Independent Review of Status of the Bobcat Population in ...



Independent Review of Status of the Bobcat Population in the

Northern Lower Peninsula of Michigan

Prepared by: Thomas M. Gehring, Timothy S. Preuss, and Bradly A. Potter

Department of Biology, Central Michigan University

Mount Pleasant, MI 48859

Summary:

The number of bobcats harvested by hunters in the northern Lower Peninsula of Michigan (NLP) increased from approximately 140 to 291 during 1985-2002. However, given the current data, it is difficult to determine if this increase was a function of the bobcat population increasing or an increase in the number of hunters or other variables. Statewide, an increase in the number of bobcats harvested by hunters was correlated with the increase in the number of hunters. During the same period, average pelt price for bobcats appeared to decrease. These results suggest that bobcats may now be more valued by hunters as a trophy species rather than an alternate income source and/or bobcat hunting is driven more by the recreational compared to economic incentives. Overall, when the change due to the number of hunters was removed, the bobcat population in Michigan appeared to be relatively stable. However, we could not adequately address the status of the NLP population separately. The proportion of juveniles and yearlings remained relatively high throughout the time period. Based on estimates from opinion surveys, if trapping and hunting occurred annually within in the NLP, an estimated 620 to 3,000 bobcats would be harvested each year assuming 100% filling of bag limits. An estimated 420 to 1,500 bobcats would be harvested annually assuming a 50% success rate. Without accurate population index or other independent data for the entire NLP, it is difficult to evaluate the impact that this harvest figure would have on the NLP bobcat population. Distribution of bobcat harvests in the NLP suggests there are 1 or 2 relatively disjunct regions of Zone 2 that maintain relatively high harvest intensity, whereas the areas between exhibit lower harvest rates. There also appears to be a correlation between harvest intensity and the density of streams and rivers in the NLP. Additionally, we found greater than expected harvest rates on public land compared to private land in the NLP. Investigation into the possible source-sink dynamics and spatial patterning of bobcat harvest in the NLP should be pursued further. These investigations could provide insight into possible new zoning recommendations based on landscape-level features (e.g., hydrologic patterns) and/or spatial patterning of high-quality bobcat habitat (e.g., contiguous patches of lowland forest and non-forested wetland habitat).

Introduction:

In 1975, bobcats (Lynx rufus) were listed in Appendix II of the Convention on International Trade in Endangered Species (CITES) because of concern for the viability of their populations. This listing required state agencies to provide the U. S. Fish & Wildlife Service with data on the status and viability of bobcat populations. The current geographic range of the bobcat includes all of Michigan although bobcat densities are likely greater in the northern 2/3 of the Lower Peninsula of Michigan. Currently, bobcats are harvested by hunting only in Zone 2 of the northern Lower Peninsula (NLP), with a bag limit of 1 bobcat per hunter. Zone 2 is subdivided into North and South zones with season lengths of 60 days (1 January to 1 March) and 33 days (15 January to 16 February), respectively (Figure 1).

[pic]

Figure 1. Northern Lower Peninsula of Michigan bobcat harvest units depicting Zone 2

North and Zone 2 South units (source: MDNR)

Annually, since 1985 the Michigan Department of Natural Resources (MDNR) has collected age, sex, and location data for bobcats harvested in the state. During 1991-1996, MDNR researchers conducted a mark-recapture study and scent-station survey in Crawford, Missaukee, and Roscommon counties in the NLP to evaluate bobcat survival and relative abundance (Earle et al. 2003a). Recently, Central Michigan University (CMU) began a radio-telemetry study of bobcats predominantly in Missaukee and Roscommon counties (Houghton Lake Study Area; Preuss and Gehring, unpublished data). Additionally, in 2003 we began assessing the relative abundance of bobcats using a scent-station survey in Roscommon and Missaukee counties, as well as parts of Clare, Crawford, Gladwin, Kalkaska, and Osceola counties. Collectively, these data allow an examination of bobcat population characteristics in the NLP over an approximately 20-year period.

Objectives:

1) Evaluate changes in the relative abundance and population structure of bobcats in the NLP;

2) Evaluate the designation of Zone 2 into a North and South unit; and

3) Provide recommendations for future bobcat management in the NLP.

