This is a system for evaluating habitat quality for black-tailed deer and moose on the basis of available food, its nutritional quality, and the nutritional requirements of the adult female segment of the population.  We focus on food because it clearly sets the potential upper limit on the number of animals a habitat can support.  Forage resources (vegetation and nutritional quality) can be measured in the field and can be manipulated by land management.  We focus on digestible energy and digestible protein, because they are the two most common nutritional limiting factors for wild ungulates, and their requirements are reasonably well known for black-tailed deer and moose.  We focus on adult females because they are the productive segment of the population, the animals that produce young.  Nutritional requirements vary seasonally and with reproductive status (e.g., maintenance versus lactation).  This system is suitable for any habitat and any species of deer (Cervidae) where the availability of forage, its nutritional quality, and the nutritional requirements of the deer are known.

           The analytical system provides a “snapshot” analysis at one, user-specified point of time.  It is assumed that all available vegetation is potential food, and there is no accounting for herbivore-plant dynamics (effects of overbrowsing).  This is not a simulation model.  Rather, it is a calculator:  given specific values of available forage biomass, nutritional quality, and animal requirements, it calculates the maximum number of “animal days” (one adult female for one day) that can be sustained by the forage resource. The animal-day values are best considered in a relative (comparative) sense, not as absolute values, because they are the maximum number of animal days (at one point in time) that can be supported by the food if all the suitable food were eaten then (no herbivore-plant feedback).  The instantaneous, food-based “carrying capacity” of a habitat varies daily with marked changes in plant phenology  and animal metabolic requirements during summer and with variation in snowpack during winter.  Our snapshot values are valid for the point of measurement and analysis; they are not an annual, or even seasonal, average.  They are best used for comparing two or more habitats, or the same habitat(s) in two or more different states (times or post-manipulation, succession, etc.).  Additionally, a large scale, GIS application of the system is useful for comparing various landscapes or patterns within the same landscape – where size and spatial configuration of habitats are important considerations.

            The FRESH system includes linked databases of understory biomass and forage-specific nutritional data for a variety of habitats and forages.  The databases provide users examples or reference points (in the case of habitats) and ballpark-level default estimates (in the case of forages) for data that the user might not have.  Although field data directly from the user’s study area always are best, a user can use the FRESH system to explore habitat relations with estimates of habitat variables based on the examples and default values from the linked databases stored within the FRESH system.

            The calculation of animal days is accomplished with a linear programming model.  Linear programming models are optimization algorithms, where an objective is maximized (or minimized) within a set of maximum or minimum constraints.  They are called linear programming models because they consist of a series of equations, each of which is written as a linear equation (e.g., Y > a + bc + de +fg), and the solution is found by solving all equations simultaneously.  In our case, the objective that is to be maximized is the quantity of forage biomass (kg/ha) that can be pooled from all available biomass while meeting (or exceeding) specified minimum constraints for digestible energy and digestible protein.  The FRESH-moose linear program also includes an additional constraint for time costs of feeding (ingestion) as a function of twig size; the objective (suitable biomass) must be maximized within a specified maximum amount of time spent feeding (minutes/day) in addition to the minimum concentrations of digestible energy and digestible protein.

            We are essentially answering the question, “If one were to harvest all the available current annual growth of plants in a given area and bring it into a captive animal facility for feeding, what is the maximum number of animal days that could be supported while meeting (or exceeding) a user-specified level of metabolic requirements (mean concentration of digestible energy and digestible protein)?”  We answer that question by using a linear programming model to determine the maximum amount of forage biomass that can be pooled (from the combination of all available forages) while meeting the specified constraints.  The maximum suitable biomass (kg/ha) is then divided by the user-specified daily dry-matter intake of an adult female (kg/day), yielding the maximum number of animal days (days/ha) that can be supported within the specified constraints.

