A Brief Summary of the Status of the Eastern Slopes Grizzly Bear Project (ESGBP) (2001)


Herrero, Stephen. 2001. “A Brief Summary of the Status of the Eastern Slopes Grizzly Bear Project (ESGBP).” Eastern Slopes Grizzly Bear Project, University of Calgary, Calgary, Alberta.
Note: You can also download a PDF version of this report.
Introduction
On March 10, 2001 a group of scientists from the United States and Canada met to review ESGBP research. I prepared the following brief summary for all who attended this meeting in Banff. In the following document is the first, albeit brief and preliminary, analysis of 7 years of ESGBP population data and their implications. An internationally recognized bear population biologist, Dr. David Garshelis, of the Minnesota Department of Natural Resources is leading this analysis. Preliminary findings suggest that in the Bow River Watershed there is one of the least productive populations of grizzly bears studied in North America (see especially table 2, p.7). Note the long reproductive interval, the very low reproductive rate and a low population intrinsic growth rate (lambda). This suggests that the population is delicately balanced between decline and growth. This emphasizes the importance of management actions and other influences on the fate of individual grizzly bears, especially females. Scientific reviewers were positive regarding ESGBP research progress.
Steve Herrero
March 10, 2001
Project background:
The ESGBP began in 1994 in response to a need for scientific understanding of grizzly bears within an area called the Central Rockies Ecosystem (CRE). This area encompasses about 40,000 km2 and includes Banff, Yoho and Kootenay National Parks as well as Kananaskis Country and surrounding provincial lands in Alberta and British Columbia. Grizzly bears found in this area are not isolated although the population has significant closure with unoccupied habitat on the prairies to the east and the Columbia Trench west. The Rocky Mountains filter bear movement east-west within the CRE. Lands west of the Rocky Mountain are more productive for grizzly bears and population densities are greater.
In the Alberta portion of the CRE there are about a million people living within an hour or two drive of occupied grizzly bear habitat. Banff National Park has an annual visitation of about 4.5 million. The CRE is one of the most heavily used and developed landscapes in North America where grizzly bears survive. In the Alberta portion of the study area grizzly bears were thought to exist at relatively low densities with low reproductive rates. The cumulative effects of human recreational, tourism, resource extraction, residential development and other activities were of concern. In the early 1990s three major development proposals in the Alberta portion of the Rocky Mountains were either turned down or significantly altered because of a lack of scientific information regarding potential effects on grizzly bears.
The need for scientific understanding of grizzly bear population and habitat parameters created broad support for a multi-agency, multi-stakeholder research project. The University of Calgary was selected as the project base. A series of graduate students working with professors and outside experts have conducted the research. A Steering Committee, composed of major stakeholders, offers comment and direction.
Since 1994 a trapping and monitoring team led by Mike Gibeau has captured and fitted VHF radio transmitters onto grizzly bears. Trapping and monitoring have been concentrated within the Bow River Watershed and surround, about a 22,000 km2 portion of the CRE. About 25 grizzly bears each year have had active transmitters. The sample of bears has been purposely biased to include more females than males because of the females’ critical role in reproduction. Monitoring has been conducted from air and ground (total N=10,000 plus) with an attempt to locate all bears about once per week. There also has been intensive daily (sometimes 24 hour plus) monitoring of selected female bears. The project was designed to generate at least 100 reproductive years of data regarding adult female grizzly bears. This goal has been reached. The ESGBP will conclude during the year 2002.
Research completed:
Bryon Benn in his Master’s thesis did a spatio-temporal analysis of 627 human- caused grizzly bear mortalities that occurred in the CRE between 1971-1996. Eighty-five percent of 462 such mortalities with accurate locations occurred within 500m of a road or 200m of a trail or backcountry development. Deaths were geographically concentrated. Benn and Herrero updated these data and have submitted a paper to Ursus.
Mike Gibeau’s Ph.D. examined the relation between various development features and habitat use, especially by adult female grizzly bears. Adult female habitat use appeared to be compromised by various developments and intensive human use. Mike developed a predictive GIS-based model of adult female grizzly bear security areas in the CRE and showed that females selected for these areas. Generally, habitat security according to the model was significantly lower than for grizzly bear populations elsewhere. Several published and in press papers have come from this effort.
John Kansas’s Master’s thesis explored the influence of mapping scale, disturbance coefficients and seasonal habitat on grizzly bear habitat effectiveness model output in Kananaskis Country. Five different vegetation-mapping scales were used to represent grizzly bear habitat. John found that changes in levels uof human use had significantly more influence on habitat effectiveness values than did changes in disturbance coefficients or habitat inventory or mapping.
