Electrofishing Efficiency and Sampling Design - PowerPoint Presentation

Electrofishing Efficiency and Sampling Design 6

Session Purpose The primary purpose of this module is to help biologists increase the accuracy and precision of sampling by 1 . improving efficiency of capture, and 2 . using a well-planned sampling design . Sampling design includes standardization (controlling important efficiency factors), optimizing sample unit number and location, and refining monitoring variables (should you use an index as CPUE or a population estimate?). We also will use some simple models to investigate the implications of capture probability and sampling variance,   number and placement of sampling units, and fish population characteristics on our decision to monitor abundance by CPUE alone or by using population estimates instead.

First, A Couple Concepts Definition of capture efficiencyThe Catch Equation

Definition of Capture Efficiency ??

Definition of Capture Efficiency Proportion of population captured by a particular sampling gear and 1 unit of effortExample: N = 1000 fish* Captured = 150 fish Efficiency = 15% *Population size usually estimated by capture-recapture, depletion sampling, or total recovery (rotenone, draining)

Capture Efficiency (“q”) and Capture Probability (“p”) C=N*q*EC=N*pp=q*E where, N = true number of fish in a sample area C = number of individuals captured in a sample area q = capture efficiency E= effort (e.g., proportion of sample area fished) p= capture probability

Area (A) containing the population total population (N) Area Sampled (a) Catch =N*q*E =N*q*(a/A) N = Catch/(q*E) N = Catch/(q*(a/A))

Area (A) containing the population total population (N) Area Sampled (a) Catch =N*q*E =N*q*(2*a/A) N = Catch/(q*E) N = Catch/(q*(2*a/A))

Area (A) containing the population total population (N) Area Sampled (A) Catch =N*q*E =N*q*(A/A) =N*q or N = Catch/q Effort =1 so p=q If you sample all of A, then capture probability (p) equals efficiency (q)

Improving Efficiency of Capture A strong approach is outlined in Framework for Increasing Sampling Efficiency and Precision of a SpeciesThis outline incorporates much of what has been learned to this point in the course. This protocol combines field and lab work to identify most effective electrical waveforms and electrode designs. Derived power standardization tables are coupled with equipment power analysis.

Sampling Design Sampling design is about reducing bias and increasing accuracy and precisioni.e., results in estimation of a population parameter (as abundance) that is likely representative and has narrow confidence intervalsThis function is accomplish bywhere and how much to sample standardization refining the variables monitored

Sampling Design Where and how much to sampleThree common frameworks: simple random sampling (SRS), stratified random sampling (STRS), and systematic sampling (SYS). (There are others as adaptive cluster sampling, etc.)

Sampling Design NOTE: calculation of parameter estimates as total abundance, means, and confidence intervals depends upon the sampling design.For a tool to learn about these 3 common designs and to use for analysis of your data, see Sampling Design Analysis

Sampling Design Another tool to learn about how sample size, capture probability, and target fish population size influences point estimates and confidence intervals of catch-per-unit-effort , seeCPUE Analysis Planner

Sampling Design For a slightly different purpose of estimating the ability to detect rare species, again with considerations of sample size, capture probability, and target fish population size, seeDetection Probability

Sampling Design StandardizationStandardization is a set of approaches to control the influence of important efficiency factors.

Sampling Design We have theoretical and empirical basis to approach controlling the effects of water conductivity (“power standardization”)The influence of other efficiency factors identified as important for certain water bodies can be addressed bySampling only above, below, or within a range of values (as only sample when water temperature is above a particular reading) Incorporate into efficiency equations for adjusting catch to a population estimate (this is not a way of standardization within a sampling design and will be discussed later in this module).

Sampling Design The following section is a in-depth view of potential efficiency factors and how the sampling variation introduced by these factors can be minimized by standardization.

Three Categories of Efficiency Factors Biological Environmental Technical

Biological FactorsFish factorsSize Habitat preferenceBehavior Population size or densityEffective conductivity

Biological Factors More efficient with larger-sized fish Several studies have shown greater proportion of larger-sized individuals in a population are captured (higher q) by EF. Capture efficiency models often have fish length as a factor.

