Allocation of Sampling Effort to Optimize Efficiency of Watershed-Level Ichthyofaunal Inventories

—Sampling protocols for ichthyofaunal inventories have received considerable study at the stream reach level; however, no studies have examined the optimal allocation of sampling effort for watershed-level inventories. To address this question, we used data from surveys of four Great Lakes watersheds to compare rates of species accumulation among several watershed-level scenarios of sampling effort allocation (sampling strategies). To allow for quantitative comparisons among strategies, we (1) divided each sampling site into smaller sampling units and separately recorded fish capture data at the smallest unit; (2) stratified the allocation of sampling effort throughout each watershed by stream order; and (3) recorded all sampling and travel costs (i.e., time spent on different tasks). We used a Monte Carlo simulation program to resample the data from each of the four watershed surveys and calculated rates of species accumulation. This technique allowed us to quantitatively compare sampling strategies. Using this program, we determined the optimal values for (1) the length of stream to sample at each reach; (2) the allocation of sampling effort (time) among stream order strata; and (3) the allocation of effort between seasons (spring and summer) for each of the four watersheds. On average, sampling 15, 9, and 9 stream widths per reach in first-, second-, and third-order reaches, respectively, maximized the rates of species accumulation across watersheds. Allocating 50–70% of the sampling effort to third-order strata of each watershed yielded the highest rates of species accumulation without increasing the number of species systematically missed in either the lower or upper portions of the watershed. Focusing sampling effort in only one season, either spring or summer, did not consistently increase the rates of species accumulation. Our results provide useful guidance for the design of watershed-scale ichthyofaunal surveys in Great Lakes basin watersheds. We recommend similar assessments in other regions with contrasting patterns of fish diversity. Watershed-level conservation and management decisions require data on fish species composition (Peterson and Rabeni 1995; Fausch et al. 2002). Systematic ichthyofaunal surveys are critical for attaining these data; however, such surveys are often limited by cost. Guidelines for optimizing allocation of sampling effort should improve the quality and reduce the costs of these inventories. Although several studies have been conducted to design efficient protocols for ichthyofaunal surveys at the stream-reach level (Lyons 1992; Angermeier and Smogor 1995; Bowen and Freeman 1998), no studies to date have examined sampling effort allocation strategies for watershed-level ichthyofaunal surveys. Given the continued increase in watershed-level threats such as non-point-source pollution, dams and other hydrological alterations, and habitat modification, guidelines are needed for efficient watershed-level ichthyofaunal inventories. Stratifying sampling effort, sampling along diversity gradients, and adjusting sample unit size and sampling method have increased the efficiency of inventories for many taxa (Gillison and Brewer 1985; Coddington et al. 1991; Neave et al. 1992, 1997; Gimaret-Carpentier et al. 1998; Fisher 1999). These same techniques can be applied to increase the efficiency of watershed-level ichthyofaunal surveys. For example, rates of species accumulation might be improved by formally considering the trade-off between sampling few longer stream reaches instead of many shorter reaches in a watershed. Rates of species accumulation may also be improved by focusing sampling effort on the higher-order streams of a watershed, where species richness is generally highest (Paller 1994; Fairchild et al. 1998); however, focusing too little effort on small headwater streams may increase the risk of failing to detect species that specialize in these habitats. In a previous study, we calculated the sampling effort needed to detect a given fraction of the estimated number of species present in nine Great Lakes watersheds (Smith and Jones 2005). In four of these watersheds, we recorded the survey data at a finer spatial scale and recorded all sampling and travel costs. This detailed sampling scheme allowed us to apply Monte Carlo simulations of our data wherein we * Corresponding author: [email protected] 1 Present address: National Marine Fisheries Service, Silver Spring, Maryland 20910, USA. Received January 31, 2007; accepted February 26, 2008 Published online September 22, 2008 1500 Transactions of the American Fisheries Society 137:1500–1506, 2008 Copyright by the American Fisheries Society 2008 DOI: 10.1577/T07-024.1 [Article] D ow nl oa de d by [ M ic hi ga n St at e U ni ve rs ity ] at 1 1: 09 1 9 Se pt em be r 20 13 resampled the actual data from each watershed according to different strategies of effort allocation and calculated the average rates of species accumulation among these strategies. Here, we present an analysis of these data with the objective of determining the optimal values for (1) the length of stream (reach size) to sample; (2) the amount of time to allocate to sampling effort among stream order strata in each watershed; and (3) the allocation of effort between seasons (spring and summer).