Effects of measurement error on inferences of environmental conditions

Abstract The effects of measurement errors on biological inferences of stream temperature and bedded fine sediment were investigated. Single variable and multivariate logistic regression models were used to relate the occurrences of different macroinvertebrate genera and observed temperature and fine sediment. Next, a simulation and extrapolation method (SIMEX) was used to adjust regression model coefficients for the effects of measurement errors in the temperature and fine-sediment observations. It was assumed that the persistence of different stream organisms was related to long-term average environmental conditions, so measurement error in this analysis was interpreted broadly as including all the variability associated with estimating long-term averages from single measurements. On average, correcting for measurement error narrowed the breadth of genus–environment relationships and shifted optimum values toward the mean of the observations. Accounting for measurement error also improved the predictive accuracy of some of the inference models. Inferences of temperature based on single-variable SIMEX models were 28% more accurate than inferences based on naive models that used uncorrected observations. However, inferences of fine sediment based on single-variable SIMEX models were only slightly more accurate than inferences based on naive models. Multivariate models, in which both stream temperature and fine sediment were modeled simultaneously, exhibited the strongest improvement in predictive performance when measurement error was taken into account. The accuracy of temperature inferences improved by 39%, whereas fine-sediment inferences improved by 8%.

[1]  Donald F. Charles,et al.  Distribution of benthic diatoms in U.S. rivers in relation to conductivity and ionic composition , 2003 .

[2]  Nicholas John Anderson,et al.  Predicting Epilimnetic Phosphorus Concentrations Using an Improved Diatom-Based Transfer Function and Its Application to Lake Eutrophication Management , 1996 .

[3]  R. Hall,et al.  Diatoms as indicators of climatic and limnological change in Swedish Lapland: a 100-lake calibration set and its validation for paleoecological reconstructions , 2002 .

[4]  C.J.F. ter Braak,et al.  Diatoms and pH Reconstruction , 1990 .

[5]  Lester L. Yuan.  Sources of bias in weighted average inferences of environmental conditions , 2005 .

[6]  J. Kuha,et al.  Covariate Measurement Error in Quadratic Regression , 2003 .

[7]  Andrew R. Solow,et al.  ON FITTING A POPULATION MODEL IN THE PRESENCE OF OBSERVATION ERROR , 1998 .

[8]  Kenneth H. Pollock,et al.  Use of simulation-extrapolation estimation in catch-effort analyses , 1999 .

[9]  James R. Karr,et al.  Assessing biological integrity in running waters : a method and its rationale , 1986 .

[10]  Lester L. Yuan.  Assigning macroinvertebrate tolerance classifications using generalised additive models , 2004 .

[11]  J. R. Cook,et al.  Simulation-Extrapolation Estimation in Parametric Measurement Error Models , 1994 .

[12]  Annika Kangas,et al.  Effect of errors-in-variables on coefficients of a growth model and on prediction of growth , 1998 .

[13]  M. Diekmann.  Species indicator values as an important tool in applied plant ecology – a review , 2003 .

[14]  D. Ruppert,et al.  Measurement Error in Nonlinear Models , 1995 .

[15]  W. Marcus,et al.  Pebble Counts and The Role of User‐Dependent Bias in Documenting Sediment Size Distributions , 1995 .

[16]  D. Carlisle,et al.  Estimation and application of indicator values for common macroinvertebrate genera and families of the United States , 2007 .

[17]  C. Braak,et al.  Weighted averaging, logistic regression and the Gaussian response model , 2004, Vegetatio.

[18]  P. Huttunen,et al.  Inferring the minimum pH of streams from macroinvertebrates using weighted averaging regression and calibration , 1996 .