Study on Size-based Efficiency Efficiency of electrofishing as a function of total length for largemouth bass in 27 ponds. Primary sampling by boat electrofishing one lap of shoreline. Secondary sampling by rotenone. Reynolds & Simpson (1978) AFS Spec Pub #5.

Biological Factors Less power needed per fish volume with larger-sized fish Also, the main factor in susceptibility to electroshock was size (volume or length) Dolan & Miranda 2003

Dolan and Miranda (2003) Power vs. Fish Length Pt.95 = 95% probability power threshold for immobilization

Dolan and Miranda (2003) Power vs Fish Length Pt.95 = 95% probability power threshold for immobilization

Dolan and Miranda (2003) Voltage Gradient vs Fish Length Est. Volt Grad = voltage gradient required for immobilization

Dolan and Miranda (2003) Voltage Gradient vs Fish Length Est. Volt Grad = voltage gradient required for immobilization

Biological Factors Note: in many cases, lowest efficiencies have been observed for extreme sizes (very small and large individuals); although less power may be required for larger fish and they are easier to see, larger fish may have a greater ability to escape the electric field (“fright bias”)

Biological Factors Habitat Use (behavior, anatomy) Sculpins are benthic with absent swim bladders and often reside under rocks in shallow water More power required to immobilize and need careful technique to recover specimens

Biological Factors Habitat Use Lower efficiencies possible with deep water benthic fishes, especially those species with reduced or absent swim bladders Lower efficiencies possible for wide-ranging pelagic species (gizzard shad, striped bass)

Biological Factors Habitat Use Species # of Ponds Detection Success % Bluegill sunfish 27 100 LM bass 24 96 Crappie 8 25 Green sunfish 9 67 Redear sunfish 4 100 Channel catfish 5 20 Bullhead 5 20 Golden shiner 5 20

Biological Factors Behavior Spawning aggregations: often high catchability because fish individauls are larger, maybe more territorial, and often in shallower water Beware of the possibility of injury and reproductive impairment issues

Biological Factors Population size or density Gear saturation: catchability declines at high fish abundances or density Hansen et al. (2004) found that catchability of Age-0 walleyes ( Sander vitreum ) to electrofishing declined with increasing population density. Steeves et al. (2003) observed that capture efficiency declined with higher densities of sea lamprey (Petromyzon marinus) larvae

Biological Factors Kolz suggested an effective fish conductivity of 150 µS/cm Miranda & Dolan (2004) suggested an effective fish conductivity of 115 µS/cm (eight species) Fish evaluated were primary freshwater fish. Species with other life histories, as secondary freshwater fish, may have different effective body conductivities. Effective Fish Conductivity

Effective Fish Conductivity for an Ictalurid* Catfish *considered an “outlier” family as regards EF in the U.S. Immobilization thresholds Limited size range

Required Voltage Gradients across Water Conductivities for Channel catfish Immobilization thresholds Limited size range

Biological Factors Electrofishing power requirements (Kolz & Reynolds 1989)

Environmental FactorsWater quality Ionic concentrationTemperatureConductivity (ionic concentration & temperature)TurbidityDepthLotic flow rate (discharge)

Environmental Factors Ambient water conductivity results from ionic concentration and temperature VIPE: very important piece of equipment

Effects of Conductivity Change Typical Patterns Electrical output characteristics Electrofishing requirements

Voltage Goal vs. Conductivity

Environmental Factors Water temperature: water temperature as a standardization variable, when used, is often a value selected "not too exceed", due to concerns about higher temperature induced stress/mortality, particularly of salmonids. less commonly, temperature standardization is a value selected "not to go below", due to low temperature mediated slow-down of fish floatation rates. Largemouth bass catchability has been observed to decline when water temperatures fall below 6 degrees C. Electric seine hauls in Illinois are conducted more slowly in “winter temperatures”. Burkholder and Parsons (2001) recommended that yoy walleye are sampled within a restricted temperature range (10 – 20°C) in Fall due to the curvilinear relationship between CPUE and water temperature.