In 1996 the ESGBP conducted a capture/recapture DNA fingerprint hair snagging study to try to estimate population density. This effort was unsuccessful due to low sample sizes and too large grid cells. The “best” density estimate of 14 bears/1000 km2 was consistent with a previous estimate for Kananaskis Country but the confidence interval was too large to accept the density estimate.
A study initiated by Mike Gibeau and Curtis Strobeck’s lab found that our sample of radio-tagged grizzly bears had very low mitochondrial DNA diversity and about average nuclear DNA diversity.
Three major preliminary analyses of large portions of our data were completed as input into specific planning processes or as major milestones. Reports on the population and habitat status of grizzly bears in Banff National Park (1996), Kananaskis Country (1998), and a preliminary Population and Habitat Viability Assessment (2000) were each completed. Results have been important to related planning processes.
Two landsat-based habitat maps have been completed for the CRE. One looks at “greenness” as a habitat surrogate. The other classifies vegetation. In preliminary RSF habitat modeling by Theberge both classifications, but especially greenness, seemed useful in understanding habitat use.
Current research:
Describing and understanding grizzly bear population demography is a current research priority. This analysis is based on our 7 years of survival and reproductive data focusing on adult females. Dave Garshelis of Minnesota Department of Natural Resources is the lead scientist in this analysis with Mike Gibeau and Steve Herrero as collaborators. A brief description of objectives, methods and preliminary results is in Appendix 1.
Another research priority is understanding the nature of habitat selection by adult female grizzly bears. This is the primary objective of Jen Theberge’s Ph.D. research. She is using uni- and multi-variate analyses to develop resource selection functions that are being used to model habitat selection for our sample of 27 adult female grizzly bears. Jen’s work will give us realized habitat use models. If funds are available we can do additional GIS-based work, similar to Rick Mace’s analysis, and determine a measure of habitat effectiveness by re-running the RSFs and including no human influence layers. There are many collaborators in this work. They are listed along with a summary of Jen’s research Appendix 2.
Because our radio-tagged population appears to be one of the most reproductively conservative in North America we are seeking additional understanding. Nutrition is one variable known to influence reproduction. Working in cooperation with Marc Cattet, Gord Stenhouse and Todd Shury we are comparing body weights relative to body lengths with other populations. In addition we have identified a possible sample of grizzly bears taken from our study area before 1940. Hair or bone from these samples should allow collaborator Charlie Robbins to use stable isotope analysis to see if historically bears had more meat in their diets and hence may have been better off nutritionally.
We are continuing to work with Mike Proctor, Robert Barclay and Curtis Strobeck to understand gene flow and structure, and the genetics of reproductive success and failure for individuals. We hope to better understand gene flow with the CRE and how the CRE population links genetically with other populations.
Two additional Master’s theses are near completion. Karen Oldershaw is examining the relationships between forestry, cover and grizzly bear habitat. Cedar Mueller is focusing on modeling the relationships between human activities and developments and subadult grizzly bear recruitment and survival.
Another Master’s thesis is about half done. Saundi Stevens is looking at the relationship between habitat use, female grizzly bear security areas and greenness as derived from Landsat data.
Tentative management implications:
(based on preliminary analysis of our evolving data set)
The eastern slopes grizzly bear population appears delicately balanced between decrease and increase. The population could easily slip into decline because it has low density, a very low reproductive rate, significant potential for man-caused mortalities and removals, and numerous habitat stresses.
Significant effort has gone into trying to maintain females. Even more and continuing care is needed to minimize man-caused mortality/removal of sub-adult and adult female grizzly bears. Loss of several of these, as occurred with the known death or removal of at least 3 adult female bears last year in the Bow River Watershed, probably decreased the intrinsic growth rate of the population.
Decreasing mortality/removal of sub-adult and adult females will be difficult. The causes of mortality/removal are diverse and not easily turned off. One exception is managing to keep all human food, garbage and other attractants from being available to bears. This must continue being world class.
Thus far mortality in the adult female cohort is concentrated in Banff National Park (4 of 5 known mortalities). Without the greater survivorship in Kananaskis Country the intrinsic growth rate of the population would probably be negative. This places additional focus on female grizzly bear mortality/removal management in Banff National Park.
Identifying and protecting productive habitat, and linkages between productive units, is fundamentally important to maintaining reproductive output. The ESGBP will help to identify such units and possibly the connections between them. Part of effective habitat management is providing accepted levels of security for adult female grizzly bears.
Within the CRE habitat management to increase production for grizzly bears is a possibility that could yield benefits. This is easier to write or say than to do.
Appendix 1:

