Wetland Fish Monitoring and Assessment

Abstract

Fish sampling is an important component of wetland research, management, conservation, monitoring, and assessment programs, and studies of fish abundance, productivity, and community structure can provide important information about wetland condition and health. In this chapter, we discuss considerations specific to wetland sampling, including issues such as the phenology of wetland use by transient fishes and sampling constraints in hydrologically-dynamic habitats. We review both active techniques, which involve moving the gear to the fish, and passive techniques, which involve the fish moving to the gear, and differentiate gears based on their ability to provide qualitative or quantitative data. Because wetlands vary considerably in their hydrology, physicochemistry, habitat structure and biotic community composition, we review and recommend a wide variety of collection techniques, including seines, minnow traps, gill and entrapment nets, electrofishing, throw and drop traps, weirs, and trawls. Problems and solutions related to gear calibration and gear bias also are addressed, and we provide examples and exercises that demonstrate common approaches to sampling wetland fishes.

Laboratory Activities and Problem Sets

In the following sections, we offer field and laboratory activities and additional exercises to illustrate topics discussed in this chapter. Note that field fish sampling may require authorization or notification of local regulatory agencies. Please adhere to your institution’s policies regarding animal care and use during these exercises.

6.1.1 Baited Versus Unbaited Traps

Goal: To determine the influence of bait type on potential sampling bias in minnow traps.

Overview: Passive sampling is quite popular for fisheries studies. However, biases may be introduced by gear type and method of deployment. This field or laboratory exercise examines biases that may occur as the result of sampling choices.

Supplies for field trip version: Minnow traps (≥6), bait (commercial fish food and commercial fish attractant such as Berkley trout bait), rope, stakes (equal to minnow traps), standard aquarium dip nets (1–2) buckets, sorting pans (2–3), fish identification guides such as Freshwater Fishes of Virginia (Jenkins and Burkhead 1994).

Supplies for laboratory version of experiment: Aquaria, no larger than 10 gal or 38 L (6), hardware cloth (fine mesh) fashioned into minnow traps (cylinder 7–10 cm in diameter, 15–20 cm long, with 5 cm, or other dimension less than the diameter, openings), bait (as above), standard aquarium dip nets.

6.1.1.1 Methods for Field Trip Version

The evening before the planned trip, minnow traps should be baited and deployed in a nearby wetland. At least two minnow traps should be used for each treatment: unbaited; baited with commercial fish food; and baited with commercial fish attractant. All six minnow traps should be secured by rope to stakes driven into the banks and deployed along the shoreline a sufficient depth to cover the trap. Deployment and retrieval times should be noted. Students should empty the contents of each trap into individual holding buckets or directly into sorting pans if few fish are caught. Students should then identify and enumerate fish and estimate catch per unit effort (CPUE) as the number of fish of each species in a trap divided by the number of hours deployed. Next, students should calculate the arithmetic mean CPUE for each of the three trap treatments and answer the questions below.

6.1.1.2 Methods for Laboratory Version

Instructor(s) will stock three aquaria with similar densities of one or more species of fish obtained from a local wetland, bait shop or pet store. Instructors or students should build small minnow traps prior to the experiment with narrow openings sufficient for fish entry. Two minnow traps will be deployed during the laboratory period in three aquarium treatments: (1) one trap with fish food and one with fish bait; (2) one trap with fish food and one with no bait; and (3) one with fish bait and one with no bait. Time of trap deployment into the aquaria should be noted. Students will observe fish movement into minnow traps and record the species and number of fish in each trap at the end of the laboratory period. CPUE will be estimated as the number of fish of each species in a trap divided by the number of minutes passed since the traps were placed into the aquaria (Hubert et al. 2012)

$$ \mathrm{ CPUE}={{{\sum {\left( {{{\mathrm{ n}}_1}+{{\mathrm{ n}}_2}+\ldots {{\mathrm{ n}}_{\mathrm{ i}}}} \right)} }} \left/ {{{{\mathrm{ t}}_{\mathrm{ i}}}}} \right.} $$

where n₁ is the number of fish of the first species, n₂ is the number of fish of the second species, n_i is the number of fish in the last species, and t_i in the number of minutes a trap was deployed. Students can calculate the arithmetic mean CPUE for each trap type (two traps each for commercial food, commercial bait, and unbaited). Students should then answer the questions below.

