Ecological Applications, 10(4), 2000, pp. 1196±1210 q 2000 by the Ecological Society of America RELATIVE SPECIES RICHNESS AND COMMUNITY COMPLETENESS: BIRDS AND URBANIZATION IN THE MID-ATLANTIC STATES EMMANUELLE CAM,1,2,4 JAMES D. NICHOLS,2 JOHN R. SAUER,2 JAMES E. HINES,2 AND CURTIS H. FLATHER3 1North Carolina State University, Department of Forestry, Raleigh, North Carolina 27650 USA 2U.S. Geological Survey, Biological Resources Division, Patuxent Wildlife Research Center, Laurel, Maryland 20708 USA 3U.S. Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado 80526 USA Abstract. The idea that local factors govern local richness has been dominant for years, but recent theoretical and empirical studies have stressed the in¯uence of regional factors on local richness. Fewer species at a site could re¯ect not only the in¯uence of local factors, but also a smaller regional pool. The possible dependency of local richness on the regional pool should be taken into account when addressing the in¯uence of local factors on local richness. It is possible to account for this potential dependency by comparing relative species richness among sites, rather than species richness per se. We consider estimation of a metric permitting assessment of relative species richness in a typical situation in which not all species are detected during sampling sessions. In this situation, estimates of absolute or relative species richness need to account for variation in species detection probability if they are to be unbiased. We present a method to estimate relative species richness based on capture±recapture models. This approach involves de®nition of a species list from re- gional data, and estimation of the number of species in that list that are present at a site± year of interest. We use this approach to address the in¯uence of urbanization on relative richness of avian communities in the Mid-Atlantic region of the United States. There is a negative relationship between relative richness and landscape variables describing the level of urban development. We believe that this metric should prove very useful for conservation and management purposes because it is based on an estimator of species richness that both accounts for potential variation in species detection probability and allows ¯exibility in the speci®cation of a ``reference community.'' This metric can be used to assess ecological integrity, the richness of the community of interest relative to that of the ``original'' com- munity, or to assess change since some previous time in a community. Key words: capture±recapture models; conservation; ecological communities; estimation; integ- rity; landscape; Mid-Atlantic states; North American Breeding Bird Survey; relative species richness; species pool; urbanization. INTRODUCTION The idea that local ecological processes determine Identifying the nature of anthropogenic alterations local richness has been dominant for years, but there of natural habitats associated with variations in animal is increasing interest in the idea that processes taking and plant diversity is of major interest for management place at larger spatial and temporal scales also play a and conservation purposes. More generally, one of the role in determining local richness (Ricklefs 1987, Tay- main questions in ecology concerns the mechanisms lor et al. 1990, Cornell and Lawton 1992, Zobel 1992, determining local species richness. This question can Cornell 1993, Eriksson 1993, Haila et al. 1993, Rick- be addressed by investigating patterns of spatial or tem- lefs and Schluter 1993, JokimaÈki and Huhta 1996, Hu- poral variation in habitat characteristics and in the at- gueny et al. 1997, Augermeier and Wiston 1998, Karl- tributes of local plant or animal communities. Many son and Cornell 1998). It has been suggested that the studies have focused directly on species richness to size of the regional species pool in¯uences local rich- investigate these relationships (Blake and Karr 1987, ness, i.e., that communities would seldom be saturated, Soule et al. 1992, Herkert 1994). However, the infer- and would be open to colonization from the regional ence that can be drawn from observed relationships pool (Cornell and Lawton 1992). Several studies fo- between local habitat characteristics and species rich- cusing on the relationship between species richness at ness strongly depends on the conception of the pro- different spatial scales have provided evidence sup- cesses determining local richness (Ricklefs and Schlu- porting this hypothesis (Cornell 1993, Karlson and ter 1993). Cornell 1998). Under this hypothesis, spatial variations in species richness may not be exclusively linked to Manuscript received 10 December 1998; revised 2 July 1999; variation in