Preliminary Demographic Analysis of Eastern Slopes Grizzly Bears Through Year 2000

Dave Garshelis, Mike Gibeau, Steve Herrero
A particularly striking feature of Eastern Slopes grizzly bears is their especially long interval between litters. Data from radiocollared bears are shown in Fig. 1. Only 8 reproductive intervals were actually known (black bars), these being cases of cub production, followed by an interval of cub rearing, and then subsequent cub production. One of these (bear 36) was a first-time mother who lost her litter and produced again after only 2 years. All other litters stayed with their mother for at least 2 years, and up to 5 years, and mothers often did not immediately produce another litter, so some litter intervals were quite long.
An obvious problem with the data in Fig. 1, however, concerns the many intervals of unknown duration (hatched bars). These situations arose either because the bear died, its radio stopped functioning, or it is still being monitored but has not yet produced another litter. These events are more likely to disrupt long intervals than short ones, so neglecting these incomplete intervals would negatively bias an estimate of the mean litter interval.
Figure 1. Litter interval data from radiocollared bears, 1992 – 2000.


ID


1992


1993


1994


1995


1996


1997


1998


1999


2000


2001

Known

interval

Min.

interval

17

?

cubs

cubs

x

3

18

cubs

cubs

?

4

24

cubs

?

5

26

cubs?

cubs

cubs

?

4

3

27

cubs

x

28

cubs

x

30

cubs

?

7

31

cubs

x

32

cubs?

cubs

x

3,3

33

cubs?

cubs

cubs

?

3

3,3

36

cubs

died

cubs

cubs

?

2,3

37

cubs?

cubs

cubs

?

4

4

40

?

x

7

41

?

cubs

?

3,4

46

cubs

?

7

47

cubs

cubs

x

4

55

cubs

x

57

cubs

?

3

61

?

x

4

63

cubs

?

3

64

?

?