Methods:

Relative Abundance

We graphically examined the change in the number of registered bobcats over time. Further, we used furbearer harvest survey results (Frawley 2001, 2003, unpublished data) to examine the fur taker population during 1985 to 2002. Because pelt price might influence trapping effort (Royama 1992), we examined this possible influence in the number of bobcats harvested via hunting. We used average bobcat pelt price data from Wisconsin (Dhuey et al. 2003) and regressed (PROC REG, SAS Institute 1994, α = 0.05) this with the number of bobcat hunters. Reliable data were not available for an estimate of the number of bobcat hunters in the NLP only due to high variance. Thus, we pooled Zone 1 and Zone 2 bobcats harvested by hunters and used estimates of the number of bobcat hunters in the State of Michigan (Frawley, unpublished data).

We conducted linear regression analysis on the number of registered bobcats hunted (dependent variable) and the number of bobcat hunters. We transformed data using the natural logarithm to gain stationary variance (Royama 1992). We examined the residuals from this regression by plotting residuals in time (i.e., sequence plot) to determine if there was a change in the number of bobcats hunted over time (i.e., a time-related effect, Neter et al. 1996). Additionally, we plotted differenced values [N difference = ln (N t) – ln (N t+1)] for number of bobcats harvested by hunters and the estimated number of hunters. Differencing provided a stationary mean in the values and allowed estimates to be examined as rates of change over time (Royama 1992; Swanson 1998).

We conducted a scent-station survey during October – November 2003 in Roscommon and Missaukee counties, as well as parts of Clare, Crawford, Gladwin, Kalkaska, and Osceola counties. The survey was patterned after Linhart and Knowlton (1975), with modifications by Roughton and Sweeny (1982). Recommendations by Sargeant et al. (1998) and Sargeant et al. (2003) also were incorporated. Scent-stations consisted of a 0.9-m diameter circle of sand with a fatty-acid scent tablet placed in the center as an attractant. Seventy transects with 10 stations along each transect were checked for 2 nights. Stations were placed 480-m apart along each transect and transects were located ≥5 km from the nearest transect. The presence of all tracks was recorded to species whenever possible, and visitation rates were calculated for the proportion of stations and the proportion of transects visited by each species.

We qualitatively compared various population indices to assess changes in the relative abundance of bobcats over time in Crawford, Missaukee, and Roscommon counties (i.e., the 1991-1996 MDNR study area and our current study site). We only used a qualitative comparison because objectives and methodology for studies were not consistent, and the data record for the total time series (1985-2002) was not complete. These data sources included: 1) number of bobcats harvested during 1985-2002; 2) MDNR scent-station survey data from 1992-1996 (Earle et al. 2003a); 3) CMU scent-station survey data from 2003 (Preuss and Gehring, unpublished data); 4) number of bobcats captured per 1,000 trap nights by the MDNR, 1991-1996; and 5) number of bobcats captured per 1,000 trap nights during 2003 (Preuss and Gehring, unpublished data).

We also attempted to estimate the range in the number of bobcats that would be trapped in the NLP if a trapping season was offered. To estimate this number, we used fur harvester opinion survey report data for 2003 (Bull and Peyton 2004), and the number of respondents that indicated that they were ‘very likely’ to trap bobcats in the NLP if it were legal. We estimated a total NLP bobcat harvest (hunting and trapping combined) by adding the mean of the number of bobcats hunted during 1998-2002 in the NLP to our estimated trapper harvest of bobcats in the NLP (i.e., trapping includes additive mortality above the average hunt-related mortality). We used 2 estimates of a trap harvest of bobcats: 1) based on the total proportion of respondents in the Bull and Peyton (2004) survey that indicated they were ‘very likely’ to trap in the NLP (i.e., estimated at 18% of furtaker license holders); and 2) based on the proportion of respondents in the Bull and Peyton (2004) survey that had previously trapped or hunted bobcats and indicated they were ‘very likely’ to trap in the NLP (i.e., 19.5%). We used the mean number of bobcat hunters and trappers during 1998-2002 in our latter estimate (Frawley 2001, 2003). We assumed a bag limit of 1 bobcat per hunter or trapper and a trapping success rate of 1 bobcat per trapper (Frawley 2001, 2003). That is, we assumed that all trappers would fill their bag limit. We also calculated estimates assuming that 50% of trappers would fill their bag limit.

