Low-head Sea Lamprey Barrier Effects on Stream Habitat and Fish Communities in the Great Lakes Basin

J. Great Lakes Res. 29 (Supplement 1):386 402 Internat. Assoc. Great Lakes Res., 2003 Low-head Sea Lamprey Barrier Effects on Stream Habitat and Fish Communities in the Great Lakes Basin Hope R. Dodd 1, 5,*, Daniel B. Hayes 1, Jeffery R. Baylis 2, Leon M. Carl 3,6, Jon D. Goldstein 2, Robert L. McLaughlin 4, David L. G. Noakes 4, Louise M. Porto 4,7, and Michael L. Jones 1 1 Michigan State University Department of Fisheries and Wildlife East Lansing, Michigan 48824 2 University of Wisconsin Department of Zoology Madison, Wisconsin 53706 3 Ontario Ministry of Natural Resources Trent University 1600 West Bank Dr. Peterborough, Ontario K9J 8N8 4 University of Guelph Axelrod Institute of Ichthyology and Department of Zoology Guelph, Ontario N1G 2W1 ABSTRACT. Low-head barriers are used to block adult sea lamprey (Petromyzon marinus) from upstream spawning habitat. However, these barriers may impact stream fish communities through restriction of fish movement and habitat alteration. During the summer of 1996, the fish community and habitat conditions in twenty-four stream pairs were sampled across the Great Lakes basin. Seven of these stream pairs were re-sampled in 1997. Each pair consisted of a barrier stream with a low-head barrier and a reference stream without a low-head barrier. On average, barrier streams were significantly deeper (df = 179, P = 0.0018) and wider (df = 179, P = 0.0236) than reference streams, but temperature and substrate were similar (df = 183, P = 0.9027; df = 179, P = 0.999). Barrier streams contained approximately four more fish species on average than reference streams. However, streams with lowhead barriers showed a greater upstream decline in species richness compared to reference streams with a net loss of 2.4 species. Barrier streams also showed a peak in richness directly downstream of the barriers, indicating that these barriers block fish movement upstream. Using Sørenson s similarity index (based on presence/absence), a comparison of fish community assemblages above and below low-head barriers was not significantly different than upstream and downstream sites on reference streams (n = 96, P > 0.05), implying they have relatively little effect on overall fish assemblage composition. Differences in the frequency of occurrence and abundance between barrier and reference streams was apparent for some species, suggesting their sensitivity to barriers. INDEX WORDS: habitat. Low-head barriers, low-head dams, Great Lakes, fish assemblage, physical * Corresponding author. E-mail: hopedodd@inhs.uiuc.edu 5Present Address: Illinois Natural History Survey, Center for Aquatic Ecology, Champaign, Illinois 61820 6 Present Address: United States Geological Survey, Great Lakes Science Center, 1451 Green Road, Ann Arbor, Michigan 48105 7 Present Address: R.L. and L. Environmental Services Ltd., 201 Columbia Ave, Castlingar, British Columbia, V1N 1A2 386

Low-head Barrier Effects on Habitat and Fish 387 INTRODUCTION The sea lamprey (Petromyzon marinus), a native of the Atlantic Ocean, invaded the upper Great Lakes following the construction of the Welland Canal (Pearce et al. 1980). This parasitic species, along with substantial fishing pressure, nearly eliminated native lake trout (Salvelinus namaycush) and populations of other large commercial fishes in the Great Lakes, resulting in the need to control sea lamprey (Lawrie 1970, Pearce et al. 1980, Smith and Tibbles 1980). Since 1950, a variety of control methods have been instituted to reduce sea lamprey abundance in the Great Lakes. The primary control method used in Great Lake tributaries is chemical treatment with 3-trifluoromethyl-4-nitrophenol (TFM). This lampricide targets ammocoetes buried in the stream bed (Applegate et al. 1957, Applegate et al. 1961, Hunn and Youngs 1980). Although TFM has little apparent effect on fish species other than lampreys (both sea lamprey and native lampreys), public sentiment and the high cost of chemical control has led the Great Lakes Fishery Commission to search for alternative control methods to reduce the use of lampricides 50% by the end of the year 2000 (Great Lakes Fishery Commission 1990). An alternative to chemical treatment is the construction of low-head barriers. These barriers are built to prevent adult sea lampreys from migrating to suitable spawning habitat in Great Lakes tributaries. Early attempts at blocking spawning migrations included installation of mechanical weirs and traps (Applegate and Smith 1951) and alternatingcurrent (AC) electrical barriers as well as low-head barriers. Mechanical weirs and AC electrical barriers were deemed as ineffective, costly, and sources of mortality to non-target species (Erkkila et al. 1956, McLain 1957) and were modified or discontinued by the 1970s (Dahl and McDonald 1980, Hunn and Youngs 1980). By the mid-1970s, the Great Lakes Fishery Commission approved construction of low-head sea lamprey barriers as part of the Sea Lamprey Control Program (Hunn and Youngs 1980). The low-head barriers in our study ranged in head height from approximately 45 to 300 cm. Although the use of barriers predates chemical treatments in several Great Lake tributaries, there has been little study on the effects low-head sea lamprey barriers have on the entire fish community, particularly at a basin-wide scale (Hunns and Youngs 1980, Kelso and Noltie 1990). While lowhead sea lamprey barriers do not appear to cause direct mortality of non-target species, they can have negative effects at several different scales, ranging from species-level changes to changes at the ecosystem or landscape scale (Pringle 1997). The most obvious impact is the blocking of fish movement during periods of spawning or seasonal movement to locate habitat and food resources (Porto et al. 1999). Low-head barriers may also indirectly affect fish communities