Research trends and spatial distributions of studies

All statistical analyses were performed using R v4.0.3 software (R Core Team 2018). We evaluated whether the geographic distribution of studies was representative of the different terrestrial biomes within the puma’s range. We rasterised the Terrestrial Ecoregions of the World polygons provided by the World Wildlife Fund (Olson et al. 2001) via the ‘sp’ (Pebesma & Bivand 2005) and ‘raster’ (Hijmans & van Etten 2015) packages and extracted the biome type for each study location. We then compared the distribution of study site biomes with the distribution of biomes within the puma’s range using an exact multinomial goodness-of-fit test (Table 1).

We also binned our search results into five-year periods from 1970 to 2019 (excluding papers after 2019 to match the five-year design for this specific analysis) to assess potential publication trends reflecting research interest. We plotted the number of papers published over time in two categories: 1) those that reported any measure of effect or strength of interaction, and 2) those that did not. We tested whether there was evidence supporting a positive linear or exponential slope over time by fitting linear models of counts as a function of time using a Poisson error distribution with either an identity or exponential link function, respectively. We used the strength of correlation (R²) to indicate fit.

Finally, we used a mixed effects logistic model to estimate the probability that effect sizes or control groups were reported in a study. We fit a model that estimated the probability that ecological effects were reported in a study using a fixed effect on research category, and a random effect on the intercept per study as independent variables. The response variable consisted of binary data indicating whether study results reported ecological effects. Some studies reported results from multiple interaction categories and this model structure accounted for interdependence. We estimated model parameters in a Bayesian framework using the ‘rstanarm’ package (Goodrich et al. 2018), using default priors and sampling 1000 iterations of four Markov chains following a 1000 iteration burn-in period. We identified differences in the probability that effect sizes or control groups were used between research categories via 95% credible intervals (CI) around pairwise contrasts.

Diet and prey regulation

The debate over whether pumas and other predators regulate or limit their prey has produced a rich literature centred on density dependence (Sinclair 2003). Core to the regulation framework is the assumption that animal populations tend towards some equilibrium and that density-dependent mechanisms generally govern these natural processes; predation is then one mechanism by which populations are guided back towards equilibrium (called regulation; Sinclair 2003). As a tangible example of regulation, carnivore predation might reduce an overabundant prey population in order to restore equilibrium, and the strength of the impact of predation on that prey population would be governed by the distance that prey population was from its equilibrium. Proponents of limitation, in contrast, make no assumption about the influence of density on population growth and tend to view predation as a mechanistic constraint on prey vital rates, and ultimately a bookend on abundance (Krebs et al. 1995). We hypothesised that pumas would exhibit strong interactions with prey species by limiting vital rates. Our search identified 24 papers addressing puma prey regulation and limitation or reporting that puma predation was an additive cause of mortality. We did not find any explicit evidence of regulation, but 17 papers provided evidence of limitation (Appendix S2). Of these, 12 (71%) were studies on rare prey impacted by apparent competition (Holt & Bonsall 2017). An important caveat is that we did not use prey species names in our searches, which may have limited our findings.

Prior reviews corroborate our finding that pumas constrain prey numbers. In a review of 32 studies, Ruth and Murphy (2010) concluded that puma predation resulted in prey limitation in 17 cases, some of which were captured in our own review, and none of which were from study areas where prey populations were above carrying capacity. In a meta-analysis of mule deer Odocoileus hemionus survival, puma predation was found to be largely compensatory (Forrester & Wittmer 2013). Puma predation had only a minor effect on female elk Cervus canadensis survival throughout the western USA (Brodie et al. 2013) and was not a major driver of calf recruitment in western North America (Griffin et al. 2011). Nevertheless, in some localities, puma predation does influence ungulate growth rates (e.g. Proffitt et al. 2020), which could have cascading effects on community structure, including the abundance of other biota (Bressette et al. 2012).

The primary mechanism influencing ungulate population dynamics is almost always weather (White 2008), or, more specifically, temperature and rainfall that translate into primary productivity and food availability. Anthropogenic influences, including ungulate hunting (Brodie et al. 2013) and artificial resource subsidies (Muhly et al. 2013), can also overshadow the effects of large carnivores on prey vital rates. Therefore, the evidence for prey limitation exhibited by pumas should be interpreted with caution. For example, a meta-analysis by Clark and Hebblewhite (2020) found that controlling carnivores generally led to increased survival and recruitment of the youngest ungulate age-class, but that controlling carnivore numbers appeared equivocal in improving adult ungulate survival or overall ungulate abundance. Thus, the effects of puma predation are likely to be small, except on less abundant, alternative prey populations via apparent competition. In apparent competition, an abundant primary prey sustains the puma population, which in turn can have disproportionate impacts on rare prey sympatric in the system (e.g. predation on bighorn sheep Ovis canadensis, in a bighorn-mule deer system; Kamler et al. 2002).

