Temporal occurrence and species composition of birds on artificial feeding sites maintained for game mammals in the Dinaric Mountains, Slovenia

Artificial feeding is a widely used management tool, but it often attracts nontarget species, including birds, to permanent feeding sites. This study used camera traps to monitor the presence of birds at selected sites used for bear management in Dinaric forest. A large number of bird species (35) were recorded, representing roughly half of all species breeding in the surrounding area. These species were grouped based on monthly and hourly presence, and corresponded to food groups, with most belonging to granivores or scavengers. Some species, such as Pigeons ( Columba sp.), Raven ( Corvus corax ) and Buzzard ( Buteo buteo ), adapted their presence to the availability of food at the feeding sites, while others were not affected by this. Both Chaffinches ( Fringilla coelebs ) and Jays ( Garrulus glandarius ) frequented the feeding sites, but their temporal presence was influenced by their biology rather than by food availability. The Sparrowhawk ( Accipiter nisus ) also adapted its presence to food availability, and its presence was closely associated with that of the Jay. This study confirms the temporal differences in the use of feeding sites by birds and mammals, which is likely due to their different biology and past management. This can be used to make wildlife management more efficient and reduce the undesirable effects of artificial feeding.


Introduction
Human activities result in the provision of large quantities of food for wildlife, either unintentionally (Perkins et al. 2007, Plaza & Lambertucci 2017 or intentionally for various purposes (Robb et al. 2008). Some wildlife populations, such as northern populations of the Great Tit (Parus major) or vultures, rely heavily on artificial feeding (Orell 2008;Cortés-Avizanda et al. 2016), while others are affected in all aspects of their ecology, including reproduction, behaviour, demography and distribution (Robb et al. 2008).
Artificial feeding of wildlife, including birds, occurs in many forms. For example, bird feeders are common in urban areas worldwide (Robb et al. 2008). In some parts of the world, carrion feeding stations are a common practice for vulture conservation (Cortés-Avizanda et al. 2016). In addition to intentional feeding (e.g., bird feeders), birds are offered artificial food in large quantities at feeding stations maintained (usually by hunters) for mammals for the purposes of hunting, monitoring, conflict prevention and/or wildlife watching. This type of feeding is a common practice in much of the world, has a long tradition and can be fairly intensive. For instance, in Slovenia, about 12.5 tonnes of corn per 100 km 2 are fed to wildlife annually by hunters (Krofel & Jerina 2016). This artificial feeding is usually intended for only a few species. In Europe, such species include ungulates (i.e., Milner et al. 2014), mesopredators and Brown Bears (Ursus arctos) (Graf et al. 2018). However, numerous non-target species also visit these feeding sites, including several bird species (Fležar et al. 2019).
Most research on the effects of artificial feeding on birds has focused on bird feeders in urban areas (Jones & James Reynolds 2008;Robb et al. 2008). Although bird feeders are designed to provide sustenance during harsh conditions and to foster a connection between people and nature (Robb et al. 2008), unanticipated ecological and behavioural consequences can arise. For example, birds can colonise areas with feeders beyond their natural range (Robb et al. 2008) or at densities not seen in nature (Wilson 1994). Artificial feeding can also allow some non-native species to survive (Clergeau & Vergnes 2011) or even dominate over native species (Galbraith et al. 2017) in new areas. Furthermore, artificial feeding may also alter breeding through increased breeding success (Harrison et al. 2010) or increased nest predation rates (Selva et al. 2014;Oja 2017).
Although artificial feeding sites intended for hunting have been studied primarily from the perspective of ungulate game species (Wirsing & Murray 2007;Sorensen et al. 2014;Pedersen et al. 2014;Milner et al. 2014) and Brown and Black Bears (U. americanus) (Bowman et al. 2015;Selva et al. 2017;Graf et al. 2018;Candler et al. 2019;Fležar et al. 2019), effects on birds have received little attention. Previous studies have primarily focused on nest predation (Cooper & Ginnett 2000;Selva et al. 2014;Oja 2017) and breeding success (Pedersen et al. 2007). Most studies have noted the presence of birds among non-target species without further investigation (Lambert & Demaris 2001;Selva et al. 2014;Bowman et al. 2015;Fležar et al. 2019), and surprisingly, even basic information on bird use of these feeding sites is lacking.
The main objective of the present study was to investigate the temporal occurrence of birds at artificial feeding sites maintained by hunters for game mammals. The temporal availability of the food offered at these sites differs from that of natural food. In some cases, such as in Slovenia, artificial food may be available almost continuously (Fležar et al. 2019), which could potentially impact the seasonal and circadian activity of birds. Despite its potential effects, the topic has received limited attention, with no existing data on the influence of artificial feeding sites for game animals on the temporal occurrence of birds. The study aimed to (i) analyse the species composition of birds visiting feeding sites; (ii) examine the seasonal and circadian use of feeding sites by selected bird species, with an emphasis on the timing of occurrence and the overlap of different species at artificial feeders; and (iii) compare this use of the sites with that of better-studied game mammals (Candler et al. 2019).

