nena masthead
NENA Home Staff & Editors For Readers For Authors

A Comparison of Three Macroinvertebrate Sampling Devices for Use in Conducting Rapid-Assessment Procedures of Delmarva Peninsula Wetlands
T. Peter Lowe, Kerry Tebbs, and Donald W. Sparling

Northeastern Naturalist, Volume 23, Issue 2 (2016): 321–338

Full-text pdf (Accessible only to subscribers. To subscribe click here.)


Access Journal Content

Open access browsing of table of contents and abstract pages. Full text pdfs available for download for subscribers.

Issue-in-Progress: Vol.30 (1) ... early view

Current Issue: Vol. 29 (4)
NENA 29(4)

All Regular Issues


Special Issues






JSTOR logoClarivate logoWeb of science logoBioOne logo EbscoHOST logoProQuest logo

Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 321 2016 NORTHEASTERN NATURALIST 23(2):321–338 A Comparison of Three Macroinvertebrate Sampling Devices for Use in Conducting Rapid-Assessment Procedures of Delmarva Peninsula Wetlands T. Peter Lowe1,*, Kerry Tebbs1, and Donald W. Sparling1,2 Abstract - Three types of macroinvertebrate collecting devices, Gerking box traps, D-shaped sweep nets, and activity traps, have commonly been used to sample macroinvertebrates when conducting rapid biological assessments of North American wetlands. We compared collections of macroinvertebrates identified to the family level made with these devices in 6 constructed and 2 natural wetlands on the Delmarva Peninsula of Maryland. We also assessed their potential efficacy in comparisons among wetlands using several proportional and richness attributes. Differences in median diversity among samples from the 3 devices were significant; the sweep-net samples had the greatest diversity and the activity-trap samples had the least diversity. Differences in median abundance were not significant between the Gerking box-trap samples and sweep-net samples, but median abundance among activity-trap samples was significantly lower than among samples of the other 2 devices. Within samples, the proportions of median diversity composed of major class and order groupings were similar among the 3 devices. However the proportions of median abundance composed of the major class and order groupings within activity-trap samples were not similar to those of the other 2 devices. There was a slight but significant increase in the total number of families captured when we combined activity-trap samples with Gerking box-trap samples or with sweep-net samples, and the per-sample median numbers of families of the combined activity-trap and sweep-net samples was significantly higher than that of the combined activity-trap and Gerking box-trap samples. We detected significant differences among wetlands for 4 macroinvertebrate attributes with the Gerking box-trap data, 6 attributes with sweep-net data, and 5 attributes with the activity-trap data. A small, but significant increase in the number of attributes showing differences among wetlands occurred when we combined activity-trap samples with those of the Gerking boxtrap or sweep net. Introduction Public concern in the US over the destruction and/or degradation of the nation’s water resources led to the passage of the US Clean Water Act of 1972, and later revisions of the act mandated that the biological integrity of all surface waters of the US shall be maintained (Resh et al. 1995). Passage of the act led the US Department of Agriculture and US Environmental Protection Agency (USEPA) to establish programs for restoring wetlands converted to agricultural lands and for creating wetlands as replacements of natural wetlands destroyed through development. It also partially served as an impetus to modify a century-old effort to incorporate 1USGS Patuxent Wildlife Research Center, 10300 Baltimore Avenue, Beltsville, MD 20705. 2Cooperative Wildlife Research Center, MS 6504, Life Sciences II, Southern Illinois University, Carbondale, IL 622901. *Corresponding author - Manuscript Editor: David Yozzo Northeastern Naturalist 322 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 biological factors in status assessments of water resources and to monitor trends in the condition of biological communities (Plafkin et al. 1989, Resh 1995, Resh and Jackson 1993). These efforts resulted in the development of rapid assessment procedures (RAPs) that rely on evaluating the status of various qualitative biological attributes instead of exhaustive quantitative approaches such as detailed abundance and diversity studies. The quantitative studies are often quite expensive because they require repeated samples (Plafkin et al. 1989, Resh 1995, Resh and Jackson 1993). RAP surveys are sampling efforts on groups of wetlands conducted during a relatively small window of time during a sampling season. RAP surveys using wetland macroinvertebrates (hereafter invertebrates) should be conducted when invertebrate diversity is at its highest (Gernes and Helgen 2002) and use the same sampling protocol in each wetland. The protocol should identify the collection devices to be used and their characteristics, such as such as dimensions and mesh sizes, how many samples should be taken from each wetland, where and how long the samplers would be deployed, and how collected samples should be processed. Attribute values derived for each wetland are then compared among the sampled wetlands. The RAP approach can also be used to monitor changes in wetlands by applying the protocol over time throughout the duration of an established monitoring program for constructed wetlands, or for natural wetlands used as reference sites. RAPs have an advantage over more-detailed studies in reducing the cost and effort involved in conducting site surveys and summarizing survey results in a manner that is useful for conservation organizations and is understandable by non-specialists (Plafkin et al. 1989, Resh 1995, Resh and Jackson 1993). The efficacies of various invertebrate-sampling devices have been studied to establish protocols for conducting RAPs in wetlands. The sweep net, a common semi-quantitative sampling device (Cheal et al. 1993, Florencio et al. 2012, Meyer et al. 2011), is a fine-mesh net mounted on a half circle or rectangular metal frame that is attached to a 1.5–1.8-m (5–6-ft)-long pole handle. Mesh size may vary, but we used 1-mm2 mesh for this study. Sampling is performed by sweeping the net through the water, sometimes brushing the bottom, along a measured distance (Murkin et al. 1983, Muzaffar and Colbo 2002, Turner and Trexler 1997) or for a measured length of time (Bercerra Jurado et al. 2008, Furse et al. 1981, Menetrey et al. 2011, Oertli et al. 2005, Sychra and Adamek 2010). Enclosure-type samplers, such as stovepipes, open-bottom boxes, drop frames, throw traps, Gerking box-traps, and benthic corers also are frequently used. These devices provide quantitative measures of invertebrate abundance and diversity for a specific bottom area and water volume (Brinkman and Duffy 1996, Meyer et al. 2011, O’Conner