333.V Yl5lrl3l rn3 eepoirrNQ.l^ EFFECTS OF URBANIZATION ON PHYSICAL HABITAT FOR TROUT IN STREAMS PLEASE RETURN ITATE DOCUMENTS COLLECTION DEC -3 1984 AAONTANA STATE f iY 1515 E. 6rh A\r. HELENA, MONTANA 59620 Montam Umersity Joint Water Resources Research Center MONTANA STATE LIBRARY S33391W31rl39clWhne Etlecw 01 urtj.mz.twn on ptiytic.l h.b. 3 0864 00048221 9 EFFECTS OF URBANIZATION ON PHYSICAL HABITAT FOR TROUT IN STREAMS by Ray J. White Department of Biology Montana State University Bozeman, MT 59717 Jerry D. Wells Montana Department of Fish, Wildlife & Parks Bozeman, MT 59715 Mary E. Peterson Graduate Student Department of Biology Montana State University Bozeman, MT 59717 Research Project Technical Completion Report The work upon which this report is based was supported in part by fed- eral funds provided by the United States Department of the Interior, as authorized under the Water Research and Development Act of 1978, as amended, through annual cooperative program agreement number 14-34-0001- 2128. This constitutes the final technical completion report for project A-134-MONT. Montana Water Resources Research Center Montana State University Bozeman, MT 59717 October 1983 urbanisation an I'hysical i.ubitat tor troat i« fe> treams WAY 1 6)980 -<^^r^ i)/y<:i "APR 5 WO^ 3o.^, ^^__ i^P/? 12004 The contents of this publication do not necessarily reflect the views and policies of the United States Department of the Interior, nor does mention of trade names or commercial products constitute their endorse- ment or recommendation for use by the United States Government. ABSTRACT Non-urban were more favorable than urban stream sections as habitat for trout and held more trout. The major habitat difference was amount of instream solid overhead hiding cover. Urban land modifications had created unnaturally straight, narrow channels with high, unstable banks with little of the undercuts and woody debris that provide shelter for fish. Urban and non-urban sections did not differ significantly with respect to water velocity, dissolved nitrate, or amount of pools or water turbulence. Per unit stream length, non-urban sections averaged 54% more trout larger than 20 cm (8 inches) and 74% greater total trout biomass than urban sections. In both urban and non-urban areas, trout abundance as kg/ha was generally below the level predicted by the Wyoming Habitat Quality Index (HQI). This could have been due to effects of angling or other unmeasured factors, to measurement errors or to inapplicability of the HQI method to the areas studied. There is evidence that altering the HQI method to consider solid overhead hiding cover and pool-turbulence hiding cover as separate variables rather than as a total cover index will enhance predictiveness. Implications for urban stream fishery management are discussed. ACKNOWLEDGEMENTS This study was suggested in 1980 by Ron Marcoux, then MDFWP Region 3 Fishery Manager, and he participated in early stages of project planning. Chris Clancy, MDFWP District Fishery Biologist at Livingston helped set up study sites on Fleshman Creek. Assisting in construction of electrofishing gear were Daniel Gustafson and Almut White. The following particpated in electrofishing: Daniel Gustafson, Jeff Bagdanov (MDFWP), Wayne Black (MDFWP), Tom Simpson, Sven White, and Jim Hampton, the latter as a volunteer who received neither pay nor university course credit. Helping with habitat measurements were Daniel Gustafson and numerous other fellow graduate students in the Department of Biology, as well as Esther White, Sven White and Sonya White. Daniel Goodman and Andrew Dolloff advised on various aspects of data analysis. TABLE OF CONTENTS ABSTRACT i ACKNOWLEDGEMENTS ii LIST OF TABLES iv LIST OF FIGURES v PROJECT SUMMARY 1 MAIN BODY ^ Introduction ^ Description of Study Areas 5 Methods 10 Selection of Study Stations 10 Estimates of Trout Population and Biomass H Habitat Measurements 12 Data Analysis 13 Results and Discussion 13 Comparisons between Urban and Control Areas 13 Application of the Wyoming HQI 17 Multiple-regression Modelling of Trout and Habitat Relationships 21 Other Observations on Fish Populations 24 Comparions with Other Streams 29 Further Analysis Planned 32 REFERENCES CITED 35 APPENDIX I 37 APPENDIX II 38 APPENDIX III 40 LIST OF TABLES 1. Number of study stations, total length of stream studied, and trout species present in the study streams. 6 2. Unweighted mean values for habitat and trout population variables in urban (U) and control (C) stations of study streams. Values of p are from Mann-Whitney U tests. 15 3. Comparison of trout standing crop predictions by the Wyoming HQI method (Model II) against measured values. 18 ^. Adjusted r squares from stepwise multiple regression of four estimates of logs of trout abundance against logs of habitat variables, before and after mean width was removed as an independent variable 23 5. Estimated number of over-20-cm trout/km for individual stations. 