We initially attempted to parameterize the Minnesota Furbearer Population Model (e.g., Rolley et al. 2001) with new harvest estimates and existing estimates of other population parameters; however, our simulations of population trends were very sensitive to the value of the initial population size. Furthermore, small changes in the initial population size resulted in vastly different population trends. Rolley et al. (2001) suggested that this model should be calibrated to either an independent source of population trend data or an estimate of absolute population size. We chose not to include these analyses because of concerns about the lack of accurate population indices or absolute counts of the number of bobcats in the NLP over the time period 1985-2002.

Population Structure

We sorted MDNR databases to isolate the age and sex of registered hunted bobcats. We used the sex identified based on the MDNR’s assessment using lower canine tooth maximal thickness and width of the root (Friedrich et al. 1983) when given. If this lab-determined sex was not provided, we used the field-determined sex. The MDNR estimates age of registered bobcats by counting cementum annuli in cross sections of the lower canine root (Crowe 1975). We classified age as juvenile (0.5 year), yearling (1.5 year) and adult (2.5+ year). Animals that did not have the sex or age identified were excluded from further analyses. Initially, we graphically examined the change in estimated age and sex ratios based on harvested bobcats in Zone 2. Significant changes in ratios over time were determined by linear regression (PROC REG, SAS Institute 1994) with an α-value of 0.05.

Preliminary Modeling of Potential Bobcat Population Size and Distribution

We obtained a broad estimate of bobcat abundance by determining the number of male and female bobcat home ranges that could theoretically fit in the NLP using ArcView GIS (ESRI, Redlands, California). Average minimum convex polygon (MCP) home ranges were calculated for male and female bobcats radio collared in the Houghton Lake Study Area (HLSA). We captured bobcats during March-June 2003 using No. 3 Victor Soft-Catch padded foot-hold traps (Earle et al. 2003b). We immobilized bobcats via an intramuscular injection of 10 mg/kg ketamine hydrochloride and 1.5 mg/kg xylazine hydrochloride (Kreeger 1999) in order to determine sex, age, reproductive condition, ear tag and radio collar animals. Radio-collared bobcats were located a minimum of 2-3 times weekly from May – December 2003 using a vehicle-mounted 4-element Yagi directional antenna and electronic compass (Lovallo et al. 1994). Triangulations were made from telemetry stations with 3-4 bearings obtained as quickly as possible to reduce telemetry error (White and Garrott 1990). Bobcat locations were estimated using the microcomputer program LOCATE II (Nams 1990). We estimated home-range size using the MCP method (Mohr 1947) using the Animal Movement Analysis Extension to ArcView GIS (Hooge and Eichenlaub 1997).

Estimates of home-range size (Table 1) were consistent with those of bobcats in Wisconsin (Lovallo and Anderson 1996) and Minnesota (Fuller et al. 1985). We divided the total area of the NLP by the average home-range size of male ([pic]= 46.5 km2 = 18 mi2, sd = 27.3 km2 = 10.5 mi2) and female ([pic]= 10.4 km2 = 4.0 mi2, sd = 5.9 km2 = 2.3 mi2) bobcats to obtain an estimate of the number of male and female bobcats that could potentially reside in the NLP. This model assumed that there was no intrasexual home-range overlap, all bobcats were resident (i.e., bobcats without established home ranges were not accounted for), and bobcats occurred in all landcover types (i.e., bobcats exhibited no selection of habitat).

Table 1. Preliminary data on summer home-range size for bobcats monitored during 2003 in the Houghton Lake Study Area as part of a Central Michigan University Study (Preuss and Gehring, unpublished data).

|Bobcat ID |Sex |Minimum Convex Polygon (km2) |Minimum Convex Polygon (mi2) |

|M01 |Male |26.15 |10.10 |

|M02 |Male |77.44 |29.90 |

|M03 |Male |35.86 |13.85 |

|F01 |female |9.25 |3.57 |

|F02 |female |7.01 |2.71 |

|F03 |female |6.24 |2.41 |

|F04 |female |19.08 |7.37 |

We also developed sex-specific models of bobcat abundance relative to preferred landcover types since bobcats exhibit habitat selection (e.g., Lovallo and Anderson 1996; Preuss and Gehring, unpublished data). We used the 2001 landcover GIS theme available from the MDNR. Within ArcView GIS, 2 hexagonal grids were constructed and overlaid on the NLP (Figure 2). The area of each hexagon in the first grid was equivalent to the average home range of male bobcats. The area of each hexagon in the second grid was equivalent to the average home range of female bobcats. We calculated the percentage of streams, lowland conifer forest, and lowland conifer forest/non-forested wetlands within each hexagon of the sex-specific grids. Lowland conifer forest and non-forested wetland cover types were chosen because radio-collared bobcats in the HLSA used these cover types in greater proportion to their availability based on chi-square analysis (χ2 > 29.4, P < 0.001; Neu et al. 1974). We modeled percentage of streams because data from radio-collared bobcats, scent-station surveys, and harvest data suggested a possible relationship between bobcats and streams. We then modeled abundance using 2 approaches.

