Instance segmentation and tracking of animals in wildlife videos: SWIFT - segmentation with filtering of tracklets

Distribution of Lesions in Red and Fallow Deer Naturally Infected with Mycobacterium bovis

Mesenchymal stem cells positive to surface antigen STRO-1 were isolated from regenerating antlers of red deer (Cervus elaphus) and fallow deer (Dama dama) using a magnetic-activated cell sorting (MACS) method. In this study we analysed factors potentially affecting the number of STRO-1+ cells in the cell cultures. With regard to the STRO-1 antigen, we evaluated data from 188 MACS separation procedures of cell cultures cultivated in Dulbecco’s Modified Eagle Medium and 10% fetal calf serum of four fallow deer males (130 procedures) and four red deer males (58 procedures). The analysed factors were the sampling site of the antler or the pedicle, cell passage and type of the cell culture (mixed or STRO-1 negative cell cultures). The percentage of obtained STRO-1+ cells varied greatly from 0.4 to 38.9% for fallow deer and from 1.8 to 16.5% for red deer. We have not found any significant influence of the sampling site. The passage and the type of culture were significant factors for both fallow and red deer cells. The highest numbers of STRO-1+ cells were obtained from the second passage from both fallow and red deer cell cultures (24.6 and 5.5%, respectively). Our experiment revealed that we can maximise the number of STRO-1+ cells in the cultures by manipulating the cultivation factors.

Correspondence has appeared in the literature concerning the relationships between red deer (Cervus elaphus), roe deer (Capreolus capreolus), fallow deer (Dama dama) and some other species, such as Japanese deer (Cervus nippon), which are feral in the British Isles. Competition, described as antipathy or antagonism by various writers, appears to be the most widely accepted mechanism whereby relative abundance of one deer species is often correlated with relative scarcity of others. Other explanations involve geographical limits of distribution, two examples being the relative abundance of fallow deer in the south of England, and of roe deer in the Lake District (Delap 1955, Carne 1955). Carne (1954) has stated that Japanese (syn. sika) and fallow deer are, having been introduced into Britain, a powerful deterrent to the roe; he claims support for this contention from historical records which he considers show that as fallow deer have increased in an area, roe deer have decreased, often to extinction. Having observed in the Lake District of England that fallow deer have never established themselves in forests occupied by roe deer, Carne confirmed (1955) a statement by Delap (1955) that there is antipathy between them. Delap also stated that roe deer and red deer or red deer and fallow deer can 'associate happily', but that antagonism exists between roe deer and fallow deer. However, even co-existence of red deer and roe deer does not appear to be a constant feature; Wildash (1951) records that roe deer are seldom found where red deer are present in any numbers in Austria, that the red deer is a 'natural enemy' of the roe, and also that Austrian foresters could not give any reason for this antipathy. The purpose of this paper is to examine these ideas in the light of further evidence. It will be shown that roe, red and fallow deer are characteristically most numerous at different stages in the development of several British forests; when two or more of these three species are present in the same forest area, the number of each of them increases and decreases in a successional manner, which is related to development of the forest habitat towards a climax of mature trees. Such successional changes of the deer fauna are considered to be similar in nature to the marked changes which were shown to occur to the species composition of the bird fauna when a heathland area in East Anglia was afforested (Lack & Lack 1951).

We conducted a series of studies in the Dobris Forest, Czech Republic, to determine whether competition between white-tailed deer (Odocoileus virginianus) and sympatric cervid species could limit expansion of the white-tail population. We used grazing time among species as an indication of potential interspecific competition and predicted that grazing time on an open pasture would decrease if deer other than conspecifics were present or entered the pasture. We used 710 observational records in our analyses: 409 for white-tailed deer, 225 for fallow deer (Dama dama), 66 for red deer (Cervus elaphus), and 10 for roe deer (Capreolus capreolus). The time that white-tailed, fallow, and roe deer spent grazing was not affected by whether the pasture was harvested or whether any deer were present when the focal animal entered. We observed a strong tendency for grazing time by red deer to be directly related to the numbers of whitetails or female fallow deer visible before red deer entered the pasture, Grazing time of white-tailed and fallow deer increased if the focal deer was joined by another animal. For white-tailed and fallow deer, the anti-predatory strategy of joining a group may have been balanced by avoiding the most competitive classes of animals. Our results suggest that interspecific cooperative behavior, rather than interspecific competition, occurs among these sympatric cervids.

