Upper Midwest Environmental Sciences Center
Methods
Sampling Methods
The Long Term Resource Monitoring Program's (LTRMP) fish monitoring design and sampling protocols, including historical changes, are given in Gutreuter et al. (1995). Readers requiring detailed descriptions should refer to that report. An abbreviated description of the LTRMP design and protocols follows; a list of common and scientific names of fish used in this report is found in Table 1. As water levels are often suspected of affecting fish populations and community stratum, hydrographs are provided for each study area and each year sampled.
We summarize the annual increment of fish data obtained by the LTRMP from stratified random and fixed-site sampling by year. The LTRMP converted to a stratified random fish sampling design in 1993, augmented with limited sampling at a few permanently fixed sites. Selected aquatic areas, chosen for their enduring geomorphic features (Wilcox 1993), were used as sampling strata. Each aquatic area is artificially partitioned into 50-m2 sampling grids beginning with a random origin for each LTRMP study reach (Gutreuter et al. 1995) using a geographic information system. Beginning in 1993, sampling sites were randomly chosen from this lattice of square grids. Whenever it is discovered that a randomly selected site cannot be sampled because of environmental constraints (e.g., limited physical access or high flow), the nearest accessible site from a list of randomly selected alternate sites is sampled within the same aquatic area class.
After an evaluation of the fish sampling gear deployment scheme (Ickes and Burkhardt 2002), the LTRMP eliminated the use of night electrofishing, seining in all strata, and offshore netting in impounded and backwater strata. Major changes for the sampling design are summarized in the LTRMP Fish Component Sampling History section. The following is a summary of the sampling gears according to Gutreuter et al. (1995):
Electrofishing
Electrofishing is conducted with pulsed direct current; boat configuration and power output are standardized (Burkhardt and Gutreuter 1995; Gutreuter et al. 1995). Electrofishing effort is of 15-min duration and is paced so that the boat covers a rectangle of about 200 × 30 m. Day and night electrofishing data from these two methods were combined for length analysis. The unit of effort is a 15-min run.
Fyke Net
The LTRMP uses Wisconsin-type fyke nets (trap nets) that contain three sections: the lead, frame, and cab. All netting is 1.8-cm mesh (bar measure). Leads are 15 m long and 1.3 m high. The spring steel frames are 0.9 m high and 1.8 m wide with two internal wing throats. The cabs are constructed of six steel hoops (0.9 m in diameter) containing two throats. These nets are fished singly from shoreline or from beds of dense vegetation or in tandem (with leads connected) offshore. The unit of effort is a net-day, where each frame is one net. Fyke and tandem fyke netting data were combined for length distribution analysis.
Mini Fyke Net
Mini fyke nets are small, Wisconsin-type fyke nets. Mesh size is 3-mm Ace-type nylon. The leads are 4.5 m long and 0.6 m high. The spring steel frames are 0.6 m high and 1.2 m wide with two internal wing throats. The cabs are constructed of two steel hoops (0.6 m in diameter) with one throat. These nets are fished singly from shoreline or from beds of dense vegetation or in tandem (with leads connected) offshore. The unit of effort is a net-day, where each frame is one net.
Hoop Net
The LTRMP uses two sizes of hoop nets. The large nets are composed of seven fiberglass hoops with diameters of 1.1–1.2 m. These nets are 4.8 m long, contain two finger-style throats, and are constructed of 3.7-cm nylon mesh (bar measure). The small nets are composed of seven fiberglass hoops with diameters of 0.5 to 0.6 m. The small nets are 3 m long, contain two finger-style throats, and are constructed of 1.8-cm nylon mesh (bar measure). Hoop nets are deployed separately but in pairs within sampling sites. Both nets are baited with 3 kg of soybean cake. Because of gear inefficiency, hoop net sets in BWCO areas were optional during 1999. For this report, the estimates from pairs of nets are pooled and, therefore, treated as a single gear for consistency with the 1990–92 data. The unit of effort is a net-day, which is 24 h of effort by a pair of nets.
Seine
The LTRMP uses 10.7-m-long seines constructed of 3-mm Ace-type nylon mesh. These seines are 1.8 m high and have a 0.9-m2 bag in the centers. Seines are extended perpendicularly to shorelines and then swept in a 90 arc downstream to the shoreline.
Gill Net
In 1993, gill nets became an optional experimental sampling gear. This option was included to improve monitoring capabilities for some large riverine species. Gill nets are 91.44 m long and consist of four, 22.86-m panels of monofilament mesh. The panels are 2.44 m deep. Each panel consists of different mesh of 10.2-, 20.3-, and 25.4-cm stretch measure. The 10.2- and 15.2-cm mesh are woven from No. 8 (9.07-kg [20-lb] test) transparent nylon monofilament. The 25.4-cm mesh is woven from No. 12 (13.61-kg [30-lb] test) transparent nylon monofilament. The top line is floating foam-core rope and the bottom line is 29.5-kg lead-core rope. Gill nets are set either perpendicularly (preferred) or parallel (in high-flow conditions) to the shoreline. The standard unit of gill netting effort is the net-day, where a day is 24 h.
