twitter

Wednesday 21 November 2018

Seasonal and predator-prey effects on circadian activity of free-ranging mammals revealed by camera traps Research articleAnimal BehaviorEcologyZoologyPopulation Biology

PEER-REVIEWED Zoological Science section Anthony Caravaggi​1,2, Maria Gatta3, Marie-Claire Vallely1,4, Kayleigh Hogg1, Marianne Freeman1, Erfan Fadaei1,5, Jaimie T.A. Dick1,5, W. Ian Montgomery1,5, Neil Reid1,5, David G. Tosh1,6 November 21, 2018 Author and article information Abstract https://peerj.com/articles/5827/ Endogenous circadian and seasonal activity patterns are adapted to facilitate effective utilisation of environmental resources. Activity patterns are shaped by physiological constraints, evolutionary history, circadian and seasonal changes and may be influenced by other factors, including ecological competition and interspecific interactions. Remote-sensing camera traps allow the collection of species presence data throughout the 24 h period and for almost indefinite lengths of time. Here, we collate data from 10 separate camera trap surveys in order to describe circadian and seasonal activity patterns of 10 mammal species, and, in particular, to evaluate interspecific (dis)associations of five predator-prey pairs. We recorded 8,761 independent detections throughout Northern Ireland. Badgers, foxes, pine martens and wood mice were nocturnal; European and Irish hares and European rabbits were crepuscular; fallow deer and grey and red squirrels were diurnal. All species exhibited significant seasonal variation in activity relative to the timing of sunrise/sunset. Foxes in particular were more crepuscular from spring to autumn and hares more diurnal. Lagged regression analyses of predator-prey activity patterns between foxes and prey (hares, rabbits and wood mice), and pine marten and prey (squirrel and wood mice) revealed significant annual and seasonal cross-correlations. We found synchronised activity patterns between foxes and hares, rabbits and wood mice and pine marten and wood mice, and asynchrony between squirrels and pine martens. Here, we provide fundamental ecological data on endemic, invasive, pest and commercially valuable species in Ireland, as well as those of conservation importance and those that could harbour diseases of economic and/or zoonotic relevance. Our data will be valuable in informing the development of appropriate species-specific methodologies and processes and associated policies. Cite this as Caravaggi A, Gatta M, Vallely M, Hogg K, Freeman M, Fadaei E, Dick JTA, Montgomery WI, Reid N, Tosh DG. 2018. Seasonal and predator-prey effects on circadian activity of free-ranging mammals revealed by camera traps. PeerJ 6:e5827 https://doi.org/10.7717/peerj.5827 Main article text Introduction Animal activity patterns are influenced by a variety of environmental pressures, including food availability (Larivière, Huot & Samson, 1994; Pereira, 2010), foraging efficiency (Lode, 1995; Prugh & Golden, 2014), predator/prey activity (Fenn & Macdonald, 1995; Middleton et al., 2013), human disturbance (Van Doormaal et al., 2015; Wang, Allen & Wilmers, 2015), mate availability and activity (Thompson et al., 1989; Halle & Stenseth, 2000), and ecological competition (Rychlik, 2005; Monterroso, Alves & Ferreras, 2014). Circadian (i.e. recurring every 24 h) and seasonal patterns of activity are adaptive behavioural traits which allow species to effectively exploit their environment and the resources contained therein (Hetem et al., 2012; Phillips et al., 2013). Mammals exhibit a great diversity and flexibility in their activity patterns. A recent study of 4,477 mammal species classified 69% as nocturnal (i.e. night-active), 20% diurnal (i.e. day-active), 8.5% cathemeral (i.e. active throughout the 24 h cycle) and 2.5% crepuscular (i.e. dawn- and/or dusk-active, e.g. lesser mouse deer; Bennie et al., 2014). Activity patterns evolved in response to cyclical changes in the environment that encouraged organisms to respond on a physiological and behavioural basis (Daan, 1981; Kronfeld-Schor & Dayan, 2003; Roll, Dayan & Kronfeld-Schor, 2006; Bennie et al., 2014). Activity patterns are frequently related to daily oscillation in illumination (e.g. changes in sunrise/sunset; Halle & Stenseth, 2000) and, hence, the time(s) of the day during which species are active may vary according to season (i.e. spring, summer, autumn and winter). Indeed, it has been suggested that photic cues are the dominant factor underlying behavioural rhythmicity and that a species’ potential to adapt to non-photic cues (e.g. ecological competitors, predators) may be constrained such that responses are manifest within the normal active period rather than as a shift to a different rhythm (Kronfeld-Schor & Dayan, 2003). The capacity for adaptive behavioural plasticity, while limited in some species (Kronfeld-Schor & Dayan, 2003), is demonstrated in others by observations of intraspecific variation of activity patterns (Ashby, 1972; Hertel et al., 2016) that can result in temporal niche switching (Fenn & Macdonald, 1995; Ensing et al., 2014). For example, Ensing et al. (2014) found that red deer (Cervus elaphus) were mostly diurnal in Canada, while conspecifics in the Netherlands were mostly nocturnal. The difference in activity patterns was attributed to higher levels of human disturbance and a lack of natural predators in the Netherlands. The island of Ireland has a depauperate terrestrial mammalian community due to its prolonged isolation since the last Glacial Maximum (Montgomery et al., 2014). The behavioural ecologies of mammals in Ireland are almost entirely unknown and assumptions of behavioural equivalence between Ireland and locations elsewhere in the species’ range may not apply given differences in land use, human activity and ecosystem composition. Quantifying fundamental ecological parameters such as activity patterns has direct relevance to the management of endemic (e.g. Irish hare, Lepus timidus hibernicus, Bell 1837; Reid & Montgomery, 2007), invasive (e.g. European brown hares, L. europaeus, Pallas 1778; Caravaggi, Montgomery & Reid, 2015), pest (e.g. red foxes, Vulpes vulpes, Linnaeus 1758; Baker & Harris, 2006) and commercially valuable (e.g. deer; Carden et al., 2011) species, as well as those that could harbour diseases of economic (e.g. badgers, Meles meles, Linnaeus 1758; Griffin et al., 2005) and/or zoonotic relevance. The relative dearth of mammalian herbivores and their predators in Ireland also makes the island an ideal study system in which to investigate predator-prey relationships. However, recording and quantifying daily activity patterns of wild, free-ranging mammals presents significant challenges, including overcoming the observer effect whereby the presence of an observer influences the behaviour of the subject (Stewart, Ellwood & Macdonald, 1997), and collecting sufficient data to address scientific and conservation questions (sensu Cagnacci et al., 2010). A number of methodological techniques have been used to overcome such challenges such as radio-tracking, GPS collars and live trapping, each with varying degrees of success (Bridges & Noss, 2011). Radio-tracking has inherent limitations, including periodic (i.e. non-constant) sampling (Lovari, Valier & Lucchi, 1994) and the application of considerable survey effort (Palomares & Delibes, 1991; Reid, McDonald & Montgomery, 2010). Furthermore, they may result in small sample sizes (Bridges & Noss, 2011), capture a limited proportion of the population (Sadlier et al., 2004), and may be subject to signal-based error and/or omission (Cagnacci et al., 2010) or alter the behaviour of tagged animals (Wilson et al., 2011). GPS collars have similar constraints to radio-tracking, particularly with regards to sample sizes and potential signal issues. Moreover, inferences made from GPS collar data can lead to misleading results and much depends on the frequency with which location fixes are obtained (Merrill & Mech, 2003). Live trapping has been used to investigate activity patterns of small mammals, where each successful capture (i.e. the presence of an animal in a trap) is taken as indicating activity (e.g. Elton et al., 1931; Bradley, 1967; Hoogenboom et al., 1984). However, live trapping requires considerable time and effort, is relatively inefficient, may have implications with regards to animal welfare (Torre, Guixe & Sort, 2010), and is subject to species- and trap-specific variations in capture probability (Leso & Kropil, 2010). Remote-sensing camera traps (i.e. remotely activated cameras that are activated via motion or infra-red sensors or light beams; Swann et al., 2004) are increasingly popular in conservation and ecological studies (Kucera & Barrett, 2011) as they are non-invasive and are subject to continuing technological improvements and decreasing costs (Tobler et al., 2008a). They have been used in studies investigating population densities (Trolle & Kéry, 2003; Karanth et al., 2006; Caravaggi et al., 2016), behaviour (Maffei et al., 2005), ecosystem biodiversity (Silveira, Jácomo & Diniz-Filho, 2003; Tobler et al., 2008b), and site occupancy of rare or cryptic species (Linkie et al., 2007). Camera traps afford researchers the means to conduct long-term surveys while minimising in situ survey effort and disturbance of the focal species (Kays & Slauson, 2008). As such, data derived from camera trap surveys of common species and/or conducted at high camera densities are well suited to investigations of wildlife activity patterns (Di Cerbo & Biancardi, 2013; Carbajal-Borges, Godínez-Gómez & Mendoza, 2014; Allen, Peterson & Krofel, 2018). The size and scope of camera trap surveys are limited only by the cost of equipment and personnel to conduct fieldwork and review and analyse data, while the length of time cameras can be left in situ is restricted by available memory, battery life and the possibility of mechanical failure. There