(Ws) Equations and Standard Length Categories for 18 Warmwater Nongame and Riverine Fish Species
Total Page:16
File Type:pdf, Size:1020Kb
North American Journal of Fisheries Management 20:570±574, 2000 q Copyright by the American Fisheries Society 2000 Proposed Standard Weight (Ws) Equations and Standard Length Categories for 18 Warmwater Nongame and Riverine Fish Species TIMOTHY J. BISTER,DAVID W. W ILLIS,* AND MICHAEL L. BROWN Department of Wildlife and Fisheries Sciences, Box 2140B, South Dakota State University, Brookings, South Dakota 57007, USA STEPHEN M. JORDAN AND ROBERT M. NEUMANN Department of Natural Resources Management and Engineering, Box U-87, University of Connecticut, Storrs, Connecticut 06269, USA MICHAEL C. QUIST AND CHRISTOPHER S. GUY Kansas Cooperative Fish and Wildlife Research Unit,1 205 Leasure Hall, Division of Biology, Kansas State University, Manhattan, Kansas 66506, USA Abstract.ÐRelative weight (Wr) is one of several con- ®sh species, Ws equations must ®rst be available. dition indices used to assess the general health of ®shes. Anderson and Neumann (1996) reported accepted Standard weight (Ws) equations are required to calculate W equations for 33 ®sh species. Most of these W , but are unavailable for many nongame and riverine s r species are considered to be sport ®shes. To further ®sh species. Therefore, we developed Ws equations for the following taxa: longnose gar Lepisosteus osseus, assist biologists in ®shery assessments, we devel- spotted gar L. oculatus, common carp Cyprinus carpio, oped Ws equations for the following nongame and bigmouth buffalo Ictiobus cyprinellus, river carpsucker riverine ®sh species: longnose gar Lepisosteus os- Carpiodes carpio, shorthead redhorse Moxostoma ma- seus, spotted gar L. oculatus, common carp Cy- crolepidotum, smallmouth buffalo I. bubalus, white prinus carpio, bigmouth buffalo Ictiobus cyprinel- sucker Catostomus commersoni, black bullhead Ameiu- rus melas, brown bullhead A. nebulosus, ¯athead cat®sh lus, river carpsucker Carpiodes carpio, shorthead Pylodictis olivaris, white cat®sh A. catus, yellow bull- redhorse Moxostoma macrolepidotum, smallmouth head A. natalis, white perch Morone americana, yellow buffalo I. bubalus, white sucker Catostomus com- bass M. mississippiensis, green sun®sh Lepomis cyanel- mersoni, black bullhead Ameiurus melas, brown lus, rock bass Ambloplites rupestris, and warmouth L. bullhead A. nebulosus, ¯athead cat®sh Pylodictis gulosus. These Ws equations were evaluated with statis- olivaris, white cat®sh A. catus, yellow bullhead A. tical validation approaches similar to previously de®ned natalis, white perch Morone americana, yellow regression-line-percentile equations. Standard length categories were developed or revised for 10 of these 18 bass M. mississippiensis, green sun®sh Lepomis cy- species (longnose gar, spotted gar, bigmouth buffalo, anellus, rock bass Ambloplites rupestris, and war- river carpsucker, shorthead redhorse, smallmouth buf- mouth L. gulosus. falo, white sucker, brown bullhead, white cat®sh, and The objectives of this study were to develop or yellow bullhead) to allow for comparison of mean Wr revise W equations for these 18 species, which values within and among length categories instead of s would allow for expanded use of W in population assessing only population mean W . r r and community assessments for these nongame and riverine species. Standard length categories Introduction were also developed or revised for 10 of these species (longnose gar, spotted gar, bigmouth buf- Relative weight (Wr 5 the ratio of a ®sh's weight, W, to the weight of a ``standard'' ®sh of falo, river carpsucker, shorthead redhorse, small- mouth buffalo, white sucker, brown bullhead, the same length, Ws) is a commonly used ®sh con- dition index (Blackwell et al. 2000). Before ®shery white cat®sh, and yellow bullhead) so that con- dition assessment could be evaluated for incre- biologists can assess the utility of Wr for various mental length-groups rather than population mean Wr values (Murphy et al. 1991). * Corresponding author: [email protected] 1 The Unit is jointly supported by Kansas Department Methods of Wildlife and Parks, Kansas State University, the U.S. Geological Survey Biological Resource Division, and Letters soliciting contributions of weight±length the Wildlife Management Institute. data for 30 warmwater ®sh species were sent to Received July 26, 1999; accepted December 22, 1999 ®shery administrators of state and provincial con- 570 MANAGEMENT BRIEFS 571 servation agencies and several U.S. Fish and Wild- the Ws equations, a test (Ho:û5 0; P , 0.05) was life Service regional of®ces. Suf®cient data were conducted to determine if there was a consistent received to develop Ws equations for 18 warm- length-related bias in Wr values calculated with the water