et al. 2004, Sychra and Adamek, 2010, Turner and Trexler 1997). Captured invertebrates are removed from the stovepipe, open-bottom box, drop frame, and throw trap samplers by sweeping the enclosed space with a hand-held or bar-framed small-mesh net until no more invertebrates are captured. The Gerking box-trap is similar except it has a sliding mesh bottom that is closed after the box is placed in the water (Sychra and Adamek 2010, Tolonen and Hamalainen 2010). After closing the bottom, the device is shaken in the water causing small, grain-sized debris to escape through the mesh. The trap is then lifted out Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 323 of the water and the remaining contents are dumped into a container. Core samplers are used when the emphasis is on collecting benthos in soft-bottomed water bodies. These devices remove plugs of sediment up to several centimeters in length that are placed into sieves and washed to separate the invertebrates (Hyvonen and Nummi 2000, Milbrink and Wiederholm 1973, Whiteside and Lindegaard 1980). A disadvantage of sweep-net and enclosure-type devices is that they give semiquantitative or quantitative estimates of abundance of only those taxa that can be readily captured at the time of day samples are taken. Highly mobile or burrowing taxa that can easily escape capture, and taxa that are not active when sampling occurs may be underrepresented in samples taken with these devices (Becerra Jurado et al. 2008, Florencio et al. 2012, Meyer et al. 2011). This disadvantage can be partially circumvented by using passive-sampling devices such as activity or funnel traps and fyke nets when conducting RAPs (Florencio et al. 2012). Activity traps can be mounted horizontally on stakes with the axis of the trap parallel to the water surface to capture highly mobile taxa or mounted vertically with the funneled end directed downward. Downward-directed traps collect taxa that migrate up and down through the water column at specific times of the day. The traps are usually mounted at specific distances below the water surface or above the bottom (Brinkman and Duffy 1996, Muscha et al. 2001, Whiteside and Lindegaard 1980). Fyke nets are placed on the wetland floor where the water is no deeper than the vertical distance across the net mouth. The trapping durations of passive devices are usually between 24 h and 168 h (1–7 d; Verdonschot 2010). Passive devices provide integrated samples of taxa that are active any time of day (Florencio et al. 2012) as well as active taxa that may escape a sweep net or an enclosure-type device. They are not useful for estimating relative invertebrate abundance and diversity per unit volume of water or unit area of bottom substrate. We conducted this study to evaluate 3 devices commonly used in wetland studies—activity traps, D-shaped sweep nets, and Gerking box-traps—for conducting RAPs of wetlands on the Delmarva Peninsula, which includes portions of Delaware, Maryland, and Virginia. Our 3 primary study objectives were to: (1) determine during which of 5 sampling periods diversity was greatest, (2) compare the diversity and abundance of invertebrates collected by each device from 8 Delmarva Peninsula wetlands through the sampling season, and (3) calculate the values for selected attributes from the data collected with each device during the collection period when diversity was greatest, and determine for each device which macroinvertebrate attributes varied significantly among wetlands. Other investigations have shown that a better representation of the taxonomic diversity of an aquatic system will result from combining the invertebrates collected with an active device and a passive device (Becerra Jurado et al 2008, Florencio et al 2012, Muzaffar and Colbo 2002, Whiteside and Kindegaard 1980). We will also perform these calculations by combining Girking box trap and sweep net samples individually with activity trap samples. We based attribute selection on those used by other investigators in similar studies and on the comparisons of diversity and abundance of invertebrates collected with each device. While processing samples, we also Northeastern Naturalist 324 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 Table 1. Location, type, area, and age of wetlands. Age given in years. Wetland name County Type Area (ha) Age when studied Barnstable 1 Queen Ann, MD Constructed 1.01 11 Barnstable 3 Queen Ann, MD Constructed 1.25 8 Barnstable 10 Queen Ann, MD Constructed 4.01 7 Braun Queen Ann, MD Constructed 1.38 7 Dwyer Kent, MD Constructed 1.66 5 Powerline Kent, MD Natural 0.12 Unknown Stoltzfus Kent, MD Constructed 3.68 5 Wood, 03 Kent, MD Natural 0.28 Unknown Figure 1. Locations on the Eastern Shore of Maryland of the constructed and restored wetlands used in our study. gained some insight in the role plant debris may play in deciding which devices may be preferable when conducting RAPs of Delmarva Peninsula wetlands. Field-site Description For this study, we selected 8 wetlands located on the Maryland portion of the Delmarva Peninsula (Table 1, Fig. 1). We chose these wetlands because a previous investigation determined that the range in taxa diversity and relative abundance among them were similar to those found in other natural and constructed wetlands of the Delmarva Peninsula (T.P. Lowe and D.W. Sparling, unpubl. data). Two were Delmarva Bays—elliptical natural wetlands lying in closed depressions Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 325 with no obvious surface-water inlet or drainage channels (Tiner and Burke 1995). One (Wood 3) was located in a relatively undisturbed forest setting, and the other (Powerline) was located ~200 m from Wood 3 and lay partially in the same forested area and partially in a power-line corridor that was dominated by grasses. Cephalanthus occidentalis L. (Buttonbush) and various grasses were the principal vegetation components in the bays and Liquidambar styraciflua L. (Sweetgum) and Acer rubrum L. (Red Maple) saplings occurred along the edges (Phillips and Shedlock 1993). The remaining 6 wetlands were constructed wetlands located on the sites of natural wetlands that were drained many decades ago for agricultural purposes (Tiner and Burke 1995). The presence of hydric soils was used to indicate where each constructed wetland would be placed. The shapes of the constructed wetlands varied according to the outlines of the areas individual landowners chose to convert into wetlands. Most were sited on gently sloping land bounded with low dikes along the lower elevations. Barnstable 1 and Braun had ditches supplying surface inflows, whereas the remainder had no obvious features for supplying inflows. Barnstable 1, Barnstable 3, Barnstable 10, and Braun had ditches or culverts for removing overflow. Both the natural and constructed wetlands had open-water areas and areas that supported aquatic plants (Tiner and Burke 1995). Methods We sampled the wetlands with 3 types of devices: Gerking