Puma predation on alternative prey is also often driven by intraspecific variation in prey selection and is therefore stochastic rather than constant (Festa-Bianchet et al. 2006). When predation on rare prey (those occurring at low densities) is stochastic, prey limitation is mitigated, and prey populations persist much longer than they would under constant predation rates (e.g. puma predation on huemul Hippocamelus bisulcus; Wittmer et al. 2014). Nevertheless, when alternative prey are ecosystem engineers or other keystone species, even a small impact on their abundance could result in strong trophic cascades (e.g. predation of American beavers Castor canadensis, North American porcupines Erethizon dorsatum, and nine-banded armadillos Dasypus novemcinctus; Monroy-Vilchis et al. 2009, Elbroch et al. 2017b). For example, Sweitzer et al. (1997) reported puma prey limitation on porcupines in Nevada’s Great Basin, which could lead to cascading impacts on plant communities.

Fear effects on ungulates

Risk perception is a major driver of animal behaviour that can result in cascading effects on lower trophic levels. We hypothesised that pumas would affect prey risk perception and, in our review, we found evidence that cervids inhabiting North American temperate forests (e.g. Bacon & Boyce 2016, Kohl et al. 2019) and desert sky islands (Lowrey et al. 2019), and camelids inhabiting temperate South America exhibit spatial avoidance of pumas (Donadio & Buskirk 2016, Smith et al. 2019a). Predation risk varies across space and time. Therefore, some prey alter their daily activity patterns to access high-quality habitat during specific time periods when predators are less active (Smith et al. 2019b). In their analysis of elk movement, Kohl et al. (2019) found that individuals tended to avoid risky areas mainly when pumas and grey wolves were most likely to be hunting. Smith et al. (2019b) similarly found that vicuña Vicugna vicugna avoided risky habitats at night when puma were more likely to hunt, but would use these areas to forage during safer daylight hours. In Florida, white-tailed deer Odocoileus virginianus responded more to risk from Florida panthers during breeding and birthing than at other times of the year (Crawford et al. 2019).

Predator-induced changes to prey density or prey behaviour can result in community-wide effects by affecting the biomass, diversity, and distribution of primary producers (Schmitz et al. 2000). By avoiding dangerous places, animals concentrate their browsing pressure (Laundré et al. 2010, Atkins et al. 2019), and this spatial heterogeneity in foraging patterns may ultimately affect plant diversity or architecture, and also the species dependent on the plants (Yovovich et al. 2021). In total, we found 22 species that indirectly benefited from fear effects induced by pumas. The research documenting this included correlative studies, indicating that pumas benefit plants, small vertebrates, and insect communities where they suppress ungulate foraging (Ripple & Beschta 2006, 2008). Exclosure experiments focused on vicuña also indicated weak positive effects from fear-induced browsing suppression on grassland communities (Donadio & Buskirk 2016). Similarly, deer heavily browse plants and influence plant architecture and fitness in communities in central California, where they are at less risk of puma predation (Yovovich et al. 2021).

Overall, relatively few studies have investigated these topics for puma, and none employed strong inference via replication or controls for confounding abiotic factors (e.g. weather, nutrient cycles). Yet, findings from these few studies indicate a strong potential for pumas to affect plant communities. While outside the scope of this review, Suraci et al. (2019) simulated human presence in a natural area via experiments and found that reductions in the presence of large and medium-sized carnivores (including pumas) led to a ‘human-induced cascade’, where small rodents benefitted from reduced risk and increased foraging around treatment sites. This suggests that prey behavioural effects linked to pumas are likely to be modified. Beyond these fear effects in prey, several studies documented avoidance of pumas by smaller predators, which we address in the section ‘Effects on other carnivores’.

Carrion effects

Carrion is an energy-rich, ephemeral resource that drives the population dynamics of scavengers and decomposers and supports local biodiversity via nutrient deposition (Wilson & Wolkovich 2011, Moleón et al. 2015, Sebastián-González et al. 2020). Carrion also increases linkages in food webs, defined as pathways for energy flow, and contributes to community stability and resilience (DeVault et al. 2003, Wilson & Wolkovich 2011, Peers et al. 2020). We hypothesised that pumas would exhibit strong interactions by providing access to carrion for other species, and found 14 studies documenting pumas provisioning carrion, all but three of which occurred in the western USA (Fig. 1). Evidence suggests that pumas and other large, solitary felids that kill prey larger than themselves and are subordinate to other apex predators may provide a disproportionate amount of carrion to ecological communities (Selva & Fortuna 2007, Elbroch et al. 2017c). For example, Elbroch and Wittmer (2012) estimated that pumas in Patagonia contributed more than three times as much carrion to ecosystems as grey wolves did in Yellowstone National Park, USA, even though pumas occurred at lower densities. One study conservatively estimated that pumas contribute 1507348 kg of meat per day to their communities throughout their range in North and South America (Elbroch et al. 2017c).