3

cubs ? = year of birth unknown, but no later than
1992

sample size >

8

16

The average duration of the 8 known intervals is only 3.4 years. Adding in the 16 incomplete intervals that exceeded 2 years, and presuming cub production occurred exactly at these minimum intervals, increases the mean to 3.8 years. However, this is still an underestimate because many of the incomplete intervals would likely be longer than the minimum known period.
To reconcile this problem we used the procedure described by Garshelis et al. (1998, Ursus 10: 437–447), designed to deal with an analogous difficulty in calculating mean age of first reproduction. The calculations, shown in Table 1, yield a mean duration between litters of at least 5 years. This value is still a minimum, because three bears that did not produce cubs for 6 years (numbers 30, 40 & 46, Fig. 1) were counted as 7-year intervals (these could be longer).
Table 1. Calculation of mean litter interval, incorporating data on the 8 complete as well as 16 incomplete intervals (from Fig. 1). The incomplete intervals ultimately drop out as “lost”. The only necessary assumption is that the three longest incomplete intervals (minimum 7 years) represent known intervals. The shaded square in the bottom right corner is a less biased estimate of the mean than the more conventional estimate attained by averaging just the known intervals.

Interval

(years)

Number

in sample

Produced

cubs

Lost before

next interval

Proportion

producing

%of population

available

to produce

Years weighted

by % of pop

producing

%of population

producing

2

24

1

8

0.04

100

4.2

0.08

3

15

3

3

0.20

95.8

19.2

0.58

4

9

4

2

0.44

76.7

34.1

1.36

5

3

0

0

0

42.6

0.00

0.00

6

3

0

0

0

42.6

0.00

0.00

7

3

3

0

1.0

42.6

42.6

2.98

Sum

11

100.00

5.00

Mean interval >

4.36

Mean interval >

5.00

The same procedure was used to calculate the average age of first reproduction. The mean of just the known ages of first reproduction was 6.5, whereas the mean incorporating bears that were lost before they had a chance to reproduce was 6.7.
If we had used the conventional estimates for litter interval and age of first reproduction, estimates of population growth (lambda, derived from Eberhardt’s modification of the Lotka equation) would have been biased high by about 2%. This is significant, as our lambda estimates hover around 1.0. Unfortunately, this modified method of calculating mean age of first reproduction and litter interval precludes use of Hovey and McLellan’s (1996, Can. J. Zool. 74:1409–1416) bootstrapping program for calculating the confidence interval around lambda. Additionally, this procedure was not used in other grizzly bears studies, so the results are not strictly comparable. With these caveats in mind, we present a comparison of some grizzly bear populations for which population growth rates have been estimated (Table 2). The range of estimates for lambda shown for the East Slopes population are due to varying assumptions, related to special situations that affected estimates of reproduction or survival (Kaplan-Meier analysis), especially two bears that disappeared but whose fate was unknown. In one case they were considered alive at the time of disappearance and in another scenario they were considered to have died (considered more probable).
Table 2. Comparison of demographic characteristics among some grizzly bear population


Flathead

(Hovey & McLellan)


Yellowstone

(Eberhardt

et al.)


Kugluktuk

(Case & Buckland)


Selkirks

(Wielgus

et al.)


Kananaskis

(Wielgus & Bunnell)


Eastern Slopes

(this study)

Age 1st repro

6.4

5.7a

8.7

7.3

5.5

6.7

Cub litter size

2.35

2.0

2.26

2.22

1.40

1.88

Litter interval (yrs)

2.8

3.0

2.6

3.0

3.0

5.0

Repro rate (m)

0.42

0.33

0.43

0.37

0.23

0.19

Cub survival

87%

84%

81%

84%

78%

78%

Yearling survival

94%

76-84%

100% (?)

100% (?)

88%

Subad F survival

93%

89%

~ 80% (?)

78%

~ 89–93%

89–95%

Adult F survival

95%

92%

98%

96%

93%

95–96%

Lambda

1.085

1.046

1.026

1.0

0.99–1.01

0.99–1.01

All of the above analyses are preliminary inasmuch as more data will be collected in the coming year, which will hopefully close out some of the unknown reproductive intervals and ages of first reproduction. Also, we hope to pursue the following additional analyses:
Obtain bootstrapped CIs for lambda, either via modification of Hovey’s Booter program, or creating a different program.
Estimate total sustainable human-related mortality, not only as a percent of the population, but also as a number of female bears. To achieve this will require melding these data with tabulated data on all known deaths in the ecosystem.
Generate possible scenarios for population growth among the male segment of the population, as males have much lower survival than females (subadults 66–70%, adults 88–91%). This will require modeling, using an estimate of the population sex ratio (derived from capture and hair-snag data).
Appendix 2:

Summary of Research Approach

Jen Theberge
Title of Ph.D. Dissertation:
Influence of Spatial Pattern upon the Selection of Landscape Characteristics by Female Grizzly Bears in the Rocky Mountains of Alberta
This research is being conducted within the home ranges of individual bears.
I am using an RSF approach (with logistic regression) to investigate habitat use and availability by grizzly bears within their home ranges.
I am working with 27 adult female grizzlies.
These individual bears have been monitored anywhere from 3 to 6 years.
Seasonal home ranges were created for each bear using adaptive kernel estimators. So for an individual bear there are 3 home ranges – preberry season, berry season, composite (all locations). The seasonal home ranges are multi-annual.
The number of bear locations in each home range (pooled across years) ranges from 17-46 locations in the pre-berry season, 19-67 locations in the berry season, and 36-113 locations in the composite (all locations across all years.)
Random locations have been generated within each seasonal home range. There are 3000 random locations for each seasonal home range.
Chapters and Research Questions:
Selection of Landscape Characteristics by Adult Female Grizzly Bears
Question: What landscape characteristics do female grizzly bears prefer? Do resource selection functions change across the spatial extent of individual grizzly bear home ranges?
Variables included in analysis: elevation, slope, aspect, land cover type, distance to water, greenness, distance to human use/ intensity of use, distance to ecotone
Conditions under which tests will be run: habituated and non-habituated, with and without cubs, season (berry and preberry), core and periphery of home range
Preliminary results show that for wary bears in the preberry season: greeness is highly significant (positive coeffecient), distance to edge is highly significant (positive coefficient). When greeeness is removed from the model, vegetation is highly significant (specifically shrub, grass, steep-slope-mix-with-avalanche, rock (-ve)).
Influence of Heterogeneity and Spatial Pattern on Site Selection by Adult Female Grizzly Bears
Question: Is site selection by female grizzly bears influenced by heterogeneity and spatial pattern?
RSF using variables within 1.5 km diameter window, 300 m diameter window
Variables: land cover diversity, richness, fragmentation, dominance, terrain ruggedness, density of human use (total, motorized, non-motorized), average greenness, greenness variance, distance to ecotone, distance to cover (i.e. closed environment)
Biological Basis for Grizzly Bear Management using Security Areas
Question: Females select for security areas (Gibeau 2000) that are 3-km diameter in size. Are females selecting for specific landscape characteristics in areas of security area size (i.e. 3km-diameter window size)? Are some characteristics being selected for at a finer scale (i.e., 1.5km or 300m diameter window size)?
RSF for variables similar to #2 in window sizes 300-m, 1.5-km, and 3.0-km.
Research Support and Advice:
Committee:
Dr. Steve Herrero
Dr. Robert Barclay, University of Calgary, ecology
Dr. Nigel Waters, University of Calgary, GIS, modeling
Dr. Susan Glenn, Univeristy of British Columbia, landscape ecology
GIS Support:
Scott Jevons, GIS consultant
Jack Wierchowski, GIS consultant
Statistical Support:
Dr. Tak Fun, University of Calgary, biostatician
Dr. Trent McDonald, WEST Inc, biometrician, RSF specialist
Dr. Mark Boyce, University of Alberta, RSF specialist
Current Status:
Running statistical tests. Going to Wyoming to work on model selection with Trent McDonald.
Outcome:
From the dissertation will be a set of RSF models that detail what characteristics female grizzly bears select. These are Realized Habitat Models.
Time Frame: Ph.D. defense anticipated in June 2001.

0.00 avg. rating (0% score) – 0 votes