25 6. Numbers of over-20-cm trout/ km for grouped stations (sections). 26 7. Species composition of the trout populations in streams of the Bozeman Creek system, August 1982 30 8 . Trout abundance and stream characteristics in various small streams of the Gallatin, Madison, and Jefferson River drainages, southwestern Montana 31 N LIST OF FIGURES 1. Streams in the vicinity of Bozeman, Montana 7 2. Fleshman Creek and Livingston, Montana 8 3. Relationship of standing crops of trout as measured in the study stations and as predicted by the Wyoming Habitat Quality Index Model II 19 4. Numerical density of over-20-cm trout in study stations of Bozeman Creek, August 1982. Error bars shown are at 95% level 2? 5. Numerical density of over-20-cm trout in study sections (stations grouped) in Bozeman Creek, August 1982. Error bars shown are at 95% level 28 6. Relationship of standing crops of trout to stream gradient in creeks of the Gallatin, Madison and Jefferson River drainages in southwestern Montana (data from Table 8). 33 1 PROJECT SUMMARY Trout populations and habitat were analyzed in 16 urban and 14 non-urban areas of four streams in and near Bozeman and Livingston, Montana. The non-urban stream sections had generally undergone less artificial alteration. Non-urban stream on average held 74% greater weight (biomass) of trout per unit length of channel than did urban stream. Number of over-20-cm (over-8-inch) trout per unit channel length was 54% greater in non-urban than in urban areas. The urban parts of streams were clearly less favorable habitat for trout in certain key respects. The most striking habitat difference detected between urban and non-urban areas was that urban stream had significantly less hiding cover for trout — hiding cover measured as amount of solid material in the water or within a few centimeters above it that could provide overhead concealment for fish. Also, in urban stream, channel width was less and amount of eroding bank greater. It was apparent that urban landfill along stream banks had created straight, narrow channel with high, unstable banks. Such channels tended to lack undercut banks that could provide shelter for trout. Within such straightened and constricted waterways, the greater forces during high water may have swept away logs and other woody debris that would have formed trout cover. Also, people may have removed debris to tidy the appearance of urban channels or to help them conduct flood water more rapidly, thereby destroying trout habitat. 2 Urban and non-urban stream did not differ significantly with respect to water velocity, nitrate content, or amount of cover that was in the form of pools or water-surface trubulence. Trout abundance expressed as kilograms per hectare was generally below the level indicated as potential according to the Wyoming Habitat Quality Index (HQI) in both urban and non-urban areas. This, as well as poor coorelation betwen HQI values and measured trout abundance may have been due to various possible circumstances: (1) suppression of trout abundance by some unmeasured factor or factors, such as angling, pollution or artificial diminution of streamflow discharge, (2) inapplicability of the HQI method to the kinds of stream in the study, (3) error in estimation of habitat and/or trout population variables, or (4) inappropriateness of biomass per unit stream surface area as a measure of trout abundance. A second type of regression analysis of association between trout abundance and habitat variables indicated that the lumping of all cover types (pool, turbulence and overhead) which is involved in the HQI method may have been a major source of poor predictiveness. Multiple correlation was far higher when overhead cover and pool-and-turbulence cover were treated as separate variables. This analysis indicated that 71% of variation in number of over-20-cm trout/km was attributable to the combined variation in the following habitat factors in this order of importance: (1) mean water velocity in stream reach, (2) ratio of late summer streamflow discharge to mean annual discharge, (3) amount of nolid overhead cover, ('I) maximum summer water temperature, and C?) amount of aquatic vegetation in the stream. 3 Trout abundance in some of the relatively unaltered parts of Bozeman Creek compared favorably with that in the more densely populated small trout streams of the same geographic area. The results suggest that trout abundance can be maintained in streams flowing through areas under urban development if the natural form and vegetation of the stream banks and stream bed are not altered in the ways commonly associated with urbanization. If urban changes in land form are kept well away from the immediate riparian area, the natural channel shape and its natural accumulations of living and dead vegetation will furnish cover for substantial populations of trout. It is especially important not to straighten channels , not to remove certain kinds of bank vegetation (such as high grass and low brush), and not to conduct excessive removal of downed logs and other woody debris from stream channels. Much could be done to increase trout abundance in physically damaged urban streams by restoring channel form and vegetation to resemble the natural situation. Even in parts of streams that remain unaltered, habitat may often be enhanced and trout populations increased by creating more instream cover for trout than presently exists. 