Variation in red fox <i>Vulpes vulpes</i> diet in five continents

A key defining feature of a species' niche and ecological roles is its diet (Pocheville 2015). What an animal species eats influences trophic position, how it moves around, which other interacts with, among many behaviours. While diets are flexible, predators have evolved to forage optimally on certain prey types or (Hayward et al. 2011, 2016). The mechanisms that determine optimal foraging behaviour include adaptative responses availability, food quality, energy required handle (Pyke 1977, Sundell 2003), predation risk from larger co-occurring (Haswell 2018). Ecological climatic conditions distributions and, in turn, composition, so the study intraspecific variation using large biogeographical datasets provides powerful means understand ecology Carnivora. For instance, regional- continental-scale studies dietary been conducted for feral cats Felis catus (Doherty 2015), wildcats silvestris (Lozano 2006), badgers Meles meles (Goszczyński 2000), polecats Mustela putorius (Lodé 1997), common genets Genetta genetta (Virgós 1999), otters Lutra lutra (Clavero martens Martes spp. (Zhou 2011), dingoes Canis dingo 2019). Knowledge about spatial differences feeding can contribute understanding strategies used by generalist exploit wide range resources optimally. Among medium-sized carnivores, red fox Vulpes vulpes (hereafter ‘fox’) prime example with adaptive allows alternative when abundance main decreases (Kjellander & Nordström 2003). In addition, foxes able survive environments, including highly modified urban agricultural areas where they anthropogenic foods (Harris 1981, Saunders 1993, Contesse 2004, Bateman Fleming 2012) domestic poultry pets (Lewis 1993). As availability varies habitat environmental factors, not surprising composition geographic location. regional-scale review Iberian Peninsula found invertebrates were most frequently reported followed fruit/seeds, small mammals, lagomorphs, carrion/garbage, birds, reptiles (Díaz-Ruiz 2013). Throughout Europe, rodents principal foxes, plants, invertebrates, reptiles, amphibians (Soe 2017). Australia, has introduced, mainly comprises livestock, vegetation (Fleming 2021). Notwithstanding these diet, quantitative describing worldwide patterns undertaken. Understanding at global scale be predict this widespread will respond adapt future land use climate change. ↑ Small fruit ↓ Medium-sized Diet richness mammals Invertebrates, Invertebrates Fruit Birds, Large birds We carried out literature select focused published before during 2018. (“Vulpes vulpes” OR fox) AND (diet ecology) as keywords ISI Web Science, JSTOR, Google Scholar. each study, we downloaded title, abstract, authors, year, journal name. Additionally, examined reference lists all articles identified our initial dataset ensure was missed. selected reporting frequency occurrence (FO, i.e. number individual samples item present percentage total samples) items consumed scats, stomachs, both scats stomachs (Appendices S1 S2). FO comparison metrics consistently widely. To comparability metrics, excluded 40 only relation contents, e.g. relative weight/volume, FO, volume, around dens, hair sampled. included ≥16 collected data either single year over years, well location geographical region (e.g. county district). limit pseudoreplication, several locations, seasons sampled same pooled category across sites <80 km apart, years seasons. chose site distance based home size (median = 3.25 km2; Main 2019), distances travelled (11 km, Coman 1991; up 8 Tsukada maximum dispersal (>80 km; Trewhella 1988, Newsome 2017), reduce probability could travel between sites. This resulted 217 analyses (Fig. 1). consistent set 13 categories report diet: 1) (adult weight <500 g), 2) (500–6999 3) (≥7000 4) unidentified 5) 6) 7) 8) amphibians, 9) fish, 10) fruit, 11) vegetation, 12) garbage (i.e. human-related materials discarded food), 13) food. These chosen because widely parts world (Abe 1975, Catling Jankowiak 2008, Drygala 2014). If authors