Pumas <i>Puma concolor</i> as ecological brokers: a review of their biotic relationships

Large, terrestrial carnivores are key regulators exerting strong and irreplaceable effects on biological community assemblages, their absence can lead to fundamentally downgraded ecosystems (Estes et al. 2011, Enquist 2020). Due relatively low reproductive rates high energetic requirements, large predators also among the most imperilled species Earth highly sensitive human-induced environmental change (Ripple 2014). In fact, in human-dominated systems, tolerance for imposes social carrying capacities generally lower than (Bruskotter & Wilson 2014, Knopff 2016). Predators play outsized roles they inhabit, primarily via prey regulation limitation (Sinclair 2003). structure communities through a variety of indirect effects. For example, fear alter behaviour induce physiological costs that affect fecundity fitness (Sheriff 2009). Predator-induced changes smaller carnivore abundance precipitate ‘trophic cascades’ affecting faunal structure, floral communities, other trophic levels (Preisser 2007, Prugh Sivy Additionally, energy flow by leaving carcasses scavengers (Allen 2015, Elbroch 2017c, Sebastián-González 2020), which affects soil chemistry, nutrient cycling (Bump 2009, Wolkovich Barry 2019, Pumas Puma concolor, known as mountain lions, cougars, Florida panthers, largest iconic Americas. They exhibit widest geographical range any native mammal Western Hemisphere, inhabiting land between Canadian Yukon southern Andes (Iriarte 1990). The International Union Conservation Nature (IUCN) classifies pumas ‘Least Concern’, but indicates is declining Latin America (Nielsen 2017). Nevertheless, Dickman (2015) ranked felids highest conservation priority based upon intrinsic characteristics, extrinsic ecological variables, IUCN category, potential act umbrella species. Like grey wolves Canis lupus, African lions Panthera leo, apex carnivores, puma populations may be limited human (Treves Bruskotter legally illegally hunted because compete with humans ungulate resources (Elbroch 2017a), pose both real perceived risks people, pets, livestock (Herrmann 2013, Wolfe Guerisoli 2021). Eradication campaigns 19th 20th Centuries North reduced west central plains, except small, isolated population (Puma concolor coryi; Iriarte 1990; Fig. 1). Throughout puma’s range, main threats persistence unregulated hunting poisoning, retaliatory killing, habitat loss, fragmentation, vehicular collisions, disease (Miotto Schwab Zandbergen Vickers Pauli 2018, Kechejian van de Kerk 2019). While recolonising parts historical midwestern (LaRue 2012), others, facing increasing pressure from growth (Benson Several suffer genetic diversity (Castilho 2012, 2019), severe health consequences long observed Endangered panther coryi more recently California (Beier 2003, Onorato 2010, Sahagun We aimed synthesise published literature interactions associated advance our understanding biotic identify gaps need addressed enable evidence-based, strategic management. More specifically, we hypothesised strongly interactive (as defined Brodie 2018), via: 1) or regulation, create direct cascading (Schmitz 2000, Preisser 2007); 2) behaviours fitness, impact (Laundré 2010); 3) carrion production impacts food webs (Wilson 2011); 4) carnivores; 5) ecosystem services, benefits essential services (Millennium Ecosystem Assessment 2005). Ultimately, this information support managers practitioners communicating overall sometimes sceptical public, well aid designing wildlife management promotes healthy, integrated human–wildlife (Brodie 2018). conducted searches using Web Science Core Collection database Google Scholar empirical papers presenting new data 1950 2020. largely methodology Guidelines Standards Evidence Synthesis Environmental Management (Pullin Our search excluded book chapters, conference proceedings, reviews, meta-analyses. Specifically, exclude used had already been published, thus creating duplicates. Criteria inclusion required focus species' interactions, broad functioning (including cycling). Research solely focused size demographics, selection, was considered outside scope review. Initial were English, these identified few Spanish language papers. Since significant portion America, adapted retrieve well. keywords each utilised combinations names (English: puma, cougar, lion, panther; Spanish: león montaña, pantera Florida) either prey, predator, mesopredator, scavenger; presa, depredador, mesodepredador, carroñero) ecosystem, biodiversity, web, keystone, landscape fear, risk effect, cascade, cycling, disease; ecosistema, biodiversidad, regulación, red