Collectively, studies of puma-provided carrion identified 65 vertebrate and 215 invertebrate scavengers, contributing to evidence that puma-provided carrion supports additional food-web linkages that are likely to bolster ecosystem health. Notably, puma kills support a high percentage of species in the Greater Yellowstone Ecosystem, USA, where 28% of local mammal species and 11% of local bird species scavenged from puma kills (Elbroch et al. 2017c). Nevertheless, we did not find any studies that utilised control areas to test directly whether the presence of pumas increased food-web linkages.

Puma kills are likely to be essential for numerous carrion-dependent species, such as Andean condors Vultur gryphus (Elbroch & Wittmer 2012, Perrig et al. 2017). Pumas have also been identified as ecosystem engineers because their kills provide invertebrate habitat (Barry et al. 2019). By killing prey repeatedly in the same areas, pumas create ephemeral, nutrient-rich hotspots via nitrogen and carbon deposition that increases δ15N in soils and nearby plants (Peziol 2020). Peziol (2020) speculated that over time, the nutrient heterogeneity resulting from puma foraging may be akin to gardening, in that it is likely to increase the future use of these sites by ungulate prey seeking nitrogen-rich forage, and therefore increase the probability that pumas will hunt in these areas again.

When they are present at carcasses, pumas structure scavenger communities by suppressing mesocarnivore access (e.g. bobcats Lynx rufus, coyotes Canis latrans; Allen et al. 2014), and increasing foraging opportunities for small carnivores (e.g. western spotted skunks Spilogale gracilis; Allen et al. 2015). However, human presence can change these relationships. Wang et al. (2015) placed experimental carcasses across a gradient of human development in California and treated some of these carcasses with puma sign. The authors found that bobcats avoided carcasses with puma sign, but that scavenging coyotes did not. Yet, where human housing and activities were higher, carcasses were avoided by most mesocarnivores, except raccoons Procyon lotor.

Effects on other carnivores

Due to risks of intraguild predation, smaller competitors are expected to avoid encounters with apex predators (Prugh & Sivy 2020). Thus, we hypothesised that pumas could affect the abundance or fitness of other carnivores. Theory suggests that the presence of apex carnivores such as pumas can lead to ‘intraguild cascades’, in which the suppression of medium-sized carnivores releases smaller species occupying more distant niches (Berger et al. 2008, Roemer et al. 2009, Wang et al. 2015). Mesocarnivores are in fact commonly recorded as part of the diet of pumas (e.g. Monroy-Vilchis et al. 2009), although they are killed less frequently than herbivores. Nevertheless, most research to date suggests that medium-sized carnivores mitigate competition with pumas via avoidance (e.g. coyotes, bobcats, ocelots Leopardus pardalis; Koehler et al. 1991, Hass 2009, Massara et al. 2016, Santos et al. 2019, Ruprecht et al. 2021). We found only one study that suggested that predation by pumas was a significant cause of mortality of a medium-sized carnivore (the coyote; Ruprecht et al. 2021). Ruprecht et al. (2021) reported that pumas killed 23% of a local coyote population in eastern Oregon, USA, each year. One additional study provided evidence via camera sampling over 14 years that pumas may limit the abundance of culpeo foxes Lycalopex culpaeus, but not South American grey foxes Lycalopex griseus, in Argentina (Díaz-Ruiz et al. 2020).

The defining question is whether food provided by pumas has an overall positive effect and facilitates small and medium-sized carnivore populations, or a negative effect as an ecological trap for subordinate species that may be killed by pumas while visiting their kills (Prugh & Sivy 2020). Based on the literature to date, puma food provisioning is likely to provide greater benefits than negative consequences to smaller carnivores, with some caveats. In temperate climates, where all research on this subject has been conducted (Fig. 1), smaller carnivores, including red foxes Vulpes vulpes, bobcats, and coyotes, increase their scavenging at puma kills in winter (Koehler et al. 1991, O’Malley et al. 2018). Thus, subsidies provided by pumas are likely to be important during times of food scarcity and higher energetic demand (i.e. according to the Stress Gradient Hypothesis; Barrio et al. 2013). Ruprecht et al. (2021), however, found that coyotes were often killed by pumas at puma kill sites in northeast Oregon, highlighting the potential negative effects of provisioning. In systems with red foxes or grey foxes Urocyon cinereoargenteus, coyotes are infrequent scavengers of puma kills, suggesting that, for reasons yet unknown, coyotes may expand their activities to exploit this niche in the absence of foxes (Allen et al. 2015, O’Malley et al. 2018). Allen et al. (2015) found that when pumas are present at their kills, they generally suppress medium-sized carnivores, which, in turn, increases access to carrion resources for small species such as ringtails Bassariscus astutus and western spotted skunks Spilogale gracilis. This study was the only evidence we found supporting the intraguild cascade hypothesis (Berger et al. 2008); however, research such as that by Ruprecht et al. (2021) could be extended to determine whether pumas reducing coyote abundance also releases smaller carnivore populations.

