Assessment of Ecological Condition in Headwater Streams of the Central Plains: Evaluation of Multimetric and Predictive Modeling Approaches
Total Page:16
File Type:pdf, Size:1020Kb
ASSESSMENT OF ECOLOGICAL CONDITION IN HEADWATER STREAMS OF THE CENTRAL PLAINS: EVALUATION OF MULTIMETRIC AND PREDICTIVE MODELING APPROACHES by R. William Bouchard, Jr. B.A., Franklin & Marshall College, 1998 Submitted to the Department of Entomology and the Faculty of the Graduate School of the University of Kansas in partial fulfillment of the requirements for the degree of Master of Science ABSTRACT ASSESSMENT OF ECOLOGICAL CONDITION IN HEADWATER STREAMS OF THE CENTRAL PLAINS: EVALUATION OF MULTIMETRIC AND PREDICTIVE MODELING APPROACHES R. William Bouchard, Jr., M.S. Department of Entomology, August 2002 University of Kansas Multimetric analysis is the most common technique currently employed to assess the health of aquatic systems in North America. However, predictive modeling, a standard technique in the United Kingdom, has been tested in some regions of North America with promising results. Using a watershed assessment procedure to predict stream condition, the abilities of multimetric and predictive modeling techniques to distinguish reference from impacted conditions in headwater streams within the Western Corn Belt Plains ecoregion were compared. Different taxonomic groups (i.e., macroinvertebrates and fishes) were tested to determine which taxonomic group(s) provided the best determination of stream condition. Eighteen reference sites were used to develop models with replicate samples from these sites and eight test sites were used to evaluate the models. Additionally, data from eight reference sites collected during a separate, or independent, study were also used to evaluate the models. Predictive modeling techniques were better at identifying impacted and reference sites from the main dataset. However, there was no difference between approaches when comparisons were made using reference samples from the independent study, and multimetric approaches performed better using a combination of fish and macroinvertebrate data. Genus resolution models performed better or as well as family models for all datasets. Assessment difficulties for macroinvertebrate models most likely arose from a lack of reference sites in the eastern portion of the WCBP ecoregion. Spatial gaps could be corrected with the addition of more reference sites and, if warranted, the development of finer scale models. Also, due to the relatively small size of the dataset, elements of the predictive modeling approach were omitted, which could increase the predictive ability of the models. Both methods have shown they are useful tools in the biological assessment of streams, and could be used in conjunction to provide additional capabilities and to enhance the biological assessment of streams. i ACKNOWLEDGEMENTS ii TABLE OF CONTENTS LIST OF TABLES…………………………………………………………………………...………v LIST OF FIGURES………………………………………………………………………..……….vii LIST OF APPENDICES…………………………………………………………………………....viii INTRODUCTION…………………………………………………………………..………………..1 STUDY OBJECTIVES……………………………………………………………………...…….….6 STUDY AREA……………………………………………………………………………...……….7 Description of WCBP ecoregion………………………………………………………..…….7 Selection of study sites………………………………………………………………………...8 Tristate Sites…………………………………………………………………………...…….8 Regional Reference Sites……………………………………………………………………10 METHODS – DATA COLLECTION………………………………………………………………..10 Introduction…………………………………………………………………………………..10 Tristate Data Collection………………………………………………………………..……11 Water quality…………………………………………………………………………..……11 Geomorphology and Habitat (in-/near-stream)…………………………………..…..……12 Biota………………………………………………………………………………………...12 Land use/land cover……………………………………………………………………..….14 Regional Reference stream data collection…………………………………...…………….15 DEVELOPMENT OF THE WATERSHED INDEX………………………………...………………….16 Introduction…………………………………………………………………………………..16 Methods……………………………………………………………………………………….18 Results………………………………………………………………………………………...19 Discussion…………………………………………………………………………………….20 EVALUATION OF BIOLOGICAL ASSESSMENT APPROACHES AND TAXONOMIC RESOLUTIONS….21 Introduction…………………………………………………………………………………..21 Comparison of multimetric and predictive modeling approaches……………………..…22 Data collection………………………………………………………………...……………23 Classification of reference sites…………………………………………………………….25 Selection of reference groups for test sites comparison……………………………..……..25 Test site assessment…………………………………………………………………………26 Faunal groups……………………………………………………………………………...…27 Taxonomic sufficiency……………………………………………………………….………28 Methods……………………………………………………………………………………….29 Macroinvertebrate taxa compositions.…………………………………………….……….32 Development of multimetric approaches………………………………………………...