Carnivoran populations have historically decreased globally due to overexploitation, persecution, and habitat loss (Ripple et al. 2014). Nonetheless, following legislative protections, improved logging practices, and active management some species have rebounded across Europe (Chapron et al. 2014) and North America (Linnell et al. 2001). In humandominated landscapes, ecological generalists are more likely to recover and persist (Erwin 1998, McKinney and Lockwood 1999), while specialists are particularly vulnerable to extinction (Boyles and Storm 2007). The fisher Pekania pennanti is a forest specialist mesocarnivore associated with northern forests in North America (Powell et al. 2017). Following habitat loss and overexploitation in the 1800s, the range of fishers contracted considerably (Powell 1993). With trapping closures and reintroduction efforts, fisher populations recovered throughout many parts of their historical range in eastern and central North America (Lewis et al. 2012), possibly due, in part, to fishers' diet flexibility (LaPoint et al. 2015).

Although forest specialists, fishers have the flexibility to be dietary generalists, consuming small and mid-sized mammals, birds, fruits and carrion (Powell 1993, Powell et al. 2017). Where their ranges overlap, snowshoe hares Lepus americanus are fishers' most common diet item (Powell and Zielinski 1994). Additionally, fishers possess the uncommon ability to prey efficiently on North American porcupines Erethizon dorsatum when available (Powell 1993, Pokallus and Pauli 2015). The availability of either of these important prey species can lead to stable fisher populations (Bulmer 1974, Bowman et al. 2006). Squirrels and small mammals appear to be used as important alternative prey when preferred prey are scarce (Kuehn 1989, Martin 1994, Bowman et al. 2006, Jensen et al. 2012). Prey consumption varies markedly across their range and fishers in the Great Lakes region generally consume more large-bodied prey than eastern or western populations (LaPoint et al. 2015).

American martens Martes americana, a close relative, can be sympatric with fishers and share many dietary items (Buskirk and Ruggiero 1994, Zielinski and Duncan 2004). Recent research suggests that martens in northern Wisconsin are consuming fewer optimal prey items, relying more on smaller prey like shrews, Blarina brevicauda and Sorex spp., as well as prey like deer carrion that is likely riskier due to competition with other predators (Carlson et al. 2014). This finding may be evidence that the small prey community in northern Wisconsin is changing, resulting in different prey availability and enhanced interspecific competition. Unlike martens, fishers exploit porcupines (Bowman et al. 2006). In fact, they are energetically more efficient when consuming larger-bodied prey (Powell 1979) and may, then, be more resilient to a declining small prey base with minimal effects on their body condition (Kuehn 1989, Helldin 1999) and survival. As active predators, fishers and other mustelids rely on adequate fat depots as an important energy source (Leonard 1980, Buskirk and Harlow 1989, Robitaille and Cobb 2003). Such fat depots are critical for survival and can be used as an index of nutritional condition reflecting rates of energy intake (DeCalesta et al. 1975, Schulte-Hostedde et al. 2005). The capacity to store adipose tissue varies widely within the weasel family — American martens are lean (2.4— 5.6% body fat, Buskirk and Harlow 1989), whereas fishers can carry larger reserves (2—32% body fat, Garant and Crête 1999). Across mustelids, within a species, a comparatively low percent body fat can indicate an animal is in poor condition (Mustonen et al. 2015). Fat dynamics can vary from a variety of factors including prey availability (Robitaille and Jensen 2005). For example, Kuehn (1989) found that the amount of fat in fishers from Minnesota increased as deer harvest levels increased, indicating the potential importance of carrion to fishers.

Although historically fishers were distributed across the Great Lakes region, extensive logging and high harvest pressure resulted in severe range contraction and fishers were extirpated in Wisconsin and Michigan by the 1930s (Williams et al. 2007, Lewis et al. 2012). Reintroductions in the 1950s, coupled with natural recolonization, resulted in the recovery of fishers to the region so that regulated harvests were again implemented in Michigan, Minnesota, and Wisconsin (Berg and Kuehn 1994, Williams et al. 2007, Lewis et al. 2012). Nevertheless, recent harvest and winter track survey data (2002—2013) suggest that the fisher population in northern Wisconsin is again decreasing (Rolley et al. 2013). Research has also documented a northward shift regionally of the southern boundary of snowshoe hares (Sultaire et al. 2016), as well as many small mammals associated with northern forests (Myers et al. 2009). We hypothesized that fishers have sub-optimal prey resources and we investigated limited energy intake as a possible mechanism of their population decrease. We quantified fisher diet in northern Wisconsin and contrasted individual body condition. We estimated diet components of the contemporary population via stable isotope analyses (δ¹³C and δ¹⁵N) and compared body condition between contemporary, historical and regional populations of fishers via analysis of fat depots.