4 INTRODUCTION Expanding urbanization and its effects of unfavorably reshaping the channels of urban streams is a serious fisheries problem that has long seemed obvious to biologists but may not be as apparent to others. Urban stream alterations typically regarded as detrimental to fish habitat include channel straightening and other unnatural relocation; excessive stream widening or channel constriction through landfill, bridges and culverts; impoundments; elimination of pools, riffles and biologically productive side channels; removal of instream cover used by fish; and construction of unnatural structures such as bulkheads, walls and certain kinds of deflectors in and along streams. This situation has not been thoroughly studied in quantitative terms, in large part perhaps because it has seemed so blatantly obvious to specialists. Yet, when it comes to making policy decisions on urban development along streams, quantitative data are needed. Determining components of fish habitat that are adversely affected by practices similar to these, although not necessarily classifiable as urban, has been addressed in several studied. Whitney and Bailey (1959) and Elser (1967) both found that channel alterations involved in highway construction caused significant decrease in the salmonid carrying capacity of two Montana streams. In particular, Elser determined the amount of cover per unit area of stream to be substantially less and the occurrence of areas with more shallow, fast reaches greater in altered stream sections. In a stream- straightening project in Rocky Creek, Gallatin County, Montana, changes in channel morphometry and loss of in-streara and bank cover were responsible 5 for reduced trout abundance (Wells 1977). Still other studies have found trout populations to be limited by quality of physical habitat (Boussu 1954, Kalleberg 1958, Lewis 1969, and Newman 1956). Our study was intended to evaluate habitat quality by comparing measurements of various habitat attributes between urbanized and non-urbanized sections of stream and analysing them for correlation with abundance of trout. While it is almost impossible to locate pristine, unaltered streams in and near urban areas, one can probably make adequate analyses by comparing altered and "less altered" parts of streams in and near towns within generally urbanized areas. In urban creeks of the Bozeman area in Montana, stream problems have been studied from water quality/pollutional standpoints (Anderson 1977; Blue Ribbons of the Big Sky Country APO 1979). The emphases of these studies were on sediment pollution and on pollutants hazardous to human health. The urban reaches of streams were not analysed with respect to physical suitability as habitat for fish. The hypotheses selected for this study were (1) that key habitat characteristics are less favorable to trout in altered than in less-altered parts of urban-area streams and that (2) trout are less abundant in the more altered parts of urban-area streams. DESCRIPTION OF STUDY AREAS Thirty channel reaches (stations) totalling 3,772 meters of stream were analysed for trout habitat attributes and trout abundance in four creeks that have undergone varying degrees of artificial alteration in the course of urban- ization (Table 1). Three of the streams flow through the city of Bozeman (Fig. 1) Table 1. Number of study stations, total length of stream studied, and trout species present in the study streams. Bozeman Cr. Mathew Bird Cr, Figgins Cr. Fleshman Cr. Total Number of ition length (m ) Trout study sta Species Prevalence 18 2,064 rainbow brook brown numerous numerous few 7 739 rainbow brook numerous numerous 2 379 brook numerous 3 590 brown brook rainbow numerous few few Total 30 3,772 EAST GALLATIN RIVER MATHEW BIRD CREEK Figure 1. Streams in the vicinity of Bozeman, Montana. Figure 2. F'' eshtnan Creek and Livingston, Montana. 