by changing the geomorphology and water quality of the stream (Pringle 1997). In this paper, evidence is provided for an impact of low-head sea lamprey barriers on stream fish communities throughout the Great Lakes basin. A priori, streams containing low-head barriers were expected to contain fewer species and have a greater loss of species upstream of the barrier when compared to upstream sections of nearby reference streams (those without a barrier). Abundance of some non-target species was also expected to decrease upstream of the barriers due to habitat alteration or blocking of movement upstream, thereby altering the fish community structure and population abundance and size composition. The main focus of this study was to examine fish assemblages in streams with low-head barriers that were primarily built for sea lamprey control, however, the results of this study may apply to other types of small dams or barrier structures. STUDY SITES Forty-seven tributaries were sampled across the Great Lakes basin in the summer (June-August) of 1996, and 14 streams were re-sampled in the summer of 1997 (Table 1, Fig. 1). Streams were re-sampled in 1997 within 2 weeks of the 1996 sampling dates in order to reduce annual variability in fish assemblage data. Streams were paired, with each pair containing a stream with a low-head barrier (barrier stream) and a nearby reference stream (without a barrier). Due to the lack of suitable reference streams, South Otter Creek in the Lake Erie drainage was used twice in the pairings (Table 1). Stream pairs were selected with the advice of sea lamprey control agents and technical experts. Reference streams were selected based on proximity and similarity to the barrier stream in terms of stream size, geology, and geography (Table 1). The majority of streams were sampled at six locations, three sampling sites above and three below the barrier, or a corresponding location on the reference stream. Location of barriers upstream of the mouth

388 Dodd et al. FIG. 1. Location of streams sampled in the Great Lakes Basin (Streams identified in Table 1). varied among streams. Therefore, some streams were sampled with fewer sites when the barrier was too close to the mouth to allow placement of three sampling sites downstream. Site location was determined primarily by access to streams with each site separated by at least five to seven times the stream width. The small impoundment just upstream of the barrier was excluded because water depth was too great to sample with our equipment, and the plunge pool directly downstream of the barrier was excluded due to the potential of a localized effect of the barrier to aggregate fish in this location. METHODS Each sampling site contained a downstream, upstream, and middle transect perpendicular to flow. The downstream transect was marked where the thalweg crossed the stream. The upper transect, marking the end of the site, was placed five to seven times the stream width from the downstream transect, and a middle transect was placed at approximately half the length of the site. At each transect, stream width, maximum depth, and a pebble count of 50 stream bed particles were measured to determine in-stream habitat characteristics. Pebble counts were taken by starting at one side of the stream bank and walking along the transect. At each step, the observer would reach down and determine the size of a random stream bed particle (Kondolf and Li 1992). In addition to physical habitat measurements, temperature and conductivity were also measured at the downstream transect to aid in setting the electroshocking unit. A single upstream pass with a Smith-Root backpack electroshocker was used to assess fish species composition, richness, and relative abundance (Simonson and Lyons 1995). Most fish were identified in the field and total length was measured. Fish that could not be identified in the field were fixed in 10% formalin and preserved in 70% isopropyl alcohol for further identification in the laboratory. Voucher specimens that could not be identified due to their small size or to damage during transport and preservation were excluded from our analysis. Because the primary interest was detecting differences between upstream and downstream fish assemblages in streams with low-head barriers, sites were classified into above and below stream sections. An α value (Type I error) of 0.05 was used for all statistical tests. To determine differences in width, maximum depth, and particle type between barrier and reference streams, a nested mixed model analysis of variance (ANOVA, Littell et al. 1996) design was used treating stream pairing, stream, and position (Above or Below) within each stream as random effects and stream type (Barrier or Reference) as the fixed effect. The relationship between stream habitat characteristics and species richness (number of species caught) was examined with a nested mixed model analysis of covariance (ANCOVA) by comparing differences in richness among stream types (Barrier vs. Reference) and stream positions (Above vs. Below) using average width, maximum depth, and substrate size as covariates. The net decline in species richness (impact value) due to the barrier was calculated using the formula: I = (BA BB) (RA RB) (1) where I is the net decline of species for a stream pair and where other variables refer to species richness within a stream section for a stream pair (BA = Barrier Above, BB = Barrier Below, RA = Reference Above, and RB = Reference Below). A twotailed t-test was used to compare our observed species loss to our expected value of no net loss in species richness. Influence of habitat on the number of species lost above the barriers was examined through regressions of average width and maximum depth on loss of species for each stream pair. Simi-

