Netting Outperforms Pan Trapping for Surveying Bees in Temperate North American Meadows
Info: 12001 words (48 pages) Dissertation
Published: 10th Dec 2019
Tagged: BiologyEnvironmental Science
Title: Catch me if you can: Netting outperforms pan trapping for surveying bees (Hymenoptera: Apiformes) in temperate North American meadows
Abstract
When planning surveys of biological communities, researchers must make trade-offs between the cost and the completeness of their inventory, and the impact of sampling on focal taxa. The most commonly used methods for surveying bee communities are pan trapping and netting. However, there is debate among pollination ecologists over whether pan trap samples adequately represent a bee fauna and can be used in place of netting to answer questions about bee habitat use and to inform conservation decisions. In this study, I compare the diversity and genus assemblage of bees collected by net versus pan traps on 10 reclaimed coal mines and 3 restored park meadows in Ohio, USA. Over summers 2015-2016, I collected 12,795 bees from 32 genera. Netting out-performed pan trapping in rarefied diversity of bee genera collected, and accumulated genera faster with successive sampling. Pan trap yield was more variable across sample dates, more sensitive to the availability of flower resources, and added only two genera absent from net samples. Therefore, net data, or considering net plus pan together, was more informative for assessing differences in bee communities between sites based on management and site history. The most cost-effective strategy for surveying bees in this study system was to net and pan trap simultaneously, because the per bee cost was the same as for netting alone. Catch-and-release net surveys by trained observers, with supplemental pan trapping to extend the scope of the survey, is a more conservative approach to rapidly assessing bee communities than traditional high-mortality surveys.
Abstract word count: 250
Keywords: pollinator conservation; net survey; pan trap; bee bowl; bee community ecology
Introduction
Bees are important pollinators that are under threat from multiple anthropogenic stressors including habitat loss and fragmentation and increased exposure to pathogens and pesticides (Cane and Tepedino 2001; Brown and Paxton 2009; Potts et al. 2010; Roulston and Goodell 2011). However, high variation in bee yield, in both abundance and taxon composition, between sampling events makes it difficult to determine the extent and severity of bee declines (Bartomeus et al. 2013; Burkle et al. 2013) and to estimate the consequences thereof for pollination services. Without adequate information on the status of wild pollinator communities, management agencies may not as readily allocate funds to pollinator conservation. There is on-going debate among pollination ecologists over which techniques are the most effective and efficient for surveying bees and which produce repeatable results (Cane et al. 2000; Westphal et al. 2008; Droege et al. 2010; Grundel et al. 2011; Nielsen et al. 2011; LeBuhn et al. 2013; Gibbs et al. 2017). The goal of a biodiversity survey is not always simply to determine the fastest, easiest, most inexpensive way to reach a complete species inventory. Rather, the ideal survey method should be tailored to the ecological question the research is addressing, the habitat type, and the time of year. In deciding when, where, and how often to sample, researchers must make trade-offs between the cost of materials and person-hours and the completeness of their inventory, while also considering the impact of sampling on bee populations (Westphal et al. 2008; Grundel et al. 2011; Nielsen et al. 2011; Gezon et al. 2015; Gibbs et al. 2017).
Methods of surveying insect pollinators include both passive (pan trapping, malaise traps, vane traps, trap nesting, video recording) and active (netting, observation) techniques.
Passive trapping methods have the advantage of allowing the researcher to be, in effect, “in many places at once.” They are also known for yielding high volumes of specimens while minimizing research effort and bias. The most commonly used passive collection method is to place colored bowls of soapy water, called “pan traps” or “bee bowls,” near foraging habitat (Droege et al. 2005; Droege et al. 2010; Droege 2015). Malaise traps (McCravy et al. 2016) and vane traps (Kimoto et al. 2012; Gibbs et al. 2017; Griffin et al. 2017) have also been used to passively survey insect pollinator communities. However, pan trapping is more widely applicable and the most cost effective because it requires almost no specialized equipment.