box-traps (box trap), D-shaped sweep nets (sweep net), and activity traps. The box trap was 1 m long, 40 cm wide, and 50 cm from bottom to top and had a 1-mm-mesh stainless-steel screen floor that we slid shut as we collected samples. The straight side of the D-shaped sweep-net opening was 30 cm, the radius of the curved portion was 15 cm, and the mesh size of the bag net was 1.0 mm. The activity traps were composed of 20.2-cmlong clear plastic cylinders with an inner diameter of 10.3 cm. A clear plastic funnel with a minimum diameter of 1.2 cm was situated on one end of each cylinder, and a removable 12.7 x 12.7-cm flat piece of clear plastic blocked the opposite end. The funnels extended 4.0 cm into the cylinders. We collected samples from the constructed wetlands on 30 and 31 March and 1 and 5 April (Period 1), 27, 28, 29 April and 3 and 5 May (Period 2), 26 May and 3 and 4 June (Period 3), 29 and 30 June (Period 4), and 29 July and 3 and 4 August (Period 5). We collected samples from the natural wetlands on 29 April and 5 May (Period 2), 26 May and 4 June (Period 3), and 29 and 30 June (Period 4). We could not sample the natural wetlands during Period 5 because the latesummer drying process had progressed further than in the constructed wetlands. Seasonal drying precluded sampling at Barnstable 3 during period 5. During each period, we collected samples from a defined zone in the vegetated portion and a defined zone in the open-water portion of each wetland. The zones were ~20 m long and extended 5–7 m away from the water’s edge. Water depth in the zones generally ranged from 15–30 cm. To the extent wetland sizes permitted, we selected different zones during each sampling period. We collected 3 box-trap and 3 sweep-net samples from randomly selected locations within each zone; however, Northeastern Naturalist 326 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 we maintained a 3–4-m distance between the sampling locations within a zone to minimize the possibility that sampling activities at one location would disturb invertebrates at another location. We attached horizontally oriented activity traps to vertical stakes. We placed these traps ~10 cm under the water surface in 4 randomly selected locations in the vegetated and open-water zones of each wetland when we collected box-trap and sweep-net samples. We retrieved the contents of the activity traps 24 h after placement. We collected box-trap samples by dropping the trap vertically with the bottom screen open in an undisturbed area, then slowly closing the bottom screen as the trap rested on the bottom. Approximately 5 cm of bottom sediments and all of the vegetation inside of the trap shifted onto the bottom screen as it was closed. As we raised the trap, we gently swished it back and forth over a distance of ~20 cm to remove as much sediment as possible through the screen. The swishing did not disturb water beyond about 1 m from the sampling location. We retained for processing plant material, sediment that remained in the trap, and captured invertebrates. We collected sweep-net samples by extending the net to ~2.5 m away from the collector, then drawing the net along the bottom towards the collector for a distance of ~2 m. In order to collect both epibenthic and epiphytic invertebrates, the net was bumped along the bottom as it was drawn towards the collector (Cheal et al. 1993, Macan 1977). As with the box-trap samples, we kept for processing vegetation, bottom sediments, and invertebrates. We collected invertebrates captured in activity traps by removing the plastic squares blocking the rear of the traps and pouring the contents through a 1-mm-mesh plastic sieve. We immediately transferred the samples from each device to labeled 79-L plastic garbage bags in portable coolers with frozen cold packs for transport to the laboratory. At the laboratory, we transferred samples to refrigerators where they were held at a temperature of 4 °C until processing, which we usually completed within 24 h. We followed a 2-step procedure to process samples. In the first step, we removed invertebrates from accompanying sediment and vegetation (picking) and preserved them in 80% isopropyl alcohol in separate containers labeled with a preliminary identification for each recognizable taxon (sorting). The volume of plant debris and sediment in many box-trap and some sweep-net samples exceeded the capacity of the sorting pans. Through previous experience, we knew that a pan filled to the rim contained 1.90 L of plant debris and sediment, and that it would take ~1 h to pick each sample. Thus, in order to keep sample-sorting times to ~1 h, we subdivided samples containing large volumes of plant debris and sediment in approximately equal portions until the volume of each portion was no larger than the capacity of the pan. We made the subdivisions by spreading the sample contents on a clean surface as a 2–3-cm-deep rectangular mass, and delineating approximately equal piles until each subsample was about the size of the sorting pan. We assumed that subsamples contained representatives of each taxon present in the original sample and that they were distributed in equal density throughout the sample. We made estimates of the numbers of individuals of each taxon per sample by using the proportion of the sample picked. Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 327 The second step was making final identifications and enumerations of individual invertebrate taxa. We used recognized keys to identify individuals of most classes to the genus level (Peckarsky 1990, Pennak 1989). We identified Collembola, Hirudinea, and Oligochaeta no lower than to class. We treated the subcohort Hydrachnidia as a class, and we did not identify individuals therein to lower levels. There were individuals among the Crustacea, Gastropoda, and Insecta that we could identify to order, and among the Odonata to suborder but not to family, and others that we could identify to family but not to genus. We assigned a generic identity of “unknown” to all individuals identified to family but not to genus. Similarly, individuals that we identified to class, order, or suborder that could not be identified to family were given a family identity of “unknown”. Thus, for the sake of convenience in data analysis, the term “unknown” is treated as the family identity for Crustacea, Gastropoda, and Insecta that could not be identified to family. Also, “unknown” is used as the family identities for Colembola, Hirudinea, Hydrachnida and Oligochaeta represented in this study. We were unable to determine the genus for 9125 individuals or ~27% of the total among these classes and could not identify the family for 373 individuals or ~1% of the total among these classes. We performed statistical analyses in SAS (Version 9.1; SAS Institute, Inc., Cary, NC) software. We developed data summaries for numbers of organisms in each sample (sample abundance) and numbers of families (family numbers) represented in