Reply from M.S. Boyce, L.L. McDonald and B.F.J. Manly

Abstract

We strongly support the assertion of Mysterud and Ims1xMysterud, A. and Ims, R.A. Trends Ecol. Evol. 1999; 14: 489–490Abstract | Full Text | Full Text PDF | PubMedSee all References1 that resource selection function (RSF) models2xResource Selection by Animals: Statistical Design and Analysis for Field Studies. Manly, B.F.J., McDonald, L.L., and Thomas, D.L. CrossRefSee all References2 can depend on the availability of resources. This point might apply at any spatial scale3xJohnson, D.H. Ecology. 1980; 61: 65–71CrossRefSee all References3, not just for the selection of habitats within a home range. However, in some populations RSFs are independent of availabilities, as we explained in our review4xBoyce, M.S. and McDonald, L.L. Trends Ecol. Evol. 1999; 14: 268–272Abstract | Full Text | Full Text PDF | PubMed | Scopus (352)See all References4. Understanding why and how selection changes when availability changes is an important problem.We also concur with Mysterud and Ims1xMysterud, A. and Ims, R.A. Trends Ecol. Evol. 1999; 14: 489–490Abstract | Full Text | Full Text PDF | PubMedSee all References1 that building overly simplistic RSF models could be naive because of complex behaviors and interactions between resource requirements. One of the strengths of the RSF approach for modeling habitat selection is that it easily accommodates such complications. We believe that habitat selection seldom can be sufficiently characterized using only two habitat types, as in the two examples offered by Mysterud and Ims5xMysterud, A. and Ims, R.A. Ecology. 1998; 79: 1435–1441CrossRefSee all References5. There are no barriers to the development of models relating multivariable RSFs to availability. Examples with more than two prey types have been published6xThe Statistics of Natural Selection. Manly, B.F.J. CrossRefSee all References6, for example, for rudd (Scardinius eryophthalmus) feeding on the aquatic hemipteran Sigara distincta7xGreenwood, J.J.D. and Elton, R.A. J. Anim. Ecol. 1979; 48: 721–737CrossRefSee all References7. Such RSFs for selection of a prey type can be readily applied to habitat selection. We explicitly reviewed4xBoyce, M.S. and McDonald, L.L. Trends Ecol. Evol. 1999; 14: 268–272Abstract | Full Text | Full Text PDF | PubMed | Scopus (352)See all References4 an alternative approach where RSFs can be constructed for each individual’s home range; a general RSF can then be obtained by modeling RSF coefficients (as response variables) as a function of availabilities. This is a more general and more flexible approach than estimating a ‘habitat functional response’5xMysterud, A. and Ims, R.A. Ecology. 1998; 79: 1435–1441CrossRefSee all References5.We take issue with the suggestion that using RSFs assumes that use is proportional to availability1xMysterud, A. and Ims, R.A. Trends Ecol. Evol. 1999; 14: 489–490Abstract | Full Text | Full Text PDF | PubMedSee all References, 5xMysterud, A. and Ims, R.A. Ecology. 1998; 79: 1435–1441CrossRefSee all References; if this were so, animals would be showing no selection whatsoever. As Mysterud and Ims1xMysterud, A. and Ims, R.A. Trends Ecol. Evol. 1999; 14: 489–490Abstract | Full Text | Full Text PDF | PubMedSee all References1 point out, habitat use patterns are seldom proportional to availability. We recognize that our contention relates to different definitions of availability. Given the spatial scale defined by Mysterud and Ims5xMysterud, A. and Ims, R.A. Ecology. 1998; 79: 1435–1441CrossRefSee all References5, we assume that availability refers to the total or mean resources available within an individual’s home range. But, regardless of the availability definition, RSFs generally are not proportional to availability – RSF modelling is flexible and virtually any function can be used as long as the model is proportional to the probability of a resource unit being used. As indicated by Mysterud and Ims5xMysterud, A. and Ims, R.A. Ecology. 1998; 79: 1435–1441CrossRefSee all References5, one can easily incorporate polynomial or nonlinear functions into an RSF where the RSF is not proportional to availability. Indeed, an RSF is constructed for a particular availability regime2xResource Selection by Animals: Statistical Design and Analysis for Field Studies. Manly, B.F.J., McDonald, L.L., and Thomas, D.L. CrossRefSee all References2, and the issue in question is whether varying the availability alters the RSF. Indeed, sometimes it does.Mysterud and Ims1xMysterud, A. and Ims, R.A. Trends Ecol. Evol. 1999; 14: 489–490Abstract | Full Text | Full Text PDF | PubMedSee all References1 make reference to our suggestion that an index of habitat importance, based on caloric value, can be calculated. This is taken out of context. Although it might be worthwhile to have such an index of the food value of various habitats, such an index is not an RSF and cannot be used to extrapolate population size.