9 in Gallatin County: Bozeman Creek (18 stations of ca. 100-200 m each, distributed over 6.4 km of stream), Mathew Bird Creek (7 stations) and Figgins Creek (2 stations). One stream, Fleshman Creek (3 stations) flows through the city of Livingston (Fig. 2) in Park County. Details of station locations are in Appendix I Urban stations did not include the most drastically altered parts of some streams — those parts that had been completely enclosed in culvert for one or more city blocks. It would have been too dangerous to electrofish and perhaps also to do other sampling in these sections. Study station lengths ranged from 89 to 201 m, except for one that was 300 m long. Seventy-two per cent of stations were between 90 and 100 m long. Bozeman (Sourdough) Creek originates at the outlet of Mystic Reservoir (elevation 1950 m) in the north end of the Gallatin Range in southwestern Montana and flows northwest for 26.6 km, dropping 510 m (19.2 m/km) to its mouth (elevation 1440 m) at the East Gallatin River below Bozeman. The Bozeman 2 Creek drainage basin covers about 168 km . Average annual precipitation for the basin is 6l cm (24 inches). Mathew Bird Creek (Spring Creek on the Bozeman topographic quadrangle of 1953) is tributary to Bozeman Creek at about stream km 3-5 at elevation 1475 m. Elevation of the source is 1585 m. Stream length is about 3.3 km. Figgins Creek (Middle Creek Ditch), enters Mathew Bird Creek at stream kilometer 1.6. It originates from Hyalite Creek in T3S,R8E,Sec. 3- Fleshman Creek heads in eastern foothills of the Bridger Range in Park County and flows southeast for 14.8 km before discharging into the Yellowstone River in the city of Livingston. Its elevations at origin and mouth are 1707 m and 1372 m, respectively. The average annual precipitation for the Fleshman Creek drainage basin is 40.6 cm (16 inches). 10 Large parts of these four streams are bordered by agricultural, industrial and municipal lands. Land uses include ranching, small grain production, small industry, and urban development — and in the headwaters of Bozeman and Fleshman Creeks, forest uses. A shift from agricultural to urban-municipal use is rapidly occurring in many parts of the bordering lands, with increase in artificial alterations of stream channels and banks. The range of approximate annual mean discharges over all study sites was 0.04 m /sec to 1.46 m /sec. Average critical period discharge (late summer flow from August 1 to September 15) ranged from 0.034 m /sec to 0.80 m /sec. METHODS Selection of Study Stations Potential study streams were suggested by fishery biologists of the Montana Department of Fish, Wildlife and Parks. Identification of deliberate stream course relocations done in recent decades was accomplished by inspection of aerial photos (scale: 12 inches/mile) taken in 1937, 1954 and 1977. Each stream was divided into reference-station reaches by measuring and marking of 100-m reaches contiguously from the mouth upstream for several kilometers including both urban and non-urban areas, but with some gaps in the system of stations, owing to access problems. Within each stream, the stations were classified as being either urban (definitely within an urban-impacted area) or control (non-urban — although perhaps "suburban" in some cases — and relatively less impacted by artificial alteration). Within each category, a sample of stations was selected for study. On Bozeman Creek, many of these stations were contiguous in groups of two or three. For some analyses, contiguous stations were treated as combined areas, called "sections," with the data lumped within sections. 11 Estimates of Trout Population and Biomass Between August 1 and September 8, 1982, electrofishing for mark-recapture estimates of trout populations was done in each of the 30 stations selected for study. The electrofishing unit consisted of a 240-volt, 3500-watt alternator and Coffelt control box (150 to 600 volts DC) mounted in a canoe and towed upstream by a crew using two positive electrodes ahead of the unit, a negative electrode trailing in the water behind the canoe — except in very shallow stations, where long-line gear was used instead of the canoe-mounted unit. Collected fish were anesthetized in MS-222 (Tricane Methanesulfonate ) , weighed to the nearest 2 grams, and measured to the nearest 1 mm in length. The lower caudal fin was clipped; when two 300-meter sections were contingous, the upper caudal fin of the upstream section trout was clipped. Fish were carried in large frame nets downstream to the bottom of the station from which they had been caught and released. Two weeks were allowed between the marking and recapture runs. The population estimate for each station (and 3-station section) was calculated by the modified Petersen mark-recapture formula of Ricker (1975). Fish were grouped into 20-mm size classes, 100 to 119 mm being the smallest. Where numbers were too low to calculate a reliable pouplation estimate for each species within a station, data were lumped either by species or by combining several stations, and the estimates calculated then reapportioned according to total of fish marked on first run plus the unmarked (new) fish caught on the second run. Biomass was calculated by multiplying length-group estimates of trout numbers by mean weight for the group. Mean weight of length group was determined from length-weight regression equations for each species in each 12 station. Standing crop in kilograms and