…33 Invertebrate Community Index (ICI)…………………………………………………….33 Index of Biotic Integrity (IBI)…………………………………………………………...34 Combination of macroinvertebrate and fish multimetric procedures……..………………..36 Development of predictive modeling classification……………………………………...…37 Comparison of multimetric and predictive modeling…………………………………...….40 Comparison of taxonomic levels and groups…………………………………….…………42 Results………………………………………………………………………………..……….43 Taxonomic composition…………………………………………………………………….43 Comparison of predictive modeling and multimetric analyses and taxonomic sufficiency..47 Discussion…………………………………………………………………………………….56 iii Predictive and multimetric analyses………………………………………………………..56 Performance of predictive and multimetric approaches……………………………...….56 Comparison of predictive modeling and multimetric approaches……………………….57 Required dataset……………………………………………………………………….59 Sampling period…………………………………………………………………….…60 Faunal group………………………………………………………………………..…62 Rare taxa………………………………………………………………………………63 Reference sites……………………………………………………………...…………64 Development of reference criteria and calculation of models………………………...65 Site classification………………………………………………………………...……68 Basis for assessing impact…………………………………………….………….……71 Communication……………………………………………………….……………….71 Taxonomic sufficiency……………………………………………………..………….…….72 FUTURE CONSIDERATIONS……………………………………………………………………....75 SUMMARY……………………………………………………………………………………..…80 LITERATURE CITED……………………………………………………………………………...82 APPENDICES………………………………………………………………………………..…….92 iv LIST OF TABLES Table 1: Field measures of water quality, geomorphology/habitat, and biota from the Tristate and Regional Reference studies (* = measures also sampled in the Regional Reference study)……………………………………………………………………………………..11 Table 2: Land use/land cover measures generated from GIS data for the Tristate study (* = 30 and 100 m buffer zones generated for these parameters)……………………………..…16 Table 3: Variables and scoring ranges used for development of independent criteria with the expected response of biotic integrity to an increase in the value of the Watershed Index parameter. (CPOM = coarse particulate organic matter; FPOM = fine particular organic matter)……………………………………………………………………………………20 Table 4: Study sites for Tristate and Regional Reference studies with watershed area…………31 Table 5: Selected macroinvertebrate biological criteria metrics used in the WCBP ecoregion Invertebrate Community Index (ICI), their expected response to stress, and the scoring criteria used for family and genus resolution models……………………………………33 Table 6: Selected fish metrics used in the WCBP ecoregion IBI, their expected response to disturbance, and scoring criteria…………………………………………………………36 Table 7: Calculation of the predictive model at the >0% probability level for Walnut Creek (Jasper Co., Iowa) sample #1 using autumn 1992 data (* = taxa present; taxa in parentheses occurred at fewer than 50% of reference sites, taxa occurring at less than 50% of reference sites and not present at the Walnut Creek were removed from the table for space considerations)…………………………………………………………………39 Table 8: Numbers of macroinvertebrate taxa for “family” and “genus” treatments from all Tristate and Regional Reference samples († = not identified to family; ‡ = not identified to genus; * = resolution varied among taxa)……………………………………………..44 Table 9: Number of species for each fish family collected in autumn 1992 from the 26 Tristate reference and test sub-watersheds………………………………………………………..45 Table10: Forty macroinvertebrate taxa with the largest absolute difference in abundance from reference and test sites (RD = reference site macroinvertebrate density, and TD = test site macroinvertebrate density; log((RD-TD)+1) is sensitive to the absolute difference in abundance between reference and test sites, and log((RD+1)/(TD+1)) is sensitive to relative differences in abundance between reference and test sites)……………………..46 Table 11: Mean and standard deviation information from macroinvertebrate models………….48 Table 12: Mean and standard deviation information from fish models…………………………48 v LIST OF TABLES (CONT’D) Table 13: Mean and standard deviation information from combined macroinvertebrate and fish models……………………………………………………………………………………48 Table 14: Classification accuracy and precision of multimetric and predictive models for macroinvertebrate family and genus level resolutions and fish from autumn 1992 Tristate test (i.e., impacted) site samples (R = reference (incorrect assessment); I = impacted (correct assessment); fish data were pooled from three samples)………………………..51 Table 15: Classification accuracy for multimetric and predictive models using family and genus resolutions macroinvertebrate samples from autumn 1992 for Tristate reference sites (R = reference (correct assessment); I = impacted (incorrect assessment))…………………..52 Table 16: Classification accuracy for multimetric