Low-head Barrier Effects on Habitat and Fish 389 TABLE 1. Location and physical characteristics of study streams sampled in summer 1996 and re-sampled in summer 1997 (designated by *). Note: South Otter was used twice as a reference stream. Particle sizes were classified as follows: 1 = clay, 2 = silt, 3 = sand, 4 = gravel, 5 = cobble, 6 = boulder, 7 = bedrock. Stream Location Mean Mean Mean Head pair Stream (State/ width depth particle height no. Stream name type Province) Lake (m) (cm) size (cm) 1* East Branch AuGres Barrier Michigan Huron 10.2 69.3 3.4 67 1* West Branch Rifle Reference Michigan Huron 8.6 77.8 3.5 2* Albany Barrier Michigan Huron 6.1 51.2 3.6 77 2* Beavertail Reference Michigan Huron 3.9 65.5 2.9 3* Echo Barrier Ontario Huron 16.7 98.8 3.6 54 3* Root Reference Ontario Huron 10.2 52.4 4.5 4 Koshkawong Barrier Ontario Huron 10.6 66.5 5.1 43 4 Brown Reference Ontario Huron 3.6 32.9 4.0 5 Manitou Barrier Ontario Huron 15.0 72.8 5.0 83 5 Blue Jay Reference Ontario Huron 10.3 57.3 4.9 6 Sturgeon Barrier Ontario Huron 8.8 78.2 2.8 60 6 Mad Reference Ontario Huron 11.1 93.7 2.5 7 Betsie Barrier Michigan Michigan 18.3 95.6 3.3 70 7 Upper Platte Reference Michigan Michigan 17.7 61.1 3.6 8 Kewaunee Barrier Wisconsin Michigan 20.0 65.3 4.8 107 8 Ahnapee Reference Wisconsin Michigan 14.1 47.7 3.8 9 East Twin Barrier Wisconsin Michigan 11.3 57.7 4.0 180 9 Hibbards Reference Wisconsin Michigan 6.1 43.6 3.3 10* West Branch,Whitefish Barrier Michigan Michigan 20.4 60.6 5.3 97 10* East Branch, Whitefish Reference Michigan Michigan 17.0 49.8 4.8 11* Miners Barrier Michigan Superior 8.8 72.0 3.8 80 11* Harlow Reference Michigan Superior 5.8 61.6 3.4 12 Big Carp Barrier Ontario Superior 10.1 103.4 3.1 45 12 Little Carp Reference Ontario Superior 5.1 45.6 3.2 13 Stokely Barrier Ontario Superior 8.0 60.2 3.9 52 13 Pancake Reference Ontario Superior 13.1 79.2 4.4 14 Days Barrier Michigan Michigan 9.8 55.8 4.6 55 14 Rapid Reference Michigan Michigan 14.6 43.3 5.6 15 Misery Barrier Michigan Superior 9.6 69.9 3.4 77 15 Firesteel Reference Michigan Superior 14.2 72.2 3.5 16* Middle Barrier Wisconsin Superior 11.7 46.6 5.0 152 16* Poplar Reference Wisconsin Superior 7.4 35.1 5.2 17 Neebing Barrier Ontario Superior 11.8 87.8 3.2 65 17 Whitefish Reference Ontario Superior 15.7 69.1 4.4 18 Clear Barrier Ontario Erie 4.8 49.8 2.4 47 18 South Otter Reference Ontario Erie 2.7 33.9 2.9 19* Forestville Barrier Ontario Erie 3.9 23.4 2.9 35 19* Fishers Reference Ontario Erie 4.4 30.1 4.1 20 Youngs Barrier Ontario Erie 8.6 65.3 3.4 70 20 South Otter Reference Ontario Erie 2.7 33.9 2.9 21 Duffins Barrier Ontario Ontario 12.8 77.3 3.7 75 21 Lynde Reference Ontario Ontario 8.2 30.9 4.1 22 Grafton Barrier Ontario Ontario 4.4 32.8 4.3 45 22 Salem Reference Ontario Ontario 3.2 37.9 3.0 23 Little Salmon Barrier New York Ontario 13.4 54.6 5.6 300 23 Grindstone Reference New York Ontario 11.1 33.8 5.1 24 Shelter Valley Barrier Ontario Ontario 8.9 54.6 3.8 55 24 Wilmot Reference Ontario Ontario 7.5 45.0 4.1

390 Dodd et al. lar to the calculation of an impact value for species richness, an average decline in fish assemblage mean size was estimated (i.e., mean length of all fish at a site) above low-head barriers relative to reference streams and a two-tailed t-test was used to compare differences in mean length of the community due to the barrier. Effects of low-head barriers on assemblage composition was examined using Sφrensen s similarity index (Sφrensen 1948) which is based on presence/absence data, and is computed using the formula: S = 2C / (A + B) (2) where S is similarity of fish assemblages between two sites, A is the number of species in the first site, B is the number of species in the second site, and C is the number of species in common. Tukey s Studentized Range test was used to determine differences in mean similarity values. Species sensitivity to low-head barriers was determined by comparing frequency of occurrence, mean catch, and mean length for above and below sections of barrier and reference streams. RESULTS Habitat Analysis Both barrier and reference streams ranged widely in size (Table 1). Streams with low-head barriers had an average width of 11.0 m (n = 24, se = 0.9) and an average maximum depth of 65.4 cm (n = 24, se = 3.9), while the mean width and maximum depth for reference streams was 9.4 m (n = 23, se = 1.0) and 52.2 cm (n = 23, se = 3.7), respectively. The average difference in width of 1.9 m and maximum depth of 13.9 cm between barrier and reference streams was significantly different from zero (ANOVA, df = 179, P width =0.0236, P depth = 0.0018), indicating barrier streams were wider and deeper on average than reference streams (Table 1). Both barrier and reference streams consisted mainly of gravel with no significant difference in predominant substrate type between stream types (ANOVA, df = 179, P = 0.999). Mean water temperature for barrier streams was 17.5 C (n = 24, se = 0.5) and for reference streams was 18.1 C (n = 23, se = 0.5) with no significant difference between stream types (ANOVA, df = 183, P = 0.9027). To study spatial patterns of habitat alteration by low-head barriers, mean width, maximum depth, particle size, and temperature was calculated for the FIG. 2. Longitudinal trends in mean (± 1 standard error) stream width for barrier and reference streams for all streams and years combined. six sites across reference and barrier streams. Average width and maximum depth gradually increased in a downstream direction for barrier and reference streams, but barrier streams were generally wider and deeper at all sites (Fig. 2, Fig. 3). At sites directly upstream of the barriers, mean maximum depth was on average 15 cm greater than in the reference streams, suggesting that some effect of the impoundment extended upstream to these sites. Unlike width and depth, mean particle size and temperature did not show a longitudinal trend, with FIG. 3. Longitudinal trends in mean (± 1 standard error) maximum depth for barrier and reference streams for all streams and years combined.