each sample. The distributions of the sample abundance and family numbers and logarithmic transformations of these values were skewed (P < 0.0001). Therefore, we employed the Kruskal-Wallis non-parametric test to determine the optimal sampling time, significance between the different devices, and differences between wetlands. The test involves a 2-step procedure in which individual values of sample abundance and family numbers are ranked and the ranking scores are analyzed using a one-way analysis of variance (ANOVA). We used the Type I sums of squares option along with the least significant difference (LSD) means-separation procedure to further assess the influence of each variable. Tests of differences in attribute values among wetlands included Type I and Type II sums of squares option. We set P ≤ 0.05 to determine significance for the results of all analyse s. The use of the Kruskal-Wallis test of individual variables in the sampling design runs the risk of there being one or more Type I errors because of potential interactions among the variables. We used the Holm-Bonferroni method to modify the rejection criteria for each variable in order to control the possibility of committing Type I errors at our predetermined rejection level (P ≤ 0.05) (Holm 1979). We used the median sample abundance and median sample family numbers for each collection device to prepare graphic data summaries. The proportions of each median representing classes and orders composed of 4 or more families are based on the proportion of the total abundance and family numbers captured with each device. Five insect orders (Ephemeroptera, Hymenoptera, Lepidoptera, Megaloptera, and Trichoptera) were represented by 3 or fewer families. We treated these orders as one group called “insect orders with few families”. We identified as a group called “other non-insect groups” all non-insect classes and orders, excluding Gastropoda. Northeastern Naturalist 328 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 We evaluated the usefulness of the data collected with each device and the data from activity traps combined with the other devices in a Rap-like situation by comparing the values for 5 proportional attributes and 7 richness attributes of samples collected from each wetland. Except for family numbers of Hemiptera, the attributes were described in other studies (Barbour et al. 1996, Gernes and Helgen 2002, Lunde and Resh 2012, Resh et al. 1995). We included Hemiptera because this order had the largest number of families among the orders represented in this study. We collected the samples used to calculate attribute values during period 4. We used Kruskal-Wallis tests including Type I and Type II sums of squares with the LSD means separation test to assess differences for each attribute among wetlands. Results Box-trap samples generally included the greatest amounts of plant and sediment debris; there was no plant debris in any of the activity traps (Table 2). We collected the greatest number of samples that required subdivision among both box-trap and sweep-net samples during the 2nd sampling period and the fewest during the 5th sampling period. The number of box-trap samples requiring subdivision was generally about 3 times the number of sweep-net samples requiring subdivision during all 5 sampling periods. During our study, we collected representatives of 55 identified and unidentified family groups among all macroinvertebrate classes, and a total of 53,985 individuals. Forty-three family groups were in the class Insecta, 9 of which were unknown. Hemiptera included the greatest number (12), followed by Coleoptera (10), insect orders with few families (9), Diptera (7), and Odonata (5). Crustacea included 2 families of Amphipoda, 1 family of Decapoda, and 1 family of Isopoda. There were 4 families of Gastropoda. The taxa that accounted for the greatest proportion of the total numbers collected were in Insecta (27,558), Oligochaeta (19,779), Gastropoda (3577), Crustacea (2195), Hydrachnidae (528), Hirudinea (343), and Collembola (71). We captured representatives of 50 invertebrate families in sweep-nets, 48 families in box traps, and 41 families in activity traps. Median family numbers captured per sample were significantly different among the 3 devices; the sweep nets captured the most families and the activity traps the least (Fig. 2). We captured 19,854 (36.8 % of total) individuals in box traps, 27,101 (50.2 % of total) with sweep nets, and 7020 (13.0 % of total) in activity traps. We were unable to Table 2. The numbers of samples collected with each device that required subdividing for processing because of excessive volumes of plant and sediment debris. Portion of sample processed Total number Trap type Whole Half sample Quarter Eighth of samples Box trap 96 60 42 2 200 Sweep net 169 20 10 0 199 Activity traps 212 0 0 0 212 Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 329 make family identifications for 21,175 (39% of total) individuals, of which 19,779 were Oligochaeta. The number of unidentified individuals among the Insecta was 445 individuals or 1.3% of the total, and among non-insect classes, excluding Oligochaeta, was 951 individuals or 2.8% of the total. Median invertebrate-sample abundances were not significantly different between box-trap and sweep-net samples, but the median activity-trap sample abundance was significantly lower than those determined for the other 2 devices (Fig. 2). We captured members of every class/order combination represented in the study with sweep nets, and members of 5 families—Scirtidae (Coleoptera), Hydrometridae and Salidae (Hemiptera), Corydalidae (Megaloptera), and Phryganeidae (Trichoptera)— were only captured using sweep nets. We captured no representatives of Megaloptera in box traps, and no members of Megaloptera or of Hymenoptera in activity traps. We detected members of an unknown family group in each suborder of Odonata only among box-trap samples. Similarly, we captured members of an “unknown” family group in Ephemeroptera only among activity-trap samples. We captured representatives of 26 invertebrate families (including unidentified families of Collembola, Hirudinea, Hydrachnida, and Oligochaeta) in ≥5% of sweep-net samples, and 19 and 17 families in ≥5% of box-trap and activity-trap samples, respectively (Table 3). More families were represented in greater numbers Figure 2. Median numbers of families (FN) and abundances (A) in samples. The shaded zones on each median reflect the proportion of all the samples collected with each device that were composed of different major taxonomic groups. Orders of few families include Ephemeroptera, Hymenoptera, Lepidoptera, Megaloptera, and Trichoptera. Gastropoda are not included in other non-insect groups. Median family numbers and abundances capped with different letters were significantly different (P ≥ 0.05). Northeastern Naturalist 330 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 Table 3. Taxa represented in ≥5% of