Optimal foraging of lions at the human wildlands interface

Optimal foraging and landscape of fear theories provide frameworks which can be useful for investigating animal's space prey use decisions. Predators, such as African lions Panthera leo, are likely to respond abundance, accessibility, profitability potential risks, often anthropogenic in nature, while making decision. Identifying the relative role these processes has important conservation implications. We investigated responses a pastoralist within lion feeding spatial ecology at human-wildlands interface. collected predation data from 12 GPS-collared ungulate count transects, along South Africa–Mozambique border, including parts Kruger, Limpopo Banhine National Parks. calculated Jacobs' Index values 80 kills investigate selection wild domestic ungulates prey, used maximum entropy modelling predict multi-season occurrence resource functions estimate probability areas by lions. All had access livestock their home ranges. Lions showed strong large-bodied taking waterbuck, zebra, kudu buffalo more frequently then predicted availability. slight avoidance cattle with outnumbering larger across much study area. greatest habitats high occurrences specifically kudu, nyala buffalo, during dry seasons wet season; season when kept closer settlements thus better protected easier avoid. These results suggest that select optimally, yet show response areas. This duality behaviour suggests efforts mitigate human–lion conflict preserve vulnerable populations should focus on both increasing well exploiting lion's humans careful consideration risks presence acting an ecological trap populations. Les théories de la recherche nourriture optimale et du paysage peur fournissent des cadres qui peuvent s’avérer utiles pour étudier les décisions d'utilisation l'espace proies animaux. prédateurs, tels que d'Afrique sont susceptibles réagir à l'abondance proies, l'accessibilité, rentabilité aux risques potentiels, souvent nature anthropique, lors prise relatives nourriture. L'identification rôle relatif ces processus d'importantes conséquences conservation. Nous avons étudié le réactions un pastoral dans l'alimentation l'écologie spatiale sur, -l'interface terres en friches-humaines. recueilli données spatiales prédation chez munis colliers GPS comptage d'ongulés partir long frontière entre l'Afrique Sud Mozambique, y compris parties parcs nationaux Banhine. calculé valeurs l'indice Jacobs abattages sélection par ongulés sauvages domestiques comme utilisé modélisation désordre maximal prédire l'occurrence sur plusieurs saisons, fonctions ressources estimer probabilité zones Tous avaient accès bétail domestique leur domaine vital. ont présenté une forte grande taille prenant cobes defassa, zèbres, koudous buffles plus fréquemment prévu disponibilité. léger évitement bovins étant nombreux grands partie zone d'étude. d'habitats avec fortes sauvages, particulier koudous, puis buffles, pendant saisons sèches poussé saison pluies, laquelle est gardé près habitations donc mieux protégé facile prévoir éviter. Ces résultats suggèrent sélectionnent leurs manière optimale, mais présentent réaction vis-à-vis bétail. Cette dualité comportement suggère visant atténuer conflits l’homme lion, préserver vulnérables devraient se concentrer fois l'augmentation l'exploitation envers humains tenant soigneusement compte présence agissant piège écologique vulnérables. How animals make movement decisions relation threats is fundamental question importance species theory each animals' (Haswell al., 2019; Oriol-Cotterill 2015). predicts, inter alia, animal preferentially exploit habitat containing most abundant accessible unless movements constrained other factors (MacArthur & Pianka, 1966). diet predicts based abundance (Stephens Krebs, 1986). Landscape will avoid risky so heterogenous perceivable (Gaynor Laundré 2010). therefore they may encounter decision (Oriol-Cotterill occupy position top food web but recent times have themselves become limited pressures, loss persecution retaliation depredation (Bauer 2015; Ripple 2014). 