Woodland caribou scientific review to identify critical habitat: chapter 9

Appendix 6.2

6.2 Delineating Units of Analysis for Boreal Caribou Critical Habitat Identification

Background


Application of the Critical Habitat Identifi cation Framework for boreal caribou and associated decision tree requiresdelineation of population analysis units and their associated ranges. These analysis units form the basis for analysis to determine probability of persistence, based on range quality and population parameters.

For the purposes of Critical Habitat Identifi cation, units of analysis were provided by jurisdictions and accepted as the best available knowledge. Several jurisdictions with large continuous areas of occupied habitat have not completed local population delineation and therefore only provided extent of occurrence for continuous distribution areas. Local population delineation for these areas is a high priority as indicated in the schedule of studies.

During the Science Review process, it became evident that there was a need for a standardized protocol for identifying local populations and delineating range. There is also a need to reconcile methods for the delineation of local populations and range with variation in local population patterns, habitat fragmentation, and data availability across and within jurisdictions. The discussion below provides guidance that should be used for development of a protocol for local population identifi cation and range delineation as part of the schedule of studies.


Local Population Pattern


Populations often function demographically at scales that are different from those suggested by genetic indicators, therefore we need to distinguish units of analysis that are based on demography from those that are genetically distinct (e.g., Esler et al. 2006). Based on simulation modeling, Hastings (1993) suggested a threshold of <10% migration for defi ning independent demographic units. Dey et al. (2006) also used simulation modeling to demonstrate that sub-populations act as one large population once migration rates reach 20%. Although the question is fundamental to understanding population processes, this topic has received limited study (Waples and Gaggiotti, 2006).

For the purposes of the Critical Habitat Identifi cation Framework, we have defi ned local population as a group of caribou occupying a defi ned area that can be distinguished spatially from areas occupied by other groups. (Note that in most cases, the unit of analysis is the local population.) Local populations experience limited exchange of individuals with other groups, such that population dynamics are driven by local factors affecting birth and death rates, rather than immigration or emigration among groups. Ecological conditions, as well as patterns and intensity of anthropogenic disturbance, vary tremendously across the national distribution for boreal caribou in Canada, resulting in variation in local population patterns.

A local population is the smallest demographic unit with an annual rate of emigration and immigration of ≤10%. Some local populations may be spatially discrete and experience little or no exchange of individuals (≤5%). Local populations may also exist as part of a broader, continuous distribution where periodic exchange of individuals may be greater (> 5% but ≤10%). Alternatively, a local population could occupy a large continuous distribution whereby regular exchange of individuals occurs (> 10% immigration and emigration).


Therefore, there are three possible hypotheses proposed for local population patterns for boreal caribou, based on movement patterns:


1) discrete local populations with spatially discrete ranges
2) multiple local populations within a large area of relatively continuous habitat
3) one large local population across a large area of relatively continuous habitat


From a population dynamics perspective hypothesis 1 and 3 are the same, differing only in the extent of area occupied by a single population. However, there are implications for Critical Habitat Identifi cation and delineation of the units of analysis that require differentiating these two situations as different population patterns.


There are examples in the literature of the use of animal movement data to determine immigration and emigration rates that can be in turn used to assess population patterns. McLoughlin et al. (2002) were able to determine annual exchange rates of 3.4 - 13% for females and 7-35% for males and concluded that grizzly bears populations (determined by cluster analysis of movement data) in their study area should be considered a continuous (open) population. Bethke et al. (1996) concluded, from their analysis of polar bears in the western Canadian Artic, that three populations identifi ed in their study were relatively closed (e.g., with little immigration/emigration of radio-collared females among populations that overlapped for part of the year). The examples presented above are based on short-term studies. Dynamics of boreal ecosystems and caribou biology would need to be addressed in studies designed to assess immigration and emigration rates for boreal caribou over the long term. It should also be noted that populations fi tting one hypothesis may be reassigned to an alternate hypothesis under changed environmental conditions, such as large burns and barriers imposed through human activity, or if new information is provided. Therefore, it is important that population patterns and range be periodically re-assessed and updated.

From a practical perspective, the lack of caribou movement data in some regions will preclude the ability to determine immigration/emigration rates for the purpose of determining spatial population structure. In the absence of suffi cient immigration/emigration data, available animal movement and survey data and the degree of geographic separation of area of occupancy should be used to determine the most plausible hypothesis for local population pattern. The amount and quality of data used to delineate local populations and associated range varies across and within jurisdictions, and the level of certainty to support local population delineation varies accordingly. Uncertainty should be addressed through a schedule of studies and adjustments should be made to local population identifi cation and associated unit of analysis over time.

How does range relate to the unit of analysis for application of the critical habitat decision tree?


