OUP user menu

Pitfall Trap Size and Capture of Three Taxa of Litter-Dwelling Arthropods: Implications for Biodiversity Studies

(CC)
Timothy T. Work, Christopher M. Buddle, Luisiana M. Korinus, John R. Spence
DOI: http://dx.doi.org/10.1603/0046-225X-31.3.438 438-448 First published online: 1 June 2002

Abstract

Cost-effective and ecologically sensitive monitoring techniques are required to assess effects of anthropogenic disturbances on biodiversity. Pitfall trapping is widely used in biodiversity monitoring programs to measure the diversity of organisms active within leaf-litter. We compared catch rates and species richness of ground beetles (Coleoptera: Carabidae), rove beetles (Coleoptera: Staphylinidae), and spiders (Araneae) across five different diameters of pitfall traps (4.5, 6.5, 11, 15, and 20 cm) and three sizes of rain covers (64, 79.2, and 225 cm2) to determine optimal trap size for studying litter-dwelling arthropod biodiversity. In general, larger pitfall traps collected more individuals, and more species, of all three taxa. Further tests on data standardized to trap circumference showed that catch rates are not directly proportional to trap size, and even the smallest traps collected a disproportionately high number of certain taxa. When catch rate data were standardized by trap circumference smaller traps collected more small-bodied carabid and staphylinid species and large traps collected more wolf spiders (Lycosidae) than smaller traps. Roof size had no effect on species richness or catch rate of beetles or spiders. For the purposes of ecological monitoring, using more small pitfall traps would be the most efficient sampling technique to characterize the dominant epigaeic arthropod fauna; small traps collect few nontarget vertebrates, and sorting the samples involves generally less processing time. From a conservation perspective, however, including several large pitfall traps in the sampling regime would help detect rare species.

  • Carabidae
  • Staphylindae
  • Araneae
  • sampling
  • pitfall trap size
  • biodiversity

With increasing interest in human impacts on ecosystems, such as large-scale forestry, land conversion for agriculture, global climate change, and introduction of exotic species, detection of changes in native biodiversity has become critical. Indeed, large-scale programs aimed at monitoring such changes are being incorporated into land management, conservation and restoration activities. Thus, sampling techniques must be studied to ensure both ecological sensitivity and cost effectiveness (Southwood 1994). The usefulness of results from both scientific studies and monitoring are contingent on sampling methods and their inherent biases (Southwood 1994). Consequently, to make relevant comparisons between studies and among disturbance types, biases in sampling methods need to be standardized, or at least understood (Southwood 1994, Spence and Niemelä 1994).

Arthropods are frequently used as ecological indicators because they represent >80% of global species richness (Wilson 1992); fulfill essential roles in ecosystems (e.g., pollination, soil structure and function, decomposition and nutrient cycling, natural enemies of pest species, prey for highly valued vertebrates) (Mattson and Addy 1975, Feinsinger 1983, Seastedt 1984, Pettersson et al. 1995, Madden and Fox 1997, Schowalter et al. 1998); and have short generation times and respond quickly to ecological changes. Furthermore, various arthropod taxa have been used to detect anthropogenic impacts including forest harvesting (Niemelä et al. 1993, Buddle et al. 2000), pollution (Hewlett 2000), mining (Majer 1983), agriculture (Eyre et al. 1989), urbanization (Niemelä and Spence 1991), climate change (Parmesan 1996), and exotic species (Work and McCullough 2000).

Pitfall trapping is among the most frequently used methods to sample surface-active terrestrial arthropod communities. Typically, pitfall traps are open containers sunk into the ground, flush with the substrate surface and covered with a roof to prevent flooding and disturbance (Southwood 1994, Spence and Niemelä 1994). These traps passively collect organisms moving across the ground, and thus provide measures of activity rather than absolute density (Southwood 1994). Numerous studies have examined biases associated with pitfall trapping, and have shown that factors such as trap material, shape, preservative type, and placement influence the number, type and relative abundance of species collected (e.g., Greenslade 1964, Luff 1975, Uetz and Unzicker 1976, Curtis 1980, Bostanian et al. 1983, Niemelä et al. 1986, Topping and Sutherland 1992, Spence and Niemelä 1994, Digweed et al. 1995, Lemieux and Lindgren 1999).