samples collected by 1 or more devices. The first number is the number of samples in which the taxon is represented and the second (in parentheses) is the median abundance in those samples. Medians followed by different letters are significantly different (P ≥ 0.05). Gerking box trap Sweep net Activity traps Class or Order and Family (n = 200) (n = 199) (n = 212) Gastropoda Physidae 63 (3.0)AB 72 (5.0)A 50 (2.0)B Planorbidae 37 (4.0) A 47 (2.0) A 15 (1.0)B Hirudinea 36 (2.0)A 34 (2.0) A 13 (1.0)B Oligochaeta 50 (100.0)A 48 (110.0)A 6 (2.5)B Hydrachnida 14 (5.0)A 39 (2.0)A 32 (2.0)A Amphipoda Talitridae 6 (6.0)A 10 (2.5)A 1 (11.0)A Isopoda Asellidae 17 (12.0)A 26 (4.0)A 14 (2.5)B Collembola 10 (3.0)A 10 (1.0)A 1 (1.0)A Odonata Aeshnidae 14 (2.0)A 17 (1.0)AB 7 (1.0)B Libellulidae 26 (2.0)A 40 (1.0)A 17 (1.0)A Ceonagrionidae 14 (2.0)A 35 (2.0)A 13 (1.0)A Lestidae 9 (2.0)A 27 (2.0)A 25 (1.0)A Ephemeroptera Baetidae 8 (2.5)A 30 (3.0)A 14 (1.0)B Hemiptera Belostomatidae 19 (2.0)A 19 (1.0)B 17 (1.0)B Corixidae 142 (3.0)A 169 (2.0)A 168 (3.0)A Notonectidae 77 (2.0)A 86 (2.0)A 61 (1.0)B Pleidae 5 (3.0)A 11 (4.0)A 4 (1.0) B Coleoptera Dytiscidae 134 (2.0)A 148 (2.0)A 219 (1.0)B Haliplidae 12 (3.0)A 22 (1.0)B 11 (1.0)B Noteridae 8 (3.0)A 27 (2.0)B 44 (2.0)AB Unknown 8 (2.0)A 15 (1.0)AB 17 (1.0)B Diptera Ceratopogonidae 33 (2.0)A 57 (2.0)AB 4 (1.0) B Chaoboridae 3 (2.0)A 13 (6.0)A 1 (1.0)A Chironomidae 141 (4)A 185 (4.0)A 30 (2.0)B Tabanidae 13 (2.0)A 17 (1.0)A 1 (1.0)A Unknown 31 (2.0)A 46 (20.0) A 5 (2.0)A of sweep net samples than in those collected with the other devices. In addition, most of the families were represented in fewer activity trap samples than in those of the other two devices; notable exceptions were in the numbers of activity trap samples with Dyticidae, Noteridae, and Coleoptera with unknown family identifications. However, median activity trap sample numbers for these taxa were generally lower than the medians for the other two devices. Median abundances of families captured with the box trap samples were generally higher than or equal to the median abundances of the other two devices. The results of LSD tests generally followed the median sample abundances of each device. Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 331 Overall, the median abundance among activity-trap samples was significantly lower than those of the other 2 devices (Fig. 2). The proportions of the median sample abundance composed of the various class/order combinations were similar among box-trap and sweep-net samples; other non-insect groups represented the greatest proportion due to the high percentage of Oligochaeta in the samples. Diptera composed the next-highest proportion of the median abundance among box-trap and sweep-net samples followed by Hemiptera. The contribution of Diptera to median abundance may have been due to the high numbers of Chironomidae collected in samples from these 2 devices. Odonata and other Insecta comprised the smallest, and about equal, proportions of the median abundance of box-trap and sweep-net samples (Fig. 2). Hemiptera and Coleoptera comprosed the greatest proportions of the median abundance of activity-trap samples. Gastropoda, Diptera, insect orders with few families, and non-insect groups, except Gastropoda comprised the smallest proportions of the median abundance of activity-trap samples (Fig. 2). Odonata, insect orders with few families, and other non-insect groups composed such small proportions of the median abundance among activity-trap samples that they had to be shown collectively under the category “other non-insect groups” in Figure 2. Some investigators have found that combining invertebrates collected with an active collection device, such as a box trap or sweep net, with that of a passive device, such as activity traps, fyke nets, or fine-meshed minnow traps, can yield more complete estimates of taxa richness than using only an active device. Combining samples collected in this fashion may partially circumvent underrepresentation of highly mobile taxa that may escape an active device or that are inactive during daylight hours when sampling with an active device probably would occur (Florencio et al. 2012, Meyer et al. 2011). To determine if better estimates of family numbers and sample abundance could have occurred in this study, we combined the invertebrate data from activity-trap samples with the corresponding numbered box-trap and sweep-net samples collected from each wetland during each sampling period. Combining box-trap and activity-trap data resulted in 50 families represented or 91% of the 55 families represented in the study. This total was 2 and 9 more families more than were represented only among the box-trap and activity-trap samples, respectively. Combining sweepnet and activity-trap data resulted in 53 families represented, or 96.4% of the 55 families observed in the study; a total of 3 and 12 more families than were represented only among the sweep-net and activity-trap samples, respectively. The number of families per sample of the combined box-trap and activity-trap data was significantly lower than that of the combined sweep-net and activity-trap data (data not shown). The per sample abundance of the combined box-trap and activity-trap data also was lower than that of the combined sweep-net and activity- trap data but the difference was not significant. The analysis to determine the optimum sampling period was based on family numbers and showed that median values for samples collected during periods 3–5 were not significantly different (data not shown). However, because seasonal Northeastern Naturalist 332 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 Table 4. Description of attributes employed. Attribute Description Proportional attributes % Gastr. + Crust.1 % of total sample abundance composed of Gastropoda and Crustacea. % EOT4 % of total sample abundance composed of Ephemeroptera, Odonata, and Trichoptera. % Corixidae2 % of sample abundance of Coleoptera and Hemiptera composed of the family Corixidae. % Chironomidae1 % of total sample abundance composed of Chironomidae. % Oligochaeta1 % of total sample abundance composed Oligochaeta. Richness attributes No. Gastr. + Crust.1, 3 Number of families of Gastropoda and Crustacea captured in a sample. No. EOT4 Number of families of Ephemeroptera, Odonata, and Trichoptera captured in a sample. No. Coleoptera1 Number of families of Coleoptera captured in a sample. No. Diptera3 Number of families of Diptera captured in a sample. No. Hemiptera Number of families of Hemiptera captured in a sample. No. Odonata2 Number of families of Odonata captured in a sample. Total no. families2, 3 Total number of families of all classes captured in a sample. 1Barbour et al. 1996. 2Gernes and Helgen 1999. 3Resh et al. 1995. 