Comparison of Methods for Processing Sweep-Net Samples of Aquatic Invertebrates From Forested Wetlands

Aquatic invertebrates influence nutrient cycling and distribution within and among wetlands (Cummins et al. 1989; Malmqvist 2002). In addition to being prey for other invertebrates, fish, herptofauna, birds, and mammals, aquatic invertebrates also provide other important ecological services, such as decomposition and nitrification (Batema et al. 2005). Aquatic invertebrates occur worldwide in wetlands and deepwater habitats, such as flooded bottomland hardwood forests (Batema et al. 2005; Fredrickson 2005; Mitsch and Gosselink 2007). Most invertebrates that inhabit flooded bottomland hardwood forests acquire nutrients directly or indirectly from leaf litter and thereby interconnect producers, consumers, and decomposers (Fredrickson and Reid 1988, Batema et al. 2005).

Although collecting invertebrates in these and other wetlands can be relatively efficient and economical, processing the samples is laborious and costly because invertebrates often must be separated from leaf litter and sediments in samples (Murkin et al. 1994; Doğramacı et al. 2010). Consequently, many researchers have used subsampling approaches or techniques that expedite processing of invertebrate samples from bottomland hardwood forests and other wetlands (Wehrle et al. 1995; Batema et al. 2005; Foth 2011).

Extensive literature exists on sampling invertebrates in streams, lakes, marshes, and other wetlands, but much less is known regarding efficient and reliable methods for processing samples from forested wetlands (Murkin et al. 1994; Murkin and Ross 2000; Sherfy and Kirkpatrick 2003). Researchers have added supersaturated solutions of sugar or salt and water to samples, so that invertebrates float and thereby expedite investigators' detection and recovery of the organisms (Flannagan 1973; Kaminski and Prince 1981). However, lentic forested wetlands often contain high proportions of leaf litter and other organic matter to which invertebrates cling, and invertebrates are, therefore, less likely to be disassociated from such material and suspend in solutions. Also, we are unaware of an evaluation of solutions used to facilitate recovery of aquatic invertebrates from sweep-net samples collected in flooded bottomland hardwood forests. Therefore, we conducted an experiment to evaluate differences in invertebrate recovery from sweep-net samples from flooded bottomland hardwood forests. Specifically, we tested the null hypothesis of no difference in abundance, dry mass, Order-level richness, and Order-level diversity of recovered invertebrates among samples processed using 1) tap water alone (control), 2) a supersaturated solution of sugar and water, or 3) a supersaturated solution of salt and water. In addition, we report the effectiveness of our subsampling technique relative to the aforementioned invertebrate metrics and minutes saved by subsampling.

We collected sweep-net samples of invertebrates from plots within flooded bottomland hardwood forests in Delta National Forest in the Mississippi Alluvial Valley and at Noxubee National Wildlife Refuge in the Interior Flatwoods, both in Mississippi. We established plots within one greentree reservoir at each study site. Greentree reservoirs are seasonally flooded stands of mature bottomland trees enclosed by a levee (Figure 1; Reinecke et al. 1989). We sampled within greentree reservoirs to ensure availability of samples from these forested wetlands during winter 2008–2009. Detailed descriptions of Delta National Forest and Noxubee National Wildlife Refuge have been reported elsewhere (Wehrle et al. 1995; Foth 2011).

We collected invertebrate samples from 10 spatially balanced research plots (0.2 ha; Foth 2011), established using the grts design option of the SPSURVEY package (Kincaid and Olson 2011) in R 2.11.0 (R Development Core Team 2006), in each of the aforementioned greentree reservoirs. To collect aquatic invertebrates, we pushed a rectangular sweep net (23 × 45 cm, 500-µm mesh) through the water and leaf matter in contact with substrate for a distance of 0.5 m (Wehrle et al. 1995; Foth 2011). We collected 30 samples monthly (December 2008–February 2009). We placed all samples in Ziploc bags, then iced samples on site for transport to the laboratory at Mississippi State University, and kept samples frozen at −10°C until they were processed (Murkin et al. 1994).

We mixed sugar or salt with 8 L of 50°C tap water to create supersaturated solutions. We added 1 kg of solute at a time to the 50°C tap water until supersaturation occurred. We found that 17 kg of sugar and 6 kg of salt created supersaturated solutions. We used different masses of each solute based on their ability to dissolve in water. We assumed supersaturation occurred when additional solutes no longer dissolved in the 50°C tap water (Kaminski and Prince 1981).