Study area
The study area includes a large part of the Dinaric Mountains in southern and central Slovenia (Fig.  1). The region is the core habitat area for several large ungulate species, including Red Deer (Cervus elaphus) and Wild Boar (Sus scrofa) as well as large carnivores such as Brown Bear, Wolf (Canis lupus) and Lynx (Lynx lynx). The study area spans an altitude range from 300 to 1,796 m a.s.l. and is primarily covered by mixed forests dominated by Silver Fir (Abies alba) and Common Beech (Fagus sylvatica). This area is the most forested in Slovenia (over 80% covered by forest), with settlements located in Karst fields and a few river valleys. The area has a high vertebrate diversity, especially that of mammals (Fležar et al. 2019). However, permanent water sources are scarce, especially at higher elevations, and open habitats are limited to the edges of the mountain massifs. As a result, breading bird density and diversity is relatively low, consisting mainly of forest specialists (Mihelič et al. 2019).
The practice of artificially feeding wildlife in the Dinaric Mountains is widespread and has a long tradition, with some feeding sites dating back to the late 19th century (Garshelis et al. 2017). The feeding sites are managed by local hunters and serve various purposes. As observed in other countries (Selva et al. 2014;Bowman et al. 2015;Oja 2017), the sites typically offer two main types of food: corn (or other grain) and carrion. While corn is widely available and often available yearround at many feeding sites, carrion is supplied more opportunistically and in smaller quantities (mostly offal from hunted game species and road carcasses of ungulates; see Mohorović et al. 2015 andGraf et al. 2018 for a detailed description).

Selected feeding sites
As part of the Life DinAlp Bear project, 23 representative feeding sites were selected primarily to study the effect of artificial feeding on brown bears . Corn was available year-round at all feeding sites, with smaller amounts (or none at all) in winter, but methods of delivery differed between some sites. Corn was either distributed in the afternoon using automatic dispensers or delivered manually during the day. In both cases, a few kilograms of corn were distributed to feeding sites daily. During the 2016-2017 study period, carrion was provided to the monitored feeding sites (Graf et al. 2018). Carrion in the form of hunting remains and road carcasses of ungulates were provided throughout the year, with a peak during the hunting season (fall). The selected study sites were located in small clearings within the forest, and were on average 2 km from the forest edge and 2.6 km from the nearest settlement. The clearings varied in size, and vegetation was mowed and removed at least once a year, usually in summer, to ensure good visibility and prevent overgrowth.

Methods
The selected feeding sites were monitored 24 hours per day for 2 years with automatic photo/ video cameras (UOVision IR PLUS BF HD UV 565). Camera traps were placed so that the feeder was in front of the camera and the view was not obstructed by woody vegetation. A 30-second video was recorded after each photo. The cameras were set so that the next possible photo could not be taken until a five-minute period had elapsed. If the camera took more than one photo within this five-minute interval, we analysed only the first photo and omitted the others (11.0% of photos). The cameras were checked approximately every two to three weeks. If a system malfunction occurred or the batteries were depleted between checks, the date and time were automatically reset to default settings. Unless they were corrected in the field, we considered these date settings to be incorrect. Periods with incorrect date settings were excluded from the temporal occurrence analysis (6.4% of photos). Temporal occurrence in this paper includes both seasonal occurrence expressed in monthly presence and circadian occurrence expressed in hourly presence. The study and feeding sites were set up for Brown Bears as the primary target species. At one feeding site, the feeder was only suitable for bears, and no food was available for other species. Therefore, we did not include the data collected at this feeding site in the study. Several feeding sites experienced problems with feeders, such as a malfunction of the automatic dispenser or damage to the feeder from Brown Bears or falling trees. Periods when cameras or feeders were not working resulted in gaps in our data set. When gaps extended over several months, the entire dataset for that year from that feeding site was excluded from temporal analyses. Only feeding sites with complete data sets (16 sites) were used for circadian and monthly presence analyses. Due to the inaccessibility and poor maintenance of feeding sites and cameras during the winter months (December-February), there were many gaps in the data. In addition, during this period, food is supplied at a lower frequency. For this reason, we did not include data gathered during the winter (4.5% of all photos). More detailed descriptions of camera settings, maintenance of feeding sites and cameras, data collection, photo review and species identification from photos can be found in Graf et al. (2018) and Fležar et al. (2019).