Where natural geographic boundaries and/or habitat alteration have resulted in discrete local populations, and range boundaries have been delineated based on animal movement data and forest dynamics data, local populations and associated range are identifi ed andconstitute the unit of analysis for purposes of Critical Habitat Identifi cation.


In areas where caribou local populations are not restricted by natural geographic boundaries or habitat alteration and are distributed across large areas of relatively continuous habitat, the delineation of range for local populations becomes more diffi cult. The entire extent of occurrence for a relatively continuous habitat distribution area should be included in the delineation of units of analysis. This addresses the concern that defi ning discontinuous ranges would eventually result in fragmentation of the continuous distribution, with loss of connectivity among local populations. In the absence of evidence to the contrary, it is also consistent with a precautionary approach.


Range Identifi cation


Range identifi cation can be confounded by multiple factors, which may differ among local populations, and may not be fully understood:


■ the defi nition of range includes factors that constrain vital rates such as predation, food abundance, and other features of habitat quality;
■ caribou often occupy distinct seasonal ranges, especially during summer and winter, so conditions required to maintain connectivity among landscapes used during different seasons needs to be understood;
■ caribou may occupy different and shifting areas within ranges over relatively short time periods, although 'core areas' may be consistent over these periods;
■ caribou may occupy different and shifting ranges over long time frames due to factors that are likely related to disturbances, food supply, and predator abundance, as well as direct and indirect anthropogenic disturbances and other stressors that contribute to alter natural disturbance regimes and food availability;
■ ranges of local populations of caribou have changed historically and contracted in
many parts of the country, so a decision needs to be taken about a 'start date' for range delineation;
■ boreal forests change in response to natural perturbations (fi re, insect outbreak, wind) and through natural vegetation succession with age. Climate change may also infl uence
range conditions, in the short- and long-term. Even in the absence of humans, caribou
respond to such changes by shifting their ranges. Hence range should not be viewed as a static condition in time or space;
■ small remnant caribou populations may exist within smaller ranges than they require for long-term persistence; therefore, future range may be larger than the present range; and

■ cow and bull caribou may have different range use strategies. Sexual differences in range (or total range) will not be understood unless both sexes are observed.

Therefore, 'range use' is a dynamic concept and delineation will require regular assessment and updating (e.g., Racey and Arsenault 2007).


Owing to these dynamics, range could be defi ned in probabilistic terms, based on various sources of information, but especially including data for animal locations. Ranges should be assigned in a manner analogous to home ranges derived for individual animals from location data. In this case guidelines are required for some minimum number of animals from which data are collected, including data from both sexes, and dispersal of collars across the suspected range. There are three commonly used methods for delineation of range based on location data: minimum convex polygon, a parametric kernel estimator of probability, or a non-parametric kernel estimator of probability. The former (MCP) is the least conservative and the latter two methods provide 'probability of occupancy surfaces' that are affected bythe number of observations in the dataset. Where range is based on limited number of observations, it may be possible to assign surrounding habitat components a probability of occupancy based on niche modelling and then to use this information to improve the range estimate. Future work is required to ensure that range delineation protocols adequately address large-scale factors infl uencing the movement and occupancy behaviour of caribou.


Considerations in defi ning range for a local population


1. How is 'current range' defi ned?


Current range is defi ned as a geographic area within which there is a high probability of occupancy by individuals of a local population that are all subjected to the same infl uences affecting vital rates over a defi ned time frame. This defi nition incorporates the idea of probabilistic occupancy, functional infl uences, time, and space.


2. How many observations, over what time period, are required to provide a high probability that the defi ned range is accurate?


The number of observations required likely changes depending on the size of the population and its circumstances. An approximate answer can be determined by plotting range size against the number of observations and looking for an asymptote (Figure 1). The probability that range has been accurately defi ned will also depend on the quality of the data used (casual observation and aerial survey vs. radio-telemetry study, dispersion of observations across the population spatially and between sexes).


Observations made over the past 20 years should be accepted as evidence to delineate range of a local population. That amount of time allows for temporal variability in areas occupied among years and lag effects due to change. However, the fi nal decision onappropriate time period for inclusion of observations should be based on landscape dynamics for the local population of interest.

 



Appendix 6.2 - Figure 1. Hypothetical plot of range size against a number of animal observations required to accurately

Figure 1. Hypothetical plot of range size against a number of animal observations required to accurately defi ne range

define range.



3. Among the area estimators (MCP, parametric kernel, non-parametric kernel), which method should be used (subject to sample size considerations)?


Application of the parametric kernel and non-parametric kernel would defi ne a smaller area that an MCP due to the removal of outliers. A precautionary approach would use the minimum convex polygon method to provide a conservative estimator of range. It isimportant to note that all methods are infl uenced, to some extent, by the number of locations (Girard et al. 2005).