With the exception of Luff (1975) and Brennan et al. (1999), little attention has been given to variation in catch rates in relation to trap size, a factor that frequently varies between studies. Nonetheless, this factor is directly related to cost-effective sampling and implementing effective ecological monitoring. For example, small traps may catch fewer arthropods, thus decreasing sample processing time. Luff (1975) compared the capture and retaining efficiency of several different sizes and materials of pitfall traps for carabid beetles. Luff (1975) found that glass pitfall traps with 6.5 cm diameter caught a greater total number of Coleoptera than did glass traps of 2.5 cm diameter. Brennan et al. (1999) compared abundance and the number of morpho-species collected in a standard number (15) of pitfall traps with diameters of 4.3, 7.0, 11.1, and 17.4 cm. These authors also compared trapping intensity by comparing abundance and morpho-species richness of these same four trap diameters standardized to cumulative circumference of 206 cm. In this study, optimal sized traps were judged as traps that maximized morpho-species richness and minimized overall handling time (including installation of traps, collecting samples, rough sorting, and morpho-species designations) (Brennan et al. 1999). Brennan et al. (1999) concluded that large pitfall traps maximized morpho-species richness of spiders (Araneae) particularly through increased collections of salticid spiders although their study included only a few species of spiders (33) and individuals (138). These authors also concluded that more intermediate sized traps could be processed in a given amount of time than larger sized traps and thus maximized the number of species processed per unit of time. For these reasons and ethical concerns regarding collection of nontarget vertebrate and invertebrate species these authors concluded that 11.1-cm traps were optimal for biodiversity studies.

No previous study known to us has considered simultaneously the effects of trap size on captures of the several main taxa found in forest litter. Nonetheless, monitoring programs will seek the best information possible about whole communities per unit investment. Thus, our primary objective was to compare catch rates of the dominant epigaeic families of beetles (Carabidae and Staphylinidae) and spiders (Araneae) in five different-sized pitfall traps. A secondary objective was to compare the effects of varying roof sizes on catches of beetles and spiders. Our overall goal was to provide information about the best trap size and configuration for monitoring these taxa in boreal forest ecosystems where they have been widely suggested as appropriate indicators (Niemelä et al. 1986, Buddle et al. 2000). We also seek to provide a meaningful basis for comparing terrestrial monitoring studies that have employed a range of trap sizes.

Materials and Methods

Site Description.

This work was conducted at the George Lake Field Site (53° 57′ N, 114° 06′ W), located ≈75 km north-west of Edmonton, Alberta, Canada. The site includes ≈180 ha of continuous hardwood forest (>100 yr old), and is bordered by agricultural land to the south and west, a lake to the east, and continuous younger hardwood forest to the north. The old forest is dominated by trembling aspen (Populus tremuloides Michx.) and balsam poplar (Populus balsamifera L.), with smaller components of birches [Betula papyrifera Marsh. and B. neoalaskana (Sarg.)], white spruce [Picea glauca (Moench) Voss] and black spruce [Picea mariana (Mill.) BSP] (see Niemelä et al. (1992) for additional details). Approximately 4 ha of homogenous Populus forest, located in the southeast section of the field site, was selected for the current study.

Sampling Protocol and Experimental Design.

Epigaeic arthropods were sampled with pitfall traps made of similar white plastic which and were obtained from food-supply companies. Traps were embedded in the ground so the lip of the trap remained flush with the substrate surface. Traps contained 2-3 cm of silicate-free ethylene glycol (GM Dex-Cool, Oshawa, Ontario) as preservative and were covered with an elevated plywood roof held 2-5 cm above the trap by nails placed in the trap corners; lids are commonly used to reduce flooding and accumulation of leaves and litter in the preservative.

Pitfall traps were installed on 25 May 1999 and samples were recovered every 2 wk until 20 July 1999. This period was chosen because it encompassed the period of maximum arthropod activity in these forests (Spence and Niemelä 1994).

Seven different types of traps were used, which varied by roof size and trap size (i.e., circumference and depth). In addition to our ‘standard' trap (a 1-liter container, 11 cm diameter and covered with a 15 by 15-cm square roof), which has been used extensively for epigaeic arthropod studies in western Canada (e.g., Niemelä et al. 1993, Spence and Niemelä 1994, Spence et al. 1996, Buddle et al. 2000), four additional trap sizes were selected for study (Table 1). Trap depth increased in proportion to trap diameter. Trap sizes were chosen to include traps smaller than those typically used in studies of arthropod biodiversity (4.5 cm diameter), a size comparable to those commonly used in studies of carabids in Fennoscandia (6.5 cm traps) (e.g., Niemelä et al. 1986, Pajunen et al. 1995, Niemelä et al. 1996), and two sizes that were generally much larger than those typically used to sample epigaeic arthropods (15- and 20-cm-diameter traps), although McIver et al. (1992) used pitfall traps measuring 15 cm in diameter.