4Lunde and Resh 2012. drying of 2 wetlands precluded their sampling during period 5, we elected to evaluate wetland differences in invertebrate-attribute values using samples collected during period 4. We detected the largest significant differences among wetlands for 6 proportional attributes among sweep-net samples followed by differences among activity-trap samples and box-trap samples (Tables 4, 5). We detected significant differences among wetlands for more attributes of sweep-net samples than for samples collected with the other 2 devices. However, we observed significant differences among wetlands for more richness attributes among activity-trap samples than for the other 2 devices (Table 5). The number of wetland groups identified with LSD tests of the data from each device suggests that sweep-net samples may have been more sensitive to differences among wetlands. Calculations of the Type I and II sums of squares for all device and attribute combinations were equal to that of the model used in the ANOVAs. Discussion Gerking box-traps and sweep nets represented similar and higher proportions of the abundance and diversity of macroinvertebrates captured than activity traps. However, in terms of using macroinvertebrate attributes to detect differences among wetlands, the performance of the activity traps was similar to that of the other 2 devices. Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 333 Table 5. Differences among wetlands in macroinvertebrate attribute values calculated using Kruskal- Wallis tests of data collected with each device during period 4. Differences were significant if the probability (P) of the associated F statistic was ≤0.05. The least significant difference (LSD) meansseparation test was used to determine the number of wetland groups (Group) with significantly different attribute values. Group number is not given for attributes with P > 0.05. Gerking box trap (n = 40) Sweep net (n = 42) Activity traps (n = 48) Attribute F P Group F P Group F P Group Proportional attributes % Gastr. + Crust. 3.10 0.0161 3 2.67 0.0303 2 2.89 0.0192 2 % EOT 4.14 0.0033 2 2.00 0.0923 0.26 0.9513 % Corixidae 1.96 0.1006 6.66 less than 0.0001 3 3.69 0.0050 3 % Chironomidae 4.74 0.0014 3 3.65 0.0064 3 2.13 0.0702 % Oligochaeta 1.25 0.3083 13.60 less than 0.0001 2 Richness attributes No. Gastr. + Crust. 1.65 0.1654 8.98 less than 0.0001 2 0.83 0.5537 No. EOT 1.10 0.3832 1.10 0.3832 1.18 0.3372 No. Coleoptera 1.82 0.1247 1.52 0.2007 3.52 0.0066 2 No. Diptera 2.17 0.0711 2.26 0.0596 1.95 0.0963 No. Hemiptera 1.19 0.3382 2.11 0.0771 4.24 0.0021 2 No. Odonata 2.66 0.0326 2 2.17 0.0694 0.23 0.9639 Total family no. 1.38 0.2535 4.96 0.0009 4 3.40 0.0082 2 The apparent success of sweep nets in capturing significantly greater diversity than box traps may have been due to our moving the net along longer distances (~2 m) than has been common in other studies. It seems unlikely that a location effect would have influenced the success of the sweep net because we randomly selected sampling locations for all the devices within sampling zones. Water depths were reasonably consistent (15–30 cm), and the vegetation composition and density appeared to be similar within zones and among wetlands. Also the times wetlands were studied and the number of locations within vegetated and open-water areas were consistent throughout the study except for Barnstable 3 in period 4. Box traps, which are 50 cm tall, may collect a greater abundance and diversity of invertebrates than the sweep net when used in wetlands deeper than those we sampled. However, the dimensions of both the box trap and sweep net were adequate to collect from the whole water column of wetlands in this study. The results of other studies, however, have shown that sweep nets may not perform better than other devices when drawn across shorter distances than we covered. In a Florida study comparing the efficacy of 6 sampling devices in 3 types of vegetation (Turner and Trexler 1997), the abundance and diversity of invertebrates collected with sweep nets was less than that captured by activity traps in 3 vegetation types, and was also less than that collected from 1 type of vegetation with the stovepipe sampler. Like the box trap, the stovepipe sampler confines invertebrates in a specific volume of water, although it has about half the capacity. Turner and Trexler (1997) followed a protocol similar to ours in which researchers bounced the sweep net along the bottom to capture epibenthic and near-surface benthos as well as epiphytic invertebrates. However, their sweeps were only 0.5 m Northeastern Naturalist 334 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 in length. In a study of 3 Irish turloughs, investigators compared the performance of sampling with a pond net, similar to the sweep net, and sampling with a box (O’Connor et al. 2004). Like the box traps we used, their box confined invertebrates to a specific volume of water although the area enclosed was ~⅓ that of our box trap. Species diversity was significantly higher in box samples than in pondnet samples from 3 of the turloughs, and total abundance was lower in pond-net samples from 2 of the turloughs and equal to that of the box sampler in samples from the 3rd turlough (O’Connor et al. 2004). The investigators noted that the species richness of Coleoptera was significantly lower among pond-net samples than in the box samples, but they collected greater numbers of Corixidae with the pond net than with the box sampler. In our study, the diversity of Coleopteran families was nearly equal among all 3 devices, while family abundance of Corixidae was greater among sweep-net samples. Investigators in the Irish study swept in 1 direction along a 1-m path, then swept the same path in the opposite direction. They also grazed the bottom as they moved the net along the path. Water depths in most of the wetlands in these other studies (O’Connor et al. 2004, Turner and Trexler 1997) were similar to the depths of our wetlands. In vegetated and non-vegetated habitats of a wetland slough in the Platte River Valley, NE, Meyer et al. (2011) compared the diversity and abundance of invertebrates in samples collected with a D-frame sweep net, stovepipe sampler, and a drop frame. The mesh size of the sweep net and drop frame was 500 μm. The researchers moved the sweep net along a 0.5-m path 1 time as they vigorously agitated the bottom. The sweep net consistently captured lower diversity and abundance of invertebrates in both vegetated and open-water wetland areas than either of the 2 enclosure-type samplers. Water depths in the study of the Platte River slough were not given. In studies comparing different sampling devices used to collect nektonic invertebrates, the sweep net did not appear to have an advantage over either the box trap or activity traps. A Canadian study assessing the abundance and diversity of nektonic invertebrates found no significant differences in the performance of a sweep net drawn vertically from bottom to top through the water column and samples taken with a modified box trap (Kaminski 1981). In another Canadian study, investigators compared the sampling efficacy of