Environmental Factors Turbidity or clarity Turbidity to an intermediate range can increase catchability

Environmental Factors Quiz question: high or low water conductivity?

Environmental Factors Depth, how low can you go? Physical Habitat

Environmental Factors High discharge: fishes spread out and can be more difficult to sample

Environmental Factors SubstrateSize distribution and conductivityLentic water body shapeRound vs. many coves Coarse woody debris (e.g., downed logs)

Environmental Factors Substrate size and conductivity Conductive substates can help by lowering cathode resistance (e.g., trailing cathode) Non-conductive substrates do not shield buried eggs from electric fields

Environmental Factors Water body shape Catchability of some species higher in more elongated lakes with coves

Environmental Factors Coarse Woody Debris Catchability of some species higher in structure

Environmental Factors Coarse Woody Debris Electrofishing catchability higher in ponds with structure; all ponds stocked with same number of fish. Chick et al. (1999) found that catch rates were higher in areas with greater emergent-stem density within the Florida Everglades.

Technical Factors Equipment (power capacity, electrode design, waveform, type)Crew experienceTime of daySampling design

Technical Factors Power source: may reach limitation in conductivity extremes Electrodes: can be modified for the conductivity regime; lower cond. = larger higher cond. = smaller (moderate cond.: try to make as large as system will drive and logistics allow while keeping a favorable power allocation).

Technical Factors Waveform : AC can be more effective in high and low conductivities. DC (maybe PDC) can give good taxis. Fish usually more susceptible to higher frequencies. Waveform shape can have relatively small effect (little unstudied, however).

Voltage gradient thresholds for common carp attraction or immobilization across three PDC waveform shapes. D.J. Bird & I.G. Cowx. 1993. Fisheries Research 18:363-376 50 Hz PDC; Vertical line = 1 standard error Pulsed Direct Current Waveform Shape

Technical Factors Fish usually more susceptible to higher frequencies. Frequency The power threshold for immobilization of a 5 cm fish by 15 pps PDC is six times that needed using 110 pps PDC

Technical Factors Frequency Burnet, A.M.R. 1959. Electric Fishing with Pulsatory Direct Current. New Zealand Journal of Science 2:46-56.

Effects of Duty Cycle Miranda & Dolan (2004) found that electrofishing effectiveness was strongly associated with duty cycle. Duty cycles of 15 – 50% required the least peak power to immobilize fish.

Minimum Power at 31% duty cycle What is a frequency and pulse width setting that would give a ~31% duty?

Minimum Power at 30% duty cycle

Standardization by WaveformMay be a factor of 4 characteristics: Waveform type (AC, DC, PDC)PDC waveform shape (exponential, ¼ sine, rectangular, etc.) Water & Fish conductivity mismatch (standardizing by power) Duty cycle for PDC, which includes frequency and pulse width (e.g., 30%)

Technical Factors To minimize sampling bias, should collect fishes with the most efficient method or combination of methods (for which bias is known – best) - many gear comparison studies, see references for examples May need specialized equipment (electric seine vs. backpack, pre-positioned area shockers vs. boat, deepwater cathode deployment); May need to use multiple gear types (electrofishing and non-electrofishing gears; e.g., see Ruetz et al. 2007) Small-bodied species selected by fyke nets Large-bodied species selected by boat electrofishing In the Colorado River (large river), electrofishing was more effective for collecting juveniles and trammel nets more appropriate for adults (Paukert 2004 J. Fish Biology 123:1643-1652) May need to change sampling design (e.g., point-transect sampling with a boat or backpack)

Technical Factors Inexperienced crews, or crews inexperienced with a particular water body, have been shown to have lower catch rates than experienced crews.