Material and methods

Stable isotope preparation and analysis

Stable isotope analysis has become a useful technique in wildlife foraging studies because not only can diet contributions be estimated from consumer tissues such as hair, it also avoids biases of digestibility associated with traditional diet reconstruction methods (Ben-David and Flaherty 2012). We prepared all fisher and diet samples using standard techniques described in Pauli et al. (2009). We provide results as ratios in parts per thousand relative to international standards, Vienna-Peedee Belemnite and atmospheric nitrogen. We corrected for trophic discrimination with values established by Roth and Hobson (2000) for red foxes Vulpes vulpes, 2.6‰ for ¹³C and 3.4‰o for ¹⁵N.

Isotopic analysis is only useful when diet groups are distinct (Ben-David and Flaherty 2012); thus, we first determined which prey could be distinguished isotopically from one another. We used a K nearest-neighbor randomization to define isotopically distinct prey groups (all pairwise p < 0.05; Rosing et al. 1998) as: 1) shrews, mice, squirrels, and porcupines; 2) deer, grouse, and snowshoe hares; 3) nuts and berries; and 4) other birds. To estimate the proportional importance of each prey group to fisher diet we used Bayesian mixing models in the package ‘stable isotope analysis’ in R ver. 4 (SIAR; Parnell et al. 2010). We ran uninformed models, parameterized with uniform priors (evenly distributed among prey groups), and incorporating trophic discrimination and concentration dependence. We expressed outputs as means of probability density functions, representing each prey group's probable contribution to diet (Parnell et al. 2010). We further compared raw isotope values between sexes using a Welch's t-test, and across ages with linear regression. We also report fisher diets from Ontario, Michigan and Minnesota. Ontario (n = 270, Clem 1977; n = 57, deVos 1952) and Michigan (n = 35, Powell 1977) fisher diets were reported as percent occurrence of each prey type in GI tracts or scats. Minnesota fisher diets (n = 1649, Kuehn 1989) were reported as the aggregate percentage of total prey weight in stomachs. We grouped prey type into similar categories as identified in this study, and due to differences in diet reconstruction methods among studies, we present a qualitative, rather than quantitative comparison.

Figure 1.

Isotopic signatures of fisher Pekania pennanti hair (circles) collected during 2013 and signatures for prey groups (mean ± 1 SD) from northern Wisconsin, USA. Fisher samples were corrected with trophic discrimination factors of 3.4‰o for δ¹⁵N and 2.6‰ δ¹³C.

Body condition analysis

To assess the body condition of fishers and to explore its relationship to diet, we compared the condition of a contemporary population to that of a historical population and two regional populations. We weighed thawed fresh mass of omentum fat samples for contemporary fishers from northern Wisconsin, and estimated percent body fat for each individual using a previously developed model (Robitaille and Jensen 2005). Although Robitaille and Jensen (2005) examined percent omentum fat for both males and females, only the male-specific model was significant. Additionally, because females are constrained by their smaller structural size, males are more likely to experience changes in body fat with different food availabilities (Powell 1979, Holmes and Powell 1994). Consequently, we only considered males in subsequent analyses. We then compared these mean estimates of percent fat to fishers collected from Wisconsin during the 1990s (Gilbert 2000) and to fishers collected from two regional populations in Canada (Garant and Crête 1999, Robitaille and Jensen 2005). All fishers were sampled mid-fall through the winter during trapping seasons. Though the body condition methods differed among studies, Robitaille and Jensen (2005) demonstrated that different fat depots (e.g. omentum, kidney) are reliable estimates of percent body fat for individuals. Consequently, we used each study's estimated percent body fat for comparisons among populations. Gilbert (2000) derived percent body fat via kidney fat depots for 250 male fishers collected throughout their range in Wisconsin in 1992. To maintain similar sample sizes in our comparison, we randomly subsampled 30 individuals from this historical Wisconsin population. Garant and Crête (1999) and Robitaille and Jensen (2005) calculated percent body fat from whole carcasses (Québec 1991–1992, n = 34, Garant and Crête 1999; Ontario 1998–2000, n = 86, Robitaille and Jensen 2005). Finally, to examine whether diet is related to nutritional condition, we compared nitrogen isotopic signature and percent body fat within the contemporary fisher population using linear regression.

Table 1.