local ecological conditions: differences in accepted 21 July 1999. local richness may also re¯ect differences in the size 4 Address correspondence to: U.S. Geological Survey, Bi- ological Resources Division, Patuxent Wildlife Research of the regional pool. In conservation biology, the ques- Center, Laurel, Maryland 20708 USA. tion of spatial variation in species richness is often 1196 August 2000 COMPLETENESS OF BIRD COMMUNITIES 1197 tackled in relation to human-induced modi®cation of Tilman et al. 1994, Kareiva and Wennergren 1995, habitat characteristics, and fewer species in a site is Boulinier et al. 1998b, Huhta et al. 1998). often interpreted as evidence of impoverishment of the It has been suggested that the in¯uence of urbani- community of interest, associated with habitat alter- zation on species richness depends on the relative im- ation. However, if the pool of available species differs portance of modi®cation of original habitats and the for areas being compared, fewer species in one area increase in new habitats created by urban development may simply indicate a smaller species pool. A com- (JokimaÈki and Suhonen 1993, Blair 1996). Among the munity harboring fewer species should not necessarily factors associated with fragmentation, only the poten- be considered as ``less complete.'' Similarly, temporal tial increase in habitat diversity may have a positive variations in local richness may partly re¯ect modi®- in¯uence on species richness (AndreÂn 1994). Lancaster cations of the regional species pool. Under this view, and Rees (1979) suggested that lower levels of urban- it is desirable to take the available species pool into ization lead to habitat diversi®cation, whereas higher account when comparing communities over space or levels lead to loss of habitat diversity. In highly ur- time, and to focus on ``relative'' species richness rather banized areas, the other factors associated with frag- than species richness per se, when addressing questions mentation (i.e., reduction in patch size, increasing iso- about local factors. lation of habitat patches, and increase in edges; re- In this paper, we consider the estimation of a relative viewed in AndreÂn 1994) are likely to add to loss of species richness metric. This metric can be viewed as habitat diversity. Under that hypothesis, one might ex- re¯ecting the degree of ``completeness'' of animal pect that the relationship between urban development communities, relative to a species pool. We consider and species richness is not linear. The studies showing this estimation in situations in which not all species that species richness ``peaks at moderately disturbed are necessarily detected during sampling sessions, i.e., sites'' (JokimaÈki and Suhonen 1993, Blair 1996) sup- where species detection probabilities are , 1. In such port that hypothesis. a sampling situation, naive estimates computed as the Several studies examined the composition of urban communities and established that these communities number of species detected in samples of the com- comprise mostly native species found in predevelop- munity of interest divided by the number of species in ment habitats that have the ability to colonize and per- the reference community or the potential species pool sist in urban habitat, plus a few ``new'' exotic species will be biased, and will be a function of both sampling not recorded in less urbanized habitat (Batten 1972; variation and the true quantity of interest. We propose Lancaster and Rees 1979, Germaine et al. 1998). Not an estimator designed to eliminate this bias and the all of the species from the ``predevelopment'' bird com- confounding in¯uence of detection probability. We use munities remain in urban areas, and relatively few ex- this estimator to address a question about the possible otic species not belonging to the predevelopment pool in¯uence of urban development at the landscape scale are likely to colonize urban habitat. Lower levels of on avian communities in the Mid-Atlantic states of Del- urbanization would increase richness through a com- aware, Maryland, New Jersey, Pennsylvania, Virginia, munity characterized by the coexistence of species as- and West Virginia. sociated with original and new habitat (i.e., an initial Most previous studies have established that higher increase in species richness), whereas higher levels levels of urbanization are associated with a lower num- would lead to lower species richness (Lancaster and ber of breeding species (Batten 1972, Emlen 1974, Rees 1979, JokimaÈki and Suhonen 1993, Zalewski Huhtalo and JaÈrvinen 1977, Hohtola 1978, Beissinger 1994, Blair 1996). and Osborne 1982, Bezzel 1984, Rapport et al. 1985, Predictions about the in¯uence of urbanization on JokimaÈki and Suhonen 1993,