Technical Factors Night fishing often more effective in reservoirs and lakes (see McInerny & Cross (2000) for LMB). Predators move into shallows. Fish assemblage in area can change composition from day to night.

Technical Factors (sampling design) Point-sampling (15 min. with chase boats) for large Ictalurids *Also, boats with long dropper cathodes drift down-stream followed by a chase boat for large Ictalurid catfish

Refining Monitoring Variables This is the final aspect of sampling design covered in this course.The main questions are Should you use an index as catch-per-unit-effort (CPUE) or a population estimate to assess your fish populations? How do you use CPUE to assess population status? How do you adjust catch (CPUE) to calculate a population estimate?

Indices in Fisheries Catch-effort measures (e.g., relative abundance and catch per unit effort [CPUE]) are indicesan index is a count of fish number or species richness without an estimate of the ability to make the counts desirable attributes usually require less effort to collect the data can be more precise than population estimators (as depletion estimation) often used to assess fish population size or assemblage structure for determining biotic integrity

Relating Indices to Fish AbundanceValid use of indices for assessment of fish populations or assemblages requires that the relationship between an index (CPUE) and abundance is relatively constantacross the observable range of index values;through time when evaluating trends at a single location; and across space when making comparisons among locations

Relating Indices to Fish AbundanceWhat if efficiency (q) is NOT relatively constant and we assume it is?

Why Worry? Thought Experiment 1 Suppose we want to estimate the rate of change in population size of LMB in Lake Conway: This is a question if we are monitoring population trend. What should we use for the N’s?

Why Worry? Thought Experiment 1 Suppose we use Index data to estimate rate of change: Remember, C = N*q*E so, C/E = N*q

Why Worry? Thought Experiment How likely is this to be true? What could cause it to be false? What if this relationship is false?

Why Worry? Thought Experiment 2 Suppose we want to know whether the density in Lake Conway and Lake Maumelle are the same: ?

Why Worry? Thought Experiment How likely is this to be true?What could cause it to be false? What if it is not true?

Capture Efficiency (“q”) Hall (1986) Higher q Lower q Slope = 1 ∕ q

Capture Efficiency (“q”) Example : two sites are sampled; equal q is assumed but in reality Site A has a higher q than Site B. Although CPUE is lower in site B and the conclusion from the index is lower abundance, the true abundance is the same for both sites. Higher q Lower q Site A Site B N = CPUE ÷ q Slope = 1 ∕ q

Relating Indices to Fish Abundance CPUE Abundance Hyper-stable Hyper-depletion Proportional

Relating Indices to Fish AbundanceProblem: one or more of these constraints unlikely to be met due to changing conditions the relationship between CPUE and abundance at a sample site may vary across timethe relationship between CPUE and abundance among sample sites may be different the relationship between CPUE and abundance may not be constant across range of abundances hyper-stability or hyper-depletion

Relating CPUE to Fish Density(Single Site Across Time) Some studies have demonstrated a strong relationship between CPUE and fish density. Largemouth bass density across size classes in a 0.66 ha pond over 5 yrs. (#/ha vs. CPUEhr ; r = 0.97) (Gabelhouse 1987)

Relating CPUE to Fish Density(Across Sample Sites) Reynolds & Simpson (1978) found that boat electrofishing catch per 30.5 m of shoreline in ponds (0.5 ha average) was correlated with abundance per 30.5 m shoreline for ≥ 8” bass (r = 0.6 to 0.7); and ≥ 3” bluegill (r = 0.40 to 0.48) Abundance was estimated by rotenone census

Relating CPUE to Fish Density(Across Sample Sites) Hall (1986) derived linear regression prediction model for LMB (> 199 mm) in Ohio impoundments log10 (LMB # per hectare) = -0.55 + 1.23(log 10 (CPUE hr )) R 2 = 0.83 Water temperature, conductivity, and turbidity were measured but not included directly in the statistical analysis