View this table:
Table 1.

For the 4.5-, 6.5-, 15-, and 20-cm-diameter traps, proportion of trap size to roof size in the standard trap was kept constant. Roof sizes were all 2-2.4 times larger in area than the trap area (Table 1). We evaluated the possible effect of roof size using traps with 6.5 cm diameter by including replicates with both a proportionately smaller and a proportionately larger roof size than that based on the standard trap ratio (Table 1).

The resulting seven trap size-roof combinations were randomly assigned to one position in a grid (Fig. 1). Each trap was separated from a neighboring trap by 10-15 m, as suggested by Digweed et al. (1995) to reduce trap-to-trap interference. Ten such grids were distributed throughout the study area located ≈30-40 m apart. Thus, total sampling effort was 70 traps over 56 sample days (3,920 trap-days).

Fig. 1.

Design of pitfall trap-placement (1-7, open circles) within each sampling location,

Species Identification and Measurements.

Samples were sorted in the laboratory and the three most abundant predatory arthropod taxa, ground beetles, rove beetles and spiders, were removed for subsequent identification. These taxa have been used extensively for research on arthropod biodiversity, in particular with studies relating habitat change in this region (e.g., following clear-cutting; McIver et al. 1992, Niemelä et al. 1993, Niemelä et al. 1996, Spence et al. 1996, Hammond 1997, Buddle et al. 2000). On the final sampling date (6 July), vertebrates (mammals and amphibians) were also removed from the samples and counted to establish if trap size affected catches of these taxa.

Specimens from all arthropod higher taxa were identified to species; only adult specimens were examined because it is difficult to accurately identify immature specimens of these groups. Voucher specimens from this study are deposited in the Strickland Entomological Museum (University of Alberta) located in Edmonton, Alberta.

The body size of each species was assigned from information in the literature or by measuring specimens directly. Total body length for most carabid and staphylinid species, respectively, was taken from Lindroth (1969) and Smetana (1995), using the mid-point between upper and lower sizes when a range was given. The size of spiders was estimated from various sources (e.g., Kaston 1948, Dondale and Redner 1990). Size (lengths of carapace plus abdomen) of some spiders was measured directly under a dissecting microscope with an ocular micrometer.

Analyses.

Data were pooled over the entire sampling period because the questions of interest were related to overall response of epigaeic arthropods, as it would relate to inventory studies and to differences in trap size or roof size. One-factor analysis of variance (ANOVA) was used to test for effects in trap size or roof size, with 10 replicates (i.e., locations) for each trap type. For tests of trap size, there were five levels of the factor (see A in Table 1). There were three levels of the single factor for tests on roof size (small, regular, and large roof size, see B in Table 1). Differences in trap size and differences in roof size were analyzed separately.

Dependent variables in the analyses included the total catch for each taxon, catch of the most commonly collected species (i.e., species representing >5% of the total number of individuals collected for each taxon), dominant families of spiders, and species richness for each taxon. Additionally, for each taxon, possible differences among small-, medium-, and large-bodied species were tested using a one-factor ANOVA for tests on trap size only. Size classes were determined by dividing the range (smallest to largest species for each taxon) into equal thirds, and then grouping species as small, medium, or large-bodied. A first set of analyses was performed on raw data, and the analyses were repeated on data standardized to trap circumference.

Luff (1975) suggested that larger pitfall traps will always collect more individuals because the rate of encounter is closely related to the circumference of a trap. To test whether differences in catch rates are due to factors beyond simply the trap circumference, all catch data were standardized to circumference for a portion of the analyses. Standardization was done in relation to the standard pitfall trap size (11-cm-diameter traps, circumference of 34.54 cm). Standardized catch (y) was determined by multiplying the raw catch (r) by the standard circumference (34.54 cm) divided by the circumference of the different trap sizes (c = 14.13, 20.41, 47.1, and 62.8 cm for traps with diameters of 4.5, 6.5, 15, and 20 cm, respectively): y = (r × 34.54)/c.