Bird species occurring at feeding sites
For the photo analysis, we focused on all identifiable bird species down to the size of tits (Paridae), including smallest species such as the Coal Tit (Periparus ater) and Marsh Tit (Poecile palustris). We did not separate individuals in the different photos, but considered each photo as a separate event, even if it may have captured the same individual. This approach was adopted based on previous studies that showed that temporal autocorrelation in camera trap data diminished after one minute (Kays & Parsons 2014, Kays et al. 2017, Kellner et al. 2022). Therefore, we grouped consecutive pictures taken five minutes (or greater) apart into sequences that were considered independent records. In this way we ensured that the capture events are more likely independent and the pseudo-replication impact is mitigated. We used the independent records of birds to model the temporal activity of birds and describe the relative degree of site use (i.e., the amount of time birds spent at feeding sites). Given the similarities in feeding station characteristics, such as the use of camera traps over a long period of two years and the provision of the same food supplement (corn), we expected that, for a given species, the temporal activity pattern at feeding stations would be similar.
For species with greater than 10 records, we used hierarchical clustering analysis to group species with similar activity patterns (both seasonal and circadian, see below). To this end, we applied an unweighted pair-group clustering algorithm based on the arithmetic averages (UPGMA) to the dissimilarity matrix. This matrix was calculated based on the Kulczynski distance of the abundance data. The optimal number of clusters (i.e., groups) was determined by applying the Kelley-Gardner-Sutcliffe penalty function (KGS) (Kelley et al. 1996). This analysis was performed using the "vegan" R package (Oksanen et al. 2017).

Seasonal and circadian occurrence of birds at feeding sites
For species with over 100 records, temporal activity for each species was estimated by pooling data across years and summing the number of independent records of a given species per hour and month. We then related the temporal data to relative solar time (Nouvellet et al. 2012) using the "SunTime" function in the "Overlap" R package (Meredith & Ridout 2020). We then compared temporal activity patterns between species statistically using the Watson-Wheeler test (Zar 1999) and graphically (Fig. 3). This test, a common approach to assess differences between two circular distributions (Frey et al. 2017;Massara et al. 2018), was performed using the "hms2rad" function implemented in the "astroFns" R package to convert species activity time from angular format (hh:mm:ss) to radians where 1 hour = π / 12 (Harris 2012). By doing this, we created a vector of activity time for each species and used the Watson-Wheeler test to compare the mean time of day in which species were active.

Comparison of temporal occurrence between birds and mammals
We measured the daily activity overlap between mammal and bird species using the sum of individuals in five-minute intervals using the framework developed by Ridout and Linkie (2009), which fits a kernel density to temporal data. We then estimated the degree of overlap between the two density curves by calculating the coefficient of overlap (Dhat1), which is most appropriate when the sample size is small (at least 10 records/species) (Linkie & Ridout 2011;Frey et al. 2017). The value of Dhat varies between "0" (i.e., no overlap) and "1" (i.e., complete overlap) (Linkie & Ridout 2011). This analysis was performed using the "Overlap" R package (Ridout & Linkie 2009).

Bird species occurring at feeding sites
We detected a total of 35 bird species at all sites ( Table 1, Appendix 1). Of these species, eight were recorded with only one photo and five species were present in five or fewer photos. Only 14.8% of all photos (36,686) had a bird, representing 1.63% of all possible five-minute time intervals when camera traps were in Table 1. Number of photographs on selected artificial feeding sites in Slovenia for eight species of birds with more than 100 photographs.

