Here, we present a novel camera trapping device, the Mostela, developed to detect small mustelids, and tested it during two field seasons in a study site in the Netherlands to quantify its functioning as a monitoring tool. We detected weasels with a daily detection probability between 0.1 and 0.2 during most of the year. Furthermore, we were able to estimate daily activity patterns based on time of detection, determine the sex of the majority of “captured” individuals, and identify some individuals. The Mostela thus showed great promise as a monitoring tool that can both determine presence/absence of species and provide data on activity and sex ratio. Further development of individual identification could allow for (spatially explicit) capture-recapture or mark-resight models to estimate weasel population density. However, we did not detect any stoats in this study, likely due to the absence or very low population density of the species in our study site. Thus, further studies are needed to determine the effectiveness of the method to monitor this species.
Both site use probability and detection probability in this study were highest in summer and the beginning of autumn. This is to be expected, as weasel populations show the highest densities in these seasons due to the addition of juveniles born in spring and summer (King and Powell 2010). Furthermore, land management and farming practices have been described as being of importance for the spatial distribution and behavior of small mustelids, as they influence both habitat features, such as cover, and prey availability (Macdonald et al. 2004; Zub et al. 2008; King and Powell 2010; Mougeot et al. 2020). As most Mostelas in our study were placed in hedgerows that only provide cover during the growing season (Figs. 2 and 3), we speculate that, apart from changes in population density, mowing practices might have influenced the seasonal pattern in site use and detection probabilities we observed. In spring, the vegetation on the pastures surrounding the hedgerows increased in height, supplying both food and cover for voles and cover for weasels. This could lead to increased use of this part of the landscape and reduced use of the hedgerows. After mowing in late spring, the cover disappears and weasels are bound to the hedgerows, leading to higher site use and detection probabilities. Unfortunately, we lacked both sufficient information on mowing dates and deployments in the open part of the landscape to test this hypothesis.
We found a daily detection probability between 0.1 and 0.2 during most months, which is relatively high compared with most camera trapping studies, where detection histories are often combined for multiple days due to low detectability (Burton et al. 2015). Only Evans et al. (2019) found a similar daily detection probability of 0.24 for stoat, but they used both bait and lure. A previous publication using the Mostela had a much lower number of detections, showing that there likely are differences between study areas (Croose and Carter 2019). Croose and Carter (2019) had trapping rates of 1.4 weasels per 100 trap nights and 0.1 stoat per 100 trap nights, which are much lower than the number we obtained for weasels (19.9 and 8.6 in 2017 and 2018, respectively). Similarly, Soininen et al. (2015) obtained a trapping rate of 0.2 stoats per 100 trap nights in the camera trapping device they developed to study small mammals under the snow. Our estimates were more similar to, but still higher than, those obtained by tracking tunnels: 0–6 weasels per 100 trap nights (Graham 2002). Thus, we had high success in detecting weasels, especially considering we did not use bait or lure, which might be caused by a combination of a well-performing method and a study site where weasel activity was concentrated, making it easier to target locations with relatively high use of weasels.
We found that a Mostela equipped with a tracking tube of 10 cm had a 1.6–1.9 times higher detection probability compared with a Mostela equipped with an 8-cm tube. This was in contrast to our expectation, as a larger entrance tube could potentially increase the likelihood of larger carnivores, such as stoats, to enter the box, which could result in weasels avoiding a Mostela. Thus, our result might be dependent on the apparent absence of stoats in our study area. As tubes with a diameter of 10 cm have been used to monitor weasels previously (Graham 2002), we would recommend the use of tubes with a diameter of 10 cm for studies of both weasel and stoat.
In this study, weasels were found to be predominantly active by day, consistent with previous studies (Jedrzejewski et al. 2000; Sundell et al. 2000; Brandt and Lambin 2005; Zub et al. 2013). Jedrzejewski et al. (2000) found similar differences over seasons, with slightly more uniform activity patterns in spring compared with summer, while Brandt and Lambin (2005) found a similar more crepuscular pattern with two peaks around sunrise and sunset in summer. Both these studies were based on radio-tracked weasels while we are, to our knowledge, the first to detect similar activity patterns using non-invasive methods. The difference in activity pattern between spring and summer might be a response to high midday temperatures in summer as weasels adjust their activity to avoid heat stress (Zub et al. 2013).