Low-head Barrier Effects on Habitat and Fish 391 both habitat characteristics being similar among barrier and reference sites. Fish Community Composition and Size Structure Overall, barrier streams contained a greater number of fish species than reference streams. A total of 14 more species was caught in barrier streams compared to reference streams with six more species occurring in above sections and 12 more species in below sections of barrier streams (Table 2). When comparing differences in average richness between barrier and reference streams, above sections differed by 0.7 species on average. However, the change in average richness between below sections was much greater, differing by an average of 3.8 species. Within a stream type, 20 fewer species were found upstream in barrier streams (4.7 fewer species on average above the barrier) while only 14 fewer species were found upstream in reference streams (1.6 fewer species on average in upstream sections). Although species richness varied across years in individual streams, there was little difference in average species richness when comparing only those seven stream pairs that were sampled in both 1996 and 1997 (Table 3). For these seven stream pairs, species richness averaged 9.4 species in above barrier stream sections in 1996, and averaged 8.9 in 1997. Average richness was 15.1 in sections below barriers in 1996 and 12.0 in 1997. Reference streams likewise showed relatively small changes; richness in above sections was 10.3 in 1996 and 9.1 in 1997, and richness in below sections went from 12.1 in 1996 to 9.9 in 1997. Although the composition of the fish community changed between years in these streams, the overall patterns of richness were remarkably stable, suggesting that these results are relatively robust to the year-to-year variability observed. Species richness was examined at the site level to detect spatial patterns in richness between barrier and reference streams. In reference streams, average richness generally increased in a downstream direction (Fig. 4). Within the barrier streams, average species richness was similar among sites above the barrier. However, barrier streams showed a distinct peak of 10.8 species directly below the barrier (the Below 1 site) that then declined toward the mouth whereas reference streams showed a gradual increase downstream. Comparing the longitudinal patterns between barrier and reference streams, the TABLE 2. Total and mean (in parenthesis, n = sample size) number of species collected in sections above and below actual or hypothetical lowhead lamprey barriers for barrier and reference streams in summer 1996 and 1997 combined. Stream Species richness section Barrier stream Reference stream Above dam 54 48 (11.3, n = 72) (10.6, n = 69) Below dam 74 62 (16.0, n = 65) (12.2, n = 68) Total 79 65 (18.6, n = 137) (14.8, n = 137) upstream sites were more similar in average richness than downstream sites. Because stream width and depth differed between barrier and reference streams, an ANCOVA was used to test if habitat alterations due to barriers explained differences seen in species richness. Results of this analysis indicated that width and depth are significant covariates (df = 173, P width = 0.0046, P depth = 0.0091), but that average species richness was still significantly different between barrier and reference streams (df = 173, P barrier = 0.0334), even with the effects of these covariates removed. Using a similar analysis comparing species richness in above and below sections of barrier and reference streams, significant differences in average richness was found among the four stream positions (df = 43, P stream position = 0.0334) with average richness in the below barrier section being significantly higher than the above barrier and the below reference sections (t-test, df = 43, P ba-bb = 0.0001, P bb-rb = 0.0057). In this analysis, stream width was the only significant covariate (df = 43, P width = 0.0219, P depth = 0.2807). To further examine the effect of low-head barriers on species richness, a decline in species above the barriers (impact value) was calculated for each stream pair. On average, barrier streams lost 4.1 species from below to above segments while reference streams lost only 1.5 species. The overall impact of the barriers on species richness was a net loss of 2.5 species above the barrier relative to reference streams (Table 3). This loss of species was significantly different from the expected value of zero under the null hypothesis of no impact on species richness (t-test, n = 24, P = 0.0126). Be-

392 Dodd et al. TABLE 3. Total number of species caught in sections above and below low-head lamprey barriers in barrier streams and equivalent locations in reference streams, and mean number of species lost (impact value) for summer 1996 and 1997 combined. BA, BB, RA, RB represent species richness in above (A) and below (B) sections of the Barrier (B) and Reference (R) streams. Dates sampled in 1996 and 1997 included in table. Summer 1996 Summer 1997 Mean Stream Dates Barrier Barrier Reference Reference Dates Barrier Barrier Reference Reference impact value pair Sampled above below above below Sampled above below above below (BA-BB)-(RA-RB) 1 12/6 21/6 9 18 18 17 23/6 25/6 10 18 15 12 10.5 2 24/6 26/6 10 13 9 16 17/6 19/6 11 9 5 8 4.5 3 18/7 21/7 14 21 10 9 21/7 24/7 9 9 9 9 4 4 27/7 31/7 10 14 7 12 1 5 23/7 25/7 13 13 7 8 1 6 17/7 18/7 8 21 5 5 13 7 5/8 13/8 14 19 8 14 1 8 9/7 16/7 20 20 11 14 3 9 10/7 16/7 18 27 4 8 5 10 30/7 6/8 9 20 14 14 26/7 12 17 10 16 5 11 6/7 9/7 10 10 8 11 7/7 9/7 4 9 10 9 1.5 12 22/7 28/7 10 9 9 10 2 13 24/7 1/8 5 9 9 5 8 14 2/8 4/8 14 16 12 12 2 15 26/7 28/7 9 10 11 14 2 16 5/8 7/8 8 13 10 13 2/8 10 15 13 9 5.5 17 20/7 24/7 12 16 11 8 7 18 4/7 8/7 3 11 6 12 2 19 2/7 3/7 6 11 3 5 7/7 8/7 6 7 2 6 0 20 5/7 9/7 3 8 6 12 1 21 25/6 28/6 15 15 14 13 1 22 17/6 20/6 8 17 13 12 10 23 7/30 8/1 13 12 19 17 1 24 6/20 6/26 11 13 9 12 1 Mean 10.5 14.8 9.7 11.4 8.9 12.0 9.1 9.9 2.5 Standard Error 0.9 1.0 0.8 0.7 1.1 1.7 1.7 1.2 0.9