Integrating the interaction between cell wall components through a lignin and neutral detergent fibre ratio as a cell wall digestibility (CWD) index may provide a suitable method for diet quality and fibre digestibility estimation. Predictive faecal-Fourier-transform (FT)-near infrared reflectance spectroscopy (NIRS) equations useful for CWD determination were developed as an integrative assessment of the digestibility and quality of the diet selected by red (Cervus elaphus hispanicus) and fallow (Dama dama) deer in southern Spain. A chemometric regression was used from spectra of faeces and wet chemical rumen contents analysis as reference data coupled with partial least-squares. The accuracy and feasibility of the equations obtained for red and fallow deer samples separately were evaluated, as well as after pooling all samples for both species. The predictive equations had a high linearity with correlation coefficients (r) ranging between 0.89 and 0.99, as well as a reliable accuracy considering the errors of prediction (RMSEP 0.57–0.87), calibration (RMS 0.04–0.45) and cross-calibration (RMSECV 0.80–1.12) in relation to the range of values for which the NIRS calibration was set for each parameter, giving very good RER (residual error value) indices (7.44–13.23). The ability of the calibration equations to predict new samples from the same deer populations was also assessed through an external validation. Equations obtained for the red deer data were successfully applied not only to red deer but also to fallow deer, and vice versa. Feasible CWD simultaneous determination in red and fallow deer was also obtained by using the broad-based equations. There was variation of the CWD index throughout the year and differences between the two deer species. A lower CWD of the forage was recorded in red deer, while a lower CWD ratio was detected at the end of winter coinciding with the highest browsing levels in both species. This technique has potential to enable closer examination of the non-grass variables and distinguish between browse and herbage feeding behaviour. The faecal-NIRS method of CWD ratio determination is a useful, non-invasive tool for monitoring variations in the diet quality of Mediterranean deer in their natural environment.

Behavioral analysis of animals in the wild plays an important role for ecological research and conservation and has been mostly performed by researchers. We introduce an action detection approach that automates this process by detecting animals and performing action recognition on the detected animals in camera trap videos. Our action detection approach is based on SWIFT (segmentation with filtering of tracklets), which we have already shown to successfully detect and track animals in wildlife videos, and MAROON (mask-guided action recognition), an action recognition network that we are introducing here. The basic ideas of MAROON are the exploitation of the instance masks detected by SWIFT and a triple-stream network. The instance masks enable more accurate action recognition, especially if multiple animals appear in a video at the same time. The triple-stream approach extracts features for the motion and appearance of the animal. We evaluate the quality of our action recognition on two self-generated datasets, from an animal enclosure and from the wild. These datasets contain videos of red deer, fallow deer and roe deer, recorded both during the day and night. MAROON improves the action recognition accuracy compared to other state-of-the-art approaches by an average of 10 percentage points on all analyzed datasets and achieves an accuracy of 69.16% on the Rolandseck Daylight dataset, in which 11 different action classes occur. Our action detection system makes it possible todrasticallyreduce the manual work of ecologists and at the same time gain new insights through standardized results.

BackgroundHunting constitutes an important industry in Europe. However, data on the prevalence of vector-borne bacteria in large game animal species are lacking from several countries. Blood or spleen samples (239 and 270, respectively) were taken from red, fallow and roe deer, as well as from water buffaloes, mouflons and wild boars in Hungary, followed by DNA extraction and molecular analyses for Anaplasma phagocytophilum, haemoplasmas and rickettsiae.ResultsBased on blood samples, the prevalence rate of A. phagocytophilum infection was significantly higher in red deer (97.9%) than in fallow deer (72.7%) and roe deer (60%), and in all these compared to mouflons (6.3%). In addition, 39.2% of the spleen samples from wild boars were PCR positive for A. phagocytophilum, but none of the buffalos. Based on blood samples, the prevalence rates of both Mycoplasma wenyonii (Mw) and ‘Candidatus M. haemobos’ (CMh) infections were significantly higher in buffaloes (Mw: 91.2%; CMh: 73.3%) than in red deer (Mw: 64.6%; CMh: 45.8%), and in both of them compared to fallow deer (Mw: 30.3%; CMh: 9.1%) and roe deer (Mw: 20%; CMh: 1.5%). The prevalence of Mw and CMh infection significantly correlated with the body sizes of these hosts. Furthermore, Mw was significantly more prevalent than CMh in buffaloes, red and roe deer. Mycoplasma ovis was detected in mouflons, M. suis in wild boars, R. helvetica in one fallow deer and one mouflon, and an unidentified Rickettsia sp. in a fallow deer.ConclusionsForest-dwelling game animal species were found to be important carriers of A. phagocytophilum. In contrast, animals grazing grassland (i.e. buffaloes) were less likely to get infected with this Ixodes ricinus-borne pathogen. Water buffaloes, deer species, mouflons and wild boars harbored haemoplasmas that may affect domestic ungulates. Evaluated animals with larger body size had significantly higher prevalence of infection with haemoplasmas compared to smaller deer species. The above host species rarely carried rickettsiae.