            The data-entry requirements for analyzing any given habitat are the following:  (1) a list of all available (a forage is a plant part with a unique nutritional composition – e.g., different species are different forages; shrub leaves are different forages from shrub twigs, even within the same species); (2) the available biomass (kg/ha, dry weight) of each forage; (3) the concentration of digestible energy (kJ/g) of each forage; (4) the concentration of digestible protein (percent, dry weight) of each forage; (5) the daily energy requirement (kJ/day) of an adult female; the daily digestible protein requirement (g/day) of an adult female; and the daily dry-matter intake (g/day) of an adult female.  In the case of the FRESH-moose calculations, the forage data also must include the distribution of twig sizes (basal diameters, mm) and the total time available for feeding by the moose (min/day).  We provide guidelines for all the animal data inputs; the forage variables (biomass and nutritional quality) are the chief data requirements from the user.  If the user does not have their own plant nutritional data, they can use our plant nutritional database (automatically linked database) to provide estimates.

            Animal requirements for digestible energy and digestible protein are well known because energy and protein are such basic currencies, and common limiting factors, in animal nutrition that they have been studied extensively.  Energy is needed for maintaining body heat, fueling activity and growth, and conducting basal metabolic processes.  Protein is needed for building and maintaining body tissue.  Not all energy and protein consumed by an animal, however, is metabolically available to the animal.  “Gross energy” (kJ/g) is the total energy contained in the food – the amount that would be released by burning it in a fire.  Only the energy in the digestible portion of a food, however, is available to the animal.  Energy in the indigestible fraction passes from the animal in excretory products (feces and, to a lesser extent, urine), unused by the animal.  The energy in the digestible fraction is called “digestible energy.”  Some of that is lost as heat in conversion to “metabolic” energy, but the metabolic energy coefficient for most forages is nearly constant (at about 0.85), so we have focused the analysis on digestible energy rather than metabolic energy per se.

            Similarly, “crude protein” (which is calculated simply as 6.25 times the nitrogen concentration of a forage) is only a rough index of the true protein content of a forage.  More importantly, however, the digestible fraction of the protein in a forage can vary greatly, especially for wild forages commonly eaten by deer and moose, which may contain tannins and other protein digestion-reducing compounds.  Thus, the “digestible protein” concentration of forage is much more important than its crude protein concentration when evaluating forages for deer and moose. 

            Because digestible energy and digestible protein are so important to animals and have been so widely studied, there exist good laboratory analytical techniques for measuring or estimating their values in forages, some specifically designed for deer and moose.  Similarly, much research has focused on energy and protein requirements of animals, including deer and moose, so those requirements also are reasonably well known.  Thus, digestible energy and digestible protein are our most sound currencies for evaluating forage quality for deer and moose.

            The nutritional quality of plant material, however, is far more complex than simply that of digestible energy and digestible protein.  Animals also require vitamins, minerals, and micronutrients.  Wild forages for herbivores like deer and moose, moreover, also contain many noxious compounds – some affecting palatability, some affecting digestibility, and some even toxic.  The bacterial flora in the ruminant stomach of deer and moose synthesizes vitamins, so vitamins are not an important limiting factor.  Minerals and micronutrients are important, but usually they are in sufficient concentration in forage, so they are less likely a limiting factor than is energy or protein.  However, that is not always true, and in some cases, minerals and/or micronutrients may be a very important constraint – either by being in too short supply, or by being so abundant as to be toxic.  For most deer and moose ranges, however, that is not the situation.  Therefore, mineral and micronutrient requirements of deer and moose have received much less scientific investigation than have energy and protein, and they are much more poorly understood.  We have not included mineral or micronutrient requirements as constraints in our linear programs.

            Noxious organic compounds abound in wild plants, especially in forbs, ferns, shrubs, and trees – all the forages commonly consumed by deer and moose.  Ecologists have frequently termed these compounds “secondary compounds” (secondary to basic plant metabolism) and have considered them as defensive mechanisms protecting plants from herbivory.  They are considered in two groups, differing functionally in the herbivore:  (1) digestion-reducing compounds, which decrease the value of a forage by decreasing its energy and/or protein digestibility; and (2) toxins, which produce acute debilitating effects in the herbivore.  The two groups are not necessarily exclusive.  Plant secondary chemistry is an enormously complex subject, with thousands of compounds in the environment and different compounds in virtually every forage.  The effects of one major class of compounds (tannins) have been studied well enough to be incorporated into laboratory analytical techniques for estimating digestible energy and digestible protein, but most effects of most compounds are very poorly understood.  Plant secondary chemistry is far too complex to model, yet it has very real effects on herbivore use of plants – either through learned behavior or through innate palatability preferences/avoidances of specific forages.