Low-head Barrier Effects on Habitat and Fish 393 FIG. 4. Longitudinal trends in mean (± 1 standard error) number of species caught (species richness) in barrier and reference streams for all streams and years combined. cause low-head barriers are constructed to prevent passage of sea lamprey, their loss was expected above the barrier in streams where they were observed below the barrier. When sea lamprey was excluded from the assessment of a barrier impact on decline in species richness, upstream sections of barrier streams still lost 2.4 species on average, a significant loss compared to the expectation under the null hypothesis of no species loss above barriers (t-test, n = 24, P = 0.0139). The effect of possible habitat alteration by low-head barriers on the degree of impact was explored through regressions of mean width and mean maximum depth on loss of species above barriers. These regressions were not significant (ANOVA, df = 22, P width = 0.4194, P depth = 0.7535) and showed substantial scattering of the data, indicating that habitat differences between stream types did not explain the net species decline in individual streams. Low-head barriers selected for this study ranged in age from two to 26 years and in head height (height from water level in the impoundment to water level in the tailrace) from 45 to 300 cm, influencing the size of the impoundment as well as ease of fish passage. The effects of barrier age, head height, and time of last breach on the decline in species upstream were analyzed and found to be poor predictors of species loss above low-head barriers (ANOVA, df = 22, P age = 0.7952, P height = 0.7175, P breach = 0.2938). Not all low-head barriers in this study were constructed specifically for sea lamprey control and two of our low-head sea lamprey barriers (Big Carp River and Albany Creek) were variable crest barriers that serve as barriers only in the spring during sea lamprey migration. When the larger non-sea lamprey barriers and the variable crest barriers were excluded, age, height, and time of last breach still had no significant effect on loss of species. Sφrensen s similarity index based on species presence/absence data was computed to compare fish community composition between upstream and downstream sections of barrier and reference streams. The greatest similarity in species composition occurred between upstream and downstream sections of reference streams with a mean index value of 0.65 (Fig. 5). Above and below sections of barrier streams were found to be the second highest in mean species similarity (0.57). Comparing composition between barrier and reference streams, the below stream sections were slightly more similar in species composition (0.53) than the above sections (0.49), although differences in species richness was greatest between the below sections (Table 2). A Tukey s Studentized Range test performed on mean similarity showed a significant difference only between the highest (within reference stream) and lowest (between above sections) similarity values (df = 92, P = 0.0316). There was no detectable effect of barriers on mean length of the fish community. Differences in mean community size composition between barrier and reference streams were determined by calculation of an impact value for each stream pair. The difference in community size composition between above and below barrier sections was 4.3 mm while reference streams showed a difference of 3.4 mm (Table 4). Overall, the fish community above the barrier was 1.8 mm smaller relative to the reference stream and was not significantly different from the expectation of zero (t-test, N = 24, P = 0.7302), indicating barriers had little or no effect on the size of the fish assemblage upstream. For each species, frequency of occurrence and relative abundance data were examined to give an indication of the sensitivity of individual species to low-head barriers (Appendix 1, Appendix 2). Impact values for relative abundance were calculated to help indicate species that were positively (impact score > 0) or negatively (impact score < 0) affected by barriers, although frequency of occurrence may have remained unaffected for that species. Three species appeared to be negatively affected by lowhead barriers, meaning that a species was seen less frequently or was less abundant upstream of the

394 Dodd et al. FIG. 5. Distribution of Sφrensen s Similarity Index comparing species composition between the four stream positions sampled (BA = Barrier Above, BB = Barrier Below, RA = Reference Above, RB = Reference Below.). low-head barriers compared to their frequency or abundance in the remaining stream positions (Barrier Below, Reference Above, and Reference Below). Sea lamprey, yellow perch (Perca flavescens), and trout-perch (Percopsis omiscomaycus) were not caught above any barrier in the study streams (Appendix 1), but were found frequently in below barrier sites as well as in above and below sites in the reference. Fish species seen more frequently or that were more abundant in above sections of barrier streams compared to the below barrier and reference stream locations included blacknose shiner (Notropis heterolepis) and brook stickleback (Culaea inconstans) (Appendix 1, Appendix 2). DISCUSSION Based on the general habitat characteristics measured, streams with low-head barriers showed relatively little habitat alteration compared to reference streams. Average width and maximum depth were found to be significantly higher in barrier streams, but mean substrate size and temperature were similar between the two stream types. Based on the River Continuum Concept (Vannote et al. 1980), width, depth, and temperature were anticipated to increase and substrate size was expected to decrease in a downstream direction. Both barrier and reference streams follow this general trend of increased width and depth downstream, but sites directly above the impoundment (Above 1 site in barrier streams) are deeper on average compared to those sites in reference streams (Fig. 2, Fig. 3). Although the area immediately upstream of the barriers was excluded from our sampling protocol, sites closest to the barrier may have been within the impacted zone upstream of the small reservoir, explaining the greater average depth at these sites. Barriers slow the flow of water entering an impoundment and often act as sediment traps (Ward and Stanford 1983). From this, sites immediately upstream of the barrier (Above 1 sites) would be expected to have a greater portion of fine substrate particles such as silt and sand and the site directly downstream to have coarser substrate. This was not evident in the data where substrate size was similar at sites above and below the barrier. Surface release dams, such as these low-head sea lamprey barriers, might also be expected to increase temperature directly below the barrier relative to that site in the reference stream (Fraley 1979) if these low-head barriers notably alter stream flow. However, average temperature was not significantly greater in streams with low-head barriers nor were sites directly below the barrier warmer on average than the reference sites. This indicates that, unlike larger surface release dams, these low-head barriers do not retain water long enough to significantly change the

