North American Journal of Fisheries Management 11:167-176. 1991 © Copyright by the American Fisheries Society 1991 Stepped-Oblique Midwater Trawling as an Assessment Technique for Rainbow Smelt RICHARD A. KIRN* AND GEORGE W. LABAR School of Natural Resources, University of Vermont Burlington, Vermont 05405, USA Abstract.—Discrete-depth midwater trawling for rainbow smelt Osmerus mordax at night in Lake Champlain revealed a delayed vertical migration of smaller fish and size partitioning vertically within the water column. These behavioral observations led to the development and evaluation of a nighttime stepped-oblique midwater trawling technique. This latter method proved effective in obtaining large catches of rainbow smelt (mean catch per unit of effort [CPUE], 215-655 fish/ 55-min trawl) and provided estimates of relative abundance with reasonable precision (CPUE ± 13-50% from June through October), as well as estimates of length and age distributions unbiased by vertical migration behavior. Rainbow smelt Osmerus mordax serve as the vide information on the sexually mature portion principal forage species for stocked salmonids in of the population, limiting the value of such sur- Lake Champlain (Kirn 1986) and many other New veys for monitoring year-class dynamics. In ad- England waters (Kircheis and Stanley 1981). The dition, this approach is not feasible for waters with desirability of rainbow smelt as a food and forage significant lake-spawning populations of rainbow fish, and the relative ease with which they are smelt, such as are found in Lake Champlain (Plo- transferred to other waters, is resulting in a con- sila 1984). In recent years, hydroacoustic sampling tinually expanding distribution of this species has proven to be a reliable technique for moni- (Maydenetal. 1987). toring the abundance of pelagic fish stocks (Argyle Knowledge of forage-fish population dynamics 1982; Heist and Swenson 1983), but requires ex- is an important component of recent freshwater pensive equipment and must be integrated with salmonid introduction and restoration programs, other sampling gear to determine species com- especially in situations such as in Lake Champlain, position and obtain biological samples. where recruitment of salmonids relies almost ex- Daytime bottom and midwater trawling in Lake clusively upon hatchery production and is there- Huron indicated that discrete portions of the rain- fore independent of forage abundance. The ability bow smelt populations inhabited both the bottom to recognize fluctuations in rainbow smelt abun- and midwater during the day, and were therefore dance before predator growth is affected is partic- unavailable to a single trawling method (Argyle ularly critical given the potential for extreme vari- 1982). Observations of size partitioning among ability in the abundance of rainbow smelt, which daytime rainbow smelt distributions have been apparently is unrelated to stock size, predation, associated with bottom depth and possibly water fishing intensity, or disease (Smith 1972; Havey temperature (MacCallum and Regier 1970; Argyle 1973; Kircheis and Stanley 1981; Selgeby 1985). 1982; Plosila 1982). Length or age compositions Sampling rainbow smelt populations has been estimated under these conditions become a func- complicated by size- or age-specific seasonal dis- tion of trawling location and effort rather than a tributions (MacCallum and Regier 1970; Argyle random sample of the population at large (Argyle 1982; Plosila 1982) and diel vertical migrations 1982). (Ferguson 1965; Gray 1979; Heist and Swenson Midwater trawling at night, when diel vertical 1983). Population dynamics of Great Lakes rain- migration of rainbow smelt provides a concen- bow smelt have been assessed from spring spawn- trated, pelagic distribution, has shown promise as ing runs, during which concentrations of rainbow an effective sampling strategy. Rainbow smelt con- smelt in tributaries are vulnerable to a variety of stituted 90-99% of the nighttime pelagic fish pop- sampling gears (Luey and Adelman 1984; Frie and ulation sampled with midwater trawls in the upper Spangler 1985). These surveys, however, only pro- 50 m of Lake Superior (Heist and Swenson 1983). Nighttime midwater-trawl sampling for cisco Coregonus artedii in Lake Champlain has also 1 Present address: Vermont Fish and Wildlife De- produced large catches of rainbow smelt (G.W.L., partment, Roxbury, Vermont 05669, USA. unpublished data). 