CARMA ’ s approach for the collaborative and inter-disciplinary assessment of cumulative e " ects

32 (1), 2012 This journal is published under the terms of the Creative Commons Attribution 3.0 Unported License Editor in Chief: Birgitta Åhman, Technical Editor Eva Wiklund and Graphic Design: Bertil Larsson, www.rangifer.no Introduction One of the most frequent concerns about the future of migratory tundra caribou, Rangifer tarandus groenlandicus or granti, are the impacts of the cumulative e"ects of changing climate and land-use activities across herd’s ranges. Assessing cumulative e"ects is typically a requirement in environmental assessment of industrial developments but policy and technical limitations have hindered development of assessment methods (Duinker & Greig, 2006). Johnson & St.-Laurent (2011) commented on the lack of a methodological framework as one of the reasons for slow progress on cumulative e"ects. #ey suggested a framework based on the scaling from individual to population, the relative frequency, and magnitudes of e"ects and their regulation. We know quite a bit about individual caribou responses to human activities – interruptions to foraging and displacement of individuals at various distances from the disturbance (Aastrup, 2000; Cameron et al., 2005; Boulanger et al., 2012). However, to scale up from the behavioral responses of individual caribou to the population scale (Johnson & St.-Laurent, 2011) requires baseline information on the ‘state’ of the individual and population giving consideration to, for example, climate, population density, and genetic structure. At both the individual and population scale, we also have to consider environmental in$uences, especially weather and climate, which will be additive or compensatory to impacts imposed by human activities. To scale up the individual’s behavioral responses to the population requires being able to estimate the costs to the individual and whether those costs will a"ect its reproduction and survival. Estimating the costs of a behavioral response is not straight forward; as well as the energy costs of movement and interruption in foraging time, there may also be an e"ect on diet (energy protein intake) if a displacement puts the individual in a di"erent habitat. Understanding and integrating those relationships between behavior, habitat selection, energy and protein intake relative to reproduction and surRangifer, 33, Special Issue No. 21, 2013: 161–166 13 Arctic Ungulate Conference Yellowknife, Canada 22-26 August, 2011


Introduction
One of the most frequent concerns about the future of migratory tundra caribou, Rangifer tarandus groenlandicus or granti, are the impacts of the cumulative e ects of changing climate and land-use activities across herd's ranges.Assessing cumulative e ects is typically a requirement in environmental assessment of industrial developments but policy and technical limitations have hindered development of assessment methods (Duinker & Greig, 2006).Johnson & St.-Laurent (2011) commented on the lack of a methodological framework as one of the reasons for slow progress on cumulative e ects.ey suggested a framework based on the scaling from individual to population, the relative frequency, and magnitudes of e ects and their regulation.
We know quite a bit about individual caribou responses to human activities -interruptions to foraging and displacement of individuals at various distances from the disturbance (Aastrup, 2000;Cameron et al., 2005;Boulanger et al., 2012).However, to scale up from the behavioral responses of individual caribou to the population scale (Johnson & St.-Laurent, 2011) requires baseline information on the 'state' of the individual and population giving consideration to, for example, climate, population density, and genetic structure.At both the individual and population scale, we also have to consider environmental in uences, especially weather and climate, which will be additive or compensatory to impacts imposed by human activities.
To scale up the individual's behavioral responses to the population requires being able to estimate the costs to the individual and whether those costs will a ect its reproduction and survival.Estimating the costs of a behavioral response is not straight forward; as well as the energy costs of movement and interruption in foraging time, there may also be an e ect on diet (energy protein intake) if a displacement puts the individual in a di erent habitat.Understanding and integrating those relationships between behavior, habitat selection, energy and protein intake relative to reproduction and sur-vival is data intensive and interdisciplinary as the understanding is based on ecology, nutritional ecology, and modeling.
Collaborations among researchers and an interdisciplinary focus are among the strengths of the CircumArctic Rangifer Monitoring and Assessment (CARMA) which is a network of shared expertise (Russell et al., 2013).CARMA has worked to develop an approach and associated tools for cumulative e ects assessment.
e principle tools are currently a spatial climate database scaled to herd seasonal ranges, an individual-based energy/protein (E/P) model (Russell & White, 2000;Russell et al., 2005;White et al., 2013) and a population model.CARMA, through international cooperation and collaboration, has also compiled herd-speci c databases on caribou condition and health that is essential as input for modelling cumulative e ects.
We have two objectives in this brief communication.Firstly, to describe the conceptual approach of using CARMA's tools and secondly, to brie y describe how the di erent types of input feed into the models and how the two models work together.e model generates corresponding outputs which are subsequently used to project cumulative e ects.
Conceptually, the approach is to track how environmental conditions and movements affect the energy and nitrogen intake of a female caribou.e model tracks energy/protein input (i.e., diet and foraging time) and then the model projects how a cow allocates her energy and protein balance for the probability of pregnancy, fetal growth during gestation, and calf growth during lactation.
e pregnancy rates and calf survival are linked to a population model, which in turn tracks vital rates and trends in abundance.

