Yong Cao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yong Cao has a total of 2 co-authors. Here are the top co-authors by number of shared publications:
Charles P. Hawkins
Department of Aquatic, Watershed and Earth Resources, Western Center for Monitoring and Assessment of Freshwater Ecosystems and the Ecology Center, Utah State University, Logan, UT 84322–5210, USA
2 shared publicationsAndrew W. Storey
Aquatic Research Laboratory, School of Animal Biology, The University of Western Australia, Nedlands, WA, Australia
1 shared publication
Latest Publications
- Open Access
- 5 September 2005
- Published by Wiley in Journal of Applied Ecology
- Vol. 42 (5) , 954-965
Simulating biological impairment to evaluate the accuracy of ecological indicators
Summary: Biological indicators are used to measure the biological health of ecosystems. Resource managers may uncritically assume that such indicators give unbiased estimates of true biological condition, but this assumption is largely untested. Use of biased indicators could lead to ineffective and potentially damaging management. We used a simulation model to compare estimated indicator values objectively with true impairment. We used the model to progressively alter the densities of 155 stream invertebrate taxa as functions of stress and taxon‐specific sensitivities to stress. We applied the model to large samples collected from five reference‐quality sites. After each stress event, we randomly collected 100, 300 and 600 fixed‐count subsamples from the remaining assemblages and recorded the densities of all taxa in both the large sample and the fixed‐count subsamples. We then examined how well two indicators, non‐metric multidimensional scaling (NMDS) and sample taxa richness, detected the simulated impairment when derived from fixed‐count subsamples. NMDS ordinations of artificially impaired samples, and samples from reference and impaired sites, showed that the simulated impairments were realistic. NMDS further showed that impairment trajectories were evident for subsamples of 100, 300 and 600 individuals. However, the discrimination of stress levels within ordination space greatly improved with increasing sampling effort. The ordination also showed that increasing stress could cause assemblages at some sites to resemble unstressed assemblages at other sites. Taxa‐richness indicators were much more problematic. Estimates of taxa richness from 100‐ and 300‐count samples often substantially underestimated true taxa loss and frequently indicated taxa gain. Apparent gains in taxa richness occurred because stress caused changes in evenness that compensated for, or even overrode, the effect of true taxa loss. Estimates of taxa loss derived from 600‐count subsamples also underestimated true taxa loss, but these estimates were strongly correlated with true taxa loss. Synthesis and applications. Bioassessements that rely on richness‐based indicators derived from small fixed‐count subsamples may substantially underestimate true biological impairment. Simulation models provide an objective means to evaluate how well different types of biotic indicators measure true biological impairment in aquatic and other types of ecosystems.
- 5 September 2005
- Published by Wiley in Journal of Applied Ecology
- Vol. 42 (5) , 954-965
Open AccessSimulating biological impairment to evaluate the accuracy of ecological indicators
Summary: Biological indicators are used to measure the biological health of ecosystems. Resource managers may uncritically assume that such indicators give unbiased estimates of true biological condition, but this assumption is largely untested. Use of biased indicators could lead to ineffective and potentially damaging management. We used a simulation model to compare estimated indicator values objectively with true impairment. We used the model to progressively alter the densities of 155 stream invertebrate taxa as functions of stress and taxon‐specific sensitivities to stress. We applied the model to large samples collected from five reference‐quality sites. After each stress event, we randomly collected 100, 300 and 600 fixed‐count subsamples from the remaining assemblages and recorded the densities of all taxa in both the large sample and the fixed‐count subsamples. We then examined how well two indicators, non‐metric multidimensional scaling (NMDS) and sample taxa richness, detected the simulated impairment when derived from fixed‐count subsamples. NMDS ordinations of artificially impaired samples, and samples from reference and impaired sites, showed that the simulated impairments were realistic. NMDS further showed that impairment trajectories were evident for subsamples of 100, 300 and 600 individuals. However, the discrimination of stress levels within ordination space greatly improved with increasing sampling effort. The ordination also showed that increasing stress could cause assemblages at some sites to resemble unstressed assemblages at other sites. Taxa‐richness indicators were much more problematic. Estimates of taxa richness from 100‐ and 300‐count samples often substantially underestimated true taxa loss and frequently indicated taxa gain. Apparent gains in taxa richness occurred because stress caused changes in evenness that compensated for, or even overrode, the effect of true taxa loss. Estimates of taxa loss derived from 600‐count subsamples also underestimated true taxa loss, but these estimates were strongly correlated with true taxa loss. Synthesis and applications. Bioassessements that rely on richness‐based indicators derived from small fixed‐count subsamples may substantially underestimate true biological impairment. Simulation models provide an objective means to evaluate how well different types of biotic indicators measure true biological impairment in aquatic and other types of ecosystems.
