Evaluation of a forage allocation model for Theodore Roosevelt National Park.

We developed a forage allocation model using a deterministic, linear optimization module in a commercially available spreadsheet package to help resource managers in Theodore Roosevelt National Park (TRNP), North Dakota determine optimum numbers of four ungulate species, bison (Bison bison), elk (Cervus elaphus), mule deer (Odocoileus hemionus), and feral horses, in the Park. TRNP staff actively managed bison, elk, and feral horse numbers within bounds suggested by our model from 1983 to 1996. During this period, we measured vegetation at 8 grassland and 12 wooded sites at 1-3 year intervals to determine if model solutions were appropriate for maintaining stable conditions in important plant communities in the Park. The data we recorded at these sites indicated minimal change in plant communities from 1983 to 1996. Changes in most vegetation categories that we expected when animal numbers exceeded model optimums for short periods (decreases in coverage/stem numbers of palatable plant species, increases in bare ground or unpalatable plant species) did not occur consistently under high or low precipitation conditions. The lack of sensitivity of our model to decreases in overall production of palatable plant species that occurred due to drought, fire, expansion of black-tailed prairie dog (Cynomys ludovicianus) colonies, and the spread of leafy spurge (Euphorbia esula) in areas of the Park where we did not have monitoring sites suggested that the model under-estimated the total number of ungulates that the Park could support. Management for population levels of ungulates defined by the model probably led to over protection of common plant communities and insufficient protection of rare plant communities. Detecting changes in rare plant communities could have been accomplished by re-designing our vegetation monitoring program, but changing emphasis to protection of rare plants would have likely promoted under use of grazing-tolerant habitat types, dissatisfaction in tourists visiting the Park to see large mammals, and large increases in cost and intrusiveness of management activities such as fencing and control of ungulate populations. The model was a flawed representation of grazing dynamics in TRNP, but we believe it succeeded in making management personnel aware of the biological constraints they face when making management decisions.