            “Nutritional wisdom” (the ability of animals to select their food on the basis of its nutritional value) is a very old concept, yet it has never been found to be true.  Palatability of forages and diet selection in large herbivores are far more complex than simply the nutritional value of the food.  Plant secondary chemistry complicates the problem even further.  We must acknowledge that forage quality is more than simply digestible energy and digestible protein, and we must account for the fact that some forages are avoided (unpalatable) regardless of their apparently high quality in terms of digestible energy and digestible protein.  But we must do this within a relatively simple model.

            We have taken a very pragmatic approach to the solution of the “unpalatability” problem – the problem of apparently good forages (in terms of digestible energy and protein) not being eaten or eaten in only small proportions by deer and moose, despite an abundant availability.  This could pose a significant problem in the linear programming solution, for example, when a forage such as alder (Alnus spp.) is relatively abundant, has high digestible energy and protein concentrations, and yet is eaten in only very small proportions by deer or moose.  The linear programming solution would include much or all of the alder, regardless if it comprised a large proportion of the solution, and therefore, would inflate the apparent value of the habitat beyond its true value.  Our approach to resolving this problem in a simple, pragmatic way is to add additional, forage-specific constraints to the linear programming model, whereby an upper limit (maximum constraint) can be specified for each forage known to be relatively unpalatable, regardless of its digestible energy and protein concentrations.  Thus, if 3.0% is specified as the maximum constraint for alder, then that will be the maximum amount of alder in the linear programming solution.  Values for the forage-specific constraints are user-specified.  However, the user can base their choice of such values on the results of diet composition studies (e.g., fecal composition or rumen analyses) of deer or moose in similar habitats.  Thus, the forage-specific constraint can be based on field data, not just professional opinion.  This provides a relatively simple and workable solution to a problem that is far too complex to model biologically.  The linear programming solution is driven by forage availability, digestible energy, and digestible protein, but it is restricted by empirically determined limits on palatability of any given forage.

            “carrying capacity” is a very ambiguous term in ecology.  It is usually meant to be the maximum number of animals (of a given species) that a given habitat can support indefinitely.  The “indefinitely” aspect requires that trophic dynamics (herbivore-plant, predator-prey) be considered and that the system maintains a stable equilibrium for long (indefinite) time.  It is a useful concept for theoretical system modeling, but it is very problematic for practical application.  In the real world, virtually no habitat is stable indefinitely.  Seasonal variations occur throughout the year; annual variations occur between years (e.g., weather); and disturbances, succession, ecological change are present over both short and long time scales.  “Maximum number of animals” can vary, depending on sex and age composition desired or assumed, as well as expectations for productivity of the population (e.g., “maximum sustained yield”).  Herbivore-plant dynamics are seldom, if ever, truly stable (even within “dynamic equilibrium bounds”).

            In practical application, carrying capacity is best determined empirically, after carefully defining exactly what is meant about location, time scale, reasonable limits of natural variation (e.g., are droughts or floods included?), and animal population demographics.  This has been done occasionally for closely managed deer populations, and it has been done extensively for managing livestock grazing.  It requires much empirical experience with the animals and the habitats, including sound data on animal demographics and vegetation dynamics, and usually, careful manipulation of the animal population.  Yet, extrapolation to other habitats than those specific lands studied involves much uncertainty.  For most large, free-ranging populations of deer and moose, empirical determination of carrying capacity is impossible.

            Theoretical calculation or estimation of carrying capacity for large herbivores is confounded by two major problems:  (1) Food quantity and quality are not substitutable for one another, and (2) the diet selection process, central to predicting diet composition, has remained an exceedingly difficult process to model or predict.  The combination of these two problems has implications well beyond estimating carrying capacity on the basis of food supplies.  It also confounds the interpretation of results from habitat use studies (e.g., habitat selection or preferences) and models of habitat quality derived from such data (e.g., “resource utilization functions”).  

