9564 - Laserfiche WebLink

Home Browse Search

March 2008 11 Herbivore{ detritivore Flange 207 krn n 4 A EXTINCTION-PRONE TRAITS OF DESERT FISHES Range = 10.2 km n=22 Trophic guild All thers R7.$ km n 18 Yes (non-guarder open substrate) Range = 3.7 km n-=12 B Parental care { 1.5 (reproductive guild) No (all others) Range = 16.0 km n 6 Diet breadth < 3.5 items Yes No Range 3.5 krn Range - 28.4 krn n- 3 n=3 C D 851 FIG. 1. Regression tree discriminating among native species of the Lower Colorado River Basin, USA, according to empirical estimates of species rarity (defined according to range, the total river kilometers occupied). According to the tree-branching topology, we also interpret the best competitive surrogate splits (indicated in parentheses) that showed similar classification power to the primary split. Note that, for continuous traits, the specific split-defining values in the tree should not be interpreted as distinct thresholds, but as a relative threshold compared to the remaining species. Letters A-D indicate terminal nodes; n is the number of species. Parental care refers to an index (range 1-3) of total energetic contribution of parents to their offspring according to the specificity of zygote placement, length of time given protection, and degree of nutrition contribution (see Winemiller 1989). were grouped into two terminal nodes located close to the root of the regression tree (Fig. 2, nodes A-B), whereas species exhibiting near or below average frequencies were situated on longer branches (nodes C-F). Trait combinations leading to the highest extirpation probabilities included either large body size at maturity (four species in node A) or smaller size at maturity coupled with extremely low fecundity (three species in node B). Of the remaining 15 species, biological traits describing diet breadth, parental care, and longevity discriminated among species with predict- ed extirpation frequencies between 0.30 and 0.52. Species having the lowest probability of extirpation exhibited either a combination of diet specialization and a reproductive strategy of higher parental care to offspring (node D), or a combination of broad diet breadth plus a relatively short life span (node E). Notable species that significantly contribute to terminal node impurity included the under-prediction of extirpa- tion frequencies for Gila trout (node B) and O. g. apache (Apache Trout, node D), and the overprediction for Gila topminnow (node B) and G. intermedia (Gila chub, node C) (Appendix Q. Level and source of extinction risk Species perceived level and source of global extinction risk were correctly classified for 19 of 22 native species (86.4% correct classification, x = 0.878, P < 0.0001; Fig. 3). The branching sequence of the classification tree indicates that traits describing body morphology, reproductive strategy, and trophic specialization were the most important predictors of extinction risk. Similar to the regression trees for species rarity and extirpation, the phylogenetic relatedness indices were not present in the final classification tree for species extinction. Maximum body size was the primary splitting variable in the classification tree. Larger-bodied species were identified as being under moderate threat to extinction (Fig. 3, node A), whereas smaller-bodied species were considered at high extinction risk from interactions with nonnative species (node B). The remaining 17 species of intermediate body sizes formed a diverse group of fishes under different levels and sources of extinction threat. These species were split into two large branches according to species reproductive strategies, specifically the degree of parental care provided to offspring.