numerical density of over-20-cm trout were calculated on per-hectare and per-stream-kilometer bases. The 95% confidence intervals (Ricker, 1975: 8l) for the estimates of numbers of trout over 20 cm were also calculated for each station and section and converted to density. Habitat Measurements Stream habitat attributes were measured during the low flow period of September 1982 to March I983, following the prcedures described by Binns (1979, 1982) for the Wyoming Habitat Quality Index. Habitat variables measured or evaluated were maximum summer water temperature, stream discharge, channel width, percent of eroding banks, length of thalweg, mid-channel length, amount of submerged aquatic vegetation (substrate), nitrate nitrogen level, mean water velocity through station, and percent cover. The cover designation included any of the following features that occurred in water at least 15 cm deep: undercut banks, instream and closely overhanging terrestrial vegetation, instream debris (brush, logs and "snags"), pockets of surface turbulence, and pools. Pools were identified as abrupt increases in water depth. The length and width of these cover features were measured, summed and converted to per cent of total stream surface area in the station. Subsequently, the categories of pool and surface turbulence pockets were, for more detailed analyses, separated from the other cover types — which could be categorized as "solid overhead cover." Mean water velocity through station was measured by timing an injection of rhodamine dye (Binns 1982). Nitrate nitrogen levels were measured by the ultraviolet spectrophotometric method as described in Standard Methods for Examination of Water and Wastewater 13 (1976). Annual stream flow variation was determined by the use of past discharge records and estimates from governmental agencies . Late summer streamf low discharge for each stream was measured with a gurley "pigmy" flow meter. Stream discharge gauging records were available only for Bozeman Creek. The USDA Soil Conservation Service operated a gauge at stream kilometer 5.6 in 1977 and 1978. Data Analysis The mean values of urban- and control-section data on habitat and trout population variables were compared. The data were examined for significant differences via the non-parametric Mann-Whitney U test. Data on habitat variables were entered into the Wyoming Habitat Quality Index Model II (Binns 1979). The resultant predictions of trout abundance in kilograms per hectare were compared with the actual measurements of trout abundance. The relationship of trout abundance to habitat variables was further analysed by multiple-regression modelling. The logarithms (base 10) of habitat data were entered as independent variables. The logarithms of four expressions of trout abundance (number of over-20-cm trout per stream kilometer and per hectare and kilograms of all trout per stream kilometer and per hectare) were entered separately as dependent variables. In the step-wise procedure, the F- value set for exclusion of independent variables from the model was 1.0. RESULTS AND DISCUSSION Comparisons between Urban and Control Areas The most striking difference detected between urban and control areas of 14 the streams was in terms of amount of solid overhead concealment cover for trout — cover submerged in the stream or within a few centimeters above the water surface. Averaged over all 30 study stations, control areas had 60% more such cover than did urban areas — 8.33% vs 5.22%, a difference significant at the 97.5% confidence level (Table 2). For Bozeman Creek alone, the difference was even more pronounced. Urban areas contained especially low amounts of cover, slightly less than 1%, and the control areas did not have particularly high amounts, 5.3%, but this was a several-fold difference or 485% more in control areas, a difference significant at the 99.8% level. It can be inferred that reduction of instream concealment cover is a major impact of urbanization along trout streams of the types included in our study. Wells (1977) also found that cover reduction was strongly associated with channel straightening in another stream near Bozeman, although that was not a consequence of urbanization. The only other measured habitat variables having significant urban-vs-control differences at the 90% confidence level or higher were channel width, which was narrower in urban areas (96.3% confidence level), and amount of eroding bank — 34% in urban areas vs 23% in control areas, a difference of 48% relative to the control figure and significant at the 90% confidence level (Table 2). Part of the difference in channel width could be an artifact of the longitudinal distribution of study stations within the stream systems. This possible effect has not yet been analysed. However, we suspect that encroach- ment on the channel by land fill involved in urban development may have been a major influence in narrowing the urban