Species
No Among all species, eight were present in more than 100 photos (Table 1).

Temporal occurrence of birds at feeding sites
When comparing temporal patterns of occurrence, most species were divided into three clusters.
The largest cluster included eight species, including Sparrowhawk (Accipiter nisus), with Sparrowhawk and Jay being the closest, followed by White-tailed Eagle (Haliaeetus albicilla) and Raven and Wood Pigeon and Stock Dove (Columba oenas) (Fig. 2). Seven out of the eight species with more than 100 photographs were present throughout the study period from March to November (Fig. 3). The only exception was Stock Dove, which was present in only nine photographs after July and completely absent in October and November. Although most species were present throughout the studied period, there were marked differences in the frequency of their presence. For example, Jay, Wood Pigeon, Raven and Stock Dove had unimodal monthly distributions; Chaffinch (Fringilla coelebs) and Great Tit had bimodal distributions; and Blackbird (Turdus merula) and Buzzard (Buteo buteo) had multimodal monthly distributions. Each species had its peak or peaks in different months (Fig. 3). Four species (Jay, Wood Pigeon, Chaffinch, Great Tit) peaked in late summer, two (Raven and Buzzard) peaked in fall and only one peaked in May (Stock Dove) and April (Chaffinch). Although all selected species were present throughout the day, the Watson-Wheeler test for differences in activity distributions showed significant differences (p = 0.001 at df = 2) in the temporal activity patterns for all species comparisons: Ravens and Blackbirds were more often present in the morning; Chaffinches and Jays in the middle of the day; and Wood Pigeons, Stock Doves and Buzzards in the afternoon (Fig. 3).

Temporal occurrence of birds and comparison with mammals
The highest percentage of birds was photographed in September (Fig. 4). The overall distribution of the monthly presence of birds and mammals was similar, but the correlation was not significant (Spearman's r = 0.53, df = 8; p = 0.1475). Birds peaked in fall while mammals were more abundant in summer (Fig. 4). The only month in which there were more birds than mammals was October, and both groups had similar presence in September. Fig. 3. Temporal distribution of selected species according to detected monthly and hourly occurrence at artificial feeding sites in Slovenia. The activity pattern is a scaled index between 0 and 100%. It was calculated by dividing the number of independent records in a given hour by the maximum number of independent records in an hour for a given species. Lines indicate an average monthly sunrise (dashed line) and sunset (dotted line).
Birds were photographed exclusively during the day (Fig. 4), with the exception of a few photos of owls taken at night. Most birds were photographed in the afternoon, with a peak between 2 and 3 pm (9.3% of photos). The occurrence of birds at feeding sites had an "inverted" hourly distribution compared to mammals (Spearman's r = -0.75, df = 22; p < 0.001). They were almost completely absent at night when most mammals were present, but abundant during the day when mammals were rare. The shift occurred just before sunrise in the morning and just after sunset in the evening (Fig. 4).

Bird species occurring at feeding sites
The present study reports the highest number of bird species (35) reported to date at a wild game feeding site. To date, most studies have reported fewer than 10 taxa (Lambert & Demaris 2001;Bowman et al. 2015;Selva et al. 2017;Candler et al. 2019;Fležar et al. 2019), but this may be due in part to the grouping of species into passerine birds (Lambert & Demaris 2001), unknown birds (Bowman et al. 2015) or small and medium-sized birds (Fležar et al. 2019). Other possible reasons for the greater species diversity of birds in our study compared to other studies include shorter monitoring periods in other studies (Lambert & Demaris 2001;Bowman et al. 2015;Fležar et al. 2019), specific baits (Candler et al. 2019) and a focus on a specific group, such as potential nest predators (Selva et al. 2014). We detected about half of all breeding species in surrounding Karst upland forests (Mihelič et al. 2019;Fležar et al. 2019). It is possible that some species were overlooked in the study because they are too small to trigger cameras (Randler & Kalb 2018), but this seems unlikely for two reasons: 1) Most smaller species that breed in the surrounding forests and were not detected at feeding sites are either insectivorous or foraging specialists (Mihelič et al. 2019) and would only be accidental at feeding sites; 2) Only a few species found in surrounding forests are smaller than Coal, Blue and Marsh Tits (e.g., Goldcrest, Regulus regulus) and most avoid open areas such as clearings. On the other hand, a few abundant species breeding in the vicinity were expected but not recorded in the study, such as several finch species (e.g., Goldfinch, C. carduelis, or Hawfinch, C. coccothraustes) and two remaining tit species (Crested Tit, Lophophanus cristatus, and Willow Tit, Poecile montanus). Also, more species could have been observed during the migration and winter periods, such as the abundant Marsh Harrier (Circus aeruginosus), which often feeds on carrion and hunts rodents (Orta et al. 2020), or Starling (Sturnus vulgaris), which feeds on corn and visits feeders (Cabe 2020). Although we divided birds into groups based on their temporal occurrence at the feeding sites, the groups actually represent the types of food the birds ate. Most detected species were either granivores (7 species) or scavengers (4 species) and were grouped as such. There were a few exceptions, the most important of which was the Sparrowhawk, which was grouped with granivorous species, indicating a predatory response to the temporal presence of its prey (Botts et al. 2020). Out of five species not included in the two biggest clusters, two formed a separate group. These two were probably not attracted to either type of food offered (corn and carrion) or to potential prey attracted to the food (mice, birds). Both feed on seeds and grains and likely eat some of the available corn, but the majority of their diet consists of invertebrates (Collar 2020;Collar & Christie 2020). Both species likely only take advantage of forest clearings that result from maintained feeding sites and do not adjust their presence to the availability of offered food.