            The problem of “nonsubstitutability” of quantity and quality of food is that poor quality food cannot be substituted for deficiencies of high quality food – in other words, much poor food is not equal to less good food.  The reason for this is that herbivores are limited by the amount of food that they can process (ingestion, digestion and passage, or both).  Once they reach their limit of intake, they cannot consume more; they cannot make up for poor quality food by eating more of it.  The most common cause is bulk-passage limitations through the gastrointestinal tract (especially problematic for ruminants, with their four-chambered stomach).  Thus, one cannot simply multiply the biomass of forages by their digestible energy (or protein) concentrations, sum for all forages, and then divide by the animal’s daily metabolic requirement to determine the number of animal days that the food can sustain.  Furthermore, virtually every forage in the habitat is unique – it has a unique combination of biomass and nutritional quality.  Thus, it is the combination (mix) of forages that must be optimized in order to quantify the quality of a habitat’s food resources.  Some of the potential foods may or may not be suitable, depending on what else and how much of it is in the diet.

            Empirical observations of diet composition of herbivores in the habitat provide a way of determining a suitable mix of potential foods.  Some models of carrying capacity apply empirically observed diet composition to the array of potential foods in the habitat to determine the maximum quantity of food that can be mixed in that same proportion.  The problem with this is that diet composition is not a static, fixed attribute of an animal-habitat interaction.  It varies with the relative availabilities of the various foods, and therefore, varies with animal population density as well.  As population density increases (“how many animals can the habitat support?”), for example, the most preferred foods decrease in availability, and diet composition shifts, yet it still might yield a diet that is well above minimal requirements.  This is the similar problem with habitat use studies – habitat preferences (and resource utilization functions), should be expected to shift with changes in relative availabilities of habitats in a landscape and with population densities of the animal, the latter because food quantity and quality are not substitutable for one another.  The best habitat for an individual animal (the level at which habitat preferences are chosen) may be high quality, low quantity food at low population densities; but at high population densities (i.e., carrying capacity), the habitat that supports the most animals may be lower (but sufficient) quality, high quantity food – i.e., exactly the opposite patterns of habitat use.   An accurate diet prediction model (e.g., “optimal foraging theory”) would enable us to account for the interaction of forage availability, nutritional quality, and herbivore population density; but no such model exists.

            We have taken a very pragmatic approach to resolving this dilemma:  we have avoided the prediction of diet composition altogether; similarly, we do not predict habitat preference or use, either.  Rather, we move directly to the question, “what is the maximum number of animals (or animal days) that can be supported by a given food resource at a given level of metabolic requirement?”  Our answer is the maximum number of animal days that can be supported at that instant of analysis (the time when food supplies were measured), without concern for an “indefinite” or “stable” concept of carrying capacity (herbivore-plant interactions) or the composition of diets at anything less than the maximum density of animals.  We calculate the dietary mix that solves the optimization problem of the linear program:  maximize the quantity of suitable forage biomass within the specified minimum constraints of digestible energy concentration and digestible protein concentration (and for moose, the maximum constraint of foraging time) and maximum, forage-specific limits for forages known to be unpalatable.  Our solution would be the “optimal diet” only at that population density that is the maximum for the habitat.  This is not a diet prediction model; it is an animal feeding capacity calculator.  It is something more akin to what a feedlot manager would use than to what an ecologist would find exciting.

The FRESH system consists of two levels of application – a web-based, stand-level module; and a GIS-based, landscape-level module. The web-based application runs on our server. The user imports their stand-level data, or types it in directly, or accesses stand-level data from our linked database. Typical applications might be comparison of results from silviculture treatments, or before-versus-after treatment, or various types of old-growth forests, or for maintenance versus reproductive requirements, etc. Data are analyzed for one stand or habitat – one array of forage availability – at a time.