Mean proportion of assimilated diet groups (95% credible intervals) in fishers Pekania pennanti in Wisconsin, fall 2013, estimated via uninformed Bayesian mixing models. Also shown are summarized fisher diet studies from Ontario (ON), Michigan (MI) and Minnesota (MN), reported by Powell 1993. Estimates are grouped into same diet categories as the contemporary study and shown as proportion of GI tracts or scat containing each prey type for the Michigan and Ontario studies, and proportion of total weight of prey from stomach contents for the Minnesota study.

Results

Fisher hair exhibited great variation in isotopic signatures (Fig. 1), though males and females did not differ in either δ¹³C (t_42.2 = 0.63, p = 0.53) or δ¹⁵N (t_31.4= 1.14, p = 0.26). We found no effect of age on δ¹⁵N (F_1,43 = 0.63, p = 0.43), but a slight positive relationship between age and 5¹³C (ß = 0.26, F_1,43 = 5.47, p = 0.02). We found no differences in overall diet between sexes or age classes, so we present the population-wide diet. Mixing models revealed that the shrew, mouse, squirrel, and porcupine group represented the highest mean proportion of fisher diet, accounting for 63% of assimilated diet. Deer carrion, grouse, and hares made up the second highest proportion of diet (25%) (Table 1). Past studies from Michigan, Minnesota and Ontario indicated higher reliances of fishers on deer, hares and grouse compared to our estimates from Wisconsin.

Contemporary Wisconsin male fishers (n = 28) averaged 6.6% body fat (± 0.78, 95% CI), which was less than the average for male fishers observed 20 years ago in Wisconsin (=11.7% ± 2.40); and for fishers in Québec (=11.4% ± 0.40) and Ontario (=8.4% ± 0.63; Fig. 2). The 95% confidence intervals for the contemporary population did not overlap with confidence intervals for the other populations. Additionally, we did not detect a relationship between percent body fat and δ¹⁵N in contemporary Wisconsin fishers (F_1,24 = 0.05, p = 0.82).

Figure 2.

Mean percent body fat (± 95% CI) of four male fisher Pekania pennanti populations (Wisconsin 2013, n = 28; Wisconsin 1992, n = 30; Ontario 1998–2000, n = 86; Québec 1991–1992, n = 34).

Discussion

Our isotopic analyses revealed that diet across individual fishers in northern Wisconsin was variable during autumn and early winter 2013, as would be expected of a generalist consumer. For the fisher population, small mammals (shrews, mice, squirrels) and porcupines made up the most used diet group, constituting the highest proportion of assimilated diet. The proportion of diet estimated for larger-bodied prey (deer, grouse, hares) has a substantially lower mean, but the credible interval overlaps with that of smaller-bodied prey, indicating uncertainty in the diet reconstruction. Thus, fishers from Wisconsin in 2013 appear to be consuming similar diets to those found in other systems, where the breadth of prey eaten by fishers was great and commonly consisted of medium to small rodents, porcupines, and hares (Powell 1993). Fishers do not appear to be relying heavily on carrion, despite a large deer population (>25 000 deer harvested each year in northern Wisconsin). This is somewhat surprising considering recent work suggested martens, the smaller relative of fishers, were consuming large quantities of deer carrion in the same area (Carlson et al. 2014).

The average percent body fat of male fishers in 2013 was lower than those from two decades ago as well as compared to fishers sampled in Canada. The lowered fat reserves would be consistent with using more small-bodied prey. Unfortunately, we are unable to separate isotopically either hares from deer or porcupines from the small mammals group to determine how much of the assimilated diet is composed of each of these prey items individually. Because the isotopic signatures of the prey species overlap in this study, traditional diet reconstruction techniques could aid in further distinguishing the contribution of important prey species like porcupines to fishers' diet.

Historical harvest records suggest that fishers were less common in northern Wisconsin than in the hardwood timber stands of central and southern Wisconsin (Schorger 1942). Thus, prior to their extirpation, fishers appear not to have been particularly abundant in the northernmost regions of Wisconsin. After reintroduction, the apparent rapid growth in the fisher population could have resulted from abundant naive prey in regenerating forests coupled with low density of competitors (Fryxell et al. 1999). The recent harvest and winter track survey data suggest that the fisher population in northern Wisconsin may again be decreasing (Rolley et al. 2013). Our study demonstrates that this population is also in lower body condition than previously reported, and that despite a highly variable diet like other populations, these fishers may be consuming fewer large-bodied prey than expected. Although our single year dataset cannot disentangle the mechanisms of the purported decline in this population, our estimates of low body fat and a diet consisting largely of small-bodied prey suggests that nutritional energetics could be a contributing factor, and warrants future monitoring.