Data that did not meet assumptions of normality for parametric statistics were transformed to natural logs [Ln(x + 1)] before analyses. Posthoc comparisons of means was done using the Bonferroni multiple comparison test (Sokal and Rohlf 1995) with α = 0.05. SPSS for Windows was used for all ANOVA tests and posthoc comparisons (SPSS 1999).

Species accumulation curves were calculated to establish the effect of increasing sample size (i.e., from 1 to 10 samples) in relation to different trap sizes. The software program EstimateS (Colwell 1997) was used to calculate species accumulation curves for each taxon separately. Estimates of species richness for each sample size were iterated 50 times, and thus standard deviations were obtained for statistical comparisons.

Results

The ‘Catch' and Body Size.

A total of 6,556 individual arthropods was collected in this study. Of these, 2,578 were carabids, 1,542 were staphylinids, and 2,436 were spiders.

Differences in the collection rates between trap sizes may be subject to trap size-species interactions, which are in part determined by factors such as overall body size (e.g., smaller traps may have an upper size limit of species that may be collected). Distribution of body sizes from the overall sample was skewed toward smaller size classes for all three groups of litter-dwelling arthropods (Fig. 2). The range of mean lengths (4.3-22.0 mm) was largest for carabid beetles. The largest size class contained only five individuals from two species, Calosoma frigidum Kirby and Carabus granulatus Linné. The medium size class contained 137 individuals representing only three species, Scaphinotus marginatus Fischer, Pterostichus melanarius Illiger, and Carabus chammisonnis Fischer. The smallest size class was the most numerically abundant with 1,186 individuals from 14 different species.

Fig. 2.

Frequency distribution of species of Carabidae, Staphylinidae, and spiders by body size (mm).

Mean length of staphylinds ranged between 1.7 and 17.5 mm. The largest size class was represented by 17 individuals of a single species, Ontholestes cingulatus (Gravenhorst). Medium sized staphylinids were represented by 535 individuals from nine species. The smallest size class contained 653 individuals from 30 species.

Spiders had the smallest range in body size with mean body length ranging from 1.5 to 11 mm. Within the largest size class, 80 individuals from five species, Clubiona canadensis Emerton, Agelenopsis utahana (Chamberlin & Ivie), Trochosa terricola Thorell, Alopecosa aculeata (Clerck), and Arctosa raptor (Kulczyñski) were collected. Medium sized spiders were represented by 1,160 individuals from 25 species. The smallest size class was comprised of 1,155 individuals from 40 species, 29 of which were in the family Linyphiidae.

Trap Size.

Catch Rates. In general, larger pitfall traps collected significantly more individuals than smaller traps, however, effect of trap size on catch varied among carabids, staphylinids, and spiders (Fig. 3A, C, and E). Significantly more carabid beetles were collected in larger diameter traps (F = 9.44; df = 4, 45; P < 0.001), with 20-cm traps catching more individuals than either 4.5- or 6.5-cm traps [Bonferonni multiple comparison (BMC) P = 0.019 and P < 0.001, respectively]. Fewer carabids were collected in 6.5-cm traps than in either 11- or 15-cm traps (BMC P = 0.001 for both comparisons). More carabids were collected in the smallest traps than in the 6.5 cm traps, although this difference was not statistically significant (BMC P = 0.314). Similar patterns in catch rate were observed for staphylinid beetles (F = 4.81; df = 4, 45; P = 0.003): 20-cm-diameter traps collected more individuals than either 4.5- or 6.5-cm-diameter traps (BMC P = 0.041 and P = 0.003, respectively). However, differences in catch rates of staphylinids were less pronounced than carabids for 6.5-cm traps. No significant differences were observed between 6.5-, 11-, and 15-cm-diameter traps. As with carabids, more staphylinids were collected in the smallest traps than in the 6.5-cm traps, but again this difference was not statistically significant. Differences in catch rates of spiders were more pronounced than for either beetle taxon. Traps of 11 cm or less collected significantly fewer spiders than traps 15 cm or larger (BMC P < 0.002 for all comparisons).

Fig. 3.