6.1.1.3 Questions for Reflection and Study

• 1. Which trap type exhibited the highest mean CPUE, combining all fish species that were collected? Did any single fish species differ from this pattern (i.e., did any species become trapped at a higher CPUE in another type of trap than the one that collected the most overall fish)? Did anything that you observed about the trap or bait suggest why CPUE was highest in this type of trap?

• 2. Do you believe that your experiment has evidence of trap bias? Did bait type matter? Explain and defend with your data.

• 3. Could your data be comparable with a minnow trap study that used another kind of bait? Why or why not?

• 4. Consult a regional fish guide, such as Freshwater Fishes of Virginia (Jenkins and Burkhead 1994), about life history and habitat requirements of the fish with the highest CPUE in each trap type. Does something about their life history or habitat requirements suggest why the fish was attracted to that trap type?

Jenkins RE, Burkhead NM (1994) Freshwater fishes of Virginia. American Fisheries Society, Bethesda, 1079 pp

6.1.2 Removal Sampling and the Influence of Increasing Sampling Effort

Goal: Introduce removal sampling and common estimators of population size associated with removal sampling.

Overview: In small, enclosed wetlands or wetlands where habitats may be segregated from the surrounding areas, researchers can use closed population methods to estimate fish population size. One historically popular method is population estimation by removal. In removal sampling, fish are collected and removed or held while additional fish are collected. A minimum of two collections are needed, and often additional collections are recommended because the additional collections enhance the population estimate. This experiment may be conducted in the field by seining or electrofishing, whichever is available, or by dip net in the laboratory.

Supplies needed for a field version: Seine or backpack electrofishing unit (depending on availability and suitability given local conductivity), gloves and waders, long-handled nets (electrofishing only), buckets or large cooler or fish basket to hold fish, block nets (if sampling a small area of a larger wetland).

Supplies needed for laboratory version: table top, simulated fish (e.g., small plastic vials or packing peanuts) and a sampling “net” (e.g., an inverted shoe box).

6.1.2.1 Methods for Field Version

A small, enclosed wetland should be selected or a small area within a larger wetland should be enclosed by block net. The ideal area would be about one-eight acre, if block nets are deployed. Students should conduct 10-m quantitative seine hauls (Fig. 6.1) or 100-m electrofishing passes (Fig. 6.3). Ideally, no fewer than four hauls or passes should be conducted; however, if very few fish are caught on the second and third haul or pass, the fourth haul or pass may be omitted. At the end of each haul or pass, fish should be identified and enumerated and then transferred to a holding bucket, cooler, or basket, until the final haul or pass when all fish can be returned to the wetland. Students should record the number and type of fish in each haul or pass to answer the questions in this section.

6.1.2.2 Methods for Laboratory Version

A small table top should be covered with random patches of “fish”, and the “sampler” should be given to a blindfolded student to randomly place on the table. After each sample is taken, the remaining “fish” should be collected and re-distributed on the table top. For best results, the sampler should probably be able to cover about 20 % of the table top. For each sample, students should record the number of “fish” and set them aside. After at least three samples, students should be able to answer the questions in this section.

6.1.2.3 Questions for Reflection and Study

The Zippin method is commonly used to estimate fish populations with two removals (Hayes et al. 2007). The Zippin method is a maximum likelihood method and differs from the regression based DeLurly method more frequently used in the past. The Zippin method assumes all fishes had equal vulnerability to being sampled by the selected gear, equal effort was expended for each sample, and the probability of capture (catchability) was equal for each sample. The Zippin method also requires that the first sample yield more fish than the second sample. The Zippin method is

$$ \mathrm{ N}\text{--}\mathrm{ hat}={{{\mathrm{ n}_1^2}} \left/ {{\left( {{{\mathrm{ n}}_1}-{{\mathrm{ n}}_2}} \right)}} \right.} $$

where N-hat is the estimate of the fish population, n₁ is the number of fish removed in the first sample, and n₂ is the number of fish removed in the second sample. When we estimate, we also desire to know the precision of the estimate. For the Zippin method, we can estimate the precision of the estimate by its variance

$$ \mathrm{ V}\left( \mathrm{ N} \right)={{{\mathrm{ n}_1^2\mathrm{ n}_2^2\left( {{{\mathrm{ n}}_1}+{{\mathrm{ n}}_2}} \right)}} \left/ {{{{{\left( {{{\mathrm{ n}}_1}-{{\mathrm{ n}}_2}} \right)}}^4}}} \right.} $$

where V(N) is the variance of the Zippin estimate, n₁ is the number of fish removed in the first sample, and n₂ is the number of fish removed in the second sample.

