科研之友
微信
新浪微博
Facebook
分享链接
摘要
Quantile regression and an extension of a Oaxaca/Blinder decomposition are applied to US EPA's Environmental Monitoring and Assessment Program (EMAP) Southwestern stream survey data to estimate the impact of physical and chemical stressors on benthic community fitness. Observations were collected in xeric and mountainous ecoregions with distinct natural environments, of diverse and confounding land uses, and exhibiting different ranges of benthic fitness. The behavior of the benthic community fitness at the extremes is of particular interest to resource managers for prioritizing stressors. Quantile regression characterizes behavior of specific response percentiles (the response for which the specified percent of population responses are at or below that value), depending on related conditions (covariates). In this application it is used to asses the change in association of fitness score and stream condition as the level of disturbance in stream segments changes. Quantile regression models the response behavior conditional on the covariates taken together. The Oaxaca/Blinder method estimates the unconditional effect of a particular covariate after accounting for explainable differences due to the rest of the covariates. Modeling the unconditional behavior provides a region-wide description of fitness-stressor associations. In this application, Melly's (2006) extension of the Oaxaca/Blinder decomposition is employed to estimate differences between quantiles of stream segments with a stressor condition above or below a dividing threshold after accounting for differences due to the other conditions. The effect of having increased concentrations of nitrogen and phosphorus and increased levels of conductivity are generally more detrimental to benthic communities of streams segments that are more compromised.
- 出版日期2010-3