The wild population of fallow deer in Central Europe has grown considerably over the last decade. However, information on feeding habits of this alien species in relation to the indigenous red deer or roe deer, in areas of their co-occurrence, is scarce. A prevailing view maintains that their food-niches are distinct, although direct comparative studies have not been carried out. Therefore, the aim of the research was to compare the diets of fallow, red, and roe deer feeding in the same habitat. Research was based on the rumen contents of 242 animals hunted in the autumn-winter season in the forests of Southern Poland. The analyses demonstrated that fallow deer are moderate grazers in such conditions and eat more graminoids in comparison to red or roe deer (36.4% vs. 16.1% or 5.5%, respectively). On the other hand, it feeds on less browse (17.2% vs. 41.4%) or dwarf shrubs (8.4% vs. 19.0%) than red deer, and on less bramble (10.9% vs. 34.6%) or forbs (4.0% vs. 7.6%) in comparison to roe deer (P=0.05). Although the diets of the three deer species differ in terms of the proportion of each food type in their diet, overlapping of their food-niches is high (52.6%).

We studied the diet composition and diet overlap in sympatric red deer (Cervus elaphus hispanicus) and fallow deer (Dama dama) throughout a whole year in order to determine variation due to season, species, sex and age class by analysing rumen contents samples of 81 red and 69 fallow deer shot monthly during 2008–09 in Sierra de Andújar Natural Park, southern Spain. We assessed diet similarity and possible inter- and intra-specific foraging competition. We found different foraging strategies for both species and sexes during constraint periods, and several theoretical considerations of specific interactions and behaviour are discussed with respect to the Mediterranean environment. In both species an annual diet dominated by grasses was recorded, peaking in spring. Browses were an important food resource at the end of winter and at the end of summer, and fruit more in autumn and winter. Red deer ingested a higher proportion of browse than fallow deer, which consumed more acorns and for a longer time showed a better ability to compensate for nutritional constraint periods. An overall decline in diet similarity in summer and at the end of winter led us to assume that exploitative competition between red and fallow deer and even between sexes was probable. Red deer females showed low diet similarity to other deer, while there was a great diet overlap between red deer males and fallow deer females at the end of summer. Differences detected between both two species and sexes do not always support predictions deriving from specific body size and morpho-physiological characteristics, but can probably be explained as a consequence of different metabolic demands. The relationship between plant nutritional attributes and food selection according to reproductive or physiological status and seasonal demands for both sexes and species should be researched in order to perform a better assessment of deer feeding behaviour.

Manipulation of reproductive seasonality of farmed red deer ( Cervus elaphus) and fallow deer ( Dama dama) by strategic administration of exogenous melatonin

Two experiments were conducted to compare digestibility of 12 diets in sheep, red and fallow deer. No differences (p>0.05) between sheep, red and fallow deer in digestibility of dry matter, organic matter and digestible energy content for all diets were found except for the sorghum diet and medic hay. Sheep and fallow deer digested the sorghum diet better than red deer. An in vitro study showed that sheep had a lower in vitro dry matter digestibility and digestible energy content than both red and fallow deer, with a significant interaction between animal species and feed ingredient. Deer digested straws and hays better (p<0.05) than sheep. In vitro digestibility was lower (p<0.05) than in vivo digestibility, but significantly correlated with in vivo digestibility for red and fallow deer. The in vitro method for digestibility estimation has potential as a rapid feed evaluation system for deer, but needs further validation. (Asian-Aust. J. Anim. Sci. 2002. Vol 15, No. 6 : 800-805)

Wild deer exert strong top–down control on forest composition by browsing on palatable trees, and these effects are exacerbated as red, fallow, and roe deer populations increase in northern temperate forests. However, the relationship between deer abundance and plant recruitment remains poorly documented. Here, we combined camera trap and vegetation plot data to quantify the shape of the relationship between habitat utilization by deer (red, fallow and roe deer) with different components of the forest understory in ten sites distributed across a temperate mixed forest in the Veluwe, the Netherlands. The list of forest attributes included the density, richness and diversity of saplings, the proportion of conifers to broadleaves, the number of browsed broadleaves and conifers, the forest basal area, the understory cover of shrubs, moss and bare soil and the depth of the litter layer. When applying General Linear Mixed Models to those camera trap data, six of the eleven forest variables were related to the utilization level (UL) by deer. With increasing UL by fallow deer, there was a decrease in sapling species richness (β = −0.26). With increasing UL by red deer, there was a decrease in litter depth (β = −0.14) and an increase in browsed broadleaf stems (β = 0.40). With increasing UL by roe deer, there was a decrease in species richness of sapling plants (β = −0.26), Shannon diversity (β = −0.11) and shrub cover (β = −0.36), whereas there was an increase in stem density (β = 0.06). When combining all deer species into one guild, a negative relation was found between UL by deer and sapling richness (β = −0.21), diversity (β = −0.09) and litter depth (β = −0.14) and a positive relation with sampling density (β = 0.04). The relationship between UL by deer and different forest attributes followed a linear log10 shape. When back transforming, the response was curvilinear with large changes at a low UL and small changes when the UL was high. Yet, the exact shape of the curve varied according to (a) biotic factors specific to each study location. The conservation of temperate forests can benefit from limiting the size of deer populations and the time deer spend browsing in forest patches with high biodiversity value.