Comparison of mean catch (±SE) and mean catch standardized by trap circumference of five different diameter (cm) pitfall traps for Carabidae (A and B, respectively), Staphylinidae (C and D, respectively), and spiders (E and F, respectively). Histograms with different letters are significantly different (BMC, test P < 0.05)

When catch rate was standardized by trap circumference, only carabids and staphylinids showed significant differences in abundance across trap sizes (F = 4.43; df = 4, 45; P < 0.004 and F = 4.33; df = 4, 45; P < 0.005, respectively) (Fig. 3B and D). More carabids were collected per centimeter of trap perimeter in the 4.5-cm trap than either the 6.5- or 15-cm-diameter traps (BMC P = 0.005 and P = 0.036, respectively). More staphylinids were collected per centimeter of trap circumference in the 4.5-cm trap than in traps with diameters of 11 cm or larger (BMC P < 0.03 for all comparisons). No significant differences in catch rates of spiders were observed among different traps sizes once catch rate was standardized by trap circumference (Fig. 3F).

When catch rates were analyzed separately by body size, differences in trap size reflected the lower limit of trap size needed to collect larger-bodied species. For example, all five individuals in the largest size class of carabids were collected in the 15-cm-diameter traps (Fig. 4A). Medium-sized carabids were significantly more abundant in traps larger than 11 cm than in the 4.5- and 6.5-cm-diameter traps (F = 8.81; df = 4, 45; P < 0.001, BMC P < 0.05) (Fig. 4A). Small carabids [e.g., Platynus decentis (Say), Calathus ingratus Dejean, Agonum retractum Leconte, Pterostichus pennsylvanicus Leconte] were also more abundant in larger sized traps (F = 8.70; df = 4, 45; P < 0.001) but were collected in relatively high abundance in all trap sizes (Fig. 4A): more individuals were collected in 20-cm traps than in 4.5- or 6.5-cm-diameter traps (BMC P < 0.001 and P = 0.045, respectively). Fewer small carabids were collected in the 6.5-cm than in the 11- and 15-cm traps (BMC P = 0.001 and P = 0.002, respectively). When abundance of medium sized carabids was standardized by trap circumference, observed differences in trap size were no longer apparent (Fig. 5A). However, when abundance of small sized carabids was standardized by circumference, the smallest trap collected significantly more individuals than all other trap sizes with the exception of 11-cm-diameter traps (F = 5.35; df = 4, 45; P < 0.001, BMC P < 0.01).

Fig. 4.

Comparison of mean catch (±SE) of three size classes of (A) Carabidae, (B) Staphylinidae, and (C) spiders collected in five different diameter (cm) pitfall traps. Size classes were determined by dividing the range in body size of a given taxon into three equal fractions.

Fig. 5.

Comparison of mean catch (±SE) of three size classes of (A) Carabidae, (B) Staphylinidae, and (C) spiders standardized by circumference of five different diameter (cm) pitfall traps.

Staphylinids in the largest size class were collected in all but the 4.5-cm-diameter traps (Fig. 4B). Medium-sized individuals were significantly more abundant in the 20-cm traps than in 4.5- and 6.5-cm-diameter traps (F = 6.66; df = 4, 45; P < 0.001, BMC P = 0.001 and P = 0.005, respectively). Likewise, the 11 cm trap collected more medium sized individuals than the 4.5-cm trap (BMC P = 0.026). Size of pitfall traps also significantly affected capture rate of small sized staphylinids (F = 2.62; df = 4, 45; P = 0.048), although variability in abundance of small staphylinids was greater than medium or larger sized staphylinids (Fig. 4B). As a consequence, posthoc comparisons of trap size showed no significant differences in unstandardized capture rate for small sized staphylinids. Differences in catch rates of medium sized staphylinids were not significant after data were standardized to trap circumference (Fig. 5B). Small sized staphylinids [e.g., Tachinus fumipennis (Say) and Quedius rusticus Smetana] were more abundant in the smallest trap once abundance was standardized by circumference (F = 4.73; df = 4, 45; P < 0.003, BMC P < 0.03 for all comparisons).

Fifteen- and 20-cm-diameter traps collected more individual spiders than 4.5- and 6.5-cm-traps for all three size classes (Fig. 4C). Within the largest size class, more individuals were collected in traps 11 cm or larger than in smaller traps (F = 18.10; df = 4, 45; P < 0.001, BMC P < 0.02). The same pattern was observed for medium sized spiders (F = 15.09; df = 4, 45; P < 0.001, BMC P < 0.02). Pitfall traps >15 cm diameter collected more small sized spiders than 4.5-, 6.5-, or 11-cm traps (F = 12.12; df = 4, 45; P < 0.001 BMC P < 0.05). When spider size classes were standardized by trap circumference only abundance of small spiders differed by trap size (F = 12.12; df = 4, 45; P < 0.001), with 4.5- and 6.5-cm-diameter traps having higher relative efficiency than 11-cm traps (BMC P < 0.03 for both comparisons) (Fig. 5C).

