥25 species), birds, amphibians, and reptiles, with 11 foods commonly consumed (occurring in >10% of coyotes).Frequently used food items included Virginia opossum, feral hog (Sus scrofa), northern raccoon (Procyon lotor), rabbits (Sylvilagus spp.), skunks (striped skunk [Mephitis mephitis] and spotted skunk [Spilogale putorius]), deer (adults and fawns), insects, mast, and other vegetation (Table 2).Young adult coyotes were less likely than adults to eat medium-sized mammals and small game, and juveniles were less likely than adults to eat birds.Age was not a factor determining use of any other food groups.Compared to females, male coyotes were less likely to consume small mammals and more likely to consume medium-sized mammals and small game.Use of any other food groups was not determined by sex.Small coyotes were less likely than larger animals to use large game, and mass was not a factor determining use of any other food groups.While animal condition appeared in 2 supported models predicting bird use by coyotes, it was an uninformative individual predictor variable, and condition was not a factor determining use of any other food groups (Table 3 and Table 4).
Coyotes were less likely to ingest mast during the spring and winter than fall, and calendar season was not a factor determining use of any other food groups.Coyotes were more likely to eat small mammals, large mammals, deer, and livestock and less likely to consume reptiles and amphibians during the deer hunting season than during the deer fawning season.Outside of the deer hunting and fawning seasons, coyotes consumed more small mammals and mast and less deer than during the fawning season.While deer season appeared in 2 supported models predicting vegetation use by coyotes, it was an uninformative individual predictor variable, and use of any other food groups was not determined by deer season.Consumption of insects was more likely and small mammals, large mammals, large game, and livestock by coyotes was less likely outside of the livestock calving season than during the calving season.While livestock season appeared in 1 supported model predicting bird use and 1 supported model predicting mast ingestion, it was an uninformative individual predictor variable.Livestock season was not a factor determining use of any other food groups (Table 3 and Table 4).
Trapped coyotes were more likely to consume large mammals, birds, vegetation, and livestock and less likely to eat reptiles and amphibians, insects, and mast than hunted coyotes.While collection method appeared in 1 supported model predicting large game use by coyotes, it was an uninformative individual predictor variable, and coyote collection method was not a factor determining use of any other food groups.Coyotes in northern and southern Florida were more likely to eat birds than coyotes in central Florida.In northern Florida, consumption of reptiles and amphibians and mast were more likely and livestock less likely than in central Florida.While collection location appeared in 2 supported models predicting small mammal use, 1 supported model of large mammal use, and 1 supported model of deer use, it was an uninformative individual predictor variable, and use of any other food groups was not determined by location of coyote collection (Table 3 and Table 4).

Discussion
Coyotes had a diverse diet, consuming vegetation, insects, mammals (≥25 species), birds, amphibians, and reptiles.Similar to other studies (e.g., [13] [16] [18] [33]).These differences may be due to differences in food availability among regions, limited spatial and temporal extent of the contrasting studies, and examination of scats to determine diets in most studies.The latter will lead to differing diet composition as some items are digested fully and do not appear in scats as opposed to GI tracts, and may reflect the preference of a single or a few animals, rather than the habits of the entire population.In addition, our study is not subject to predator misidentification, a substantial problem with most scat-based studies, as they do not genotype scats to species [34].Furthermore, we established the importance of coyote age, sex, and body mass, season (i.e., calendar, livestock calving, and deer fawning and hunting), coyote collection method (i.e., hunted, trapped, or road-killed), and location (i.e., region of Florida) as determinants of coyote food use, which may also explain differences relative to other studies that typically could not examine these factors.Physical characteristics of coyotes (i.e., age, sex, and body mass) were important factors determining food use.The lower consumption by young adults of medium-sized mammals and small game and juveniles of birds compared with adults, the reduced use by males of small mammals and greater use of mediumsized mammals and small game compared with females, and the lower consumption by small coyotes of large game compared to larger (i.e., medium and large) animals suggest that coyotes may be partitioning resources to reduce intraspecific competition [17] [35].Additionally, foraging ability may differ depending on the age, sex, and size of coyotes, which may lead to differential effects on important prey (e.g., rabbits, deer, etc.) depending on coyote population structure.The lower consumption of birds and medium-sized mammals and small game such as armadillos, Virginia opossums, rabbits, squirrels, skunks, weasels, and northern raccoons, may reflect the inexperience of younger animals at finding and capturing prey [36].Further, the greater use of large game species (i.e., deer and hogs) by larger than smaller coyotes likely reflects smaller individuals not being able to physically handle such large prey and exclusion by dominant (i.e., larger) individuals of deer and hog kills or carcasses.In contrast with other studies, we did not find that coyote age and sex were important in determining the use of other food groups.For example, Metzger et al. (2017) noted an overall difference in diet between males and females, and Albers (2012) noted juveniles using less mast and more deer than adults.