We randomly selected samples from each month for our experiment and analyses (n  =  72). We randomly assigned each sample to one of three processing treatments: 1) 50°C tap water alone (i.e., control; n  =  28), 2) the sugar and water solution (n  =  22), and 3) the salt and water solution (n  =  22; Table S1, Supplemental Material). Due to a laboratory processing accident where some samples in each solution were spilled, we analyzed 72 of the 90 collected samples. We placed thawed samples in a plastic bucket that contained 2 L of one of the three above treatments and hand-agitated each sample for 30 s to disassociate invertebrates from leaf and other litter. We poured contents from the bucket through a plastic half-cylinder, with 1.5-cm-diameter apertures (hereafter, the sieve; Figure 2), which was placed atop a 500-µm sieving bucket. We placed invertebrates from the sieving bucket that had passed through the sieve (hereafter, the subsamples), and those that remained on the sieve or attached to leaf litter, each in separate plastic bags for processing. We defined whole samples as those containing invertebrates in subsamples plus those that remained atop the sieve. For sub- and whole samples, we determined 1) number of individuals, 2) dry mass (g), 3) Order richness (i.e., number of Orders), and 4) Order-level Shannon diversity (H′; Hagy and Kaminski 2012) of all recovered invertebrates. We dried invertebrates in each sub- and whole sample and all debris associated with the sample at 60°C for 18–24 h and weighed each to 0.1 mg (Murkin et al. 1994). We also recorded time (min) required to process sub- and whole samples.

Our dependent variables were proportions (hereafter, recovery rate) calculated as the number and mass of invertebrates from subsamples divided by the total number or mass of invertebrates recovered from whole samples. We used analysis of covariance to test for differences in mean recovery rates of invertebrates among the three treatments, with dry mass (g) of combined leaf and organic matter in each sample as a covariate. We analyzed numbers and mass of recovered invertebrates separately. We used the lm function in Program R version 2.8.1 and specified the dependent variables as a function of treatments, the covariate, and their interactions (α  =  0.05). For our full model, heterogeneity of slopes tested whether the effect of the processing treatment depended on the covariate; in our reduced model, heterogeneity of intercepts tested for processing treatment effects alone. We set statistical significance at P ≤ 0.05 for all tests. We report Order-specific recovery rates for the six most frequently recovered orders, which included fingernail clams Veneroida, daphnia Cladocera, freshwater snails Pulmonata, midge larvae Diptera, amphipods Amphipoda, and isopods Isopoda. We combined data among treatments and tested the hypothesis of no differences in invertebrate Order richness and Order diversity (H′, natural log) between whole and subsamples using a paired t-test (Program R, Version 2.8.1; α  =  0.05).

We recovered 16,396 invertebrates from 72 samples. Distributions of recovered invertebrate abundance appeared similar among sugar-solution (⁠  =  202 organisms; range  =  25–917; n  =  24), salt-solution (⁠  =  224; range  =  25–849; n  =  24), and tap water treatments (⁠  =  252; range  =  33–875; n  =  24).

Recovery rate for invertebrate abundance did not vary among treatments with mass of leaf litter and other detritus (F_2,66  =  1.281, P  =  0.284). With the reduced model, we also did not detect a difference in recovery rate among treatments for abundance (F_2,68  =  0.186, P  =  0.831). However, recovery rate was inversely related to dry mass of detritus; for every additional 10 g of detritus, recovery rate decreased 3% (F_1,68  =  12.957, P  =  0.001; Figure 3). Overall, we recovered 72% (SE  =  1.5%, n  =  72) of the detected invertebrates in our subsamples. If a whole sample contained little to no detritus, our recovery rate increased to nearly 80%. Across treatments, Order-specific recovery rates were greatest for daphnia (⁠  =  86%, SE  =  0.38%) and least for freshwater snails (⁠  =  30%, SE  =  2.91%; Table 1).

Recovery rate of invertebrate dry mass was neither influenced by mass of detritus (F_1,68  =  2.572, P  =  0.113) nor by the interaction of treatment with the covariate, (F_2,66  =  1.634, P  =  0.203). From the reduced model, we did not detect a difference among treatments (F_2,68  =  0.031, P  =  0.970). Across treatments, we recovered 48% (SE  =  7.00%, n  =  72) of the recovered invertebrate mass. Across treatments, taxa-specific recovery rates by mass were greatest for daphnia (⁠  =  83%, SE  =  1.88%) and least for fingernail clams (⁠  =  17% SE  =  2.84%; Table 1).

We consistently recovered greater proportions of the numerical abundance than mass of invertebrates. The mean difference in recovery by numbers and mass was 6.0% (SE  =  2.6%, n  =  72). Fingernail clams exhibited the greatest difference (200%) between abundance and mass, whereas recovery rates of amphipod abundance and mass were equivalent (Table 1).