Low-head Barrier Effects on Habitat and Fish 395 TABLE 4. Mean total length (sample size in parentheses) of all fish collected above and below low-head barriers in barrier and equivalent locations in reference streams and the difference in fish length for each stream pair for summer 1996 and 1997 combined. Mean total length (mm) Stream pair Mean impact value number Barrier above Barrier below Reference above Reference below (BA-BB)-(RA-RB) 1 85.0 (304) 87.0 (368) 77.5 (476) 81.4 (380) 1.9 2 60.4 (265) 76.2 (335) 60.4 (108) 64.9 (60) 11.3 3 62.9 (301) 67.3 (188) 72.0 (388) 66.8 (250) 9.5 4 69.9 (72) 67.1 (121) 69.5 (97) 67.5 (102) 0.8 5 73.9 (107) 63.8 (244) 68.6 (130) 54.7 (55) 3.7 6 68.4 (80) 82.9 (99) 68.9 (10) 82.4 (8) 1.0 7 78.4 (192) 92.1 (213) 101.6 (497) 122.1 (518) 6.8 8 91.7 (376) 96.5 (347) 87.8 (276) 132.3 (357) 39.7 9 60.4 (168) 123.8 (311) 50.4 (124) 83.9 (49) 29.9 10 68.4 (375) 79.0 (315) 88.2 (160) 75.3 (222) 23.4 11 78.7 (164) 69.6 (254) 77.7 (125) 100.2 (178) 31.6 12 73.0 (268) 69.8 (23) 80.0 (79) 57.6 (179) 19.2 13 58.5 (248) 73.4 (210) 62.8 (172) 55.8 (76) 21.9 14 72.9 (190) 71.9 (176) 67.4 (210) 69.7 (157) 3.3 15 73.7 (58) 67.0 (77) 59.5 (185) 51.4 (134) 1.4 16 80.0 (215) 87.3 (364) 78.5 (243) 81.3 (394) 4.5 17 65.8 (141) 55.6 (326) 61.4 (129) 58.1 (213) 7.0 18 36.2 (27) 85.9 (26) 65.6 (39) 78.8 (20) 36.4 19 149.8 (98) 81.2 (99) 76.9 (125) 91.2 (225) 82.9 20 80.5 (60) 80.2 (47) 65.6 (39) 78.8 (20) 13.5 21 62.3 (132) 86.4 (177) 57.2 (380) 56.4 (398) 24.8 22 90.9 (163) 72.0 (524) 93.9 (96) 80.5 (80) 5.5 23 61.6 (202) 78.2 (105) 69.6 (205) 73.8 (180) 12.4 24 86.8 (205) 123.5 (94) 62.6 (138) 62.3 (434) 37.0 Mean 74.6 (24) 78.9 (24) 71.9 (24) 75.3 (24) 1.8 Standard Error 4.1 2.8 2.6 4.2 5.3 substrate composition or to noticeably increase the temperature of the stream. Beyond the small impoundment above the barrier and the plunge pool just below, barriers did not appear to substantially affect the physical habitat characteristics measured in the study streams. Since habitat characteristics were not measured in the impoundment and the plunge pool, there may be localized effects on substrate and temperature in these areas which were not detected due to the study protocol. Species richness was higher in both upstream and downstream sections of barrier streams relative to reference streams. One plausible explanation could be that barrier streams, being wider and deeper on average, provided a greater amount of habitat, allowing more species to exist in these streams. However, deeper and wider sites within barrier streams did not consistently have higher average species richness and the sites directly below barriers which contained the largest number of species were not the widest or deepest sites on average (Figs. 2, 3, and 4). Width or depth did explain a portion of the variation seen in species richness, but the longitudinal trends in habitat and mean richness within barrier streams are not as closely linked as they are in reference streams, suggesting a mechanism of impact on the fish assemblage other than habitat differences between stream types. A significant number of species, approximately 2.4 species (excluding sea lamprey), was lost above low-head barriers, implying that these barriers are affecting species richness in these streams. Although barrier streams were significantly wider and deeper than reference streams, differences in habitat did not account for the loss of species above lowhead barriers in this study. Due to the lack of influence of habitat characteristics on the decline in species richness above barriers and the high peak in richness found directly below the barrier, trends seen in species richness within barrier streams can

396 Dodd et al. best be explained by the blocking of fish movement (Porto et al. 1999), resulting in an aggregation of species directly downstream of the structure (Benstead et al. 1999). With the potential for low-head sea lamprey barriers to impede movement of fish species that do not normally exhibit strong jumping ability, the creation of a semi-isolated community above the barriers was anticipated. Fish could emigrate to downstream areas, mimicking mortality to the upstream community, while fish below the barrier are prevented from immigrating to areas above the barrier. Based on studies of larger dams without fish passageways (Erman 1973, Bulow et al. 1988, Winston et al. 1991), a shift in the assemblage composition and size structure was expected. However, no marked shift in fish assemblages in barrier streams was found, as evidenced by the relatively high similarity indices. The similarity in fish assemblage composition suggests that isolating mechanisms had relatively small effects. Further, the fish assemblage composition was remarkably similar across all stream sections, suggesting that the function of the assemblage is not severely impacted by low head barriers in this study. In this study, effects of low-head barriers was examined by evaluating differences between upstream and downstream reaches of streams with barriers relative to reference or control streams. Although the optimal study design to assess effects of lowhead barriers may be to sample both the barrier and reference stream before and after installation of the barrier, this design was not feasible given that public concern over possible effects of these barriers was not raised until after construction. For this study, reference streams were used as indicators of the expected pattern in barrier streams if a barrier had not been constructed on these streams. Using reference streams to gauge shifts in barrier streams was justified due to their close proximity and similarity in size, geography, and geology to barrier streams. Further, both reference and barrier streams in the study had been treated in the recent past with lampricides. Thus, both stream types provide habitat and water quality conditions required for successful sea lamprey reproduction. Another limitation of this study was that only summer months were analyzed, and therefore, the magnitude of low-head barrier impacts on fish assemblages during spring or fall were not assessed (Porto et al. 1999). The more stable summer months were selected to minimize the potential impacts of spawning migrations and of fluctuations in water level that often occur during spring and fall. In spring, low-head barriers would be expected to have less of an impact due to snow melt and spring rains which increase water depths, potentially allowing fish to pass (especially for strong swimmers) over the small barriers in this study. In the fall when water levels are usually at their lowest point, the barriers would become more difficult to traverse as fish would need to jump greater heights over the barrier. Porto et al. (1999), however, found that the impact on fish movement over low-head barriers (as evidenced by mark-recapture data in a subset of the 24 barrier streams examined in this study) was greatest