Of the spiders collected, 33% were in the family Linyphiidae and 25% were in the family Lycosidae. Fewer linyphiids were collected in 11-cm-diameter traps then in 6.5-, 15-, and 20-cm diameter traps (F = 9.63; df = 4, 45; P < 0.001, BMC P < 0.04 for all comparisons). The smallest traps (4.5 cm) also collected fewer linyphiids than did the 20-cm-diameter traps (BMC P < 0.02). Linyphiids were infrequently collected in 11-cm-diameter traps after standardization for trap circumference (F = 10.11; df = 4, 45; P < 0.001, BMC P < 0.02 for all comparisons). Although differences were not statistically significant, linyphiids were most commonly caught in 4.5-cm-diameter traps after standardization for trap size. Catches of lycosids increased with trap diameter (F = 30.06; df = 4, 45; P < 0.001). After standardization to circumference, fewer lycosids were collected in 4.5- and 6.5-cm-diameter traps than in 15- and 20-cm-diameter traps (F = 8.54; df = 4, 45; P < 0.001, BMC P < 0.005 for all comparisons).

The 17 commonly collected species differed in their affinities to trap size (Table 2). For example, four species were collected in higher than expected abundance in small trap sizes after standardization to perimeter length. Only one species, Pardosa mackenziana (Keyserling), showed a strong affinity for large traps after standardization. Three species were disproportionately underrepresented in traps of intermediate size: C. ingratus was seldom collected in 15-cm-diameter traps, P. decentis occurred infrequently in 6.5-cm-diameter traps, and Bathyphantes pallidus (Banks) was notably underrepresented in our standard pitfall trap (11 cm diameter) (Table 2).

View this table:
Table 2.

Species Richness. A total of 124 species were identified, comprising 21 carabid, 33 staphylinid, and 70 spider species. Species richness was significantly affected by trap size for all three groups of litter arthropods (carabids: F = 5.36; df = 4, 45; P = 0.001, staphylinids: F = 4.80; df = 4, 45; P = 0.003, spiders: F = 15.42; df = 4, 45; P = < 0.001) (Fig. 6). Mean species richness of carabids was lower in 6.5-cm traps than either 15- or 20-cm-diameter traps (BMC P = 0.013 and P = 0.003, respectively), although the range in mean species richness was only two species. Mean staphylinid species richness was lower in 4.5- and 6.5-cm traps than in 20-cm-diameter traps (BMC P = 0.013 and P = 0.041, respectively) and the range in mean species richness was marginally greater than that observed with carabids. Differences in species richness were most apparent among spiders with 4.5-, 6.5-, and 11-cm traps having significantly lower richness than the 15- and 20-cm traps (BMC P < 0.002 for all comparisons).

Fig. 6.

Comparison of mean species richness (±SE) of (A) Carabidae, (B) Staphylinidae, and (C) spiders collected in five different diameter (cm) pitfall traps. Histograms with different letters are significantly different (BMC, test P < 0.05)

When species accumulation curves were plotted for each trap size, larger traps generally collected more species per sample. However, the increase in species accumulation was not a simple linear relationship with increasing trap size. For carabids, significant differences between trap sizes were apparent only after five samples (Fig. 7A). Fifteen-centimeter-diameter traps had the greatest species accumulation after 10 samples, and 20-cm-diameter traps did not differ from 4.5-cm-diameter traps in species accumulation after 10 samples. Eleven and 6.5-cm traps had the lowest rates of accumulation. Twenty-centimeter-diameter traps had the greatest rates of staphylinid species accumulation and 4.5-cm traps had the lowest rates of accumulation, but traps between 6.5 and 15 cm accumulated species at approximately the same rate (Fig. 7B). For spiders, traps 15 cm or greater in diameter had the highest species accumulation rates (Fig. 7C). Species accumulation in 20-cm traps could only be distinguished from that in 15-cm traps after eight samples. Traps smaller than 15 cm accumulated species at approximately the same rate.

Fig. 7.

Comparison of species accumulation curves of (A) carabid, (B) staphylinid, and (C) spider species for five different diameter (cm) pitfall traps.