92–632 kg weight range (Clements 2014; Owen-Smith Mills, 2008). overlaps adult live weights (average 366 kg) local Bos taurus, breeds (van Marle-Koster 2021). Cattle also lost anti-predator defences (Muhly 2010; Weise 2020). In addition, replaced dominant grazer Africa (Hempson 2017; 2015), (Lindsey 2017). known cattle, particularly densities low (Khorozyan Loveridge Valeix 2012). forage optimally accordance its availability (LeFlore Werhahn 2019). However, predation, heavily persecuted owners (Everatt, Kokes, Ontiri fact, one causes declines abundances continent exhibit temporal thought behavioural Suraci, Clinchy, Behavioural human-induced exhibited relatively fine scales where there exists heterogeneity encounters between (Elliot For instance, thickets detection nearby herders (Schuette 2013) or limit activities night vicinity pastoralists behaviours recognise indirect associated predating offers direct pastoralist-induced part optimal Promoting persistence prey-depleted pastoralist-impacted landscapes ultimately dependent understanding 2015) using this information identify maintain minimum thresholds available Everatt, Moore, Khorozyan Here we examined ranges, were threatened levels persecution. sought whether how impacting Assuming body size proxy profitability, tested two hypotheses examining assumed if did not induced decisions, expected would Alternatively, does impact lions, expect habitat. implications, livestock, without landscape, mitigation best achieved improving husbandry efficient strategy reduce lion–livestock conflict. was conducted north–south-oriented borderlands Kruger Park (KNP) Sabi Game (SGP) Karangani Reserve (KGR) Mozambique eastward (LNP) (BNP) (Figure 1). The area contained Greater Lion Conservation Unit (IUCN, 2006). Vegetation generally classified mixed savannah woodland grasslands (Stalmans 2004). Wildlife near carrying capacity decline eastwardly due management history environmental variables majority Mozambican section occupied free-ranging eastern border KNP LNP, SGR, KGR BNP fitted tracking collars. lured baits audio lures broadcasted through loudspeakers, alternating distress call Syncerus caffer, calf warthog Phacochoerus africanus; representing sizes among commonly predated (Hayward Kerley, 2005). immobilised following Zimbabwe Veterinary Association Group standard operating procedures (Kock Burroughs, Procedures approved Research Ethics Committee: Animal Nelson Mandela University (permits A13-SCI-ZOO-014, A18-SCI-ZOO-007) approval capturing handling granted Administração Nacional das Áreas Conservação (ANAC) 005-2016, 001-2017, 006-2018). darted Dan-inject JM dart gun (Dan-inject RSA, Skukuza, Africa). Male administered approximately 80–100 mg Zoletil® 6–8 medetomidine reversed 5× dose atipamazole, female 4–6 atipamezole qualified veterinarian KE who trained working under supervision vet ANAC. satellite collars (VECTRONIC Aerospace, Berlin, Germany; Tracking, Pretoria, programmed record locations every 2 h 18:00 6:00, active location 12:00 least (Cozzi Using fixes, able observe path identified spatio-temporal clusters sequential fixes 100 m previous fix (Tambling (<60 days old) searching centre cluster outwards expanding arcs up 50 away. When evidence kill found recorded species, sex age class, identifiable material horns, skin, jaws bones 2010, acknowledge possible bias towards recording inherent analysis increase lag time events investigation our objective produce comprehensive description rather species. included opportunistically order number samples (kills). inclusion effect sampled population collared. Data (species over 16 750 kg, see Table 1 S1 list) distribution gathered point counts made vehicle-transects adjoining collared approximate Transects driven <20 km/h observers just after sunrise before sunset; (Owen-Smith Goodall, 2014) still corresponds closely activity Ungulate (November–April) (May–October). detections unique track off-road walking surveys described al. (2019) same periods biases restricting sampling roads. Tracks ≥1 km apart counted distinct occurrences, since metric interest At observation, coordinate Cyber tracker V3.440 program (www.cybertracker.org/). Presence only points, combined surveys, build models common selected frequency per site estimating constructed discrete suitability (Duangchatrasiri Watts 2019) sufficient cattle. influence vegetation cover Normalised Difference (NDVI), precipitation, elevation, slope distance (Ryan Variables extracted Google Earth Engine (https://developers.google.com/earth-engine). variables, inputted into MaxEnt version 3.4.1 (Phillips 2006; default settings logistic output ran 10 bootstrap replicates (Merow 2013), randomly 75% training 25% model validation. jackknife variable determine overall predictive receiver curve (AUC) assess accuracy 2013; Phillips interpreted resulting outputs indexes exported function analysis. cleaned removal duplicates points any Horizontal Dilution Precision (causing errors) above (Adams 2013). estimations derived 90% Minimum Convex Polygons (MCPs) 10% outlying removed QGIS (v 3.4.4) (QGIS Development Team., (RSF) wildlife (Johnson To avoided and/or use, locations, compared habitat, estimated random 1:1 ratio Univariate regression analyses run individual (total n = 21 models). independent scaled (i.e. z-transformed) Maxtent models. account variation sample intercept) (Gillies Competing ranked AIC strength direction evaluate variable's approach 0. assessed fit normal measures pseudo-R2, sensitivity, specificity characteristic AUC). Analyses performed R 4.0.3 software lme4 (Bates caret (Kuhn, 2020) pROC (Robin 2011) packages. trackers 8 females 4 males, 9 prides coalitions, distributed Three malfunctioned batteries expired poached, stolen active. Accounting losses, tracked average 245 (range 29–485), providing total 12,291 useable 2941 Based constraints listed above, it annual ranges MCP (5 3 males). Home averaged 326.4 km2 109.5–620.3 km2) 505.8 males 290.2–721.3 km2). Female smaller (n 3, 272 higher density spanning KNP-Mozambique lower central LNP 2, 408 Solitary lionesses 2) (152.6 consisting >3 331.7 km2, 3). whose spanned border. 12,646 18 2845 vehicle transects 7292 2679 2; 1; S2). Impala 3898), followed 3136) 3101) (Table S1), however, (>30 BNP. different predicting varied seasons. There is, general trend many being negatively settlement proximity season. differed positively Six nine areas, contributing model, exhibiting kills, 41 located 39 Buffalo taken 21) 19), waterbuck 12), zebra 7), 7) hippo, Hippopotamus amphibious 4). Other duiker Sylvicapra grimmia, crocodile Crocodylus niloticus aardvark Orycteropus afer, scavenging white rhino Ceratotherium simum. rhino, crocodile, having considering them outside usual unable calculate excluded below threshold interest. Kudu killed five sites. Buffalo, three sites, impala, sites duiker, 2). Predation (kills) (±SE) mean Across area, preferential impala Figure Prey patterns significant p < 0.05) AIC, 3; Figures 5). Both accounted respective acceptable discrimination AUC > 0.70). Predicted appear increases however little support R2, accuracy, driving prey. strongly large bodied buffalo; acknowledging less effective small differ events. documented here agrees meta-analyses diets Hayward 2005), broader carnivore energetics (Carbone 2007), supports predictions profitable case faced reduced gregarious herbivores (prey depletion) shifting downwards next (large) (Vinks highest poached (offering greater rates) typically either rare (buffalo) absent (zebra, wildebeest, giraffe). Similar shifts decreased been observed Kafue national parks (Creel 2018; suggestion Creel (2018) larger-bodied causing widespread smaller-bodied Additionally, indicate group (Funston 2001). item difficult groups subdue demographic interpretation explains apparent SGP SGR all those females, contrasting coalition females. result identification non-selection even despite depletion shown above. preferred 2017), possess 2005; Muhly 2010) riparian structures facilitate capture (Mosser 2009). Considering factors, While include diet, human outweigh rewards large, defenceless each, Waza NP, Cameroon (Tumenta Makgadikgadi ecosystem Botswana (Valeix 2012) confers relationships increased corresponding seasonal dispersion northern Tanzania (Kissui, 2008) southern Kenya (Lesilau 2018). further supported exceptional study. analysis, seven killed, event herd 1000 unattended moved temporarily saturating guild. involved monitored pride, later retaliation. lone lioness. Cattle, present challenge foraging, highly profitable, predator, considerable risk predator via pathway retaliatory killing owners. A complexity lies carnivores, rarely occurs itself, occurring perceived threat than itself statistical consider fitness impacts which, mentioned led, deaths inv