in spring and fall, and that some species were able to traverse the barrier during these seasons. Depending on the amount of precipitation during the winter and early spring, impacts on migratory non-jumping species may vary from year to year with an expectation of low-head barrier effects being more pronounced in dry years. The number of species lost above the barriers in this study was not related to the head height or age of the barrier, and fish composition and size structure were similar between upstream and downstream sections within barrier streams. From this finding, blockage of fish movement is probably not continuous year around, and some fish are able to traverse low-head barriers, possibly during periods of flooding (Helfrich et al. 1999). Thus, low-head barriers may not be a complete obstruction to movement for certain species (Benstead et al. 1999, Porto et al. 1999), allowing for mixing of these populations. It is important to note that most of the barriers in this study were quite small and that this conclusion does not extend to dams larger than examined in this study. In studies of larger dams, upstream movement of most fish species was blocked, resulting in a large loss of species above these barriers and/or a large shift in the fish composition and size structure above barriers relative to downstream sites (Erman 1973, Bulow et al. 1988, Winston et al. 1991). In the analysis of individual species, certain species appeared to be sensitive to low-head barriers. Sea lamprey, yellow perch, and trout-perch were not captured from upstream reaches in any of our barrier streams, suggesting these species are negatively affected by low-head barriers. However, sampling equipment used in this study was not specifically designed for sea lamprey ammocoete collection, and sample sizes for many individual species were too small to allow us to determine the impact of low-head barriers with adequate precision to make generalizations. To adequately assess the

Low-head Barrier Effects on Habitat and Fish 397 degree of impact on specific populations, an extensive mark-recapture study of particular species thought to be negatively affected by low-head barriers will be necessary. CONCLUSIONS There was relatively little effect of low-head barriers on the habitat measurements examined. A decline in number of species seen above barriers was evident, but none of the habitat characteristics measured could explain the trend of high species richness below the barrier or the greater loss of species upstream of the barrier, even though some habitat variables were significant covariates. Therefore, one of the principal mechanisms thought to be impacting species richness is the blockage of upstream fish movement as opposed to habitat alteration. Although low-head barriers had some influence on fish species composition, community size composition was not altered significantly. In this study, indications of low-head barrier effects on non-target species were found. To reduce effects of low-head sea lamprey barriers on nonjumping or non-target species, the Great Lakes Fishery Commission has experimented with installing inflatable variable crest barriers. These barriers are in operation during the sea lamprey spawning run in spring and then deflated during the remainder of the year to allow fish passage. Inflatable variable crest barriers are expected to further reduce the barrier effects observed in this study. When the two variable crest barriers were excluded in this study (Albany Creek and Big Carp River), the results did not change substantially. However, the effects of variable crest barriers warrants further investigation. Although results of this study show low-head barriers have an impact on the fish community, this impact must be weighed against the environmental, social, and monetary costs of using chemicals in these streams to control sea lamprey. This study will aid sea lamprey control agencies and managers in making decisions on the tradeoffs between chemical and mechanical controls and determine the best strategy for controlling sea lamprey in the Great Lakes while maintaining fish diversity in it s tributaries. ACKNOWLEDGMENTS Funding for this project was provided by the Great Lakes Fishery Commission through the Sea Lamprey Barrier Task force. We thank the Ontario Ministry of Natural Resources, Fisheries and Oceans Canada, and the U.S. Fish and Wildlife Service, for assistance in selecting study streams and use of equipment. Thanks to the Michigan and Wisconsin Departments of Natural Resources and the Ontario Ministry of Natural Resources for providing collection permits. E. Holm of the Royal Ontario Museum and T. Coon of Michigan State University helped with fish identification. We also thank the many student interns who were instrumental in collecting field data. REFERENCES Applegate, V.C., and Smith, B.R. 1951. Sea lamprey spawning runs in the Great Lakes, 1950. U.S. Fish Wildl. Serv. Spec. Sci. Rep. Fish. 61., Howell, J.H., Hall, A.E., Jr., and Smith, M.A. 1957. Toxicity of 4,346 chemicals to larval lampreys and fishes. U.S. Fish Wildl. Serv. Spec. Sci. Rep. Fish. 207., Howell, J.H., Moffett, J.W., Johnson, B.G.H., and Smith, M.A. 1961. Use of 3-trifluormethyl-4-nitrophenol as a selective sea lamprey larvicide. Great Lakes Fish. Comm. Tech. Rep. 1. Benstead, J.P., March, J.G., Pringle, C.M., and Scatena, F.N. 1999. Effects of a low-head dam and water abstraction on migratory tropical stream biota. Ecol. Appl. 9:656 668. Bulow, F.J., Webb, M.A., Crumby, W.D., and Quisenberry, S.S. 1988. Effectiveness of a fish barrier dam in limiting movement of rough fishes from a reservoir into a tributary stream. N. Am. J. Fish. Manage. 8:273 275. Dahl, F.H., and McDonald, R.B. 1980. Effects of control of the sea lamprey (Petromyzon marinus) on migratory and resident fish populations. Can. J. Fish. Aquat. Sci. 37:1886 1894. Erkkila, L.F., Smith, B.R., and McLain, A.L. 1956. Sea lamprey control of the Great Lakes 1953 and 1954. U.S. Fish Wild. Serv. Spec. Sci. Rep. Fish. 175. Erman, D.C. 1973. Upstream changes in fish populations following impoundment of Sagehen Creek, California. Trans. Am. Fish. Soc. 102:626 629. Fraley, J.J. 1979. Effects of elevated stream temperatures below a shallow reservoir on a cold water macroinvertebrate fauna. In The ecology of regulated streams, eds. J.V. Ward and J.A. Stanford, pp. 257 272. New York: Plenum Press. Great Lakes Fishery Commission. 1990. Strategic vision of the Great Lakes Fishery Commission for the Decade of the 1990s. Ann Arbor, MI. Helfrich, L.A., Liston, C., Hiebert, S., Albers, M., and Frazer, K. 1999. Influence of low-head diversion dams on fish passage, community composition, and abundance in the Yellowstone River, Montana. Rivers 7:21 32.