Although common species were collected in every size of pitfall trap (Table 2), locally rare species (i.e., those represented by five or fewer individuals) were better represented in large traps than in small traps. Eleven-centimeter traps collected the fewest number of rare species (8 species), followed by 6.5-cm traps (9 species), and 4.5-cm traps (11 species). In contrast. 15- and 20-cm traps collected 36 and 47 rare species, respectively.

Roof Size.

When trap size was held constant and captures were compared under three different sizes of roof, no significant differences were observed in relative abundance or species richness. Likewise, no species-specific responses to roof sizes were observed.

Nontarget Species.

Nontarget mammals and amphibians were only collected in 11-, 15-, and 20-cm traps (Table 3).

View this table:
Table 3.

Discussion

Although sampling arthropods with pitfall traps has inherent biases (e.g., Uetz and Unzicker 1976, Southwood 1994, Spence and Niemelä 1994), this simple method is still frequently used as a major tool in ecological monitoring and biodiversity studies. As a passive sampling method, collection rates of pitfall trapping may be subject to a variety of factors including trap size (Luff 1975, Brennan et al. 1999), habitat structure (Greenslade 1964), temperature (Adis 1979), daily (Adis 1979) and seasonal activity patterns (Niemelä et al. 1992), as well as behavioral characteristics and body size of litter organisms (Halsall and Wratten 1988). In general, pitfall collections of arthropods are measures of activity rather than density, so interpretations of catches as a reflection of absolute abundance is not justified. Baars (1979), however, has suggested that whole-season records support reliable statements about relative abundance, at least for carabids. Furthermore, Raworth and Choi (2001) have recently shown that pitfall data may provide reliable density estimates for single species if they are adjusted for hunger levels and temperature.

If trap size were the only factor affecting collection rates, species composition among individual traps of different sizes should be indistinguishable as should relative abundance once catch rates are standardized by trap circumference (Luff 1975). Our results, using three dominant epigaeic arthropod taxa, suggest otherwise. Larger diameter traps collected more beetle individuals (carabids and staphylinids) than smaller traps, and significant differences in catch rates following standardization by trap circumference suggested this effect was not solely due to an increase in trap size. Clearly there were interactions between species identity and trap size. Although no differences in spider abundance were observed once catch rate was standardized by trap circumference, this pattern largely reflected the combination of higher numbers of lycosids in large traps and higher numbers of linyphiids in small traps. Lycosid spiders were trapped more commonly with large diameter pitfall traps, whereas more linyphiids were caught in small diameter pitfall traps. Thus, for all taxa studied examined in this study, the generalization that catch rates of litter-dwelling arthropods is solely, or even mainly, due to trap circumference must be rejected.

Compositional differences between catches of different sized traps can be attributed in part to body size. Large bodied carabid and staphylinid were generally not captured in traps ≤6.5 cm in diameter. The limited abundance of large bodied species in the samples collected here, however, precludes meaningful statistical comparisons of trap sizes. Higher catch rates of larger species in larger traps may reflect the positive correlation between body size and motility (Luff 1975, Thiele 1977), but additional studies have revealed no relation between body size and motility in the laboratory (Halsall and Wratten 1988) or in the field (Wallin and Ekbom 1994). Future work is required to link behavior and motility of litter-dwelling arthropods in relation to trapability.

After catch rates were standardized by the size of pitfall trap, medium sized beetles showed no affinity for different sized traps but small beetles were disproportionately captured in the smallest traps. This observation is contrary to previous studies that demonstrated small beetles are more likely to escape from smaller traps of the same material (Luff 1975). However, in the same study Luff (1975) demonstrated that small traps constructed of glass rather than plastic had increased efficiency for small beetles, presumably by limiting beetle escape. One possible explanation for the increased relative efficiency of small traps for small-bodied beetles could be that the beetles cannot perceive these traps as ‘different than background' environmental heterogeneity. Species that use soil features such as cracks and fissures as oviposition sites or resting places may be preferentially collected in small pitfall traps.

In general, larger traps collected more spiders of all sizes. However, after standardization to trap circumference, smaller traps were more efficient in collecting many small-bodied species, and the majority of these species were in the family Linyphiidae. These results differ from those of Topping (1993), who suggested that behavioral adaptations enable linyphiids to escape small pitfall traps though a combination of movement and deposition of drag-lines. Litter-dwelling linyphiids typically live among the leaf-litter matrix on the forest floor (Huhta 1971), and they are thus thought to have relatively specific micro-habitat requirements. As suggested for carabids, linyphiids in our study area may not be able to distinguish small pitfall traps from background environmental heterogeneity.