Seasonal and circadian occurrence
The availability of certain types of food at feeding sites changes over time, and the temporal activity of birds reflects these changes for some, but not all, species. Automatic feeders generally dispensed corn in the late afternoon or evening, while sites supplied by hand were visited by managers during the day. Wood Pigeons and Stock Doves, which are primarily grain feeders and frequently feed on corn (Billerman et al. 2020), peaked in the afternoon when corn was most available. Ravens and Buzzards also likely responded to carrion availability, peaking during the period of highest human hunting activity from September to November, confirming the importance of human-provided carrion for Ravens (Legagneux et al. 2014). On the other hand, the availability of artificial food does not appear to strongly influence the temporal presence of Chaffinches and Jays, although both frequently feed on grains (Clement 2020;Madge et al. 2020). Chaffinches, for example, peaked in the middle of the day and in April and late summer, most likely due to migration and dispersal (Clement 2020). In addition, seeds are only an important food for Chaffinches outside of the breeding season (Clement 2020), which at least partly explains their lower presence in May and June, when they feed mainly on invertebrates. It is less clear why they peak in the middle of the day, since most of their food is available in the afternoon. One possible explanation is competitive niche displacement (Carother & Jaksić 1984) in combination with predator pressure. Chaffinches occurred when few other species were present, so they avoided possible predators and larger and more competitive granivores that feed on the ground (e.g., pigeons). Although Jays are also granivores, they peaked at midday rather than in the afternoon like pigeons. One explanation for the observed difference could be a feeding adaptation. The Jay's ability to feed along branches allows it to feed on corn high up on artificial feeders that is inaccessible to most mammals. Although the quantities are small, corn is thus available to jays for most of the day. In fact, they were often observed feeding on corn directly from feeders. In addition, Jays occur in smaller groups (Madge et al. 2020;present study), each of which requires a smaller amount of food than a group of pigeons.
Artificial food also indirectly influences bird activity and food choice through the presence of other species. At the studied feeding sites, this is true for predators such as Sparrowhawks and owls. Constantly available corn attracts small rodents and several small to medium-sized bird species (Fležar et al. 2019). While owls attracted to rodents did not change their temporal presence (both rodents and owls are active at night), the Sparrowhawk synchronised its presence at feeding sites with its prey of small and medium-sized birds (Meyburg et al. 2020). It probably also hunts smaller species in the granivore group, especially smaller males, but its main prey at feeding sites appears to be Jays, with which it was most closely associated in the cluster analyses. Of the 31 analysed records of Sparrowhawks in our data, four photos/film clips show successful capture and another seven show a close pursuit. The Jay was the most common bird species at feeding sites, but is also at the upper size limit of Sparrowhawk prey (Meyburg et al. 2020), suggesting a shift in prey size preference due to prey availability and accessibility. Aside from the high presence of Jays at feeding sites, the habitat and position of prey may also be favourable for the Sparrowhawk. It prefers to hunt in clearings (Meyburg et al. 2020), which are not unlike typical managed feeding sites. In addition, most of the food is scattered on the ground in the middle of the clearing, allowing the Sparrowhawk to surprise its prey from above. Overall, feeding sites appear to be a highly concentrated food source for raptors as well.
Although there are a number of similarities between the selected feeding sites, there are some differences that are worth exploring. The biggest difference between feeding sites is in the method of distributing corn. Corn is distributed manually at a few feeding sites. This means that there may be slight variations in the time of distribution, and that there is also the increased presence of people on the site. Also, the presence of some species can influence the temporal distribution of other species (Carother & Jaksić 1984), as was also discussed with respect to Chaffinches. Thus, different feeding sites offer the opportunity to study a possible niche shift.

