Abstract. N and P often limit primary and secondary production in ecosystems, but they also can cause eutrophication and negatively influence sensitive species above a certain level or threshold point. Aquatic biodiversity can have negative threshold relationships with water-quality variables at large scales, but the specific mechanism(s) driving these threshold relationships are not well established. We hypothesized that resource quality (i.e., C:P) might partly drive primary consumer (grazer and detritivore) richness thresholds by altering competitive interactions among species with differing resource demands, but might have less influence on predator richness. We estimated total N (TN), total P (TP), and turbidity thresholds for macroinvertebrate richness across trophic levels and feeding groups in Central Plains (USA) streams. We also determined if mean taxon body C:P of groups with diversity losses were negatively related to TP, a pattern that would suggest that eutrophic communities were dominated by a few species with high dietary P demands. Primary consumers were more sensitive to TN and TP (threshold mean = 1.0 mg N/L and 0.06 mg P/L) than secondary consumers (threshold mean = 0.09 mg P/L), a result supporting the resource quality hypothesis. Turbidity reduced richness regardless of feeding mode (threshold mean = 4.7 NTU), a result suggesting that turbidity and nutrient thresholds were driven by different factors. The TP-richness threshold could be driven partially by changes in food quality because the mean body C:P of shredding and collector-gathering taxa declined as TP increased (threshold mean = 0.07 and 0.75 mg P/L, respectively). Mean scraper C:P was not related to TP, a result indicating other factors might be responsible for the scraper richness threshold. Our results suggest that changes in resource quality could contribute to large-scale losses in biodiversity in nutrient-enriched lotic ecosystems. Within shredder and collectorgatherer macroinvertebrate feeding groups, P-rich food might allow faster growing taxa with high body P demands to out-compete slower growing taxa adapted to lower quality food resources. This pattern suggests that biotic integrity is directly linked to nutrients in streams and that toxicity, low dissolved O 2 , and increased turbidity might not be the only mechanisms leading to reductions in diversity as nutrient concentrations increase.

No abstract

No abstract

No abstract

Evaluating the effects of toxicants in ecosystems is difficult despite numerous attempts to develop field and laboratory tests. The problem appears to be the lack of an analytical methodology capable of taking advantage of the available experimental designs. Therefore, we propose a technique for modeling ecosystems-linear structural modeling with LISREL. LISREL is a path analytic technique that is more flexible than classical path analysis. Modeling with LISREL involves placing ecosystem structure and function into a framework of concepts and indicator variables. Concepts are theoretical constructs that are placed into a cause-and-effect network to reflect true ecosystem structure. Concepts are @indicated" by indicator variables; these are the measured variables in the ecosystem. LISREL incorporates measurement error into the modeling process by establishing a portion of the variance of each indicator variable as measurement error. The framework of concepts and indicators in a cause-and-effect network becomes a hypothesis that is tested using LISREL. LISREL also provides a determination of the total, direct, and indirect effects of the variables on each other. A measure of ecosystem stability is provided as part of the modeling process.