For the first approach, we calculated the mean and minimum percentages of each landcover class occurring within the home ranges of male and female radio-collared bobcats. For each landcover class, we then modeled abundance by assigning 1 bobcat to each hexagon with a percentage greater than or equal to the mean and minimum percentages of that landcover class occurring within a home range. Lovallo and Anderson (1996) found that some home ranges of bobcats in Wisconsin were composed of 5% Lowland Forest (Lovallo) |699 |2,844 |3,543 |19 |

Models depicted in figures 16, 17, and 18 highlight similar regions of the NLP that may be core habitat areas for bobcats. These same general areas also correspond with areas that typically have the highest intensity of bobcat harvest (see Figure 20). Figures 16-18 also might offer some insight into the disjunct nature of the North and South hunting units in Zone 2. That is, 2 relatively disjunct habitat patches are distributed in the northeastern and southwestern quarters of Zone 2. A possible 3rd habitat patch might exist in the east-central region of Zone 2.

Figure 19 probably provides a maximum population size since most of the NLP is depicted as potential bobcat habitat using a lower proportion of lowland forest in bobcat home ranges. We caution, however, that these estimates are based on preliminary data and small sample size of the number of bobcats monitored via radio telemetry as well as only 1 location where scent stations have been conducted in the NLP to provide landcover type associations in these models. For example, our sample sizes for figures 16-18 were n = 42 and 69 hexagons for males and females, respectively. These models will need to be refined and validated before they might offer greater insight into potential bobcat habitat and better approximations of the size of the bobcat population in the NLP.

a)[pic]b)[pic][pic]

Figure 16. a) Male and b) female hexagonal grids predicted to potentially contain male (blue) or female (red) resident, adult bobcats in the NLP. Predictions are based on the mean proportion of lowland forest in bobcat home ranges as determined by a sample of radio-collared bobcats and bobcats detected using a scent-station survey.

a)[pic]b)[pic][pic]

Figure 17. a) Male and b) female hexagonal grids predicted to potentially contain male (blue) or female (red) resident, adult bobcats in the NLP. Predictions are based on the mean proportion of lowland forest and non-forested wetland habitat in bobcat home ranges as determined by a sample of radio-collared bobcats and bobcats detected using a scent-station survey.

a)[pic]b)[pic][pic]

Figure 18. a) Male and b) female hexagonal grids predicted to potentially contain male (blue) or female (red) resident, adult bobcats in the NLP. Predictions are based on the minimum proportion of lowland forested habitat found in bobcat home ranges (22% for males, 18% for females) as determined by radio-collared bobcats in the CMU bobcat study.

a)[pic]b)[pic][pic]

Figure 19. a) Male and b) female hexagonal grids predicted to potentially contain male (blue) or female (red) resident, adult bobcats in the NLP. Predictions are based on the proportion of lowland forest in bobcat home ranges (>5%) as determined by Lovallo and Anderson (1996) in northwestern Wisconsin.

Distribution of Harvest

The distribution of bobcats harvested during 1985-2002 is relatively widespread throughout Zone 2; however, harvest locations do appear to concentrate into 1 large patch in the northeastern portion of the NLP with a secondary linear patch in the central NLP (Figure 20). A portion of this distribution pattern may be related to the different season lengths between the North and South hunting units in Zone 2, although these same regions are identified in our habitat analysis above.

[pic]

Figure 20. Distribution of harvested bobcats across the NLP during 1985-2002. Harvest intensity of bobcats is mapped to the scale of township. Cooler colors indicate lower harvest intensity, whereas warmer colors represent greater numbers of bobcats harvested.