4. After what time period should a range should be re-examined?


Defi ned ranges should be re-examined as new data becomes available and at least every 5 years.


What additional factors should be considered when delineating ranges as units of analyses for multiple local populations within a large area of continuous habitat?


The guidance below essentially provides recommended criteria for subdividing a continuous distribution into local population ranges. Therefore, the emphasis is on considerations for decisions on where to place local population range boundaries within a large area of continuous habitat. It should be noted that in areas where data availability is low, boreal caribou populations may seem to lack any obvious structure. In such cases, this guidance would be equally relevant during the process of acquiring new knowledge.

1. Animal movement and animal survey data


The most defensible and robust method to delineate units of analysis for multiple local populations in continuous habitat is to infer "surfaces" based on monitored animal movement (where the animals have been selected from a wide variety of locations across the landscape).For multiple local populations within a continuous habitat, cluster analysis of movements can be used to defi ne group membership (Taylor et al. 2001). Bethke et al. (1996) used radiotelemetry data and cluster analysis to delineate populations of polar bears (Ursus maritimus) in the high Canadian Arctic. They tested for the presence of spatial clusters of animals based on movement data, then applied a home range estimator to identify the geographic range of populations for conservation purposes. Schaefer et al. (2001) and Courtois et al. (2007) used fuzzy cluster analyses to delineate caribou populations.

Systematic or non-systematic aerial or ground surveys can also be used to delimit seasonal and total range when telemetry data are not available. However, because forest-dwelling caribou are typically most dispersed in spring and summer, winter observations alone are prone to underestimation of range area.


The following criteria should be considered, in addition to, or in the absence of adequate collaring-type data and/or animal survey data, when delineating range for multiple local populations within a continuous occupied habitat distribution.

 

2. Spatial extent


The amount of physical area identifi ed as range for a local population within a continuous distribution is fundamentally important for providing a large enough area to support a potentially self-sustaining local population of boreal caribou.


Available animal movement or survey data should be considered fi rst in determining the spatial extent of the range. Further coarse level guidance for the spatial extent of range for local populations within a continuous distribution can be derived by determining the area required to support a persistent population under density and target population size assumptions. Literature and heuristic PVA results (Callaghan pers. com.) suggest somewhat greater than 300 animals for long-term population viability, given moderate rates for calf and female survival. As an example, if range-wide densities of boreal caribou are 2-3 per 100 km2, and if a population is 300 animals, then a reasonable guideline for a unit of analysis may be in the order of 10,000 to 15,000 km2.

Additional insight into the sizes of ranges required could be derived by examining the sizes of ranges occupied by local populations that are exhibiting λ≥1 and have population size > 300 occupying similar geographic areas or habitats.

3. Modifi ers to spatial extent


The spatial extent must be large enough to account for natural forest dynamics and the presence of alternate habitats. Frequency and size of natural disturbance events should be considered and larger areas defi ned if there is a very aggressive natural dynamics cycle.

 

4. Evidence of shared geography


Consider any evidence, collaring or otherwise that indicates caribou move from one location to another on a seasonal basis, or share common geography for part of a year. Aboriginal knowledge can be very good in determining these connections. The fact that animals share a common connection would mean they likely need to be considered as belonging to the same range.

 

5. Habitat functions and behavioural responses


Large areas sharing a similar expression of habitat functions and behavioural responses warrant being kept within the same range. This would benefi t caribou with behavioural patterns suited to specifi c landscape features or functions, and would facilitate prescription of effective protection measures. The habitat functions associated with caribou life history are expressed in many ways across Canada depending on the topography, hydrology, and surfi cial geology. Ultimately, this refl ects how the animals appear to be achieving refuge (predator avoidance), forage (resources for subsistence) and other requirements on an annual cycle. For refuge and forage, signifi cant behavioural responses of caribou to mountains, foothills, or other rugged terrain, lakes with islands, peatlands, large areas of older conifer forest, nutrient poor landforms, large areas of bedrock exposure etc. warrant consideration in delineating range. Variations in habitat functions occur at all scales and only very large and signifi cant trends should be considered here. A good example in Ontario would be the apparent linkage and interaction of animals that share Lake Nipigon and the mainland, or animals that rely on both the Hudson Bay Lowlands and the shield.

 

6. Predominant Risk Factors


Broad types of risk factors, both natural and human, should be considered in the delineation of range. Anthropogenic disturbance regimes and their cumulative contributions to natural ecological drivers should be considered but do not supersede ecological factors. Dominant risk factors can include forestry, oil and gas and associated roads; fi re or succession, predation by wolves, disease (e.g., brainworm); or aboriginal subsistence harvest. Recognizing that risk factors can exist in many combinations, consideration of broad trends that may occur over specifi c geographic areas will provide additional information for decisions on where to subdivide a larger portion of continuous distribution into local population ranges.