sweep nets and activity traps left in place 24 h in capturing nektonic invertebrates occurring in open water and in 4 types of vegetation (Murkin et al. 1983). Murkin et al. (1983) and Kaminski (1981) employed similar sweep-net techniques. The correlation coefficients in the biomass and number of taxa captured with both devices generally were significant. We found that the difference in sample abundance of all invertebrates captured with the box trap and the sweep net drawn horizontally through the water column was not significant and the proportions of the median sample abundance composed of all groups were similar between the 2 devices. The manner in which activity traps are deployed may play a role in how well they perform, although there is limited data available in the published literature to assess its importance. Investigators comparing sampling devices in Florida (Turner Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 335 and Trexler 1997), found that after 24 h, activity traps had captured greater invertebrate abundance than both sweep nets and the stovepipe sampler, and greater species diversity than the stovepipe sampler. These investigators mounted activity traps vertically with the funnel opening touching the bottom substrate in order to capture invertebrates moving upward from the substrate through the water column. Activity traps in this position may not collect as many free-swimming nektonic invertebrates as traps mounted horizontally in the same location. In our study, the box trap and sweep net captured significantly higher invertebrate-abundance and more families than did the activity traps. Brinkman and Duffy (1996) had a similar result in their study comparing box traps with vertically mounted activity traps in South Dakota wetlands. Vertically mounted traps captured greater abundance and species diversity than horizontally mounted traps in a Minnesota study (Muscha et al. 2001). However, in 2 European studies comparing the performance of horizontally mounted activity traps at mid-water depths with those touching the bottom sediment (Elmberg et al. 1992, Hyvonen and Nummi 2000), diversity at the order level was similar to that of our activity traps which were mounted 5–10 cm under the water surface but not touching the bottom. The increases in family numbers represented when we combined activity-trap data with box-trap and with sweep-net data may indicate that a sampling protocol with active and passive sampling devices would be appropriate for RAPs of Delmarva Peninsula wetlands. Investigators in Ireland (Becerra Jurado et al. 2008) found that pond netting combined with activity traps yielded more complete estimates of taxa richness in heavily vegetated and open-water areas of constructed wetlands used for wastewater disposal. Investigators in Spain (Florencio et al. 2013) sampling with dip nets and fyke nets found a total of 38 taxa represented in combined samples from both devices. Sixteen taxa were captured exclusively with the dip net, 5 taxa were captured exclusively with fyke nets and 17 were captured with both devices. For other wetland types, investigators recommend combining activity traps with core samples (Hyvonen and Nummi 2000, Whiteside and Lindegaard 1980) and combining sweep-net with rock-bag samples for conducting rapid assessments of water quality (Muzaffar and Colbo 2002). Our detection of significant wetland differences among a greater number of proportional attributes with box-trap and sweep-net data than activity-trap data, and the reverse with richness attributes may indicate that combining activity-trap data with that of the other devices could increase the number of attributes detecting significant wetland to wetland differences. To test this possibility, we performed a Kruskal-Wallis test on data of the combined devices. Combining the data produced some improvements in our ability to detect differences among wetlands. All of the proportional attributes but none of the richness attributes were significantly different among wetlands with the combined box-trap and activity-trap data. Thus combining the data from these 2 devices increased the total number of significant attributes involving box-trap data from 4 to 5 attributes. Four proportional attributes and 3 richness attributes were significantly different among wetlands with the combined sweep-net and activity-trap data. Thus, combining these data increased Northeastern Naturalist 336 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 the number of significant attributes involving sweep-net data from 6 to 7. There were also some changes in which attributes were significantly different when the data were combined, and F statistics among the proportional attributes were generally larger than those for the individual devices. There also were more LSD wetland groupings with the combined data for the proportional attributes than with the data of individual devices. Thus, although there was some improvement in detecting wetland to wetland differences with these selected attributes, more investigation is needed before a this approach should be used in a sampling protocol. In conclusion, it appears that of the 3 devices we tested, sweep nets may be best for conducting RAPs of Delmarva Peninsula wetlands. The sweep net drawn across ~2 m of the wetland bottom captured representatives of significantly more invertebrate families than did 100 cm x 40-cm box traps or activity traps. Although the median invertebrate sample abundance collected with sweep nets was higher than with the other 2 devices, the differences between sweep-net samples and box-trap samples were not significant. Sweep nets have the added advantage of being easier to manipulate than box traps. One person can carry out sampling with the sweep net, whereas box-trap sampling is difficult unless there are 2 people manipulating the device. Also, sweep nets gather less plant debris than do box traps, thus making sample processing quicker and easier. Of the 3 devices, activity traps are the easiest to manipulate and they produce the easiest samples to process because there is no plant and sediment debris. If a study objective is to evaluate wetland differences using proportional and richness attributes applied to family-level identifications, it appears that data collected from activity trap samples may be more sensitive than data collected with box traps and only slightly less sensitive than that collected with sweep nets. On the other hand, if sufficient resources are available to utilize multiple sampling techniques, combining an activity trap with passive sampling methods may result in greater ability to measure family diversity and a higher sensitivity in determining wetland differences. Acknowledgments This study was supported by the USEPA under account number DWI14937887-01-1. Assistance with collecting and processing samples was provided by seasonal employees hired through a cooperative agreement between PWRC and the University of Maryland. Comments by 2 anonymous reviewers greatly improved this manuscript. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the US Government. Literature Cited Barbour, M.T., J. Gerritsen, G.E. Griffith, R. Frydenborg, E. McCarron, J.S. White, and M.L. Bastian. 1996. A framework for biological criteria for Florida streams using benthic macroinvertebrates. Journal of the North American Benthological Society 15:185–211. Becerra Jurado, G., M. Masterson, R. Harrington, and M. Kelly-Quinn. 2008. Evaluation of sampling methods for macroinvertebrate-biodiversity estimation in heavily vegetated ponds. Hydrobiologia 597:97–107. Northeastern Naturalist Vol. 23, No. 2 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 337 Brinkman, M.A., and W.G. Duffy. 1996. Evaluation of four wetland aquatic-invertebrate samplers and four sample-sorting methods. Journal of Freshwater Ecology 11:193–200. Cheal, F., J.A. Davis, J.E. Growns, J.S. Bradley, and F.H. Whittles. 1993. The influence of sampling method on the classification of wetland macroinvertebrate communities. Hydrobiologia 257:47–56. Elmberg, J.P., P. Nummi, H. Poysa, and K. Sjoberg. 1992. Do intruding predators and trap position affect the reliability of catches in activity traps? Hydrobiologia 239:187–193. Florencio, M., C. Diaz-Paniagua, I. Gomez-Mestre, and L. Serrano. 2012. Sampling macroinvertebrates in a temporary pond: Comparing the suitability of two techniques to detect richness, spatial segregation, and diel activity. Hydrobiologia 689:1221–130. Furse, M.T., J.F. Wright, P.D. Armitage, and D. Moss.1981. An appraisal of pond-net samples for biological monitoring of lotic macroinvertebrates. Water Research 15:679–789. Gernes, M.C., and J.C. Helgen. 2002. Indexes of biological Integrity (IBI) for large depressional wetlands in Minnesota. Final Report to USEPA, Federal Assistance #CD-995525- 01. Minnesota Pollution-Control Agency, Biological Monitoring Program, Environmental Outcomes Division, St. Paul, MN. Holm, S. 1979. A simple, sequentially rejective multiple-test procedure. Scandinavian Journal of Statistics 6:65–70. Hyvonen, T., and P. Nummi. 2000. Activity traps and the corer: Complementary methods for sampling aquatic invertebrates. Hydrobiologia 432:121–125. Kaminski, R.M. 1981. Evaluation of two devices for sampling nektonic invertebrates. Journal of Wildlife Management 45:493–496. Lunde, K.B., and V.H. Resh. 2012. Development and validation of a macro-invertebrate index of biotic integrity (IBI) for assessing urban impacts to Northern California freshwater wetlands. Environmental Monitoring and Assessment 184:3653–3674. Macan, T.T. 1977. The fauna in the vegetation of a moorland fishpond as revealed by different methods of collecting. Hydrobiologia 55:3–15. Menetrey, N., B. Oertli, and J.-B. Lachavanne. 2011. The CIEPT: A macroinvertebratebased multimetric index for assessing the ecological quality of Swiss lowland ponds. Ecological Indicators 11:590–600. Meyer, C.K., S.D. Peterson, and M.R. Whiles. 2011. Quantitative assessment of yield, precision, and cost-effectiveness of three wetland invertebrate-sampling techniques. Wetlands 31:101–112. Milbrink, G., and T. Wiederholm. 1973. Sampling efficiency of four types of mud bottom samplers. Oikos 24:479–482. Murkin, H.R., P.C. Abbott, and J.A. Kadlec. 1983. A comparison of activity traps and sweep nets for sampling nektonic invertebrates in wetlands. Freshwater Invertebrate Biology 2:99–106. Muscha, M.J., K.D. Zimmer, M.G. Butler, and M.A. Hanson. 2001. A comparison of horizontally and vertically deployed aquatic invertebrate activity traps. Wetlands 21:301–307. Muzaffar, S.B., and M.H. Colbo. 2002. The effects of sampling technique on the ecological characterization of shallow, benthic, macro-invertebrate communities in two Newfoundland ponds. Hydrobiologia 477:31–39. O’Connor, A., S. Bradish, T. Reed, J. Moran, E. Regan, M. Visser, M. Gormally, and M.S. Skeffington. 2004. A comparison of the efficacy of pond-net and box-sampling methods in turloughs—Irish ephemeral aquatic systems. Hydrobiologia 524 :133–144. Northeastern Naturalist 338 T.P. Lowe, K. Tebbs, and D.W. Sparling 2016 Vol. 23, No. 2 Oertli, B., D.A. Joye, E. Castella, R. Juge, A. Lehmann, and J-B. Lachavanne. 2005. PLOCH: A standardized method for sampling and assessing the biodiversity in ponds. Aquatic Conservation: Marine and Freshwater Ecosystems 15:665–679. Peckarsky, B.L. 1990. Freshwater Macroinvertebrates of Northeastern North America. Comstock Publishing Associates, Ithaca, NY. 442 pp. Pennak, R.W. 1989. Fresh-water Invertebrates of the United States: Protozoa thru Mollusca. 3rd Edition, Wiley Publishers, New York, NY. 626 pp. Phillips, P.J., and R.J. Shedlock. 1993. Hydrology and chemistry of groundwater and seasonal ponds in the Atlantic Coastal Plain in Delaware, USA. Journal of Hydrology 141:157–178. Plafkin, J.L., M.T. Barbour, K.D. Porter, S.K. Gross, and R.M. Hughes. 1989. Rapid bioassessment protocols for use in streams and rivers. Benthic macro-invertebrates and fish. EPA/440/4-89/001. USEPA, Office of Water Regulations and Standards, Washington, DC. Resh, V.H. 1995. Freshwater benthic macroinvertebrates and rapid-assessment procedures for water-quality monitoring in developing and newly industrialized countries. Pp. 167– 177, In W.S. Davis and T.P. Simon. (Eds.). Biological Assessment and Criteria, Tools for Water Resource Planning and Decision Making. CRC Press, Boca Rat on, FL. 432 pp. Resh, V.H., and J.K. Jackson. 1993. Rapid assessment approaches to biomonitoring using benthic macroinvertebrates. Pp. 159–180, In D.M. Rosenberg and V.H. Resh (Eds.). Freshwater Biomonitoring and Benthic Macro-invertebrates. Chapman and Hall, New York, NY. 488 pp. Resh, V.H., R.H. Norris, and M.T. Barbour. 1995. Design and implementation of rapid-assessment approaches for water-resource monitoring using benthic macro-invertebrates. Australian Journal of Ecology 20:109–121. Sychra, J., and Z. Adamek. 2010. Sampling efficiency of Gerking sampler and sweep net in pond-emergent littoral macrophyte beds: A pilot study. Turkish Journal of Fisheries and Aquatic Sciences 10:161–167. Tiner, R.W., and D.G. Burke. 1995. Wetlands of Maryland. US Fish and Wildlife Service, Ecological Services, Region 5. Hadley, MA. 261 pp. Tolonen, K.T., and H. Hamalainen. 2010. Comparison of sampling methods and habitat types for detecting impacts on lake littoral-macroinvertebrate assemblages along a gradient of human disturbance. Fundamentals of Applied Limnology, Archaeology, and Hydrobiology 176:43–59. Turner, A.M., and J.C. Trexler. 1997. Sampling aquatic invertebrates from marshes: Evaluating the options. Journal of the North American Benthological Society 16:694–709. Verdonschot, R.C.M. 2010. Optimizing the use of activity traps for aquatic biodiversity studies. Journal of the North American Benthological Society 29:1228–1240. Whiteside, M.C., and C. Lindegaard. 1980. Complementary procedures for sampling small benthic invertebrates. Oikos 35:317–3320.