Oestrous synchronization, semen collection and artificial insemination of farmed red deer ( Cervus elaphus) and fallow deer ( Dama dama)

ABSTRACTAcross the northern hemisphere, ungulates are expanding in range and abundance, forming novel communities in increasingly human‐modified landscapes. These shifts drive new interactions over available food resources, but patterns of resource use and partitioning in Europe's multi‐species systems remain poorly understood. This study examined seasonal diets and resource partitioning in diverse cervid communities (moose, roe deer, red deer, and fallow deer) across two Swedish landscapes (coastal‐boreal and boreo‐nemoral) differing in deer density and land use. Based on their foraging strategies, we expected (Hypothesis 1) diet richness and dietary niche width to be greater in intermediate feeders (red and fallow deer) than in browsers (moose and roe deer), (Hypothesis 2) trophic partitioning between browsers and intermediate feeders to be driven mainly by graminoid use, and (Hypothesis 3) intra‐ and interspecific overlap to vary with season, deer density, habitat diversity, and proportion of arable land. DNA metabarcoding of 2568 fecal samples showed that deer consumed plants from over 70 families, though diets were typically dominated by fewer than 10. Vaccinium shrubs were key forages year‐round, while birch and willow dominated during the growing season. Moose consumed large amounts of pine in spring and winter (> 50% in the boreo‐nemoral, 35%–40% in the coastal‐boreal landscape), with less during summer‐autumn (~15%). Forbs were important for smaller deer, especially in spring and summer‐autumn, and more heavily used in winter in the boreo‐nemoral landscape, likely due to supplementary feeding with human‐provided food like hay or silage. Spruce use was low overall (< 5%), with fallow deer showing the highest intake. Consistent with Hypothesis 1, diet richness and niche width increased from moose to fallow deer. In partial support of Hypothesis 2, principal coordinates analysis (PCoA) revealed that graminoids contributed to trophic partitioning, but the pattern was not a strict browser–intermediate feeder divide. Moose consistently separated from the smaller deer due to avoidance of graminoids and reliance on pine and juniper, while roe deer, although a browser, sometimes overlapped with red and fallow deer through greater use of graminoids. During winter in the coastal‐boreal landscape, wavy hairgrass (Avenella flexuosa) contributed to the significant separation between browsing roe deer and intermediate‐feeding red deer diets, consistent with Hypothesis 2. Diet overlap among smaller deer varied with season and landscape. Intraspecific overlap was the highest in moose and the lowest in fallow deer, declining during summer–autumn across species. Overlap was influenced by deer density, habitat diversity, and arable land, consistent with Hypothesis 3, but effects were species‐specific and explained only limited variation. Our results highlight the dietary plasticity of red and fallow deer, which may intensify resource competition with moose and roe deer in multi‐species systems, particularly where supplementary feeding is common. These insights support adaptive, multi‐species management of deer in northern ecosystems.

Polymorphic SNPs were identified using BovineSNP50 BeadChip in three groups of cervids: farmed Red deer ( n = 3), and free range Red deer ( n = 5) and Fallow deer ( n = 2). From the total of 54 609 SNPs, 53.85% could be genotyped. Out of 28 502 successfully genotyped autosomal SNPs only 5.3% were polymorphic. The average minor allele frequency within cervids was 0.23 (number of polymorphic SNPs ranged from 467 to 686). Results of the molecular variance analysis showed that 67.38% of variation occurred within individuals and the rest was explained by a species difference ( F ST = 0.32). The value of F IT (0.33) indicated a higher proportion of homozygote genotypes in the analyzed dataset. Pairwise F ST values showed very clearly the genetic differentia - tion between Red and Fallow deer which ranged from 0.06 (farmed and free range deer) to 0.74 (farmed Red and Fallow deer). A similar result was found for Nei's genetic distances that ranged from 0.01 (among Fallow deer) to 0.79 (among farmed Red and Fallow deer). The genetic differentiation of the analyzed cervid species was evalu - ated also by the principal component analysis with the involvement of 6 other species from the family Cervidae, which showed a division of the Cervidae cluster into 7 subpopulations. The panels of SNPs primarily produced for a model species are becoming the marker of choice for the application in other species, but the best methods of their discovery, validation, and genotyping in non-model species need further investigations.