Relating CPUE to Fish Density(Across Sample Sites) Largemouth bass densities (#/ha) in 2 large reservoirs (4,900 ha & 13,160 ha) were related to CPUE (#/100 m), R 2 = 0.96; stratified random sampling used (McInerny & Degan 1993) Largemouth bass densities (#/ha) in 12 impoundments (0.4 to 8.3 ha) were modeled by CPUE hr and by water conductivity (a covariate). log 10 (density) = -2.144+1.082(log 10 (CPUE)+0.847(log 10 (Conductivity) R 2 = 0.88 - Specific water conductivity varied from 410 to 1,700 μS/cm - Water temperature was not a significant variable in the stepwise multiple linear regression analysis (Hill and Willis 1994)

Relating CPUE to Fish Density(Across Sample Sites) Chick et al. (1999) investigated whether airboat electrofishing CPUE in shallow, vegetated habitats is a useful index of overall fish (SL > 8 cm) abundance.17 speciesLog 10 (CPUE 5min ) alone vs. fish/0.1 ha R 2 = 0.84 Log 10 (CPUE 5min ) & emergent stem density vs. fish density R 2 = 0.96no relationship with floating-mat volume, water depth, or water conductivitystandardized by power and removed water conductivity as a factor.

Relating CPUE to Fish Density (Across Sample Sites) However, length-frequency and species relative abundance (assemblage structure) data differed between electrofishing and block net rotenone samplesyellow bullheads, Seminole killifish, sunfishes, and small size-classes of all species lower for electrofishing largemouth bass, Florida gars, and large size-classes of all species had greater relative abundance in the electrofishing samples

Relating CPUE to Fish Density But… McInerny & Cross (2000) in Minnesota found that “q” decreased with increasing LMB densityTurbidity, water conductivity, temperature, and percent littoral area affected CPUE depending upon season and time of day Recommended that CPUE obtained in Spring provides best index of density Effects of density on q must be determined and environmental conditions must be similar before CPUE can be a useful index of LMB density Minnesota DNR developing standardized protocols for largemouth bass sampling using factors found to affect CPUE Hansen et al. (2004): “q” decreased with increasing walleye age-0 density; variability in q related to water conductivity but not to shoreline complexity, littoral area, pH, or alkalinity; recommended that CPUE only be used as a crude index of age-0 walleye population density

To Use or Not to Use CPUE ? If q has low variance, then CPUE may serve as a reliable population density index or predictor variable of population size; reason for “ standardization ”. For a tool to compare the efficacy of CPUE vs. population estimation, see Trend Analysis Your choices (most to least rigorous): Estimate q for every survey. Appropriate for high resolution management Model q as a function of important covariates. Considered essential by some for monitoring. Stabilize q (especially for indices) Main purpose of standardization protocols. Also makes modeling q more reliable. Pray Use CPUE as an index without knowing q and hope for the best.

Estimate q for Every SurveyCapture efficiency or capture probability can by estimated by depletion sampling, mark-recapture, or census techniques as rotenone. For investigating data and estimating capture efficiency, seeDepletion sampling 4 PassDepletion sampling 3 Pass Mark-Recapture Analysis

Estimate q for Every SurveyExample: Colorado River rainbow trout

AVG p^=.17CV(p^)=0.46 Rainbow trout abundance over time p^ = estimated capture probability CV = coefficient of variation Catch per unit effort adjusted to abundance by capture probability

AVG p^=.07CV(p^)=0.34 Rainbow trout abundance over time Note that catch per unit effort can increase because of increasing abundance or increasing capture probability.

Model q as a Function of Important Covariates

Modeling Capture Efficiencies Estimate capture efficiencies (e.g., via mark-recapture or depletion) and measure important efficiency factors (as mean depth) at a number of sample sites Derive a model relating capture efficiency to the measured variables using logistic regression Use the model to adjust catch to abundance at future sample sites by measuring efficiency factors

Modeling Capture EfficienciesFor an example of three efficiency equations and how they are used to adjust catch to a population estimate, see Capture Efficiency Models

Next Step “Electrofishing Equipment” (Module 7)