- 6 May 2005
- Published by Wiley in Freshwater Biology
- Vol. 50 (6) , 1105-1115
A method for measuring the comparability of different sampling methods used in biological surveys: implications for data integration and synthesis
Summary: 1. Numerous methods have been developed to sample the biota occurring in different ecosystems. However, the comparability of data derived from different sampling methods is generally unknown and is a major concern when integrating data from different studies.2. Examination of assemblage‐level attributes such as taxa richness and biotic index scores is generally inappropriate for evaluating the degree to which different sampling methods produce comparable descriptions of entire assemblages, because these measures provide no information regarding taxonomic composition. Multivariate methods are generally more appropriate for this purpose, but some of the methods previously used are not satisfactory and others have not been tested. A useful measure of sampling‐method comparability (SMC) should be independent of sampling effort, independent of the sites sampled and have an explicit biological interpretation.3. We used simulated data to compare two potential methods of assessing SMC, the R‐value produced by anosim and a modified version of classification strength (CS‐SMC) derived from Van Sickle's Mean Similarity Analysis. Analyses were based on similarities between the assemblages captured by two different sampling methods (electrofishing and seining) employed at the same sites. Similarities were calculated two different ways: the Bray–Curtis index and the Jaccard coefficient.4. Based on simulated data, anosimR‐values were strongly affected by sampling effort, highly variable across sites and difficult to interpret biologically. In contrast, CS‐SMC values were highly stable over a range of sampling effort, across sites and easy to interpret biologically.5. Application of CS‐SMC to field data showed that seining and electrofishing produced highly comparable samples of fish in small streams: 97% comparable on average for species lists and 94% comparable for relative abundances. Kicknet and Surber samples of benthic invertebrates were also comparable after being standardised to a fixed count, but to a lesser extent than fish samples: 77% comparable on average for the taxa lists and 93% comparable for relative abundances. CS‐SMC should be of general use when integrating and synthesising assemblage data from a variety of assemblages.
- 6 May 2005
- Published by Wiley in Freshwater Biology
- Vol. 50 (6) , 1105-1115
A method for measuring the comparability of different sampling methods used in biological surveys: implications for data integration and synthesis
Summary: 1. Numerous methods have been developed to sample the biota occurring in different ecosystems. However, the comparability of data derived from different sampling methods is generally unknown and is a major concern when integrating data from different studies.2. Examination of assemblage‐level attributes such as taxa richness and biotic index scores is generally inappropriate for evaluating the degree to which different sampling methods produce comparable descriptions of entire assemblages, because these measures provide no information regarding taxonomic composition. Multivariate methods are generally more appropriate for this purpose, but some of the methods previously used are not satisfactory and others have not been tested. A useful measure of sampling‐method comparability (SMC) should be independent of sampling effort, independent of the sites sampled and have an explicit biological interpretation.3. We used simulated data to compare two potential methods of assessing SMC, the R‐value produced by anosim and a modified version of classification strength (CS‐SMC) derived from Van Sickle's Mean Similarity Analysis. Analyses were based on similarities between the assemblages captured by two different sampling methods (electrofishing and seining) employed at the same sites. Similarities were calculated two different ways: the Bray–Curtis index and the Jaccard coefficient.4. Based on simulated data, anosimR‐values were strongly affected by sampling effort, highly variable across sites and difficult to interpret biologically. In contrast, CS‐SMC values were highly stable over a range of sampling effort, across sites and easy to interpret biologically.5. Application of CS‐SMC to field data showed that seining and electrofishing produced highly comparable samples of fish in small streams: 97% comparable on average for species lists and 94% comparable for relative abundances. Kicknet and Surber samples of benthic invertebrates were also comparable after being standardised to a fixed count, but to a lesser extent than fish samples: 77% comparable on average for the taxa lists and 93% comparable for relative abundances. CS‐SMC should be of general use when integrating and synthesising assemblage data from a variety of assemblages.