Where landscape pattern of habitats is an important consideration, then the GIS-based application is needed. The GIS application is downloaded from our website directly onto the user’s computer; the user must also download a GIS dataset onto their computer; and then all calculations take place on the user’s computer. The user must have a description of the forage resources and overstory canopy cover of each habitat type in the landscape. The habitat types must be mapped in the GIS system; and for winter analyses, elevation must be mapped in the GIS system. Additionally, the user must specify a mean home-range size (ha) for the animal. The FRESH system then analyzes all GIS cells within a block the size of the mean home range, computing available forage after accounting for habitat type, elevation, and burial by snow (if winter). It compiles all available forage within the home range into one composite array of available forage, and then it uses the same linear programming algorithm as in the stand-level module to calculate an animal-days value for that “home range.” It then moves over, by a user-specified amount of space, to another, overlapping area of home range, and repeats the process. After all the landscape has been analyzed in this manner, the mean is calculated across all deer-day calculations for all home ranges. This is known as a “moving window” analysis. It accounts for the size and spatial locations of each habitat within a scale appropriate for the animal. The underlying assumption is that each animal knows an area the size of its home range and can mix forages from anywhere within that home range. The synergistic effect of combining forages from different habitats can be greater than the sum of the values of each habitat considered in isolation. However, the combination of habitats must occur at the scale of what an individual animal knows – i.e., its home range – because diet composition and nutritional requirements operate at the scale of the individual animal.

Typical applications of the GIS-based module might be comparisons of landscape-level management areas, or the same area under various potential management alternatives or temporal patterns for a landscape as it changes with plant succession, etc. Data are analyzed for all stands in the landscape, all within the same run, and all at the same point of time.

            The FRESH system can be applied to any species of large, generalist herbivore (i.e., herbivores that consume a mixed-species diet without specializing on one or a few forages) for which its nutritional requirements are reasonably well known and the food resources of its habitat can be quantified.  It can be applied anywhere.  If additional constraints are needed (e.g., concerning micronutrient limitation in a particular locale), then the FRESH system cannot be applied directly.  Its theory would still be applicable, but the system would need to be modified to incorporate the new constraint(s).     

            The current system has been developed for black-tailed deer and moose in Alaska .  The moose application is basically the same model as that for deer, except that it includes an additional constraint for foraging time.  The time constraint is needed because browse forages consumed by moose often occur as very large twigs, and the size of bite taken while feeding on such twigs involves a trade-off between nutritional quality and time costs of harvesting.  Large bites are more time efficient (g/min) than are small bites, but small bites (distal ends of twigs) have higher concentrations of digestible energy and digestible protein than do large bites.  Thus, each twig presents an array of opportunity (availability) to the moose.  Part of the optimization problem is choosing the appropriate bite size(s) for each twig (browse) species.

            Of course, forage resources of deer and moose range, and metabolic requirements of deer and moose, differ substantially between the species.  However, those are simple matters to adjust through user-specified data input in the FRESH system.

            As of January, 2006, FRESH-Deer is fully operational at both the stand-level (web-based) and landscape-level (GIS-based) scales of application.  We are currently working with the Tongass National Forest to increase the linked databases for both habitat biomass and forage nutritional quality.  We will add data to both databases as they become available.  We also anticipate updating and improving documentation and user-guide information. 

            Ultimately, we intend to expand the analytical framework of the FRESH system to include other factors affecting habitat quality at the landscape scale, beyond forage alone.  We plan to do this by considering the current FRESH food-based estimate of carrying capacity as only the maximum potential (upper limit) of the habitat, and then modifying that value by the “probability of use” of the habitat, determined from a broad “resource utilization function” model.  In other words, the current, food-based FRESH estimate is the carrying capacity that could be achieved if the habitat were fully acceptable to the animal, whereas the resource utilization function model provides an estimate of probability of use of any given patch (GIS cell) of habitat.  Resource utilization functions are calculated from habitat-use data, usually radio-telemetry studies, and are best considered descriptions of observed (past) patterns of use, rather than predictions of future patterns.  However, when multiple habitat-use datasets are available over a wide geographic area and time period, then a resource utilization function model calculated from a meta-analysis of those datasets should be reasonably robust and might offer some predictive insight.  For example, during the 1980s and 1990s in southeastern Alaska, there were four major radio-telemetry studies of habitat-use by black-tailed deer – on Admiralty Island, Prince of Wales Island, Heceta Island, and Mitkof Island .  Together, those studies cover a broad range of southeastern Alaska over a nearly 20-year period.  Resource utilization functions that are consistent across all four studies are likely robust for most of southeastern Alaska ’s islands and for deer population densities similar to those of the past two decades. 