may be a trade-off between the proximity and angle of cameras with regards to targets, and the likelihood of detecting and identifying species of varying size (Hofmeester, Rowcliffe & Jansen, 2017). Downward-facing cameras, for example, are more efficient at detecting small mammals (De Bondi et al., 2010). However, species identification is difficult where similar species occur in sympatry (Claridge, Paull & Barry, 2010; Meek, Vernes & Falzon, 2013; Oliveira-Santos et al., 2010). This is often particularly true of small mammals (Claridge, Paull & Barry, 2010; Meek & Vernes, 2016). However, Ireland is home to six species of rodent and two species of shrew, few of which occur in sympatry and all of which are uniquely identifiable. Here, we describe the first study into temporal activity patterns of 10 mammal species found in Ireland, from 14 g (wood mouse, Apodemus sylvaticus, Linnaeus 1758) to 60 kg (fallow deer, Dama dama, Linnaeus 1758), captured via camera traps. In addition, we investigate interspecific relationships, specifically whether predator-prey activity patterns demonstrate (a)synchrony, demonstrating temporal (dis)association between ecologically-linked species. We hypothesised that predator-prey pairs would exhibit non-random, interrelated distribution of activity throughout the diel cycle. This distribution would either display a considerable overlap between predators and prey (i.e. predators are attracted to prey), or asynchrony between the species (i.e. temporal avoidance of predators by prey). Materials and Methods We collated data from 10 camera trap studies conducted in Northern Ireland, where land use is predominantly agricultural (75%) and forest cover, mixed and deciduous woodland and coniferous forest plantations, is 8% of land area (Department of Agriculture, Environment and Rural Affairs, 2018; Forestry Comission, 2018). The focal species of the 10 studies were fallow deer (Dama dama, Linnaeus 1758; n = 6), Eurasian red squirrel (Sciurus vulgaris, Linnaeus 1758) and Eastern North American grey squirrel (S. carolinensis, Gmelin 1788; n = 2) and European brown hare and Irish mountain hare (n = 2). However, as is common for camera trap studies, non-target species were also detected. Therefore, in addition to the five focal species, we present additional information on a further five species: (i) European badger; (ii) European rabbit (Oryctolagus cuniculus, Linnaeus 1758); (iii) wood mouse; (iv) pine marten (Martes martes, Linneus 1758); and (v) red fox (hereafter ‘fox’). The data presented here resulted from a total of 1,164 camera deployments at 431 locations (defined herein as broad study areas, rather than individual camera placements; Fig. 1). Deer surveys were conducted from June 2013 to November 2016, squirrel surveys from January to March 2014 and January to May 2015 and hare surveys from April 2013 to August 2015, non-inclusive. Constituent surveys were independent thus methodologies were not standardised. There was no evidence of intrageneric variation in the activity patterns of hare (Fig. S1) and squirrel (Fig. S2) species, and, hence, both were grouped (i.e. ‘hares’ and ‘squirrels’) for the purposes of the current study. Locations of sites used in camera trap wildlife studies in Northern Ireland from 2013 to 2016. Figure 1: Locations of sites used in camera trap wildlife studies in Northern Ireland from 2013 to 2016. For species-specific maps, see Fig. S3. Download full-size imageDOI: 10.7717/peerj.5827/fig-1 Species surveys We included data from six independent deer studies (D1—6) in our analyses. D1 was conducted over 15 1 km2 squares with an average of 10 cameras per km2 and five additional one km2 squares set at a higher density of 20 camera traps per km2. In total 38 camera trap units were randomly deployed over 255 individual camera trap placements (Table 1) using a combination of Bushnell Trophy Cam (119467), Bushnell Trophy Cam HD (119477), Reconyx (HC600) and Scoutguard Camera (SG560P-8M)—the number of each model used differed between sites. Camera traps were set at a height of 30 cm, perpendicular to the ground. Cameras were set to capture the maximum photographs per trigger (3–10 photographs depending on camera model) and no delay between triggers. Cameras were left for 14 days before being collected and relocated. D2–D4 surveyed smaller areas of 0.02, 0.04 and 0.05 km2 using 10 Bushnell Trophy Cam HD (119677) at each site. Each camera was set at a height of 40 cm from the ground and set to capture bursts of three still pictures and a 60 s video per trigger, with a delay of 1 s between triggers. Cameras were left in situ for 7 days. D5–D6 were focussed on areas of 0.05 and 0.02 km2, respectively and used Bushnell Trophy Cam HDs (119477, 119577, 119676, 119677). Cameras were set at a height of 40 cm and set to capture either three still pictures or a 30 s video, depending on the camera model, with a 1 s delay between triggers. Cameras were deployed for 7 days.