Anchored Trammel Net
In 1994, anchored trammel nets became an optional experimental sampling gear. This option was included to improve monitoring capabilities for some large riverine species. Trammel nets may be anchored or drifted with the current.
Trammel nets are 91.44 × 2.44 m, inside netting is 10.16-cm bar of No. 8 monofilament hung about 85 m per 30.48 m of finished net. The net wall size is 35.56-cm bar of No. 9 multifilament twine hung 61 m per 30.48 yards of finished net. The net float line is 1.27-cm foam-core (two strands on the floating nets, one strand on the bottom set nets), and the lead line is lead-core (No. 20 on the floating net, No. 65 on the sinking net).
Bottom Trawl
Bottom trawl is conducted only at permanently fixed-site sampling locations in tailwater zones and unstructured channel borders. The LTRMP trawls collect mainly small, bottom-dwelling fish. The trawls are two-seam, 4.8-m slingshot balloon trawls (TRL16BC, Memphis Net and Twine Co., Inc., or the equivalent). The body of the trawl is made of No. 9 nylon with stretch mesh 18 mm in diameter. The cod end is made of No. 18 nylon with stretch mesh 18 mm in diameter. The cod end contains a 1.8-m liner consisting of 3 mm Ace-type nylon mesh. Floats are spaced every 0.91 m along the headrope, and a 4.8-mm steel chain is tied to the footrope. The trawl is equipped with 37-cm-high by 75-cm-long iron "V" doors (otter boards). These trawls are dragged downriver by small, flat-bottomed boats. Trawl speed is barely faster than ambient current speed. The standard unit of trawling effort is a haul. A minimum of six hauls are collected in main or side channel sites and four hauls at tailwater sites.
Statistical Methods
The LTRMP uses mean catch-per-unit-effort (C/f) as an index of abundance, as is conventional practice (Ricker 1975). The units of effort are specific to particular gears. For electrofishing and seining, effort is a constant, but for other gears it is somewhat variable. For example, although the effort goal for fyke netting is 1 day (Gutreuter et al. 1995), actual effort may vary between 20 and 30 h. Catch and effort are recorded for each species from individual samples (deployments of particular gears at unique combinations of time and place. Whenever a species is not caught in a sample, the catch for that species is zero. Although these zero catches are not recorded, they are reconstructed for analyses.
The estimates of pooled reachwide mean C/f were obtained from
the conventional design-based estimator for stratified random samples
(Cochran 1977). For an arbitrary random variable denoted y (for
this report y represents C/f), the pooled mean, denoted
(st
represents stratified) is given by
(1)
where Nh is the number of sampling units within stratum
h, N = 
Nh,
and 
denotes the estimator of the simple mean of y for stratum h.
The estimator of the variance of
is
(2)
where
is the usual estimator of the variance of yh and nh
is the number of samples taken in stratum h (Cochran 1977). The
standard error of
is therefore
.
For LTRMP fish monitoring, the sampling units are 50-m2 sampling
grids.
In this report, C/f statistics are reported separately for the limited, fixed-site sampling and the primary stratified random sampling. Equation (1) is used to estimate means of data obtained from fixed-site sampling to maintain computational consistency. The pooled means from fixed-site sampling are not guaranteed unbiased because there is no assurance that the fixed sites were unbiased within the stratum. Equation (1) is also used to obtain estimates of overall mean C/f from stratified random sampling. In random samples, equation (1) yields unbiased estimates of the pooled means regardless of the probability distribution of y (Cochran 1977).
Length distribution analysis was performed for 13 selected fish species
(gear used): gizzard shad (electrofishing), common carp (electrofishing),
smallmouth buffalo (electrofishing; small and large hoop netting), channel
catfish (electrofishing; small and large hoop netting), northern pike
(electrofishing; fyke and tandem fyke netting), white bass (electrofishing),
bluegill (electrofishing; fyke and tandem fyke netting), largemouth bass
(electrofishing), white crappie (fyke and tandem fyke netting), black
crappie (fyke and tandem fyke netting), sauger (electrofishing), walleye
(electrofishing), and freshwater drum (electrofishing; fyke and tandem
fyke netting). Night electrofishing was eliminated in 2002 and, therefore,
total catch may be lower for length distributions in years after 2001.
The length data are illustrated in the form of histograms. Because data
within a single sampling season are taken over a long time and size ranges
for certain species of fish can overlap (e.g., a 6-cm-long bluegill collected
early in period 1 is not of the same cohort as a 6-cm-long bluegill collected
late in period 3), interpretations in the length distributions should
be made cautiously. In some instances, meaningful biological interpretation
of these distributions may be limited by small sample size or size selectivity
of the gear (Anderson and Neumann 1996). Some fish histograms with small
sample sizes (<100) are included in this report because of local interest,
while others were omitted (reach dependent).
