9564 - Laserfiche WebLink

Home Browse Search

March 2008 EXTINCTION-PRONE TRAITS OF DESERT FISHES turn spawners, brood hiders), guarders (substratum choosers, nest spawners), or bearers (external) (follow- ing Balon 1975). We selected these attributes from a larger trait database because they reflect the main dimensions of the species ecological niches in this region (Olden et al. 2006). Trait assignments were based on a comprehensive review of state fish textbooks, primary literature, state agency reports, university reports, graduate theses, and electronic databases available on the World Wide Web (see Olden et al. 2006 for more details). Expert knowledge of regional specialists was used to assign values to a small number of trait states (<2%) that could not be obtained from the previous methods (mainly inferred from congenerics). To account for interdemic variation in biological traits, values were based on research conducted in the study region whenever possible. Ordinal and nominal trait values were assigned a single state based on a majority-of-evidence rule according to adult preferences, and median values for continuous traits were used when ranges were presented. Although we recognize the sensitivity of trait estimates to factors including sample size and geographic location, the assembled database reflects the best available information for this group of species. Phylogenetic inertia It is expected that species share similar life-history attributes through descent from common ancestry, thus necessitating the need to account for phylogeny effects when exploring patterns in ecological data (Fisher and Owens 2004). Approaches for controlling the effects of phylogeny typically involve the method of independent contrasts (Felsenstein 1985); however, this technique cannot accommodate combinations of nominal, ordinal, and continuous variables that are present in our data set. Therefore, we employed the eigenvector method pro- posed by Diniz-Filho et al. (1998) to quantify the degree of phylogenetic inertia in our species pool. This involved constructing a qualitative phylogeny of native fishes (see Appendix A) and assembling a phylogenetic distance matrix by tabulating the total number of nodes separating the species in the tree (following Webb et al. 2002). Next, a principal coordinate analysis (Gower 1966) was computed from the phylogenetic distance matrix to represent species in reduced multivariate space expressing variation in their phylogenetic relatedness. The first two principal coordinates, which accounted for 90.2% of the original variation (70.8 and 19.4%, respectively), were statistically significant based on the broken-stick model (Peres-Neto et al. 2003) and provided two indices of phylogenetic relatedness. Species' rarity, frequency of extirpation, and perceived extinction risk We obtained empirical estimates of species' rarity and frequency of local extirpation from Olden and Poff (2005) and Fagan et al. (2002), respectively (Appendix 849 B). Both studies used the Sonoran Fishes (or SON- FISHES) database; a comprehensive data source con- taining incidence, identity, and collection records for the complete holdings of major museums, numerous smaller holdings, records from the state agencies, and peer- reviewed and "gray" literature sources. Together, SON- FISHES contains 20 000+ unique occurrence records collected over a 150-year period (1843-1998) providing information on past and present distributions of fishes in the region (Unmack 2002). Olden and Poff (2005) estimated present-day range size (inversely related to species' rarity as defined by the size of a species' geographic range) as the total kilometers of stream reach occupied by each species during the modern record (1981-1998). Fagan et al. (2002) quantified extirpation probabilities as the proportion of fish occurrence records at the 5-km river-segment scale in the historic period (1843-1980) that were absent during the modern record (1981-1998). Sampling effort (num- ber of collections per kilometer of stream network) increased through time due to the intense sampling adopted post-1980 by resource managers and govern- ment agencies, therefore estimates of extirpation are unlikely to be overestimated (Olden and Poff 2005). Recent work supports the robustness of extirpation estimates based on 1980 as the threshold for delineating the historical and modern periods (Fagan et al. 2005a, b). To estimate perceived extinction risk for native species of the Colorado River Basin, we developed a question- naire to survey 20 professional fish biologists with decades of research experience in the region (see Acknowledgments). Based on the findings of Fagan et al. (2005b), we decided to survey local experts rather than rely on conservation rankings compiled in national (U.S. Fish and Wildlife Service 1999) or international assess- ments (2006 IUCN Red List of Threatened Species, available online).5 Using the SONFISHES database, Fagan and colleagues examined spatial distributions of Lower Colorado River fishes at three scales for historical and modern time periods, and compared these trends to U.S. Fish and Wildlife Service and IUCN listings. The authors found poor correspondence between long-term distributional trends and both ranking systems. Using the IUCN ranking protocols with the long-term data available from the SONFISHES database, they suggest- ed a revised IUCN ranking for 14 out of 15 endemic fish species. For this reason, we were not confident that either ranking system accurately reflects the current conserva- tion status of native species in this region. We asked each survey respondent to classify all species according to two criteria: present-day "perceived risk" of extinction using four levels (none, low, moderate, or high), and the primary source of imperil- ment (altered flow regimes, altered temperature regimes, 5 (www.iucnredlist.org)