species. proposed Ws equation (Willis 1989; Neumann and These Ws equations were developed using the Murphy 1991). Regressions of Wr on ®sh TL were regression-line-percentile (RLP) technique pro- done for each population. We expected that some posed by Murphy et al. (1990). The validity of the populations would have slopes signi®cantly dif- solicited data was assessed by plotting the weight± ferent than zero (P , 0.05); however, if the num- total length (grams and millimeters, respectively) bers of negative and positive signi®cant slopes relationships for each population separately so that were not signi®cantly different, we accepted the individual outliers could be identi®ed. Outliers, Ws equation. which were probably a result of measurement er- Finally, we determined whether suf®cient pop- rors, were deleted from the data sets. Next, the ulations were available to reliably determine a Ws RLP technique required determination of the min- equation. Brown and Murphy (1996) reported that imum total length (TL) to be used in the Ws equa- a minimum of 50 populations would consistently tion development. Primarily, we sought to avoid provide reliable Ws equations. For data sets with inaccurate or imprecise weights for small ®sh. We fewer than 50 populations, they recommended us- calculated an error estimate (variance to mean ra- ing a bootstrap approach to determine whether the tio) for mean log10weight by centimeter TL group number of populations was suf®cient. Slopes from (SAS Institute 1996), which was then plotted as a each population log10weight±log10TL regression function of TL. The minimum TL was selected were randomly selected at sequential intervals of where variance: mean error was less than 0.02, three, with 100 iterations done to determine the which corresponded to the in¯ection point in the variance about the regression slope means. The plot for each species (Willis et al. 1991). We cal- sample variance for slopes decreased as sample culated a log10weight±log10TL regression equation size increased. We used the benchmark for a min- for each population in the data set using only ®sh imum level of precision (s2 , 0.002) that was rec- longer than the minimum TL selected in the pre- ommended by Brown and Murphy (1996). vious step. Populations were removed from the Length categories were developed or revised for database at this point if correlation coef®cients (r) 10 of the 18 ®sh species in this study. Gabelhouse were less than 0.90. Slopes were then plotted as a (1984) recommended that minimum stock, quality, function of intercepts (Pope et al. 1995) to identify preferred, memorable, and trophy lengths be set at and remove population outliers caused by insuf- 20±26, 36±41, 45±55, 59±64, and 74±80% of the ®cient sample size, narrow length range, or mis- world record length for each species, respectively. identi®ed length measurements other than TL (i.e., We compared world record lengths, obtained from body length, fork length, or standard length). the National Freshwater Fishing Hall of Fame and After the minimum TL was determined and the the International Game Fish Association, with the data set ®nalized, the Ws equation was developed longest ®sh in our database for each species. We using the RLP technique (Murphy et al. 1990). based the development of the length categories for First, we predicted mean ®sh weight in 1-cm in- a particular species on the longest ®sh in our da- tervals for each population. Then, we found the tabase if that length exceeded the world record 75th percentile of mean weights in each interval length. and regressed these means on TL to determine the Results and Discussion proposed Ws equation. Some previous RLP equa- tions were developed after dividing data sets into The results from steps performed during the a development subset and a subset for testing po- RLP development of these Ws equations are tential length-related bias in Wr values (e.g., Pope lengthy and are therefore summarized in Bister et et al. 1995). This was in reality a test of the RLP al. (1999). These results include population sample method itself. Because the RLP method of Ws size, minimum length determination, population equation development has proven useful (i.e., no log10weight±log10TL regression parameters, boot- length-related bias) in many previous cases, we strap results, proportional stock density (PSD; An- did not believe it was necessary to test the RLP derson and Neumann 1996), and mean Wr by technique again. In addition, many data sets for length category for each population used in the Ws species in this study contained too few populations equation development. Parameters for log10weight to allow splitting into two data sets. To evaluate 2 log10length regression formulas for these 18 spe- 572 BISTER ET AL. TABLE 1.ÐParameters for log10weight 2 log10length regression formulas