BROAD-SCALE CORRELATES OF VERTEBRATE RICHNESS AS A BIOGEOGRAPHIC FOUNDATION FOR MAINE GAP ANALYSIS
Randall B. Boone, Department of Wildlife Ecology, and
William B. Krohn, Maine Cooperative Fish and Wildlife Research Unit, University of Maine, Orono
FIGURES 



Terrestrial vertebrate richness were response variables:  

INTRODUCTION 
Underlying the potential occurrence predictions made in Gap Analysis are biogeographic relations that limit the ranges of species across regions. Gap Analysis shows where species are whereas biogeographic analyses explain why species are distributed as they are. Thus, understanding biogeography is important to conservation.
To explore biogeogeographic relations in Maine terrestrial vertebrates, richness patterns for amphibians, reptiles, mammals, birds, and all species were compared to geomorphology, climate, and woody plant distributions. Correlation analyses were conducted using multiple linear and tree regression; variation was partitioned into spatially-structured and non-structured components.  



Explanatory variables describing geomorphology, climate, and woody plants were compiled:  

 

 METHODS  
    The ranges of 275 species that regularly breed in Maine were mapped using observations, literature, and expert review (Boone and Krohn, Unpublished manuscripts). Bird ranges were assessed using the Breeding Bird Survey, and compared well (Boone 1996). Measures of geomorphology were from a digital elevation model of Maine. Climate was modeled using regression analyses (see Boone, In press). Woody plants were drawn from dot maps in McMahon et al. (1990).
     Three methods were used to compare ranges to the environment. Individual ranges were compared to environmental data using methods akin to ranked tests, followed with cluster analyses to identify patterns in relations to the environment. Multiple linear regression and tree regression (Clark and Pregibon 1992) related richness to explanatory variables. Because these data are spatially autocorrelated, a method (Bocard et al. 1992) was used that partitioned variation into spatial and non-spatial components.



Percent variation in vertebrate species richness explained using geomorphology, climate, and woody plant distributions:  

 

 RESULTS  
     In multiple regression models, climate variation explained richness better (78% of variation explained for all species) than woody plant variation (67%) and geomorphology (56%). Reptiles were highly correlated with environmental variation (95%), followed by mammals (73%), amphibians (63%), and birds (57%). Tree regression results were similar, except tree regression models explained much more variation in richness (+24%) based upon spatially-structured environmental variation. Climatic variation was most closely associated with total vertebrate richness (92%), with woody plants and geomorphology explaining about 87%
     In individual comparisons of ranges and explanatory variables, mammalian ranges were more often related positively to snowfall and elevation than for other groups. Ten mammals and 27 birds were positively related to snowfall, elevation, and slope. All amphibians and reptiles, 9 mammals, and 47 birds were positively related to frost-free period and temperature measures:  



Variation explained by all environmental variables:  

 
Vertebrate
class
Species positively
related to productivity (n)
Species positivey
related to topography (n)
Other patterns (n)
Amphibians
4
0
2
Reptiles
12
0
1
Mammals
9
10
2
Birds
47
29
4
Total species
75
39
9
Woody plants
141
17
8
DISCUSSION  
     In general, ranking vertebrate classes as to how much variation was explained by environmental measures (from better to poorer) yields reptiles, amphibians, mammals, and birds. Reptiles were linearly associated with environmental variation, with nonlinear relations and interactions more important for amphibians and mammals, and most important for birds. Birds were least- well explained in linear regression, which spawned further analyses, described in the poster "Forest birds and woody plants: broad-scale biogeographic relations." Future research will entail predicting the occurrence of species using Gap Analysis methods, then repeating these analyses. As a test of Gap Analysis, correlations between richness and environmental variables should improve using maps of predicted occurrence.  
LITERATURE CITED 

Bocard, D., P. Legendre, and P. Drapeau. 1992. Partialling out the spatial component of ecological variation. Ecology 73:1045-1055.

Boone, R.B.  1996.  An analysis of terrestrial vertebrate diversity in Maine.  Ph.D. Thesis, University of Maine, Orono. 

Boone, R.B. In press. Modeling the climate of Maine. Northeast Naturalist.

Boone, R.B. and W.B. Krohn. Unpublished manuscripts (four volumes). Amphibian and reptile-, Mammal-, Non-passerine-, and Passerine species synopses. Maine Cooperative Fish and Wildlife Research Unit, USGS Biological Resources Division, University of Maine, Orono.

Clark, L.A. and D. Pregibon. 1992. Tree-based models. Pages 377-419 In Stastistical models in S. J.M. Chamber and T.J. Hastie (editors). Wadsworth and Brooks/Cole, Pacific Grove, California.

McMahon, J.S., G.L. Jacobson, Jr., and F. Hyland.  1990.  An atlas of native woody plants of Maine: a revision of the Hyland maps.  Maine Agricultural Experiment Station, University of Maine, Orono. Bulletin 830.