Pumas overlap with more dominant competitors in 48% of their range (Elbroch & Kusler 2018). Although pumas have occasionally been documented killing grey wolves and American black bears Ursus americanus (e.g. Cunningham et al. 1999, Elbroch et al. 2015), we did not find evidence that they negatively influence the fitness or abundance of other large carnivores (bears Ursus spp., wolves, or jaguars Panthera onca). Pumas kleptoparasitise jaguar kills (Escobar-Lasso et al. 2016, Fonseca et al. 2018), but the frequency or impacts of this behaviour are unknown. There is, however, significant evidence suggesting that pumas are subordinate to bears, wolves, and perhaps jaguars, and that pumas are likely to suffer fitness consequences where they are sympatric with these species (Elbroch & Kusler 2018). Wolves, in particular, affect puma space use, diet, and abundance (Kortello et al. 2007, Elbroch et al. 2020). Comparatively, the potential influences of bears and jaguars on pumas are not well understood (see review in Elbroch & Kusler 2018). Pumas may, in fact, have positive effects on brown bear Ursus arctos and black bear populations, as both species are frequent scavengers of puma kills (Murphy et al. 1998, Elbroch et al. 2015). Elbroch et al. (2015) speculated that if pumas subsidise bear populations, bears could lower ungulate recruitment by increasing predation of neonates. Yet, we did not find any research directly testing whether resources provided by pumas increase bear fitness or abundance.

Ecosystem services

We found few studies that explicitly examined whether biotic interactions associated with pumas provided ecosystem services, defined broadly as benefits and essential services that support human economies, health, and well-being (Millennium Ecosystem Assessment 2005). Of the five relevant studies we reviewed, three considered the effects of pumas in mitigating chronic wasting disease (CWD) in cervids. CWD is a prion disease that causes spongiform encephalopathy and can be passed across species barriers, resulting in significant economic damage to hunting industries (Rivera et al. 2019). Some evidence suggests that pumas may selectively prey on infected mule deer (Krumm et al. 2010), potentially because diseased animals have a lower ability to escape before other symptoms become apparent (Miller et al. 2008). In a study of predation and CWD on elk, both were found to be additive sources of mortality (Sargeant et al. 2011). It remains unclear whether pumas that eat CWD-infected prey pass infective prions through their guts back into the environment (as documented in coyotes; Nichols et al. 2015), which would negate the idea that pumas provide the ecosystem service of reducing the spread of this disease.

Across large scales, pumas may also provide ecosystem services via prey limitation that mitigates vehicle collisions with cervids and therefore reduces human injuries, fatalities, and economic costs incurred by victims and society. For example, the recolonisation of South Dakota, USA, by pumas is estimated to have reduced costs associated with deer-vehicle collisions by US$ 1.1 million (Gilbert et al. 2017). Extending these findings, Gilbert et al. (2017) estimated that recolonisation of the eastern USA by pumas could reduce deer-vehicle collisions by 22% over 30 years, averting 21400 human injuries, 155 human fatalities, and US$ 2.13 billion in costs.

[Correction added on 14 March 2022, after first online publication: ‘North’ has been replaced with ‘South’ and ‘billion’ has been replaced with ‘million’.]

It is also possible that pumas control invasive species. Seward et al. (2004) suggested that Florida panthers in the USA may play a role in regulating invasive feral pig Sus scrofa populations, but we did not find any evidence in the published literature. Feral pigs cause significant damage to agricultural landscapes and fragile wetland habitats (Shwiff et al. 2020), and limitation by pumas would comprise a substantial economic benefit.

Finally, researchers working in Argentina concluded that pumas provided an ecosystem service through their role in dispersing seeds from plant species consumed by eared doves Zenaida auriculata (Sarasola et al. 2016). Seed dispersal is a key process that is important for sustaining plant communities. Nevertheless, we encourage caution when interpreting these results for two reasons. First, the authors did not employ genetic tools when identifying the scats they analysed to study carnivore diet (see Morin et al. 2016). Second, the diet they reported for pumas in their study system – which was overwhelmingly made up of doves – is an outlier when considering puma diet. It seems plausible that the authors may have included the scats of smaller carnivores in the analysis.