Collection methods in which the researcher actively seeks out pollinators through observation and netting have a major advantage in that they provide valuable ecological information about plant-pollinator interactions that pan trapping cannot. Further information can be gleaned about bee resource use if the researcher identifies the pollen the bees were carrying as well (Alarcón et al. 2010). Pan trap specimens are often more difficult to identify than net-collected ones (unless they are cleaned and dried, as per instructions in Droege et al. 2015), because bees’ hair becomes matted when soaked in the killing agent, which can change its appearance and color.
On the other hand, active collection is more prone to research bias than passive surveying methods (Westphal et a. 2008). Net surveys are limited by the skill of the observer at netting and at sight-identifying insects on the wing and by the keenness of his or her eyesight, as well as by the size of the bee and its foraging behavior. Therefore, netting tends to promote the collection of larger bees that are more easily detected, captured, and removed from the net (e.g. Andrena, Anthidium, Apis, Bombus, Megachile), as well as those with specialist foraging (Ptilothrix) or nesting behaviors (e.g. the spring aggregate-nesting Colletes inaequalis, and cleptoparasiteslike Coelioxys, Epeolus, Triepeolus) (Cane et al. 2000; Toler et al. 2005; Grixti and Packer 2006; Roulston et al. 2007; Wilson et al. 2008; Grundel et al. 2011; Richards et al. 2011). Meanwhile, pan traps typically collect larger proportions of the smaller less conspicuous taxa that are more likely to go unnoticed by observers, like those in the family Halictidae (Droege et al. 2010). Bumble bees (Bombus spp.) and honey bees (Apis mellifera), two major bee pollinators that make up a large portion of plant-pollinator interactions in bee surveys in the temperate northern and eastern region of North America (Grixti and Packer 2006; Grundel et al. 2011; Richards et al. 2011; Burkle et al. 2013; McLeod 2013; Tonietto et al. 2017; Griffin et al. 2017;) are not caught in pan traps in proportion to their abundance and importance. Therefore, if the goal of the survey is to assess the functional status of a bee community, pan trapping alone does not adequately represent the bee community or add to our understanding of the quality of pollination services it could provide.
While worldwide many pollination ecologists advocate for pan trapping as an inexpensive means of assessing the status of bee communities that is standardizeable and comparable across researchers and study systems (Westphal et al. 2008; LeBuhn et al. 2013), netting in many cases yields greater diversity of bees and provides ecological information about plant-pollinator interactions (Cane et al. 2000; Roulston et al. 2007; Kwaiser and Hendrix 2008). Where flower resources are scarce and/or patchily distributed in space and time pan trapping has proven more or equally effective to netting (Wilson et al. 2008; Popic et al. 2013). However, in situations where flower resources are more homogenous in space and time (compared to xeric or tropical forest ecosystems), netting is expected to yield greater bee diversity (Grixti and Packer 2006; Roulston et al. 2007; Richards et al. 2011; McLeod 2013). In order to maximize yield and productivity, researchers often employ a mix of pan trapping and netting when assessing bee community response to habitat quality (Hopwood 2008; Nielsen et al. 2011; Williams 2011;Richards et al. 2011; Tonietto et al. 2017). However, if one method or the other will give a clear answer to ecological questions, it may be an unnecessary expenditure of researcher time and loss of bees to perform both pan trapping and netting.
As the collection bias between net versus pan trap sampling is detectable at higher taxonomic levels than species (Cane et al. 2000; Grixti and Packer 2006; Roulston et al. 2007; Wilson et al. 2008; Grundel et al. 2011; Richards et al. 2011), genus level analysis may be sufficient for addressing questions about sampling methodology and pollinator ecology. If within a genus, the majority of species share similar foraging and nesting requirements, then at the community level species diversity may be inflated compared to ecological or niche diversity. If that is the case, species-level analysis may accord disproportionate weight to highly speciose genera or families. On the other hand, genus-level analysis can reduce the number of singletons in a data set and increase statistical confidence, and, thereby, aid the researcher in understanding broader trends in pollinator communities. Bee identification to genus can also often be completed within the same day or soon after field sampling, providing land managers and researchers with more immediate information on how pollinators are responding to recent changes in foraging and nesting habitat, whereas it may take even trained bee biologists several years to identify all individuals to species in a large dataset. Since most bees can be sight-identified to genus in the field by trained observers, non-lethal netting surveys may provide rapid, reliable information about pollinator communities and habitat use without the high mortality incurred by intensive net or pan trap collecting.