multiple mammal sizes one category, classed ‘unidentified mammals’. those instances, recorded ‘not applicable’ against categories. Also, if primary source encompassed more than ‘fruit vegetation’ ‘amphibians reptiles’), considered them food’. Values <1% <0.01% 0.5% 0.005%, respectively. value comment provided coded dataset. When mentioned text accounted but zero Where sources differed adopted, combinatorial probabilities (for details, see Murphy al 2019) pool individuals occurrences sample size). created 20 circular buffer estimate mean annual temperature precipitation, elevation, HFI. Mean temperature, elevation 5 m resolution sourced WorldClim (www.wordclim.org). quantified influence HFI layer version 2, 1995–2004 (Wildlife Conservation Society – WCS 2005); database represents normalised biome realm. Global estimated population density, human use, infrastructure built-up areas, night-time lights), access coastlines, roads, railroads). Given publication dates papers (1935–2018), fitted model full (n studies) another 1995 2004 59 match temporal (Appendix S3). similar models; thus, subsequent S4). Predictor variables correlated (r ≤ 0.62). confirm choice appropriate, tested degree correlation three different radius values 5, 10, km) coefficients very high (temperature: r > 0.99; precipitation: elevation: 0.96; HFI: 0.90). indicates inferences unlikely influenced greatly distance. assessed (excluding food) within continents analysis similarity (ANOSIM) ‘vegan’ package (version 2.5-7; Oksanen 2020) R 4.1.2; Core Team Africa two met criteria. ANOSIM measure dissimilarity (R) continents. Dissimilarity −1 low dissimilarity) +1 groups). Because requires complete dataset, absences genuine absences. Euclidean ordinate dimensions 300 random starts. performed Monte Carlo randomisation significance final stress test hypothesis no difference groups, 999 permutations data, pairwise ANOSIMs. modelled relationship dependent richness, diversity, equitability) separate predictor absolute latitude, HFI, sampling method (stomach scat ratio: 0 multicollinearity variance inflation factors calculated ‘car’ 3.0.11; Fox Weisberg There collinearity latitude (variance factor >5); did models. generalised linear models predictors plus That is, response variable. ‘lme4’ 1.1.27.1; Bates 2015) fit All mean-standardised added model. quartile–quartile plot function ‘DHARMa’ (Hartig indicated overdispersion residuals categories; Tweedie fitted. alpha maximise normality ‘Tweedie’ (Dunn Smyth 2005, Dunn combinations dredge ‘MuMIn’ (Barton 2020), then weighted according Akaike Information Criterion corrected (AICc; Burnham Anderson 2002) Tweedie-AIC (t-AIC; averaged estimates units best carry predictions. 95% confidence intervals significant covariates (P < 0.05). At scale, commonly (mean FO: 45 ± 4%) (FO: 41 4%; Fig. Appendix S5). significantly (R 0.090; P 0.001), there degrees overlap comparisons (Table 2). Key Africa) lower Australia 23 4%), Asia 4 7%), higher (Europe: 36 5%; North America: 15%), 37 6%; 39 18%), 18 10%) Europe America, 10 3%; Three significant, having highest 0.092; 0.001; S6) America lowest 0.386; Table S6). model, decreased increasing precipitation (t −2.174, 0.031; 3, S7). −3.302, 0.001) −3.450, Appendices S8 S9). models, 8.662, 3.720, 7.302, increased 4a–c, 2.545, 0.011) 3.308, 4e,f), while −2.300, 0.022; 4d). −3.059, 0.002; S10). bird −2.920, 0.004) invertebrate −3.113, 0.002) 4g,h, incidence −2.805, 0.005) 4i), 2.259, 0.025) 4.727, 4j,k, results −8.067, −2.531, 0.012), −6.090, some method. −2.824, 0.005), −2.219, 0.027), −2.867, likely analysed rather stomach contents 4l,m,o; By contrast, 2.326, 0.021) 4n, Similar Based collation comparable fox, predictions drivers. varied geographic, climatic, richness. reflect influences. Moreover, type (stomach-to-scat ratio) samples. Globally, dominated similarities Europe). Our accord previous smaller continental regional scales, showing terms 2013), Consumption may related their demonstrated local scales (Pavey Cupples Spencer 2014), preferences exhibited (Randa 