We were able to identify several individuals based on body size and coat patterns from the video footage recorded with the Mostela. For weasels of the subspecies vulgaris, it was shown previously that individuals can be identified from these characteristics (King 1979; Linn and Day 2009). However, in this study, we could not identify individuals in all videos, mainly due to some issues with lack of contrast in the infrared footage from the Bushnell cameras. The contrast of the images was much better, and the recognition of individuals therefore easier, using the Browning cameras. Thus, it is important to test the contrast of images produced by different camera trap models before using them in a Mostela if individual recognition of individuals is considered important. It is important to state that not all individuals need to be identifiable as the identification of a few individuals with unique spot patterns is sufficient to allow for density estimation using spatially explicit mark-resight models (Royle et al. 2014).
We found a male-biased sex ratio of 65% males in our dataset. The ratio in this study is similar to that in previous findings based on live trapping summarized by Moors (1974), 76.5% and 59% males, and King (1975), 75% males. These male-biased captures are likely caused by (seasonal) differences in home ranges and hunting behavior between the sexes (King and Powell 2010), rather than true sex ratio. However, they could also be caused by sexual differences in seasonal mortality which can result in a male-biased sex ratio during summer (Zub et al. 2011). The identification of individuals and calculation of individual trap rates would enable the correction of differences in capture probability between the sexes which would allow the true sex ratio to be determined.
We captured footage of more than one individual weasel inside the Mostela at the same time on several occasions. A study on stoat and long-tailed weasel (M. frenata) showed that at the end of summer the young leave the den, but apparently travel together in family groups which resulted in captures of several individuals (Rust 1968). Therefore, the recording of multiple individuals at once might have been the result of mothers with young or groups of siblings moving together.
We were able to identify six small mammal species besides weasel. Among these were the pygmy shrew and harvest mouse, which shows that the Mostela can detect even the smallest species of mammal. Thus, the Mostela might also be applicable for non-invasive monitoring of small mammals, other than mustelids, which needs further exploration. We detected all species without the use of a bait or lure. Weasels can be caught without bait if the trap is well placed, because mustelids are naturally curious and investigate any hole or burrow when hunting. Previous studies have shown that bait or lure can increase the effectiveness of trapping of weasels (King and Edgar 1977). The lack of stoats in our study might be a result of the lack of lure as Croose and Carter (2019) detected this species using lure in their Mostelas. However, the use of bait has several drawbacks, such as the potential attraction of larger predators which might reduce weasel presence (King and Powell 2010; Lambin 2018), potential for habituation, and violation of several assumptions often made when analyzing camera trap data (Hofmeester et al. 2019). Nevertheless, rodents using the Mostela likely leave a natural scent lure, which is accentuated by the recording of grass snakes in Mostelas in this study and a study by Croose and Carter (2019) and a recording of a Western whip snake (Hierophis viridiflavus) in the Dordogne, France (J. Mos, personal observation).
Overall, we think that the Mostela is a promising non-invasive tool to study and monitor small mustelids, especially weasels. Furthermore, the relatively easy setup and low associated costs make it a great method to use in volunteer-run monitoring programs. Given the high daily detection probability in July–September, it is sufficient to deploy a Mostela with a 10-cm entrance tube for 2 weeks during that period to determine presence of weasel at a site. However, in early spring and autumn, and potentially in winter when we did not deploy Mostelas, longer deployments might be needed to ensure good presence/absence estimates. Similarly, the study by Croose and Carter (2019) suggests that detection probabilities for stoats are lower than those for weasel, calling for longer trapping periods when monitoring stoats. In general, it would be beneficial to test the Mostela in a larger number of study sites, to test the generality of our results, as well as the functionality of the Mostela to monitor stoats. The additional information that can be gathered with Mostelas, such as the time of detection and the recognition of individuals, will further increase the utility of the data for studies of activity patterns and density estimates. One promising future application might be for estimating density of common weasels based on (spatially explicit) capture-recapture or mark-resight models.