Seasonal variations (i.e., calendar, deer, and livestock) were the more important type of factors determining coyote food use.Coyotes were less likely to consume mast during spring and winter than fall, which likely reflects its greater availability during the fall masting season [37] [38].Although not noted in this study, Whitaker et al. (2015) suggested greater use of deer in fall and winter and hogs in spring and summer than other seasons, but statistical tests were not performed.
Similar to calendar season effects, availability likely increased use of large mammals, deer, and livestock and decreased use of reptiles and amphibians by coyotes during the deer hunting season compared to the fawning season, and led to greater mast and reduced deer use outside of both the deer hunting and fawning seasons.Deer hunting season was generally during fall-early winter and provided less than ideal conditions for most reptiles and amphibians to be active (i.e., dryer and cooler; [38]).However, a large portion of the deer season outside of both hunting and fawning seasons included parts of spring and fall masting periods [37] [38].The increased use of large mammals (comprised primarily of deer, hog, and cow), deer, and livestock in the deer hunting compared with deer fawning season may often reflect coyote scavenging habits, availability of livestock calves, hunting of deer and hogs, and landowner control programs of coyotes and hogs.During the deer hunting season, which overlaps with much of the livestock calving season, agricultural land managers kill many feral hogs and leave their carcasses in pastures while monitoring calving operations and use hog as bait while conducting coyote control programs leading up to and during the livestock calving season (J.R. Selph, Florida Cooperative Extension Service, University of Florida, Arcadia, FL, personal communication, 2015).Further, during much of this period, livestock are calving and deer are physically stressed with the conclusion of the breeding season, and hunters are harvesting (and wounding) deer and hogs during the hunting season [23].Coyotes may be taking advantage of these easily obtained food sources by scavenging the remains of hogs used as bait, hogs and deer left in the field, and opportunistically depredating livestock calves and weak or wounded deer and hogs [33].The greater use of deer by coyotes during the hunting season (26 % occurrence) than other seasons supports this conclusion [11].
Our findings regarding deer use are similar to those of Swingen et al. (2015), who found the greatest deer use in winter, but in contrast to Hidalgo-Milhart et al. (2001), Schrecengost et al. (2008), and Wooding et al. (1984), who noted greater deer use by coyotes during the deer fawning season.Although we documented adult and juvenile deer in coyote GI tracts, other items (e.g., mediumsized mammals, feral hogs, insects, and vegetation) were found more often [14] [39] [40].The relative amount of deer consumed by coyotes in Florida was less than in other studies (e.g., [13] [14] [15] [18] [28] [32]).Florida is a large state with a diversity of ecosystems [41] that offer a variety of food from which coyotes can choose, possibly explaining differences with other studies.Additionally, in more northerly portions of coyote range, deer are easily preyed on in deep snow [42] [43] [44] [45] and other foods important to Florida coyotes (e.g., hogs and Virginia opossums) may be less abundant [46], partially explaining why deer are relatively less important to coyotes in Florida.Huebschman et al. [50]).Our data partly support this conclusion, finding deer use to be greater in the deer fawning season compared with outside of both the deer hunting and fawning seasons, as fawns may be easy prey for coyotes [7].However, deer fawns were generally, infrequently consumed in Florida, which is similar to that of

Swingen et al. (2015).