stream sections. The higher degree of bank erosion in urban areas may represent decreased 15 Table 2. Unweighted mean values for habitat and trout population variables in urban (U) and control (C) stations of study streams. Values of p are from Mann-Whitney U tests. Variable Station type Bozeman Creek (n=9U,9C) Mathew Bird Creek (n=4U,3C) Figgins Fleshman Creek Creek (n=lU,lC) (n=2U,lC) Habitat Variables Channel U width (m) C 4.84 7.01 (p<.001) 2.31 2.34 (ns) 1.40 1.58 2.15 1.73 3.65 5.24 (p=.047) Solid over- U head cover (%) 0.91 5.32 (p<.002) 9.24 17.53 (ns) 30.42 13.33 3.94 2.84 5.22 8.33 {p=.025) Pool, turbu- U lence cover (%) C 7.13 4.78 (ns) 8.52 7-21 (ns) 0.50 5.47 7.38 6.38 7.09 5.46 (p=.377) Eroding banks (%) 32.85 19.07 (ns) 24.01 1.95 (ns) 12.65 9.00 70.3 126.0 33.8 22.6 (p=.101) Water velocity (cm/sec) 79.5 77.2 (ns) 43.5 22.9 (p<.02) 12.2 12.2 32.6 26.0 60.4 57.3 (p=.790) NO -N (m|/l) 0.420 0.371 (ns) 0.612 0.338 0.325 0.338 0.414 0.429 0.393 0.431 (p=.822) Trout Variables Number of Trout 20 cm and Larger Per stream km U 323 188 64 C 523 79 241 (p<.02) (ns) 244 376 (p=.154) Per stream ha U 673 827 456 C 754 346 1524 (ns) (ns) 421 483 666 702 (p=.984) Kilograms of All Trout Per stream km U 56.9 35.4 15.3 20.1 44.3 C 104.1 (p<.01) 24.7 (ns) 51.0 18.6 77.2 (p=.052) Per stream ha U 118.5 155.1 109.5 93.1 123.9 C 152.4 (ns) 107.5 (ns) 322.9 107.5 151.8 (p=.355) 16 channel stability, a hydraulic response of the streams to construction of unnaturally straight and narrow channels with high, steep and constricting banks. There had apparently also been some past artificial channel straightenings in some of the control stations. Had that not been the case, perhaps the difference in amounts of bank erosion between urban and control areas would have been even greater. Most of the erosion was at high elevation on banks, apparently coinciding with spring flood stage. Urban station 4 on Mathew Bird Creek, while having only 9.76% erosion of banks, showed obvious potential for erosion due to urban development. With construction of condominiums occurring within 30 meters or less of the stream, all of the riparian willows and high grasses had been removed. Where natural deeply undercut bank cover had been formed by the stream action, major portions of it were beginning to slump into the stream, leaving bare soil exposed. Fleshman Creek's stations (including the control station which had apparently been ditched some years ago) had the highest average percent of eroding banks (88.89%). We do not believe eroding stream margins to be of direct detriment to trout in the immediate vicinity of the erosion but suspect they may be indicative of unfavorable action of currents at high flow in unnatural channel configurations , and there may often be damage to downstream areas via siltation of spawning and food-producing stream bed. The urban areas had somewhat greater measured amounts of stream area in pools or having the water surface broken by turbulence that might provide concealment cover for fish. However the urban-vs-control differences in this variable were not statistically significant at the 95% level of confidence. There was even less difference in the water velocity and nitrate-nitrogen variables. 17 Trout abundance was higher overall in control stations than in urban stations with respect to all four expressions of abundance that were calculated (Table 2). | Trout abundances expressed per unit of stream length showed much greater statistical I significance of urban-vs-control differences than did per-unit-area expressions. In terms of kilograms per stream kilometer, control areas had lk% more trout than urban areas did, a difference significant at the 97-5% confidence level. In Bozeman Creek, the control sections averaged 83% more kg of trout per km, a difference significant at the 99% level. The urban-vs-control differences in terms of number of over-20-cm (over-8-inch ) trout per kilometer were significant at the 85% level overall and at the 98% level for Bozeman Creek, which had the majority of stations. It can be concluded that the results are consistent with the hypotheses of this study: (1) that key habitat variables are less favorable for trout in altered (urban) than in less altered stream sections, and (2) that trout are less abundant in the more altered sections. The following sections of this report further support these conclusions and provide some information toward identifying possible causative processes. Application of the Wyoming HQI Application of the Wyoming Habitat Quality Index Model II (Binns 1979) to the data from all 30 study stations yielded poor correlation (r = 0.228) between predicted and actual standing crops of trout expressed as kg/ha (Table -3, Figure 3), however it appears that if data from study sections (combinations of contiguous stations) were plotted instead of stations, variability would be reduced and the fit would be somewhat tighter. (This analysis remains to be done.) It would still be the case, however, that in the great majority of cases, predicted values fall below the actual values of standing crop. 18 Table 3. Comparison of trout standing crop predictions by the Wyoming HQI method (Model II) against measured values. ~~~ Standing Crop of Trout (kg/ha )~ Station Type* Predicted Measured Bozmean Creek 141 91.4 9 C 253 175 13 C 229 306 14 C 237 117 15 C 229 119 19 U 26.2 83.2 20 U 253 181 21 U 179 81 22 U 148 ■ 123 23 U 148 216 24 U 226 125 26 U 116 97.1 27 U 148 93.9 28 U 148 73-0 36 C 148 227 60 C 214 95.7 61 C 296 100 1 U 307 160 2 U 181 179 3 U 285 148 4 U 148 133 5 C 265 34. 