The Zippin method is not possible for more than two removals. For three removals, the following formula is used following Hayes et al. (2007) citing Junge and Libosvarksy (1965) as cited in Seber (1982)

$$ \mathrm{ N}\text{--}\mathrm{ hat}={{{6{{\mathrm{ X}}^2}-3\mathrm{ XY}-{{\mathrm{ Y}}^2}+\mathrm{ Y}\surd \left( {{{\mathrm{ Y}}^2}+6\mathrm{ XY}-3{{\mathrm{ X}}^2}} \right)}} \left/ {{18\left( {\mathrm{ X}-\mathrm{ Y}} \right)}} \right.} $$

where N-hat is the population estimate, $\mathrm{ X}\ \mathrm{ is}\ 2{{\mathrm{ n}}_1}+{{\mathrm{ n}}_2}$ and $\mathrm{ Y}\ \mathrm{ is}\ \mathrm{ n}{_1}+\mathrm{ n}{_2}+\mathrm{ n}{_3}$. Again, it is always of interest to estimate variance, which is estimated by finding q or catchability first as

$$ \mathrm{ q}\text{--}\mathrm{ hat}={{{3\mathrm{ X}-\mathrm{ Y}-\surd \left( {{{\mathrm{ Y}}^2}+6\mathrm{ XY}-3{{\mathrm{ X}}^2}} \right)}} \left/ {{2\mathrm{ X}}} \right.} $$

where q-hat is the catchability estimate, $\mathrm{ X}\ \mathrm{ is}\ 2{{\mathrm{ n}}_1}+{{\mathrm{ n}}_2}$ and $\mathrm{ Y}\ \mathrm{ is}\ \mathrm{ n}{_1}+\mathrm{ n}{_2}+\mathrm{ n}{_3}$. Then, we estimate variance as

$$ \mathrm{ V}\left( \mathrm{ N} \right)={{{\mathrm{ N}\text{--}\mathrm{ ha}\mathrm{ t}\left( {1-\mathrm{ q}\text{--}\mathrm{ ha}\mathrm{ t}} \right)\mathrm{ q}\text{--}\mathrm{ ha}\mathrm{ t}}} \left/ {{{{{\left( {1-\mathrm{ q}\text{--}\mathrm{ ha}\mathrm{ t}} \right)}}^2}-\left[ {\mathrm{ t}{{{\left( {1-\mathrm{ q}\text{--}\mathrm{ ha}\mathrm{ t}} \right)}}^2}\mathrm{ q}\text{--}\mathrm{ ha}{{\mathrm{ t}}^{{\left( {\mathrm{ t}-1} \right)}}}} \right]}} \right.} $$

where V(N) is the estimate of the variance of the fish population, q-hat is the estimate of catchability, and t is the number of removals.

• 1. Estimate the fish population size and variance by the Zippin method with the first two removals. Then estimate the fish population by the three removal method with the first, second, and third removals. Estimate catchability and the variance of the three removal population estimate.

  • 1a Do the fish population estimates differ?

  • 1b If smaller variance may be assumed to suggest greater precision, did adding another removal increase precision?

  • 1c Given your experience sampling, do you think that adding another removal is worth the difference in precision?

  For the field version only:

  • 1d Given that you do not know the actual number of fish present, do you believe that removal sampling provides a reasonable estimate of the fish present?

  • 1e Can you get insights from the variance? Defend your answer with your data.

  For the laboratory version only:

  • 1d Show your estimates to the instructor, who will reveal the actual number. Which estimate was closer to the real number?

  • 1e Given your experiences, how confident are you that removal sampling may offer reasonable estimates of fish populations?

