American marten (Martes americana) COSEWIC assessment and status report: chapter 8

• Previous page
• Table of contents
• Next page

Population Sizes and Trends

It is difficult to accurately estimate the population size of marten in Newfoundland, due to limited data from field studies and inadequate or outdated habitat suitability mapping. Population estimation is further compounded by small, widely dispersed and, possibly, isolated sub-populations, and variable habitat quality across the population range. Snyder and Hancock (1985) estimated the population at 630 to 875 marten, based on densities derived from live-trapping studies, distribution determined from a trapper questionnaire, and locations of sightings and accidental mortalities. They delineated high and low density areas in western Newfoundland, and assumed no population in the Terra Nova area despite translocations in 1982-83 (Slough 1994). The total area of occupancy was assumed to be 13,354 km², of which 4,551 km² was a high density area.

Bissonette et al (1988) estimated 150 marten in the 790 km² Environmental Assessment Area based on areas of home ranges that contained suitable habitat of 7 resident marten (6.64 km² for females and 9.19 km² for males) and the area of available habitat (561 km² of mature softwood and mixed-wood).

I. Thompson (pers. comm., reported in Forsey et al. 1995) estimated the Newfoundland population to be 300 marten, based on densities estimated by Bissonette et al. (1988), a habitat supply analysis and known distribution (600 km² of prime marten habitat, including Terra Nova). This estimate was very close to the minimum viable population estimate of 237 (Thompson and Harestad 1994).

Marten reintroductions to Terra Nova National Park in the 1980s established a population of 25-30 individuals (Gosse et al. 2005). Home range estimates of 29.5 km² for males and 15.2 km² for females are considerably larger than elsewhere in North America, reflecting the low diversity and abundance of prey (Gosse et al. 2005) and naturally fragmented landscape (Hearn et al. 2005). Fuller et al. (2006) report home ranges from Newfoundland marten of 30.8 km² and 12.8 km² for males and females, respectively. Marten demography and habitat use were studied in southwest Newfoundland between 1995 and 2000 (Hearn et al. 2005). One hundred and sixty-eight individual marten, including 97 adults, were captured in a 2,278 km² area over a 5-year period, providing a minimum estimate of population status in that area.

I. Schmelzer (pers. comm., 2007) combined information on marten distribution, a range of area-specific densities and probability of occupancy in different habitat types to estimate a range of population sizes. Distribution was based on live-trapping studies, accidental captures, radio telemetry, bait stations, and verified incidental sightings collected between 1990 and 2006. A small number of locations distal from core and peripheral areas were excluded as cases of possible vagrancy. Marten locations were classified as either ‘adult’, ‘juvenile’ or ‘unknown’ (Figures 2 and figure3). Core and peripheral areas of occurrence based on verified marten locations 1990-2006 were delineated, where core areas enclosed regions occupied by animals confirmed as adult resident animals and peripheral areas contained juveniles or animals of unknown age, likely transients. Densities were calculated for the three largest core areas (95% of the total core area) by dividing the effective trapping area (ETA) for each region into the number of unique animals captured plus the number of radio-collared animals relocated within the area. Two separate density values were calculated for each core area: the first, a ‘high density’ value was based on an estimate of population size within the ETA in 2006, and the second, a ‘mean density’ value was based on the average density over all years of trapping at each site. Mean densities ranged from 0.04 to 0.08 marten/km² and high densities from 0.09 to 0.14. Densities in the peripheral areas were assumed to be 20% of those in the core areas, or 0.016 marten/km². It was assumed that densities in study areas were representative of those on the landscape, and that they could be extrapolated across each core area.

Habitat was modelled after Fuller et al. (2006) using the habitat classes identified as ‘suitable’ in Hearn et al (2005).  This model is based on an analysis which suggests that the most important predictor of marten occurrence was the amount of suitable habitat within a potential home range.  Four occupancy thresholds ranging from 60% to 89% were chosen on the basis of observed levels of suitable habitat in studied marten, model sensitivity and specificity, and to capture a large (85%) proportion of potential marten home ranges. This model assumed no gender-based differences in habitat use, and that a dichotomous (‘used’ versus ‘not-used’) description of habitat was adequate in determining the probability of occupancy.