6 C 265 161 7 C 297 127 1 U 259 109 2 C 223 323 1 U 61.2 96.2 2 U 61.2 90.0 3 C 61.2 108 c 284 c 150 c 253 c 229 c 237 c 229 u 26.2 u 253 u 179 u 148 u 148 u 226 u 116 u 148 u 148 c 148 c 214 c 296 Mathew Bird Creek u 307 u 181 u 285 u 148 c 265 c 265 c 297 Figgins Creek u 259 c 223 Fleshman Creek 19 o BOZEfviAN CR. ° M. BIRD CR. o FIGGIN5 CR. A FLESHMAN CR Solid points = urban Y = 101..228X r = .2276 r2= 0518 100 200 PREDICTED KG/HA 300 Figure 3. Relationship of standing crops of trout as measured in the study stations and as predicted by the Wyoming Habitat Quality Index Model II. 20 This could be interpreted in several ways. The HQI model may not be applicable to the trout/habitat relationships in the kinds of streams included in our study. The poor fit of the data might also be due to deficient accuracy or precision in our measurement of some habitat variables or of standing crop of trout. Some of the habitat measurements called for in the HQI method seem more subjective than would be desirable (e.g., vegetative substrate) or anotherwise difficult to accurately determine (e.g. annual streamflow variation), but this may be compensated for at least partially by the transformation of measurements to class ratings involved in the procedure (Binus 1979). Comparison of the results of the HQI method against results of the multiple- regression analysis in the following section of the report would seem to indicate that much of the unpredictiveness of the HQI in our streams lay in the procedure of combining all forms of cover, i.e., failure to distinguish between pool-and- turbulence cover and solid overhead cover. Comparison of the predictiveness of the HQI method in our streams against the better predictiveness of the multiple- regression analyses desribed below may also point to an inappropriateness of using biomass per unit of stream surface area as an expression of trout abundance. It is far more important to consider the possible inference from the lower-than-predicted distribution of actual trout abundances that some unmeasured influence is preventing trout in these streams form being as abundant as the HQI predicts they should be, i.e., as the physical habitat would allow. Actual abundance is lower than HQI-predicted abundance particularly in the stations with highest predicted values, i.e., highest habitat rating (Fig. 3). Candidate variables for such abundance-depressing effect would be water pollution and intense angling harvest. Neither were measured in this study, however, urban water pollution problems were revealed in a previous study on Bozeman Creek (Blue 21 Ribbons of the Big Sky Country APO 1979). It could also be that streamflow variability was not adequately estimated. The summer low flow may often be less due to irrigation withdrawal than it was when measured in 1982, a fairly wet summer. Also, there may be withdrawals for municipal water supply, making winter low flow severe in the Bozeman Creek system. This was not analysed in our study. Insufficient reproduction of trout could be another factor preventing standing crop from reaching the potential indicated by the HQI. Although age-structure analyses of the trout populations have not yet been done, the size structures of the Bozeman Creek rainbow trout population, the main fish in that stream, appear to indicate fewer fish of age I and II than of age III or IV. This is the reverse of the situation that must exist in a population being replenished by local reproduction. It is likely that the rainbow trout population of our study area on Bozeman Creek consists largely of immigrants from upstream or from the East Gallatin River, downstream. The combination of immigration, body growth and whatever reproduction exists within the study may not be creating enough biomass to saturate the habitat. If low reproductive rate is also an influence for brook trout in Bozeman Creek and in the other study streams (it is the predominant fish in the other streams), it is not as strongly the case. Size distributions of brook trout in most study stations indicate that a more normal age structure probably exists. Multiple-regression Modelling of Trout and Habitat Relationships When logarithms of the habitat variables involved in the HQI (but not ~^ transformed to class ratings) were entered