Methods
e energy-protein model integrates the state of an individual caribou (e.g., body size and condition) on a particular landscape which has speci c attributes (e.g., vegetation type, forage biomass, snow cover, and insect harassment).
e approach accommodates responses to human activity as measured through displacement and/or daily activity budgets (i.e., behavioral responses).ose responses can include a reduction in foraging time for caribou close to the development based on measured activity budgets, increased activity costs (e.g., due to avoidance of human activity), and displacement away from the development that may result in foraging in di erent plant communities which a ects diet quantity and quality for the individual caribou.e energy-protein model converts the diet to protein and energy reserves by tracking the physiological steps of digestion and metabolism and then allocates protein and energy to maintenance, protein and fat reserves, body growth, fetal growth, and calf growth (based on milk production).
To describe the di erent types of input we use the example of a population on its postcalving summer range.
e rst set of data input relate to the landscape.To describe forage quantity, the model input starts with the relative abundance of plant cover types derived from a vegetation classi cation typically based on Landsat satellite imagery.e frequency of the caribou's use of those plant cover types is derived from habitat selection modelling such as resource selection functions (Manly et al., 2002).Estimates of above-ground green biomass available in the plant cover types during the growing season are available from satellite imagery (i.e., the normalized vegetation difference index (NDVI)).
e energy-protein model tracks 10 plant groups in the caribou diet (moss, lichens, mushrooms, horsetails, deciduous shrubs, evergreen shrubs, forbs, graminoids, standing dead graminoids, and cotton grass heads).e relative abundance of those plant groups among the plant cover types has been described using eld measurements which can then be applied to other landscapes on the tundra.
e inputs to the model for forage quality (e.g., nitrogen concentration, digestibility including secondary compounds of shrubs) are based on a relationship that associates published plant nutrients with phenological stage based on growing degree-days and biomass.
e model can use as input eld measurements of diet or if those data are unavailable, the model can generate a likely diet based on known nutrient requirements, forage biomass, and forage quality.
Growing degree-days, as well as other climatic variables that a ect caribou activity patterns (e.g., index of insect harassment) or diet (e.g., mushroom growth index), as input to the energy-protein model are derived from one of CARMA's other tools.We downloaded the retrospective spatial data at the scale of seasonal ranges for all circumpolar caribou herds and developed caribou-relevant variables (Russell et al., 2013).e climate data are available as a spreadsheet and a searchable database organized at the level of seasonal ranges for individual herds.e herd database is available on request to CARMA.e climate data themselves are from NASA's Modern Era Retrospective-analysis for Research and Applications (MERRA) website (http://gmao.gsfc.nasa.gov/merra/).
e next set of model inputs include daily activity budget, which can include those budgets when caribou are responding to disturbances by reduced time spent foraging and increased energy costs of walking or running away.e model inputs also require an assessment of the individual caribou's initial body condition.is is an advantage o ered by CARMA which has compiled from historic sources, extensive herdspeci c data and metadata on condition.e same databases are also useful as a validation of the model's projected probabilities of pregnancy which the model generates from fall body condition of the cow.
With these inputs, the energy-protein model can run scenarios to examine the possible range of e ects of industrial development and or climate.e scenarios can include the degree of changes in distribution as a result of displacement which are tracked through shifts in habitat type (thus diet quantity and quality) and changes in density (tracked by plant biomass) if the displacement changed the relative density of the caribou.

Results and Conclusion
ere is complexity in the modelling approach but the integration of spatial data using the habitat selection models has been successfully incorporated into the energetic model during a demonstration project for the Bathurst herd (Nishi et al., 2009;Gunn et al., 2011;Adamczewski et al., In press).A signi cant advantage of the ability to integrate spatial data is that it allows the inclusion of longer-term datasets such as those held by aboriginal elders.For the demonstration project on the Bathurst herd's range, we were able to work with the Tlicho to include longer-term information on caribou distribution across the landscape based on knowledge from the elders (Legat et al., 2001).
e original energy model (Russell et al., 2005) was linked to a population model to explore the impacts of climate change and development on the Porcupine caribou herd (Kruse et al., 2004).We are presently in the process of linking the current energy-protein model to a "Caribou Estimator", a model that projects populations into the future with the focus on assessing the impacts of harvest policies on the productivity of herds.at linkage will allow decision-makers to consider the health of populations into their harvest management planning.
In the context of cumulative e ects, CAR-MA's approach o ers four key features.Firstly, it allows the scaling up from individual to population responses to environmental changes in-cluding climate and industrial exploration and developments.Secondly, the energy-protein model is exible in data input.Predicting the body condition of the individual uses monitoring data or values through literature review and expert opinion.For those inputs with signi cant uncertainty in their values, a range (or distribution) of values can be provided.irdly, the model is adaptive in that it incorporates recent data about typical caribou responses to human development (e.g., a large open pit mine).Fourthly, during an environmental assessment, the approach can assist with decisions about cumulative e ects by allowing the relative ranking of the relative e ects of di erent response scenarios based on, for example, degrees of displacement across seasonal and annual ranges (Russell, 2012).
e adoption of individual-to populationscaled modelling in cumulative e ects is a recent development although the need for the approach has been long recognized.ere are other energy-based models for caribou (White et al., 1975;Boertje, 1985;Camps & Linders, 1989;Fancy, 1986;Bergerud et al., 2008), and Boertje's (1985) model is being used in the recent environmental assessment for a diamond mine in the Northwest Territories (Mackenzie Valley Environmental Impact Review Board public registry http://www.reviewboard.ca/).However, we are not aware of any model that tracks complete energy-protein balance and no other modeling approach designed to address both the cumulative e ects of climate change and incremental human activity.
CARMA's tools for cumulative e ects assessment work together to couple the state of an individual or population to the cumulative e ects of climate change, industrial development, and harvest on circumpolar Rangifer herds.To be useful as tools, CARMA's models have to be relatively available and so CARMA is working to ensure that the models are web-based and accessible.Steps such as graphical comparison of alternative model scenarios, modular approach for sharing parameters between herds, built-in capability to edit model inputs in Microsoft Excel®, the ability to make multi-year runs, and the capability for stochastic Monte Carlo simulations are all underway.