Given the positive relationship between probability of occupancy and the amount of suitable habitat within a potential home range (Fuller et al. 2006), it would be unrealistic to assume a constant rate of occupancy for areas with varying amounts of suitable habitat types. Therefore, population size was estimated as the sum of areas (associated with each habitat occupancy threshold) X density for ‘mean’ and ‘high’ density values for each core area and the peripheral area, multiplied by the probability of occupancy. Probability of occupancy was determined using two methods.

One method was the model-predicted Probability of Occupancy (POA) associated with each occupancy threshold (Habitat area for each of 4 occupancy thresholds from Fuller et al. (2006) X Marten density (mean and high values, respectively) X Probability of occupancy for each of the 4 probability thresholds). The estimate from this method was 561 to 852 marten (Table 1a, Probability of Occupancy).

The second method used the empirical relationship expressed in the cumulative occupancy rate of all home ranges versus the percent suitable habitat within the home range of all marten included in the study (Cumulative Decay Function) (Area X Density as above, but probability of occupancy determined by the proportion of marten retained as suitable habitat declines to the level associated with each of 4 occupancy thresholds. The estimate from this method was 438 to 661 marten (Table 1b, Cumulative Decay Function).

The estimates, while not directly comparable to those made by Snyder and Hancock (1985) and I. Thompson (pers. comm., reported in Forsey et al. 1995), are based on better information and should be more accurate. Expected marten range is more broadly defined and more extensive than in both 1985 and 1995. This suggests that gains in distribution (equivalent to 10,000 km²) have occurred in the Main River and Terra Nova areas. The sub-population of the Main River area in particular is larger and 75% of this core area contains habitat with a probability of occupancy exceeding 78%. Future surveys are planned in the peripheral areas to determine actual densities, and possible extensions to the known core areas across a range of land use activities (J. Brazil, pers. comm., 2007). The high density estimates are probably overestimates for non-protected areas where hare snaring and trapping still occur.

The population estimates include all age classes.  Non-reproducing juveniles (12 months of age and younger) make up approximately 22% of the population.  Assuming an equal sex ratio and pregnancy rate of 60% for females in their second year and 100% for older females, then the lower estimate of 438 marten would include about 150 breeding females, distributed among the 4 main core areas.  The estimate of ‘mature’ marten (reproducing females and all males older than 12 months) would range from 320 to 622.

Earlier efforts to model Newfoundland marten extinction probability based on the old-growth paradigm are no longer valid (Thompson 1991, Schneider and Yodzis 1994, Schneider 1995, 1997). Additive mortality from hare snaring, however, continues to prevent marten from colonizing unoccupied habitats.

Fluctuations and trends

A long-term population decline probably began in the mid-19^th to early 20^th century (Bergerud 1969; Snyder 1985; J. Brazil, pers. comm., 2006, summary of historical accounts). Marten sub-populations have recently increased in the Terra Nova and Main River areas after translocations in the 1980s. Habitat protection and efforts to reduce accidental capture are helping to stabilize populations, and population growth is possible under this management regime.

Marten populations typically vary in response to fluctuations of microtine prey species; alternative prey dampens the fluctuations. Newfoundland marten were thought to be highly dependent on the meadow vole, which prefers old growth forests (Sturtevant and Bissonette 1997) and undergoes population fluctuations. Snowshoe hares occupy younger conifer stands and are an important alternate prey, especially in winter (Gosse and Hearn 2005). The large home ranges and the extreme variability of home range size among years (6.37 – 67.24 km² for males and 4.35 – 46.5 km² for females; Gosse et al. 2005) likely reflects the low diversity and abundance of prey species as well as the highly fragmented habitat (B. Hearn, pers. comm., 2006).

Rescue effect

There are no neighbouring populations of the subspecies Martes americana atrata which are likely to immigrate to Newfoundland. There is no possible rescue effect for this genetically and ecologically distinct Newfoundland population.