as independent variables into stepwise multiple regression against logarithms of each of the four trout abundance expression; used separately as dependent variables, stronger correlations resulted than in the 22 HQI model. Further, when the HQI cover variable was separated into a solid-overhead- cover component and a pool-and-turbulence-cover component and these new components entered as independent variables along with the rest, much stronger correlation yet was obtained . In a set of stepwise multiple regressions involving (1) all 30 study stations, (2) control stations only, and (3) urban stations only, mean channel width was the first variable entered, based on its high initial influence relative to that of other independent variables. However, as other variables were entered, mean width rapidly lost significance, its F-value falling below the predetermined rejection level of 1.0. It is inferred that channel width was meaningless in describing the effect of habitat on trout abundance. Another stepwise multiple regression was run with the channel-width variable omitted, and the correlation improved (Table 4). Number of over-20-cm trout/km was consistently the dependent variable having tighest correlation (Table 4). We infer that the habitat variables are related to trout abundance in a linear, rather than areal fashion. Trout abundance per unit stream surface area is likely to be poorly correlated with habitat quality because trout orient strongly to instream cover (Hunt 1971, Wesche 1976, Devore and White 1978, Enk 1977) which tends to be concentrated along channel margins, hence is a rather linear variable. With the channel-width variable omitted from multiple regression, the dependent trout abundance variable most highly correlated with habitat variables was still the number of over-20-cm trout/km (Table 4). For the model involving 2 all 30 stations, the adjusted r values* indicated that variation in habitat 2 2 2 ♦Adjusted r = r - [ (K-1) / (N-k) ][l-r ], where K = number of independent variables and N - number of cases. 23 o m B e w •H (« O 2 o QD 1 1 2 rs: a: UJ Q. h- Z) o a: 1- 2 CM o 00 1 en o *" (NJ ^ cr 111 LU li 1 > cn O u U- o z. < cr :? UJ LU m M :k O Z) CD 2 1 -o- - -o -^^- - — """^ 1 0 _ ^ 0 — ^ — 0 I 1 ^2 Z) CD >4- O UJ > o CD CO < UJ CJ 2 < WVI/HSId 28 CD 3 o rH to CO -4- O 2 T3 in ON 3 ^ UJ « "" > 0) o s u? 00 < U 0 O O CM U xi LU 4-1 to O 6 0) J^ z < h- 1 M 2 (N to Q !^ O > M o w o • If) C^J U ON CO ■H •-! -a D tH 3 C^ « < U •H " z o O 2 U H vJ3ir\ Lr^ LTN LTN LTN LTn ir\ w^ ITS ir\ ITS u^ LrNir\Lr\ SSS SS° ° ° ° ° °Soo<5<=°° J^ vO ^ (%l oo vO s o g r cr- ° l«-\ J- o ~ ^ o^ ^^ oo s; CTN t^ o h^ CJs O r^ r^ r^ OO o U-N CXI C3^ O -J- o ^.^3 U^ K> VXD CS ir\ V- _3- [-^ o -r oo C3 ^r no CTv c — CD o^ o crv ctn o o o CTv ,^ 39 o o ^•^3 -J- ^£> U^ O — 1 O U^ U> c-~ O t^ O O O CD Appendix III. Multiple-regression models describing numerical density of over-20-cm trout (trout/km) as a function of stream nabitat variables. 1 . All 30 stations included (urban and controls) log trout/km = 15.289 + 1.77 log water velocity + 0.153 log critical period flow/mean flow + 0.317 log of solid overtiead cover - 0.3'»8 log max summer water temp + 0.2't3 log vegetational substrate rating r = 0.8753, simple r^ - 0.763, adjusted r^* = 0.713 Order or Adj. r F-values at step entry Variable at step log wtrvel log CPf/MF log cover log maxT log substi log wtrvel log CPF/MF log cover log maxT log substr. .5't2 .670 .703 .713 35.297 23.606 25.99't 30,856 25.552 8.768 8.6'.7 If. 902 1.15't 5.352 6.12't 5.105 3.8't5 5.902 1.922 2. Urban Stations Only log trout/km = - 64.255 - 0.762 log water velocity + 0.905 log vegetational substrate rating + 1.560 log NO -N + 0.72't log annual streamflow variation + 1.827 log max T + l.O'i't log critical period flow/mean flow - 0.'«22 log % pool & turbulence cover + 0.215 log % eroding bank r - 0.976, simple r Order or entry Variabl 0.952, adjusted r Adj. r F-value at step at step log wtrvel log substr log NO log ASFV log maxT log CPF/MF log pool loger. bu- 1 log wtrvel 59't 22.966 , 2 log substr 710 35.277 6.567 3 log NO if log ASrv 715 28.5'.0 7.687 1.252 7't6 IZ.'tO't 5.598 5.7'.7 2.'.66 1 5 log maxT 7'tl 3.051 5.828 '..509 2. 1.60 0.796 6 log CPF/MF 8'*5 0.872 15.550 5.'t63 3.519 8.785 7.'.7'. 7 log pool 858 0.829 19.159 7.55'. 1.961 10.756 't.567 1.967 8 log er.buk. 898 3.5't3 50.580 1'..650 2.778 19.088 7.7'.0 5.225 -..litS * Adjusted R r r^ - [(K- 1)/N -K)][l-r2] , where K := number of independent variables N = number of cases. 41 Appendix III. Continued. 3. Control Stations Only log trout/km = hk.Zll + 1.300 log water velocity - 0.768 log maxT - 0.3't't log annual streamflow variation + 0.298 log NO -N r = O.g'tg, simple r^ = 0.901, adjusted r^ = 0.857 Adj. r^ at step .578 .761 .837 .857 F-value at step log wtrvel log maxT log ASFV log NO 1 log wtrvel 2 log maxT 3 log ASFV h log NO 40.700 61.105 45.61'. 10.162 20.625 20.266 6.114 7.032 2.385 I