Should the range of a single local population with a large continuous distribution be sub-sampled to identify Critical Habitat?


Delineating single local population areas that are very large (e.g., the NWT distribution as one local population analysis unit, or all of Quebec) may result in a mean condition that masks the variation across the large continuous population range. This could allow for substantial occupied range to be lost (major range recession) and erosion of the national population while still supporting a self-sustaining local population. This would be contrary to the goal and objectives in the NRS (National Recovery Strategy), which stipulates maintaining the current distribution. Such large areas would also fail to have strong demographic connections across their breadth - an important practical and theoretical consideration in Critical Habitat Identifi cation. Therefore, it may be necessary to subdivide large continuous population ranges into smaller contiguous analysis units, as an application of the precautionary principle. These may be best derived along ecological boundaries.


Should Forest Management Unit (FMU) boundaries be used to delineate sampling units within the range of local populations within a continuous distribution?


Our recommendation is to delimit large units of analysis based exclusively on animal movement data and ecological factors as listed above. General objectives for caribou habitat (particularly forest composition and connectivity) could be determined at the scale of the range with specifi c objectives assigned to each FMU partly or totally included in the range. In other words, fi t FMU's into the defi ned ranges for local populations in a continuous distribution rather than the other way around. If the FMU conforms to most of the factors identifi ed as criteria for delineating range as above then it is likely a reasonable unit. As the FMU diverges from the criteria above, then it becomes less acceptable.


The rationale for this approach is supported by the following:


■ FMU's look very different from one jurisdiction to the other (and even within a jurisdiction) varying dramatically in size and shape, and seldom conform to ecological drivers. In some jurisdictions, they are surprisingly dynamic with new confi gurations andamalgamations occurring frequently. In some cases FMU's may represent more than one discrete block of land separated by large distances.
■ Social planning considerations should not override fundamentally important ecological drivers. However, if ecological drivers and social planning considerations are geographically close, boundaries of range may be reconciled with other existing management unit boundaries.


References

Bethke, R., M. Taylor, S. Amstrup, and F. Messier. 1996. Population delineation of polar bears using satellite collar data. Ecological Applications 6:311-317.

Courtois, R., J.-P. Ouellet, L. Breton, A., Gingras, andC. Dussault. 2007. Effects of forest disturbance on density, space use and mortality of woodland caribou. Écoscience 14:491- 498.


Dey, S., S. Dabholkar and A. Joshi. 2006. The effect of migration on metapopulation stability is qualitatively unaffected by demographic and spatial heterogeneity. Journal of Theoretical Biology 238:78-84.


Esler, D., S.A. Iverson, and D.J. Rissolo. 2006. Genetic and demographic criteria for defi ning population units for conservation: the value of clear messages. Condor 108:480- 483.


Girard, I., C. Dussault, J.-P. Ouellet, R. Courtois and A. Caron. 2005. Balancing numbers of locations with numbers of individuals in telemetry studies. Journal of Wildlife Management 70:1249-1256.


Hasting, A. 1993. Complex interactions between dispersal and dynamics: lessons from coupled logistic equations. Ecology 74:1362-1372.


McLoughlin, P., D. Cluff, R.Gau, R. Mulders, R. Case, and F. Messier. 2002. Population delineation of barren-ground grizzly bears in the centrally Canadian Artic. 2002. Wildlife Society Bulletin 30:728 -737.


McLoughlin, P.D., D. Paetkau, M. Duda, and S. Boutin. 2004. Genetic diversity and relatedness of boreal caribou populations in western Canada. Biological Conservation
118:593-598.


Racey, G.D., and A.A. Arsenault. 2007. In search of a critical habitat concept for woodland caribou, boreal population. Rangifer Special Issue 17:29-37.


Schaefer, J.A., M. Veitch, F.H Harrington, W.K. Brown, J.B. heberge, and S.N. Luttich. 2001. Fuzzy structure and spatial dynamics of a declining woodland caribou population. Oecologia 126:507-514


Taylor, M K., S. Akeeagok, D. Andriashek, W. Barbour, E. W. Born, W. Calvert, H. D. Cluff, S. Ferguson, J. Laake, A.Rosing-Asvid, I.Stirling, and F.Messier. 2001. Delineating Canadian and Greenland polar bear (Ursus maritimus) populations by cluster analysis of movements. Canadian Journal of Zoology 79: 690-709.


Waples, R.S., and O. Gaggiotti. 2006. What is a population? An empirical evaluation of some genetic methods for identifying the number of gene pools and their degree of connectivity. Molecular Ecology 15:1419- 1439.

Report a problem or mistake on this page
Please select all that apply:

Thank you for your help!

You will not receive a reply. For enquiries, contact us.

Date modified: