Monitoring network-design influence on assessment of ecological condition in wadeable streams

We investigated outcomes of three monitoring networks for assessing ecological character and condition of wadeable streams, Waikato region, New Zealand. Site selection was based on professional judgment, stratification within categories of watershed characteristics, or on using an unequal-probability survey design. The professional-judgment network, stratified network and all site analyses included more ≥4th-order streams than for the probability-network survey-design estimates Professional-judgment and stratified network sites and survey-design analyses incorporated higher-quality catchments with coarser substrates. Cumulative frequency distributions indicated that the stratified and/or judgmental networks yielded fewer taxa than did the probability network, and that the stratified network provided lower estimates of the macroinvertebrate community index (MCI). Compared with the probability-network survey-design analysis, the stratified network site analysis underestimated percentage stream length classed as 'Excellent' by the quantitative MCI, and the professional-judgment site and survey-design analyses overestimated the percentage classed 'Fair' by the average score per metric. We conclude that deriving reliable estimates of wadeable stream character and condition requires (1) clearly defining and quantifying the target population for which inferences will be drawn, (2) accounting for probability of site selection and (3) optimising spatial representation across dominant stressor gradient