Comparison of temporal presence between mammals and birds
The apparent difference in the circadian occurrence of birds and mammals at feeding sites is due to differences in biology and probably also to differences in management between the two groups. Most birds feed during the day, which is clearly reflected in their circadian distribution at feeding sites, where all birds except owls were observed during the day. In contrast, most mammals were detected at night, a pattern also observed at Black Bear bait sites (Candler et al. 2019). Most mammals detected at feeding sites are game species, and while some are naturally nocturnal, many became nocturnal due to long-term human disturbance and hunting (Russo et al. 1997;Marchand et al. 2014;Hertel et al. 2016). In the Slovenian Dinarics, most game species are hunted at feeding sites, making these sites areas of higher "predation risk" in the "landscape of fear" (Laundré et al. 2010) and possibly even exacerbating nocturnal behaviour at feeding sites. The difference in the circadian occurrence of mammals and birds provides an opportunity to reduce the amount of food at feeding sites, reduce costs and reduce impacts to birds (non-target species) while maintaining all of the desired effects of artificial feeding on game mammals (target species). Although birds were detected on approximately 30% fewer photographs than mammals at feeding sites and their biomass is an order of magnitude lower, birds still occur often and likely consume a significant portion of the available food. Most feeding sites are designed for mammals, not birds. Therefore, if the majority of the food is to reach the desired species, it should be available just after sunset when bird presence is decreasing and mammals can be present in notable numbers.
Although there are some similarities between birds and mammals in terms of monthly occurrence, there are also some important differences. One of the possible reasons for the earlier decline in mammals in autumn could be the hunting season, which begins in September for many game species in Slovenia (Adamič & Jerina 2010). Almost all mammal species detected at feeding sites are game species. The decline of game species at feeding sites during the hunting season has also been documented in other studies (e.g., Candler et al. 2019). On the other hand, of the birds recorded, only the Jay and the Hooded Crow (Corvus cornix) (only one record in the study) are hunted in Slovenia, and although the Jay was the most frequently recorded species, hunting interest in this species is low (up to 4000 individuals culled in the country, compared to more than 40,000 Roe Deer, C. caprelous). The avoidance of feeders by game species because of hunting is likely since preliminary results in mammal temporal distribution suggest that at least some species (e.g., Wild Boar) are less often observed during the day in the hunting season than outside of it. In addition, in years with a good beech mast, a huge quantity of preferred natural food is available for many game species. Beech masting affects the presence of birds at feeders (Chamberlain et al. 2007), but the effect at feeding sites appears to be less pronounced for birds than for game mammals. The peak of bird occurrence in early October also coincides with the peak migration period for many species at feeding sites, e.g., Wood Pigeon, Blackbird, Chaffinch and Buzzard (Billerman et al. 2020).

Conclusions
The study offers valuable insights into the temporal occurrence of birds in natural systems and the impact of artificial food on this phenomenon. For some bird species, artificial food constitutes a significant food source, causing them to adjust their temporal occurrence to the availability of this food, either directly or through an increase in the presence of the prey. For other species the influence is less direct, since they can be attracted to the food offered but their presence is also shaped by the presence of other species. The role of different types of feeding methods is worth studying, particularly for the species demonstrating greater temporal adaptation to artificial feeding. However, for some bird species, feeding sites simply provide a suitable habitat. The study raises several questions. A more in-depth investigation of bird-mammal interactions and avoidance behaviours is necessary, particularly during sunrise and sunset when the overlap is greatest. Additionally, exploring the relationship between the temporal availability of food and bird-mammal relations by manipulating the time of food availability would also be a noteworthy area of study.