2009, Less abundant preferred less naïve predation, making easier capture (Graham adaptations successfully (optimal foraging). especially true introduced represent direct (Salo 2007) indirect (Molsher 2017) threat (Woinarski 2015, Radford One recognised phenomena decline diversity (Hillebrand 2004). Many variables, also vary drive distribution patterns, therefore available carnivores such fox. however, change latitude. result accords Díaz-Ruiz (2013) who Peninsula. throughout shown decrease cold warm periods Relationships American americana Holarctic 2011). support predictions, due cooler temperatures away equator. Indeed, biological processes endotherms somewhat independent ambient ectotherms invertebrates) directly linked Caldwell 2017, Brandt 2018, Jara possibly equator (Roll Thus, endothermic (birds mammals) ectothermic reptiles) latitudes. latitudinal (mainly rodents) Peninsula, northern southern latitudes g) suggests mostly carrion. However, part densities (>250 head km−2) (livestock) (Robinson Like strong effects community (Heaney 2001, McCain 2009), turn what eat. Eurasian characterised greater proportion amphibians) elevations (Remonti 2009). level elevation. altitudinal trend consumption described Sumava Mountains Czech Republic (Hartová-Nentvichová 2010). (McCain highlight importance considering carnivore scales. Mediterranean non-Mediterranean climates bioclimatic regions 2019, Temperature (sometimes even counteracting other) depending prey. temperature. finding contrary that, under scenarios, shift changing particularly given behavioural adaptability species. shifts Endangered eastern quoll Dasyurus viverrinus (Fancourt 2018) polar bear Ursus maritimus Canada (Gormezano Rockwell Broader knowledge precipitation) hence enhance predator–prey interactions. activity vulnerability predation. Consequently, thermoregulatory strategy important determinant diets. example, (and period frogs relationships suggest contrasts (total taxa) increases 2021), lagomorphs continent-specific specific categories, warrants further investigation studies. precipitation. explained reduction families (Passeridae Columbidae; Serafini Lovari Balestrieri needed identify precise drivers Foxes benefit living (Stepkovitch level, shows negative density (Main probably subsidies (Bateman 2012, Dawson Nevertheless, prediction overall Gámez (2020), vertebrate unrelated European although replicated assess increased. increase presence wild (Underwood Kilheffer 2016) livestock) cities areas. consume juveniles (Saunders Gentle 2006) adults vulnerable giving birth), taken carrion (Catling cities, features (Aronson consequence closer (Clergeau 1998). ground-foraging ground-roosting road strike waterbirds towns. horticultural, ornamental, pest plant Africa, (Doncaster 1990, Dell'Arte Leonardi Rosalino Santos-Reis effective seed agents natural habitats (Koike landscapes, fleshy figs, grapes, melons, apples, olives, cherries; Lowe 1989, makes potential disperser role habitats, urbanised ones ornamental (McKinney invasive Digestibility considerably, analyses. prey, body digestibility (Ferreras Fernandez-de-Simon Therefore, detected derived opposite mammals. had contents. findings It consuming eat solid bones hairs, indigestible) stomachs. fact failed consider possible (Brunner Kirkwood 2000, 17% 10% fruits. result, (2021) material (including fruits parts) Australian often analysis, digestible pulp). possible, should distinct parts, order improve evaluation any (seasonal) cultivated fruiting plants). T.S.D. supported Discovery Early Career Research Award Council (DE200100157). S1. Data extraction procedure. S2. List review. S3. Effects variable (Human Footprint Index) (%) S4. S5. (CI) globally five S6. Boxplots among-group within-group dissimilarities diet. S7. Parameter standard errors describe S8. error S9. Effect S10. anthropogenic, type-of-sample Please note: publisher responsible content functionality supporting information supplied authors. Any queries (other missing content) directed corresponding author article.