Coyotes may be using deer and other food items more evenly through time as the fawning seasons in Florida can occur during 3 calendar seasons (e.g., spring, summer, and into fall; [51]), and food availability during each season may be dominated by other food items (e.g., mast in fall).Additionally, fawn remains may be digested at different rates than other prey items [52] [53], leading to a loss of evidence, and may be a reason why deer generally do not appear to be important in coyote diets.Gier (1968), and to some extent Albers (2012) and Litvaitis and Shaw (1980), suggested that livestock were more important to coyote diets, particularly in winter when calves are most vulnerable and heifers are physically stressed and most likely to abort or abandon a calf, providing inexperienced animals with an easy meal [54].The deer hunting season, when livestock use was high, includes much of the early winter period.Similar to deer hunting season, and to some extent deer fawning season, during the livestock calving season, coyotes consumed more large mammals, large game, and livestock than during the non-calving season, likely for the reasons described above for deer seasons.
Method and location of coyote collection also affected what coyotes consumed.Land managers consistently kill hogs, use their remains as bait to trap coyotes, and place traps and bait for coyotes around livestock carcasses to increase capture success.In addition, remains of harvested birds and poultry are often used as bait for coyotes.A portion of coyote consumption of hogs, livestock, and birds may be due to baited traps, as seen by the greater use of large mammals (including hogs), livestock, and birds by trapped than hunted coyotes.Further, the greater ingestion of vegetation by trapped than hunted coyotes often reflects trapped animals biting at anything within reach, while in a trap.
When compared to other regions of the Southeast, Florida may contain a greater variety of ecosystems and climatic conditions, explaining some of the observed differences in coyote food use [41].North Florida exhibits more seasonality and is more forested compared with the remainder of the state [41] [55], contributing to increased mast production, and explaining the greater use of mast by coyotes in this region compared to the central region.In addition, the majority of the state's large-scale agricultural production, including livestock, occurs outside of north Florida [56], explaining why coyotes use livestock less in the north region.A superficial comparison of coyote food habits among the many studies with local extents suggests regional differences in coyote diets, and Metzger et al. (2017) noted regional differences in coyote diets in their larger scale Pennsylvania study.
As hypothesized, coyotes were opportunistic and omnivorous foragers with a diverse diet of vegetation, insects, birds, reptiles, amphibians, and more than 25 species of mammals (including important game species and livestock), with 11 food items commonly consumed (Virginia opossum, non-mast vegetation, feral hog, northern raccoons, insects, rabbits, skunks, deer, mast, birds, and cows).Food use was determined by coyote age, sex, and body mass, season of the year, deer hunting and fawning seasons, livestock calving season, collection method, and location/region, and suggests that coyotes are opportunistic, generalist predators that forage on the most available foods.
White-tailed deer is often the most utilized food of endangered red wolves (Canis rufus), with feral hogs also consumed but to a much lesser extent.Where red wolves and coyotes coexist, they often have very similar diets.However, red wolves rely more heavily on both deer and hogs [35].To increase the availability L. N. Watine, W. M. Giuliano DOI: 10.4236/oje.2017.713045663 Open Journal of Ecology of these important food sources for red wolves, targeting for removal coyotes during the deer-hunting season may be a viable strategy.For deer managers and livestock producers, targeting for removal coyotes during the deer hunting and livestock calving seasons may relieve pressure on deer and calves, particularly in areas with little other seasonal coyote foods.As coyotes expand their range and numbers, conservationists may find it useful to understand how this opportunistic and adaptable predator uses available food sources to reduce conflict across the landscape.

Table 3 .
Supported models a of factors determining coyote (Canis latrans) food use in Florida, USA, 2011-2015.
bFood groups defined in Table1.cModelpredictor variables described in Table1.d Number of model parameters.e Akaike weight.f When P ≤ 0.10, models were considered to fit.

Table 4 .
Model-averaged variable coefficients from supported models a of factors determining coyote (Canis latrans) food use in Florida, USA, 2011-2015.