publications > reports > inventory of freshwater fishes of the big cypress national preserve > methods
We conducted the inventory using a stratified sampling design based on habitat type. To simplify access, most sites were located within 250 m of roads or trails passable by truck. Sites were also reached by boat, all-terrain vehicle, helicopter, and airboat when these means of transportation were available. Habitat heterogeneity made random selection of sites difficult, particularly during the dry season, because no spatially explicit hydrologic data were available to help us determine whether our randomly pre-selected sites were inundated. In such cases, we selected sampling sites arbitrarily to insure the presence of standing surface water.
The habitat type classification scheme we used is based on the seasonally inundated habitats identified by Duever et al. (1986). Eight broad categories based on dominant vegetation and hydroperiod were used. Six of these are subject to seasonal dry-down and are ranked by hydroperiod as follows:
Two additional habitats hold water year-round:
The diversity of habitat types in BCNP presented a considerable challenge to the development of a comprehensive sampling regime, because the effectiveness of any given methodology varied among habitats. To compensate for this, we used numerous techniques during the inventory portion of this study. These methods included a variety of fish traps, as well as electrofishing gear, gill nets, cast nets, dip nets, and angling.
Traps: Traps provide a means of sampling with a standardized unit of effort and are suitable for use in virtually all BCNP habitats. They are also relatively portable and, therefore, suitable for work in remote locations. They have the disadvantage of inherent selection biases, based both on trap construction, and the size and behavior of targeted species. To attempt to minimize these biases, we deployed a variety of small-fish traps simultaneously. These included Gee-type minnow traps, box traps, collapsible mesh traps, and Breder traps (Table 1). Soak times were generally 24 hours, although 1-hour sets were also performed. Small-fish traps were consistently fished unbaited and relied on passive encounters to generate captures.
Hoop nets were used to sample larger fishes in deeper water. They were fished unbaited, with or without leads, but were occasionally baited with cheese to selectively target catfish species that proved difficult to capture otherwise. The hoop nets used here were 1.4 m in overall length and were constructed from four 50-cm diameter fiberglass hoops and tar-coated twine with a 2.5-cm mesh size. The nets had two throats and an approximately 15-cm-diameter aperture. Typically, hoop nets were deployed for 24-hour intervals.
Electrofishing: Electrofishing was conducted in locations where habitat composition permitted. Two electrofishing setups were used; the first consisted of a boat-mounted Smith-Root Type-6A electrofisher with a maximum current output of either 1,008 volts DC at 120 pulses per second, or 720 volts AC at 60 hertz. The electrofisher was used extensively for sampling in canals but was too large to operate in other habitats. Effort was generally standardized by sampling 100-m transects, although opportunistic sampling around structures such as bridge pilings was conducted as well.
Forested habitats and marshes were sampled using a second setup, which consisted of a small barge carrying a Smith-Root Type-2.5 GPP electrofisher with a maximum current of 1,000 volts at either 120 pulses per second DC, or at 60 hertz AC. The barge drew only several centimeters of water and was a meter wide, but it was still too large to be used in heavily vegetated habitats. Samples were standardized to 300 seconds of total shock time.
Nets: We used experimental gill nets to sample fishes in canals. Two nets were fished in tandem, and each was composed of four 242-cm deep x 180-cm wide panels. The first net had mesh sizes of 1.2, 2.5, 3.7, and 5 cm; the second had mesh sizes of 6.2, 7.6, 8.8, and 11 cm. Nets were typically set from 1 to 4 hours. Reptile predation on entrapped fishes was a problem when using these nets, and encounters with alligators were particularly damaging, precluding longer sets.
Although cast nets do not provide quantitative data, we used them extensively in an opportunistic fashion to capture species sighted in canals. The cast nets had a radius of 180 cm, with a 1.2-cm mesh. In dense vegetation, we used dip nets with fine mesh (<1 mm) for collecting juvenile fishes and small-bodied species, such as the least killifish (Heterandria formosa) and Everglades pygmy sunfish (Elassoma evergladei).
Other: We conducted opportunistic sampling using light tackle, hook-and-line fishing with a variety of lures and live or dead baits. Lines of baited hooks were also occasionally deployed in canals in attempts to catch catfish, although hoop nets proved much more effective. Finally, we recorded those species observed and positively identified in the field when we could not capture the specimens.
We recorded the location of each sampling site in Universal Transverse Mercator (UTM) coordinates using a Garmin Etrex Vista GPS unit. For each sample, we identified all specimens to species, and recorded the total catch per species. We measured the total lengths of the first 20 randomly selected individuals of each species to obtain a representative size distribution. Water temperature, pH, salinity, and dissolved oxygen were measured at each site whenever possible; however, instrumentation problems prevented the collection of these data during much of the year. We subsequently borrowed a Hydrolab 4a minisonde and datalogger from the BCNP hydrology department. For electrofishing expeditions, water conductivity was determined using a YSI-33 conductivity meter.