We mapped the distribution of harvested bobcats across 3-year intervals to examine changes in the pattern of harvest intensity. We maintained the same scale of harvest intensity in Figure 21. In general, harvest intensity remained relatively constant and localized in townships of the northeastern portion of the NLP, with 8-18 bobcats harvested per township during 1985-2002. The central portion of the NLP exhibited periods of greater harvest intensity during 1985-1990, reduced harvest pressure during 1991-1996, and an apparent increasing harvest intensity since 1996 (Figure 21). Conversely, the distribution of harvest intensity for hunted bobcats in Zone 1 appeared relatively stable despite the increases in bag limit in 1994 and 1996 (Figure 22). These patterns may suggest that there are 1 or 2 relatively disjunct regions of Zone 2 that maintain relatively high harvest intensity, whereas the areas between exhibit lower harvest rates. As such, these patterns provide some support to the idea that the NLP bobcat population may exhibit source-sink dynamics and may need to be managed as 2 separate units. However additional demographic data on rate of population increase and dispersal would be needed before characterizing it as a true source-sink population (e.g., Pulliam 1988).

We also mapped harvest intensity of bobcats during 1985-2002 with respect to landscape features (hydrology and road systems) to identify potential patterns. In the NLP, bobcat harvest pressure appears to be concentrated predominantly along streams and rivers, including headwater tributaries (Figure 23). This pattern is likely linked to the lowland forest and non-forested wetland habitats (including beaver sloughs) located along water courses. These habitats appear to be preferred habitat for bobcats in the NLP (Preuss and Gehring, unpublished data). A more ambiguous pattern was noted for the relation between bobcat harvest intensity and road density in the NLP, perhaps a function of the extensive road network that exists throughout much of the NLP (Figure 23).

a) 1985-1987[pic]

b) 1988-1990[pic]

Figure 21. Distribution of harvest intensity of bobcats, mapped to the scale of township, over the NLP during 1985-2002. Lighter colors indicate lower harvest intensity, whereas darker red colors represent greater numbers of bobcats harvested.

c) 1991-1993[pic]

d) 1994-1996[pic]

Figure 21. Distribution of harvest intensity of bobcats, mapped to the scale of township, over the NLP during 1985-2002. Lighter colors indicate lower harvest intensity, whereas darker red colors represent greater numbers of bobcats harvested.

e) 1997-1999[pic]

f) 2000-2002[pic]

Figure 21. Distribution of harvest intensity of bobcats, mapped to the scale of township, over the NLP during 1985-2002. Lighter colors indicate lower harvest intensity, whereas darker red colors represent greater numbers of bobcats harvested.

[pic]

b) 1988-1990[pic]

Figure 22. Distribution of harvest intensity of bobcats, mapped to the scale of township, over the UP during 1985-2002. Lighter colors indicate lower harvest intensity, whereas darker red colors represent greater numbers of bobcats harvested. Bag limit was increased to 2 bobcats per hunter and 3 bobcats per hunter in 1994 and 1996, respectively.

c) 1991-1993[pic]

d) 1994-1996[pic]

Figure 22. Distribution of harvest intensity of bobcats, mapped to the scale of township, over the UP during 1985-2002. Lighter colors indicate lower harvest intensity, whereas darker red colors represent greater numbers of bobcats harvested. Bag limit was increased to 2 bobcats per hunter and 3 bobcats per hunter in 1994 and 1996, respectively.

e) 1997-1999[pic]

f) 2000-2002[pic]

Figure 22. Distribution of harvest intensity of bobcats, mapped to the scale of township, over the UP during 1985-2002. Lighter colors indicate lower harvest intensity, whereas darker red colors represent greater numbers of bobcats harvested. Bag limit was increased to 2 bobcats per hunter and 3 bobcats per hunter in 1994 and 1996, respective

[pic]

a)

b) [pic]

Figure 23. Distribution of harvested intensity of bobcats across the NLP during 1985-2002 mapped to the scale of township. Lighter colors indicate lower harvest intensity, whereas darker red colors represent greater numbers of bobcats harvested. a) Overlay of harvest intensity and hydrology illustrating the general pattern of greater number of bobcats harvested along streams and rivers and associated tributaries; b) Overlay of harvest intensity and primary and secondary roads illustrating no clear pattern of higher harvest in more or less roaded areas.

During 1998-2002, a greater number of bobcats was harvested on private land compared to public land in both the NLP ([pic] = 64 bobcats on public and [pic] = 155 bobcats on private land, P ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download