Wolf spiders (Lycosidae), in contrast, were more commonly collected in larger pitfall traps after standardization to trap circumference. Wolf spiders are active on the forest floor substrate and have been known to travel upwards of 10 m in a single day (Hallander 1967). When actively moving across the forest floor, wolf spiders may simply run purposely into a large pitfall trap, yet may somehow stop themselves or avoid falling into small traps.

Although species richness of beetles was significantly greater in larger traps, these differences were relatively small (i.e., 2-4 species). The most commonly collected species were represented in all sizes of traps, whereas rare species were better represented in large traps. These observations suggest that the overall epigaeic arthropod assemblages collected by small traps are predominately a subset of the assemblages collected in large traps.

The relationship between trap size and sampling effort was not sufficient to explain differences in species accumulation between trap sizes. For carabid species, accumulation curves for both the smallest and the largest trap sizes were equivalent, and 15-cm traps rather than 20-cm traps, had the highest species accumulation rate. Species accumulation curves for staphylinids and spiders showed that the largest trap size accumulated species most quickly, although rates of species accumulation are not linearly associated with trap size. Staphylinid species accumulation curves differed only by two species for traps between 6.5 and 15 cm. Increased species richness of staphylinids in 20-cm traps is largely attributed to increases in singleton species. Nonlinear increases in spider species richness have been reported elsewhere and have been attributed to large species avoiding small traps, behavioral differences in locomotion between different trap sizes, and/or decreased efficiency in retaining due to leaf "ladders" that permit escape from smaller traps (Brennan et al. 1999). However, our data show that large-bodied spider species are collected as frequently in small and large traps after standardization to trap circumference. Thus, large spiders do not appear less prone to capture in smaller traps, and their lower overall rate of capture in smaller traps may simply be a reflection of trap circumference.

In conclusion, increases in sampling effort were on a per-trap basis. Similar results (i.e., increased accumulation of species) were reported when sampling effort was incrementally increased through additional traps (Niemelä et al. 1986). Consequently, it may be possible to characterize species composition of the dominant and active litter fauna using a relatively limited number of small sized traps.

The optimal pitfall design for ecological monitoring strongly depends on the ultimate objectives of study. From a pragmatic perspective, sampling design must be cost-effective, manageable, and able to detect meaningful ecological differences. Our results suggest that at a coarse level, small sized pitfall traps characterize the dominant fauna active within the litter as well as larger traps although they require substantially less work processing and identifying samples. Currently, many forest monitoring programs overlook terrestrial invertebrates as ecological indicators in favor of larger vertebrate taxa (Dourojeanni 1990, T.T.W., unpublished data), even though their use in monitoring and conservation, and their ecological importance, cannot be overstated (e.g., Kremen et al. 1993).

From a monitoring perspective, larger traps were more effective at characterizing rare elements of an epigaeic fauna. It is not clear whether rare species that were collected in large traps would be better represented by complimentary sampling methods rather than more intensive sampling effort. Our study suggests, however, that larger numbers of small traps could be effectively employed to sample a greater range of habitats while keeping the overall sorting and processing costs reasonably low. Monitoring programs might effectively couple use of a small number of larger traps, intended mainly to document the presence of larger-bodied species, with a more extensive battery of smaller traps aimed at sampling as many microhabitats as possible.

Large pitfall traps, however, should be used sparingly and with caution as the number of ‘nontarget' mammal and amphibian species substantially increased in larger sized traps. Studies targeting specific taxa (e.g., linyphiid spiders) or species [e.g., P. pennsylvanicus and Lepthyphantes intricatus (Emerton)] will also benefit from using small versus larger traps. Furthermore, certain sized species (e.g., large-bodied carabid beetles such as C. frigidum and lycosid spiders) are better collected using larger pitfall traps. Thus, decisions about the most appropriate trapping methods, including different traps sizes, should precede the design of monitoring and/or biodiversity studies.

Acknowledgments

We thank Joshua Jacobs and Karen Cryer for their help in setting out pitfall traps at the George Lake Site and for sorting of specimens.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/), which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions{at}oup.com

References

View Abstract