All field data were recorded on paper datasheets or, during poor weather, in waterproof-paper notebooks. Datasheets were transcribed into an MS Access database, with linked tables for site location, physical parameters, and catch information. Each sample was given a unique numeric identifier to allow the automated generation of reports for each sampling expedition. We proofed geographical information by exporting Access data to Arcview GIS software and comparing the map location of a sample with its known physical location. Once the field data were entered, we compared Access site reports to field notes to ensure quality control.
Community-Dynamics Research Methods
Three identical sampling plots were constructed within each of the three sampling regions. All plots were located in forested wetland areas adjacent to long-hydroperiod ponds or sloughs. A pair of 4-m2 drop traps was constructed within each plot to sample short- and long-hydroperiod areas. The drop-trap design is analogous to that described by Lorenz et al. (1997); however, the high density of tree trunks in the sampling area required that we reduce the area of the trap from 9 m2 to 4 m2.
We also constructed a single drift-fence minnow-trap array at each plot. Each array consisted of four 15-m wings built with construction-debris fencing in an X pattern aligned so that the apex of each faced a cardinal direction. The wings direct animals into the center where four Gee-8 type minnow traps constructed of 1/8-in. (3.2 mm) steel mesh were positioned. This design has been used extensively in ENP and the Water Conservation Areas, and can provide data that are directly comparable to those regions (Loftus et al., 2001). We planned to sample large fishes in deep-water areas at each plot using either electrofishing or gill nets, depending on which of these methods exhibited greater efficiency after the first year of data collection. Small fish sampling was conducted as described in the next section.
We originally planned to sample five times per year (June/July, October, December, February, and April/May) to detect changes across the annual hydrologic cycle. However, only the June/July 2004 sampling was conducted before funding for this project was redirected to another study at the end of FY 04.
We conducted drop-trap sampling over a three-day period. On the first day, nets were transported to the sampling site and attached to the frames installed permanently at the site. The trap frames were then secured at the top of the supports and rigged to cable releases positioned several meters away. A Gee-type minnow trap was deployed in proximity to each drop trap to collect fish for efficiency measurements.
Drop traps were triggered on the second day of sampling when we arrived at the plot. Fishes from the associated minnow trap were counted and marked by fin clipping, then released into each trap, but efficiency was measured only if 10 or more individuals were caught in the minnow trap. Once this was completed for all traps within the plot, powdered rotenone was added to each trap at a dosage sufficient to kill the fish in the minnow traps. Traps were treated with rotenone in the same order they were triggered to attempt to normalize the sampling time for individuals inside. Physical parameters, including water temperature, conductivity, dissolved oxygen, and pH, were recorded from areas near the traps using a Hydrolab datasonde. After allowing enough time for the toxicant to act, we collected all fishes from each trap and preserved them for lab processing. Potassium permanganate was added to oxidize the rotenone to avoid incidental mortality outside of the traps. On the third day of sampling, fishes that had floated to the surface overnight were collected from each trap and preserved. The nets were then removed from the drop-trap frames and returned to the lab. Minnow-trap arrays were sampled during the first two days of this work. Minnow traps were placed at the center of each array on the first day and emptied 24 hours later. All fishes collected were recorded and returned to the lab for processing.
Statistical Analysis - Inventory
The extensive use of small-fish traps during this study provided a large amount of data for analysis. Although the traps used in this study have inherent selective biases, samples collected using identical methodology should be comparable to one another in this regard and permit an examination of variability among habitat types.
A total of 180 sampling expeditions were performed during this study using 24-hour, small-fish trap sets. For analytical purposes, the wet season was defined as June 1-November 30, and the dry season December 1 - May 31. Although this is a generalization of conditions that vary annually, it provides a uniform basis for analysis. Samples are not evenly distributed temporally because habitats were sampled when conditions were most favorable. For example, it was only possible to collect a few herbaceous prairie samples during the dry season because the habitat dries very soon after the rains cease. Although the uneven temporal distribution limited the statistical rigor of our analyses, acquiring information on seasonality was secondary to obtaining a complete inventory of BCNP fishes during this project.
Data for each sampling expedition were pooled to determine per-trap catch per unit effort (CPUE) values for each species at each site. No differentiation was made among different traps because all were used routinely for the trap sets. The correction to CPUE from absolute abundances was made to account for differences in the number of replicates of the various traps used, which varied for a number of logistical reasons. For each sample, the total number of species (S), total CPUE, and Shannon-Weiner Diversity were calculated.
U.S. Department of the Interior, U.S. Geological Survey
This page is: http://sofia.usgs.gov/publications/reports/fish_bcnp/methods.html
Comments and suggestions? Contact: Heather Henkel - Webmaster
Last updated: 04 September, 2013 @ 02:04 PM(TJE)