167 168 KIRN AND LABAR Considering the influences of rainbow smelt be- graduated and ranged from 20.3 cm at the head- havior on past sampling strategies, the specific ob- rope to 1.3 cm at the cod-end liner. Although this jectives of this study were to develop and evaluate net was capable of retaining some young-of-the- a nighttime midwater-trawling technique for sam- year rainbow smelt, many escaped. For this rea- pling rainbow smelt populations. Results from son, young-of-the-year fish, distinguished by size preliminary discrete-depth trawls, also presented, and transparency, were not included in data anal- led to the development of a stepped-oblique mid- yses. water-trawling technique. This technique was The net was towed at 1.1 m/s by the Vermont evaluated for consistency in estimating length and Fish and Wildlife Department's 9.4-m research age distributions and relative abundance of rain- vessel Dore. Temperature and net depth were bow smelt. The influence of the temporal and spa- monitored with an ultrasonic temperature-pres- tial movements of rainbow smelt were also eval- sure transducer attached to the port bridle near uated and used to develop an overall sampling the headrope. strategy for monitoring the population. Discrete-depth trawls.—Discrete-depth trawls were performed during June, August, and October Study Area 1984, and during June and August 1985. Prelim- Lake Champlain is a typically dimictic lake lo- inary trawls in June 1984 sampled sites in Malletts cated within a north-south, glaciated fault be- Bay and the Northeast Arm (Figure 1). Subsequent tween the Adirondack Mountains of New York trawls were confined to the Main Lake. and the Green Mountains of Vermont. Meyer and Discrete-depth trawls consisted of a series of 20- Gruendling (1979) provided a detailed account of min trawls at each of three predetermined depths: the limnology of Lake Champlain. Lake dimen- midway between the surface and the top of the sions are length, 172 km; maximum width, 19 km; metalimnion, at the top of the metalimnion, and maximum depth, 129 m; mean depth, 22 m; sur- at the bottom of the metalimnion. Because of a face area, 1,140 km2. Plosila and Anderson (1985) paucity of rainbow smelt (4% of the total catch in estimated 45% of the lake to possess deep, cold June), trawls midway between the top of the meta- water with sufficient dissolved oxygen to support limnion and the surface were replaced by trawls salmonids. 10 m below the metalimnion after June 1984. In Potash et al. (1969) described five limnologically the absence of thermal stratification, depths of 15, distinct water masses in Lake Champlain: South 25, and 35 m were sampled. All three depths were Lake, Main Lake, Malletts Bay, Northeast Arm, sampled before a second series of trawls was re- and Missisquoi Bay (Figure 1). New York and peated in sequence. Trawls conducted in June 1984 Vermont state fishery investigators have further were begun within 10 min after sunset. Trawls partitioned the lake into 15 management zones. conducted after June 1984 were generally initiated Sampling during this study concentrated on zones within 90 min after sunset. Two series of discrete- 3 A and 3B of the Main Lake (Figure 1), where the depth trawls were conducted per night except in highest densities of lake trout have been reported August 1984, when three series were conducted at (Plosila and Anderson 1985). The Main Lake ac- two of the four sites sampled, and in October 1984, counts for 82% of the entire volume and 60% of when three series were conducted at 25 and 35m. the total surface area of Lake Champlain (Potash Stepped-oblique trawls.— Stepped-oblique trawls et al. 1969). Maximum surface temperatures are were deployed at least 10 m above the lake bottom, reported to reach about 23°C, while summer hy- or to a maximum depth of 35 m. Once set, the polimnetic temperatures average 6°C (Meyer and net was fished for 5 min at each 3.3-m depth stra- Gruendling 1979). Summer stratification of the tum, in stairstep fashion, until the net reached a Main Lake is reported to generally begin in June depth of about 7 m, when it was completely re- (Meyer and Gruendling 1979). Thermal stratifi- trieved. During preliminary discrete-depth trawls cation was only observed in this study during Au- in this study and earlier work (G.W.L., unpub- gust trawl sampling, when the metalimnion had lished data) in Lake Champlain, few rainbow smelt an upper boundary at 10-12 m and a lower bound- were captured in the upper 7 m of the water col- ary ranging from 17 to 25 m. umn. Although the actual net dimensions during trawling were unknown, it was assumed that the Methods 3.3-m depth strata were sufficient to ensure that Rainbow smelt were sampled with a 5-m x 5-m the net sampled the entire water column. A rep- midwater trawl. Mesh size (stretched measure) was licate trawl was then fished along the same course MIDWATER TRAWLING FOR RAINBOW SMELT 169 FIGURE 1.—Map of Lake Champlain (left) showing water masses and fishery management zones 3 A and 3B. In the large-scale map of zone 3 A and 3B (right), the lines connecting triangles, squares, and circles indicate shallow (<45-m), intermediate (45-75-m), and deep (>75-m) trawl sites, respectively. in the opposite direction. Based on results from sured to the nearest millimeter (total length). Ad- the discrete-depth trawls, stepped-oblique trawl- ditional fish were counted.