We detected differences in H′ (paired t_1,78  =  −4.337, P < 0.001) and Order-level richness of invertebrates (paired t_1,78  =  −8.406, P < 0.001) between sub- and whole samples. On average, H′ and richness were 19% and 12% greater, respectively, from whole than subsamples.

Processing time for subsamples (⁠  =  16.6 min, SE  =  1.7, n  =  72) was > 1.6 times faster and 2.2 times less variable than the portion that remained atop the sieve (⁠  =  26.8 min, SE  =  3.8, n  =  72). Average processing time for whole samples was 43.4 (SE  =  2.2, n  =  72) min. Processing time ranged from 8 to 40 min for subsamples and from 10 to 80 min for whole samples.

We did not detect differences in recovery rates of numbers or mass of invertebrates among the three treatments even with a robust sample size (n  =  72). Although some researchers used supersaturated solutions to recover invertebrates (Flannagan 1973; Kaminski and Prince 1981), we suspect recovery rate was similar among test media, because our samples generally contained large amounts of detritus that were buoyant like invertebrates. Because leaf litter and invertebrates were buoyant, our results suggest that equivalent relative abundances and masses of invertebrates can be recovered from sweep-net samples from forested wetlands using hot tap water alone.

Our subsampling technique proved useful for separating most detritus from invertebrates. As such, this procedure aided in expediting processing subsamples because invertebrates became increasingly conspicuous. We estimated that subsampling may reduce laboratory processing time by > 1.6 times, which may allow researchers to process additional samples and thereby improve precision of estimates. However, if investigators only process subsamples, our results indicate they may underestimate Order richness and diversity by 12% and 19%, respectively. Because richness and diversity often are included in rapid environmental bio-assessments, data only from subsamples will provide conservative estimates of invertebrate community composition in lowland forested wetlands. These estimates would become increasingly conservative if researchers required estimates of species richness or diversity.

Recovery of numerical abundance and mass of invertebrates was influenced by the invertebrate taxa collected by our sweep nets. Some invertebrates collected have relatively greater mass (e.g., fingernail clams, freshwater snails) than other taxa (e.g., chironomid larvae, amphipods). Regardless of specific gravity, which made detritus less dense than the supersaturated solutions, we rarely observed these taxa floating in test media. Therefore, fingernail clams and freshwater snails had lower recovery rates than other invertebrates. Additionally, freshwater snails and fingernail clams made up the majority of invertebrate biomass in our samples, and their inability to suspend in solution explains why we recovered less invertebrate biomass than we did numerical abundances. In contrast, we and other investigators recovered midge larvae, daphnia, and other small taxa efficiently (Flannagan 1973; Kaminski and Prince 1981). Because our samples from forested wetlands consisted primarily of taxa that easily floated (Foth 2011), our overall recovery rates of invertebrates were skewed toward these taxa. Our method of using a sieve to subsample invertebrates is appropriate when relative abundances are desired and when most invertebrates encountered have relatively low mass or they float.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Table S1. Data for analysis of three processing treatments on samples collected from forested wetlands of Mississippi, winter 2008–2009. Data are organized by study site (Delta National Forest, DNF; Noxubee National Wildlife Refuge, NOX), survey plot, compass azimuth, survey round, position relative to sieve (on top of sieve, subsample), processing treatment, and Family (count, weight, and proportion).

Found at DOI: http://dx.doi.org/10.3996/022012-JFWM-020.S1 (272 KB XLS).

Reference S1. Fredrickson LH, Reid FA. 1988. Invertebrate response to wetland management. United States Department of the Interior, Fish and Wildlife Service Fish and Wildlife Leaflet 13 in Cross D, Vohs P, editors. 1988 Waterfowl Management Handbook. Fort Collins, Colorado.

Found at DOI: http://dx.doi.org/10.3996/082011-JFWM-020.S2; also available at http://www.nwrc.usgs.gov/wdb/pub/wmh/13_3_1.pdf (154 KB PDF).

We would like to thank A. Leach for his help with collecting invertebrate samples in the field, assisting in the laboratory, and providing input throughout the process. Additionally, we thank the U.S. Department of Agriculture's Forest Service Center for Bottomland Hardwoods Research for funding and Mississippi State University Forest and Wildlife Research Center (MSU-FWRC) and the U.S. Fish and Wildlife Service for facilities and additional support. Thanks also to the Subject Editor and anonymous reviewers for their assistance. Our manuscript has been approved for publication as MSU-FWRC Publication No. WF-344.

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.