Upper Missouri River Basin Aquatic GAP Fish Distribution Model Accuracy Assessment and White Sucker, Catostomus commersonii, Population Characteristics in the Upper Missouri River Basin

Fish species surveys were done at 61 sites in 6 drainages throughout the upper Missouri River Basin. A total of 21,204 fish were collected, representing 42 species and 11 families. Minnows (family Cyprinidae) dominated the catch (77.9%), followed by suckers (family Catostomidae, 6.5%), and catfishes (Ictaluridae, 5.7%). Of the 21,204 fish collected, fathead minnow Pimephales promelas (40.9%), unidentified Hybognathus spp. (9.8%), and longnose dace Rhinichthys cataractae (7.8%) were the dominant species. Important findings include: the first record of the plains minnow, Hybognathus placitus, in Canadian waters, specifically the Rock Creek drainage; and the first fish survey in the Sweet Grass River drainage, in Montana. Data provided to state and provincial agencies will be useful for future fish management and monitoring efforts as well as possibly stimulating further scientific investigations. The accuracy of fish distribution models used in the Gap Analysis Program of the United States Geological Survey was assessed by comparing observed fish assemblages in the six drainages to predicted fish assemblages. A number of accuracy metrics were calculated for each site, valley segment, species, and drainage and included: the number of true presences, true absences, omission errors, commission errors, and Kappa (K), which is the proportion of specific agreement between two classifiers after accounting for chance agreement. Models performed better than chance in each of the six drainages (i.e., positive mean K), although the mean valley segment K in each drainage was variable. The mean valley segment K (± SE) in each drainage was: Nowood River 0.3131 ± 0.0481; Beaver Creek 0.4200 ± 0.0945; Elm River 0.2167 ± 0.0684; Rock Creek 0.6367 ± 0.1197; Frenchman River 0.4282 ± 0.0319; and Sweet Grass River 0.2171 ± 0.0567. Site K values ranged from –0.04 in the Sweet Grass River to 0.86 in the Rock Creek drainage. Species K values ranged from -0.80 for northern redbelly dace (poor agreement) in the Rock Creek drainage to 1.00 (perfect agreement) for several species in the Nowood River, Beaver Creek, and Elm River drainages, although small sample sizes affected values of K. Commission errors generally outnumbered omission errors, but a high number of omission errors (N = 6 to 8) occurred in several valley segments in the Sweet Grass River. Poor agreement can be attributed to lack of nongame fish species data in the area. Differences in sampling and modeling scales, potential sampling biases, temporal variation in fish communities, variation in habitat conditions, and biases in model input data and classifications are reasons for discrepancies between model predictions and independent test data. Although agreement between model predictions and actual field observations was better than chance in most cases, potential users must decide if agreement was high enough to justify using the Aquatic GAP fish distribution models as a conservation planning tool. A common species modeled in the Aquatic Gap Analysis Program in upper Missouri River basin is the white sucker, Catostomus commersonii. White suckers are a common and sometimes abundant species throughout much of North America, but information for entirely lotic populations in lacking in the upper Missouri River Basin. Specific data voids include: precision of aging structures; age structure; length frequencies; growth; and condition. White suckers were collected in each of the six drainages during fish surveys (N = 1,107) and a subsample of fish was retained for age, growth, condition, and other population analyses. Otoliths were assumed to provide more accurate age estimates than scales and pectoral fin ray sections. Estimated ages from pectoral fin rays and otoliths were more precise (coefficient of variation 9.9) than estimated ages from scales and otoliths (coefficient of variation 17.9), indicating better accuracy of fin rays than scales. Ages estimated from pectoral fin rays were similar to ages estimated from otoliths until age 8, whereas ages from scales and otoliths were similar to age 4 years. The maximum age difference between scales and otoliths was 5 years, whereas the maximum was 3 years between fin rays and otoliths. Otoliths of white suckers are recommended as an aging structure if lethal techniques are allowed. Pectoral fin rays are the best non-lethal alternative aging structure and result in more accurate age and growth estimates of white suckers compared to scales. This information will be useful to anyone aging white suckers because it quantifies the biases of non-lethal aging structures, and may also be useful for other species of catostomids. The age structure, length frequency, weight-length relationships, growth, and relative weight of white sucker populations differed among drainages. Populations contained fish from 0 to 11 years old and from 17 to 381 mm in maximum total length. Growth estimates from scales and pectoral fin ray sections were greater than those from otoliths because of age estimation bias of scales and pectoral fin rays. The maximum growth difference (scale versus otolith) was 99.5 mm at age 4 in the Rock Creek drainage. Mean relative weight of length classes of white suckers ranged from 63 for substock fish in the Nowood River drainage to 95 for quality to preferred length fish in the Frenchman River drainage. Differences in time of collections made statistical comparisons inappropriate for length frequency, growth, and condition analyses. Differences in growth and condition are likely due to local genetic influences, food supply, time of spawning, and growing season length. Some lotic white sucker population characteristics, such as growth and age estimates, were similar to lentic white sucker populations studied in other regions of North America. This study provides baseline information on several lotic populations of white suckers that may be used for future management and monitoring efforts of the species and will also be useful for future comparisons to other lotic populations