            The combination of a food-based upper limit and a behavior-based probability of use should yield some interesting insights into habitat value for deer.  Also, the differences between the two should be insightful as well.

Thomas A. Hanley, Chief Wildlife Biologist and Team Leader, Alaska Wildlife Habitat Team (Juneau) and Boreal Ecology Cooperative Research Unit (Fairbanks), USDA Forest Service, Pacific Northwest Research Station, Juneau, Alaska

Donald E. Spalinger, Professor, Department of Biological Sciences, University of Alaska Anchorage, Anchorage, Alaska

Kenrick J. Mock, Professor, Department of Computer Science and Engineering, University of Alaska Anchorage, Anchorage, Alaska

Oran Weaver, student, Department of Computer Science, University of Alaska Anchorage, Anchorage, Alaska

Grant M. Harris, Wildlife Biologist, Supervisor’s Office, Chugach National Forest, Anchorage, Alaska

Doerr, J.G, E.J. DeGayner, and G. Ith. 2005. Winter habitat selection by Sitka black-tailed deer. J. Wildl. Manage. 69:322-331.

Gillingham, M.P., K.L. Parker, and T.A. Hanley. 1997. Forage intake by large generalist herbivores in a natural forest environment: bout dynamics. Can. J. Zool. 75:1118-1128.

Hanley, T.A. 1997. A nutritional view of understanding and complexity in the problem of diet selection by deer (Cervidae). Oikos 79:209-218.

Hanley, T.A. and J.D. McKendrick. 1985. Potential nutritional limitations for black-tailed deer in a spruce-hemlock forest, southeastern Alaska . J. Wildl. Manage. 49:103-114.

Hanley, T.A., C.T. Robbins, A.E. Hagerman, and C. McArthur. 1992. Predicting digestible protein and digestible dry matter in tannin-containing forages consumed by ruminants. Ecology 73:537-541.

Hanley, T.A., C.T. Robbins, and D.E. Spalinger. 1989. Forest habitats and the nutritional ecology of Sitka black-tailed deer: a research synthesis with implications for forest management. Gen. Tech. Rep. PNW-GTR-230. USDA Forest Service, Pacific Northwest Research Station, Portland, OR. 52 pp.

Hanley, T.A. and J.J. Rogers. 1989. Estimating carrying capacity with simultaneous nutritional constraints. Research Note PNW-RN-485. USDA Forest Service, Pacific Northwest Research Station, Portland, OR. 29 pp.

Hobbs , N.T. 1989. Linking energy balance to survival in mule deer: development and test of a simulation model. Wildl. Monogr. 101:1-39.

Hobbs, N.T., J.E. Gross, L.A. Shipley, D.E. Spalinger, and B.A. Wunder. 2003. Herbivore functional response in heterogeneous environments: a contest among models. Ecology 84:666-681.

Hobbs , N.T. and T.A. Hanley. 1990. Habitat evaluation: Do use/availability data reflect carrying capacity? J. Wildl. Manage. 54:515-522.

Hobbs , N.T. and D.M. Swift. 1985. Estimates of habitat carrying capacity incorporating explicit nutritional constraints. J. Wildl. Manage. 49:814-822.

McCullough, D.R. 1979. The George Reserve deer herd: population ecology of a K-selected species. Univ. Michigan Press, Ann Arbor , MI. 271 pp.

McArthur, C., C.T. Robbins, A.E. Hagerman, and T.A. Hanley. 1993. Diet selection by a ruminant generalist browser in relation to plant chemistry. Can. J. Zool. 71:2236-2243.

Parker, K.L., M.P. Gillingham, T.A. Hanley, and C.T. Robbins. 1993. Seasonal patterns in body mass, body composition, and water transfer rates of free-ranging and captive black-tailed deer (Odocoileus hemionus sitkensis) in Alaska . Can . J. Zool. 71:1397-1404.

Parker, K.L., M.P. Gillingham, T.A. Hanley, and C.T. Robbins. 1996. Foraging efficiency: energy expenditure versus energy gain in free-ranging black-tailed deer. Can. J. Zool. 74:442-450.