• 2. Another common assessment is to describe the taxonomic diversity of fishes in a wetland. One common and readily estimated measure of diversity is the Shannon index, also termed the Shannon-Wiener or Shannon-Weaver Index. The index ranges from a low end near one, which indicates low species richness (low number of species) and evenness (few species numerically dominate the community), to 3.5, which indicates high species richness (higher number of species) and evenness (numbers spread among the species). The Shannon Index is found by

  $$ {\mathrm{ H}}^{\prime}= - {\Sigma_{\mathrm{ i}}}^{\mathrm{ s}}\left[ {{{\mathrm{ p}}_{\mathrm{ i}}} \ln \left( {{{\mathrm{ p}}_{\mathrm{ i}}}} \right)} \right] $$

  where p_i is the proportion of an individual species. For example, if 3 species are present in a sample at 35, 25, and 15 individuals, ${\mathrm{ H}}^{\prime}=-\left\{ {\left[ \left( {35/75} \right) {\ln \left( {35/75} \right)} \right]+\left[ {\left( {25/75} \right) \ln \left( {25/75} \right)} \right]+\left[ {\left( {15/75} \right) \ln \left( {15/75} \right)} \right]} \right\}=-\left[ {\left( {-0.36} \right)+\left( {-0.37} \right)+\left( {-0.32} \right)} \right] = 1.05$ suggesting low diversity of species and dominance by 1 species.

  • 2a Estimate H’ for the first removal, then for the first 2 removals combined, and then for all removals combined.

  • 2b Does adding removals increase H’? What does this mean?

  • 2c Is H′ sensitive to the number of removals?

Hayes DB, Bence JR, Kwak TJ, Thompson BE (2007) Abundance, biomass, and production. In: Guy CS, Brown ML (eds) Analysis and interpretation of freshwater fisheries data. American Fisheries Society, Bethesda, pp 327–374

Junge CO, Libosvarsky J (1965) Effects of size selectivity on population estimates based on successive removals with electrofishing gears. Zoologicke Listy 14:171–178

Seber GAF (1982) The estimation of animal abundance and related parameters. Blackburn Press, Caldwell, , pp 672

6.1.3 Additional Exercises

• 1. Suppose a fish manager is tasked to sample a wetland in order to determine its ecological health. She assumes that a wetland with high fish diversity (a Shannon index above 2.5 for this small wetland) and high fish density (more than100 fish per acre) would indicate a healthy ecosystem. She conducts removal sampling by seine hauls in a 1-acre enclosed portion of the wetland. The hauls produced 56 fish, 32 fish, and 13 fish. The hauls produced 9 emerald shiners, 7 fathead minnows, 5 common carp, 9 largemouth bass, 4 bluegill, 7 pumpkinseed sunfish, 9 green sunfish, 3 sand shiners, 6 warmouth bass, 5 chain pickerels, 3 golden shiners, 9 Johnny darters, 9 white crappies, 5 brown bullheads, and 11 yellow bullheads. Use the estimators given in Sect. 6.4.3 to determine if the wetland would be considered healthy.

• 2. Suppose the same fish manager in question 1 visited a second wetland. This time, the hauls produced 39 fish, 41 fish, and 18 fish. Use the estimators given in Sect. 6.4.3 to determine if the wetland would be considered healthy. The hauls produced 19 emerald shiners, 7 fathead minnows, 5 common carp, 1 largemouth bass, 1 bluegill, 8 pumpkinseed sunfish, 19 green sunfish, 3 sand shiners, 3 warmouth, 2 chain pickerels, 3 golden shiners, 2 Johnny darters, 2 white crappies, 12 brown bullheads, and 11 yellow bullheads. Use the estimators given in Sect. 6.4.2 to determine if the wetland would be considered healthy.

• 3. A wetland manager decided to sample a large wetland with boat electrofishing and gill nets. The electrofishing sampled for 45 min and collected 100 fish of 7 species. The gill nets soaked for 12 h and collected 292 fish in 11 species. Which gear exhibited the higher CPUE? Why might the electrofishing have collected fewer species than the gill net?

• 4. A wetland manager sampled a boat-accessible wetland with point abundance electrofishing and a drop sampler. The point abundance electrofishing sampled 900 s in five 100 m² areas surrounded by block nets, yielding 14, 25, 9, and five fish. The drop trap of 1 m² was deployed 10 times yielding 1 fish, 0 fish, 3 fish, 0 fish, 0 fish, 1 fish, 0 fish, 0 fish, 0 fish, and 2 fish. Estimate fish density per 1 m² for both methods. Which method estimated a higher mean fish density?