alimentaria, especies clave, depredador mayor, paisaje del miedo, riesgo depredación, cascada trófica, ciclaje nutrientes, enfermedad). Prior initiating searches, conduct brief scoping exercise validate could English-language studies (100% retrieval 11 previously papers). Upon finishing initial search, ‘snowball sampling’ drawing knowledge authors examining references relevant reviews (Livoreil On several occasions, included reports theses press subsequently journals after timeframe (published study, recorded country, state/province, geographic coordinates study site. When location not explicit, searched name area found Pro (Version 7.3, major question research, methods, sample size, findings interpreted authors, whether some measure effect strength listed within explicitly interacted affected (directly indirectly). placed paper into one following interaction categories: diet plants invertebrates), effects, services. criteria assess measured magnitude effects: reported evidence regulation; assessed impacting competitor abundance; all categories, controls, comparisons treatment types, before–after designs (Appendix S1). All statistical analyses performed R v4.0.3 software (R Team evaluated distribution representative different biomes range. rasterised Terrestrial Ecoregions World polygons provided Wildlife Fund (Olson 2001) ‘sp’ (Pebesma Bivand 2005) ‘raster’ (Hijmans Etten 2015) packages extracted biome type location. then compared site an exact multinomial goodness-of-fit test (Table binned results five-year periods 1970 2019 (excluding match design specific analysis) publication trends reflecting research interest. plotted number over time two those interaction, did not. tested there supporting positive linear exponential slope fitting models counts function Poisson error identity link function, respectively. correlation (R2) indicate fit. Finally, mixed logistic model estimate probability sizes control groups study. fit estimated fixed random intercept per independent variables. response variable consisted binary indicating Some multiple categories accounted interdependence. parameters Bayesian framework ‘rstanarm’ package (Goodrich default priors sampling 1000 iterations four Markov chains iteration burn-in period. differences 95% credible intervals (CI) around pairwise contrasts. reviewed 162 criteria, across 15 countries, concentrated western USA (n = 84), Canada 16), Argentina Mexico 14; 1, Appendix S1. majority 120) ‘diet regulation’, while 23 focussed ‘fear effects’ fear-based cascades, 14 ‘carrion effects’, 12 ‘effects carnivores’, five ‘ecosystem services’. Only category. Together, 485 taxa (Fig. 2). This count connected intermediate (e.g. where herbivore browsing invertebrates). Two hundred three 148 mammals, 36 birds, amphibians reptiles, fish. eighty-one including 215 invertebrates, 33 birds. Twelve competitors, mammals order Carnivora, bird. Forty 17 studied eight plants, six fish, reptiles. Studies involved plant distributed accordance expectations (exact test, P < 0.001; Table formally services’ only located review use site; however, occurred temperate grasslands (75%, n conifer forests (25%, Notably, 36% classified ‘Tropical Subtropical moist broadleaf forest’ 2001), 12% ‘diets 13% zero Fewer interspecific 35) report 126). steep increase publications about time. trend reflected scientific interest better explained curve (R2 0.92) line 0.78). increased, fashion 0.83) at moderate 3). no categories. explanatory power 85% (95% CI [0.75, 0.94]), contrasts zero. debate regulate limit has produced rich centred density dependence assumption animal tend towards equilibrium density-dependent mechanisms govern natural processes; predation mechanism guided back (called Sinclair As tangible example might reduce overabundant restore equilibrium, would governed distance its equilibrium. Proponents limitation, contrast, make influence view mechanistic constraint vital rates, ultimately bookend (Krebs 1995). limiting rates. 24 addressing reporting additive cause mortality. find explicit S2). Of these, (71%) rare impacted apparent competition (Holt Bonsall An important caveat have findings. corroborate finding constrain numbers. 