398 Dodd et al. Hunn, J.B., and Youngs, W.D. 1980. Role of physical barriers in the control of sea lamprey (Petromyzon marinus). Can. J. Fish. Aquat. Sci. 37:2118 2122. Kelso, J.R.M., and Noltie, D.B. 1990. Abundance of spawning pacific salmon in two Lake Superior streams, 1981 1987. J. Great Lakes Res. 16:209 215. Kondolf, G.M., and Li, S. 1992. The pebble count technique for quantifying surface bed material size in instream flow studies. Rivers 3:80 87. Lawrie, A.H. 1970. The sea lamprey in the Great Lakes. Trans. Am. Fish. Soc. 99:766 774. Littell, R.C., Milliken, G.A., Stroup, W.W., and Wolfinger, R.D. 1996. SAS System for Mixed Models. SAS Institute Inc., Cary, NC. McLain, A.L. 1957. The control of the upstream movement of fish with pulsated direct-current. Trans. Am. Fish. Soc. 86:269 284. Pearce, W.A., Braem, R.A., Dustin, S.M., and Tibbles, J.J. 1980. Sea lamprey (Petromyzon marinus) in the lower Great Lakes. Can. J. Fish. Aquat. Sci. 37:1802 1810. Porto, L.M., McLaughlin, R.L., and Noakes, D.L.G. 1999. Low-head barrier dams restrict the movement of fishes in two Lake Ontario streams. N. Am. J. Fish. Manage. 19:1028 1036. Pringle, C.M. 1997. Exploring how disturbance is transmitted upstream: going against the flow. J. North Am. Benthol. Soc. 16:425 438. Robins, C.R., Bailey, R.M., Bond, C.E., Brooker, J.R., Lachner, E.A., Lea, R.N., and Scott, W.B. 1991. Common and scientific names of fishes from the United States and Canada, Fifth edition. Am. Fish. Soc. Spec. Publ. No. 20. Betheseda, MD. Simonson, T.D., and Lyons, J. 1995. Comparisons of catch per effort and removal procedures for sampling stream fish assemblages. N. Am. J. Fish. Manage. 15:419 427. Smith, B.R., and Tibbles, J.J. 1980. Sea lamprey (Petromyzon marinus) in Lakes Huron, Michigan, and Superior: history of invasion and control, 1936 78. Can. J. Fish. Aquat. Sci. 37:1780 1801. Sφrensen, T. 1948. A method of establishing groups of equal amplitude in plant sociology based on similarity of species content. K. Dan. Vidensk. Selsk. Biol. Skr. 5:1 34. Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., and Cushing, C.E. 1980. The River Continuum Concept. Can. J. Fish. Aquat. Sci. 37:130 137. Ward, J.V., and Stanford, J.A. 1983. The serial discontinuity concept of lotic ecosystems. In Dynamics of lotic ecosystems, eds. T.D. Fontaine, III and S.M. Bartell, pp. 29 42. Ann Arbor, MI: Ann Arbor Science Publishers. Winston, M.R., Taylor, C.M., and Pigg, J. 1991. Upstream extirpation of four minnow species due to damming of a prairie stream. Trans. Am. Fish. Soc. 120:98 105. Submitted: 21 December 2000 Accepted: 27 June 2002 Editorial handling: William D. Swink APPENDIX 1. Number of streams in which each species was caught above and below a low-head barrier in barrier streams or an equivalent location in the reference streams. (BA refers to Barrier Above stream sections, BB = Barrier Below, RA = Reference Above, RB = Reference Below). Scientific names follow Robins et al. 1991. Number of streams Common name Scientific name BA BB RA RB American brook lamprey Lampetra appendix 6 6 1 1 American eel Anguilla rostrata 0 1 0 0 Atlantic salmon Salmo salar 0 0 0 1 Black bullhead Ameiurus melas 2 3 1 2 Black crappie Poxomis nigromaculatus 1 1 0 1 Blackchin shiner Notropis heterodon 1 1 0 2 Blacknose dace Rhinichthys atratulus 20 21 14 15 Blacknose shiner Notropis heterolepis 5 2 2 4 Blackside darter Percina maculata 3 5 2 4 Bluegill Lepomis macrochirus 1 1 3 3 Bluntnose minnow Pimephales notatus 4 9 4 3 Bowfin Amia calva 0 1 0 1 Brassy minnow Hybognathus hankinsoni 4 4 1 3 Brook stickleback Culaea inconstans 13 7 9 7 Brook trout Salvelinus fontinalis 8 8 7 5 (Continued)

APPENDIX 1. Continued. Low-head Barrier Effects on Habitat and Fish 399 Number of streams Common name Scientific name BA BB RA RB Brown bullhead Ameiurus nebulosus 3 1 0 1 Brown trout Salmo trutta 6 5 6 5 Burbot Lota lota 0 5 0 3 Central mudminnow Umbra limi 14 9 15 13 Channel catfish Ictalurus punctatus 0 1 0 0 Chestnut lamprey Ichthyomyzon castaneus 1 0 0 0 Chinook salmon Oncorhynchus tshawytscha 0 0 1 2 Coho salmon Oncorhynchus kisutch 2 2 2 2 Common carp Cyprinus carpio 2 3 2 3 Common shiner Luxilus cornutus 11 14 9 10 Creek chub Semotilus atromaculatus 19 20 17 15 Cutlips minnow Exoglossum maxillingua 1 1 1 1 Emerald shiner Notropis atherinoides 0 2 0 0 Fallfish Semotilus corporalis 1 1 1 1 Fantail darter Etheostoma flabellare 2 4 3 3 Fathead minnow Pimephales promelas 3 8 2 2 Finescale dace Phoxinus neogaeus 1 2 2 0 Flathead catfish Pylodictis olivaris 0 1 0 0 Golden redhorse Moxostoma erythrurum 0 1 0 1 Golden shiner Notemigonus crysoleucus 0 2 0 1 Grass pickerel Esox americanus vermiculatus 0 0 0 1 Greater redhorse Moxostoma valenciennesi 1 0 0 1 Green sunfish Lepomis cyanellus 1 2 0 0 Hornyhead chub Nocomis biguttatus 4 5 5 5 Iowa darter Etheostoma exile 1 1 1 1 Johnny darter Etheostoma nigrum 15 19 14 15 Lake chub Couesius plumbeus 1 1 0 0 Lake trout Salvelinus namaycush 0 1 0 0 Largemouth bass Micropterus salmoides 1 4 1 2 Largescale stoneroller Campostoma oligolepis 0 2 0 0 Logperch Percina caprodes 3 15 7 10 Longnose dace Rhinichthys cataractae 13 19 16 20 Mimic shiner Notropis volucellus 0 2 0 0 Mottled sculpin Cottus bairdi 17 19 17 18 Ninespine stickleback Pungitius pungitius 0 1 0 1 Northern brook lamprey Ichthyomyzon fossor 2 1 0 0 Northern hog sucker Hypentelium nigricans 2 2 3 2 Northern pike Esox lucius 3 6 4 2 Northern redbelly dace Phoxinus eos 6 8 2 1 Pearl dace Margariscus margarita 4 4 1 2 Pugnose minnow Opsopoeodus emiliae 0 1 0 1 Pumpkinseed Lepomis gibbosus 5 11 6 5 Rainbow darter Etheostoma caeruleum 2 2 2 2 Rainbow trout Oncorhynchus mykiss 17 16 17 17 Red shiner Cyprinella lutrensis 1 0 0 0 Redside dace Clinostomus elongatus 0 1 0 0 River chub Nocomis micropogon 0 0 1 0 River darter Percina shumardi 0 1 0 0 Rock bass Ambloplites rupestris 9 19 10 15 Rosyface shiner Notropis rubellus 2 2 2 1 Ruffe Gymnocephalus cernuus 0 2 0 0 Sand shiner Notropis stramineus 0 2 0 0 Sauger Stizostedion canadense 0 1 0 1