Parker, K.L., M.P. Gillingham, T.A. Hanley, and C.T. Robbins. 1999. Energy and protein balance of free-ranging black-tailed deer in a natural forest environment. Wildl. Monogr. 143:1-48.

Parker, K.L. and C.T. Robbins. 1984. Thermoregulation in mule deer and elk. Can. J. Zool. 62:1409-1422.

Robbins, C.T. 1993. Wildlife feeding and nutrition, second edition. Academic Press, New York, NY. 352 pp.

Robbins, C.T., A.E. Hagerman, P.J. Austin, C. McArthur, and T.A. Hanley. 1991. Variation in mammalian physiological responses to a condensed tannin and its ecological implications. J. Mammal. 72:480-486.

Robbins, C.T., T.A. Hanley, A.E. Hagerman, O. Hjeljord, D.L. Baker, C.C. Schwartz, and W.W. Mautz. 1987. Role of tannins in defending plants against ruminants: reduction in protein availability. Ecology 68:98-107.

Robbins, C.T., S. Mole, A.E. Hagerman, and T.A. Hanley. 1987. Role of tannins in defending plants against ruminants: reduction in dry matter digestion? Ecology 68:1606-1615.

Schoen, J.W. and M.D. Kirchhoff. 1985. Seasonal distribution and home range patterns of Sitka black-tailed deer on Admiralty Island, Southeast Alaska. J. Wildl. Manage. 49:96-103.

Schoen, J.W. and M.D. Kirchhoff. 1990. Seasonal habitat use by Sitka black-tailed deer on Admiralty Island, Alaska. J. Wildl. Manage. 54:371-378.

Shipley, L.A. , J.E. Gross, D.E. Spalinger, N.T. Hobbs, and B.A. Wunder. 1994. The scaling of intake rate in mammalian herbivores. Am. Nat. 143:1055-1082.

Shipley, L.A. and D.E. Spalinger. 1992. Mechanics of browsing in dense food patches: effects of plant and animal morphology on intake rate. Can. J. Zool. 70:1743-1752.

Shipley, L.A. and D.E. Spalinger. 1995. Influence of size and density of browse patches on intake rates and foraging decisions of young moose and white-tailed deer. Oecologia 104:112-121.

Shipley, L.A. , D.E. Spalinger, J.E. Gross, N.T. Hobbs, and B.A. Wunder. 1996. The dynamics and scaling of foraging velocity and encounter rate in mammalian herbivores. Functional Ecol. 10:234-244.

Spalinger, D.E. and N.T. Hobbs. 1992. Mechanisms of foraging in mammalian herbivores: new models of functional response. Am. Nat. 140:325-348.

Spalinger, D.E. and C.T. Robbins. 1992. The dynamics of particle flow in the rumen of mule deer (Odocoileus hemionus hemionus) and elk (Cervus elaphus nelsoni). Physiol. Zool. 65:379-402.

Spalinger, D.E., C.T. Robbins, and T.A. Hanley. 1986. The assessment of handling time in ruminants:  the effect of plant chemical and physical structure on the rate of breakdown of plant particles in the rumen of mule deer and elk. Can. J. Zool. 64:312-321.

Spalinger, D.E., C.T. Robbins, and T.A. Hanley. 1993. Adaptive rumen function in elk (Cervus elaphus nelsoni) and mule deer (Odocoileus hemionus hemionus). Can. J. Zool. 71:601-610.

Van Horne, B. 1983. Density as a misleading indicator of habitat quality. J. Wildl. Manage. 47:893-901.

Van Soest, P.J. 1982. Nutritional ecology of the ruminant. O&B Books, Corvallis, OR. 374 pp.

Wallmo, O.C., L.H. Carpenter, W.L. Regelin, R.B. Gill, and D.L. Baker. 1977. Evaluation of deer habitat on a nutritional basis. J. Range Manage. 30:122-127.

Yeo, J.J. and J.M. Peek. 1992. Habitat selection by female Sitka black-tailed deer in logged forests of southeastern Alaska . J. Wildl. Manage. 56:253-261.