32 studies, Ruth Murphy (2010) concluded resulted cases, captured own review, none areas above capacity. meta-analysis mule deer Odocoileus hemionus survival, compensatory (Forrester Wittmer 2013). minor female elk Cervus canadensis survival throughout 2013) driver calf recruitment (Griffin 2011). localities, does Proffitt biota (Bressette 2012). primary influencing dynamics almost always weather (White 2008), or, temperature rainfall translate productivity availability. Anthropogenic influences, artificial resource subsidies (Muhly 2013), overshadow Therefore, exhibited should caution. Clark Hebblewhite (2020) controlling led increased youngest age-class, numbers appeared equivocal improving adult abundance. Thus, likely less abundant, alternative competition. competition, abundant sustains population, turn disproportionate sympatric system bighorn sheep Ovis canadensis, bighorn-mule system; Kamler 2002). often driven intraspecific variation selection therefore stochastic rather constant (Festa-Bianchet 2006). (those occurring densities) stochastic, mitigated, persist much longer under huemul Hippocamelus bisulcus; when engineers keystone species, even small result cascades American beavers Castor porcupines Erethizon dorsatum, nine-banded armadillos Dasypus novemcinctus; Monroy-Vilchis 2017b). Sweitzer (1997) Nevada’s Great Basin, communities. Risk perception levels. and, cervids Bacon Boyce 2016, Kohl 2019) desert sky islands (Lowrey camelids South spatial avoidance (Donadio Buskirk Smith 2019a). Predation varies space daily activity patterns access high-quality during active (Smith 2019b). analysis movement, (2019) individuals tended avoid risky mainly hunting. (2019b) similarly vicuña Vicugna vicugna avoided habitats night hunt, forage safer daylight hours. Florida, white-tailed virginianus responded panthers breeding birthing times year (Crawford community-wide biomass, diversity, producers 2000). By avoiding dangerous places, animals concentrate Atkins heterogeneity foraging architecture, dependent (Yovovich total, 22 indirectly benefited induced pumas. documenting correlative benefit vertebrates, insect suppress Beschta 2006, 2008). Exclosure experiments indicated weak fear-induced suppression grassland Similarly, heavily browse architecture California, Overall, investigated topics employed inference replication controls confounding abiotic factors weather, cycles). Yet, Suraci simulated presence reductions medium-sized pumas) ‘human-induced cascade’, rodents benefitted sites. suggests behavioural linked modified. Beyond documented predators, address section ‘Effects carnivores’. Carrion energy-rich, ephemeral drives decomposers supports local biodiversity deposition Moleón increases linkages webs, pathways flow, contributes stability resilience (DeVault Peers providing provisioning carrion, large, solitary kill larger themselves subordinate provide amount (Selva Fortuna 2017c). (2012) Patagonia contributed Yellowstone National Park, USA, though densities. One conservatively contribute 1507348 kg meat day Collectively, puma-provided 65 vertebrate invertebrate scavengers, contributing additional food-web bolster health. kills percentage Greater Ecosystem, 28% 11% bird scavenged directly linkages. numerous carrion-dependent such Andean condors Vultur gryphus Perrig (Barry killing repeatedly same areas, ephemeral, nutrient-rich hotspots nitrogen carbon δ15N soils nearby (Peziol Peziol speculated time, resulting akin gardening, it future sites seeking nitrogen-rich forage, will hunt again. present carcasses, scavenger suppressing mesocarnivore bobcats Lynx rufus, coyotes latrans; Allen 2014), opportunities spotted skunks Spilogale gracilis; 2015). However, relationships. Wang experimental gradient development treated sign. sign, scavenging housing activities higher, mesocarnivores, raccoons Procyon lotor. intraguild predation, competitors expected encounters (Prugh carnivores. Theory ‘intraguild cascades’, releases occupying distant niches (Berger 2008, Roemer Mesocarnivores fact commonly part 2009), although killed frequently herbivores. date mitigate coyotes, bobcats, ocelots Leopardus pardalis; Koehler 1991, Hass Massara Santos Ruprecht suggested mortality (the coyote; (2021) 23% coyote eastern Oregon, year. camera years culpeo foxes Lycalopex culpaeus, griseus, (Díaz-Ruiz defining facilitates populations, negative trap visiting Based date, greater caveats. climates, subject 1), Vulpes vulpes, winter (Koehler O’Malley scarcity higher demand (i.e. according Stress Gradient Hypothesis; Barrio (2021), northeast highlighting provisioning. systems Urocyon cinereoargenteus, infrequent kills, suggesting that, reasons yet unknown, expand exploit niche which, turn, ringtails Bassariscus astutus gracilis. cascade hypothesis 2008); extended determine reducing populations. overlap dominant 48% Kusler Although occasionally black bears Ursus americanus Cunningham 1999, 2015), negatively (bears spp., wolves, jaguars onca). kleptoparasitise jaguar (Escobar-Lasso Fonseca frequency