Herein, I compared the diversity, abundance, and genus composition of bees caught by pan trapping versus netting over the course of two summers at 13 meadows in southeastern Ohio, USA, in order to evaluate the following questions:
- Which method yielded greater bee genus diversity, and had a greater rate of taxon accumulation?
- How did pan trap and net samples differ in bee genus composition?
- Which method yielded a more stable bee assemblage across sample dates?
- Which method was more sensitive to flower resource availability?
- Which method was preferable for detecting differences in bee genus diversity and assemblage by site history and management characteristics?
- Which method was more cost effective?
Methods
Study system
In summers 2015 and 2016, I surveyed pollinators and wildflowers on ten former coal mine sites that had been previously reclaimed to grassland (Table 1). Sites were selected using the Abandoned Mine Land Inventory System (Office of Surface Mining Reclamation and Enforcement 2014), a federal open-access database of mining and reclamation records. Most were privately-owned and mined prior to the passage of federal reclamation standards mandated by SMCRA (1977). They have since been reclaimed to grassland by state natural resource agencies with funds from the state and federal Abandoned Mine Land programs (Office of Surface Mining Reclamation and Enforcement 2017). The majority of these reclaimed mines were originally planted with a cost-effective aggressively-growing mixture of non-native European pasture grasses and herbaceous legumes (Lotus corniculatus and Trifolium spp.), and have been unmanaged since and colonized by a variety of other native and non-native plant species. In the second year of the study, I also sampled three managed early-successional meadows in nearby park districts. All park meadows had been seeded with a mixture of native wildflowers at least two years prior to sampling and were mowed periodically to prevent reversion to shrub land (JL personal communication with Franklin County Metro Parks park managers). Study sites were located in Franklin, Fairfield, Perry, Hocking, and Muskingum Counties, in southeastern Ohio, USA. They ranged in size from 6.2 to 52.8 ha (mean 15.6 ± 12.7 SD ha). The landscape in this study region is heavily forested, consisting of 72.4 ± 19.4% (SD) forest, 14.8 ± 12.5% hay/pasture lands plus row crops, and 10.9 ± 12.4% developed lands within a 2 km buffer radius area around each site center (from analysis of modified National Land Cover Data set, Fry et al. 2011, in ArcGIS software, ESRI 2016).
Flowering plant surveys
Plant and pollinator surveys were conducted approximately once per month from May – August 2015 and 2016 in two habitat types at each site: (1) the field (either reclamation grassland for former mine sites, or planted wildflower meadow for park sites) and (2) along a 25m-wide edge zone between the field and adjacent forest (hereafter “edge”). In preliminary plant surveys I found more diverse flower species assemblages along the field-forest interface that differed from bee forage plants offered in the field, with some overlap (e.g. Lotus corniculatus, Trifolium spp.). Therefore, plants and pollinators were observed separately, but on the same day, in edge and field habitats at each site. Edge and field data were later combined for analysis.
During each visit, richness and abundance of flowering species were recorded along two 100m x 1m transects, one in the field and one along the edge. Because most sites were privately owned, transects could not be permanently marked, but instead were run haphazardly each month within an area representative of that habitat type. For plant species flowering in heads or with other compound inflorescence types (e.g. Apiaceae, Asclepiadaceae, Asteraceae, Lamiaceae, and Fabaceae), floral units were counted in “bee-walkable” clusters, where one unit consisted of the flowers a bee could reach by walking before it would have to fly to another flower cluster (Saville 1993). Only fresh, open flowers presenting pollen or nectar were counted. Each flowering species was assigned an overall abundance score in the transect, ranked 1-5 on a quasi-log scale of flower abundance (1 representing 1-10 flowering stems, 2: 11-100, 3: 101-500, 4: 501-1000, and 5: >1000) to approximate its resource value to pollinators. Plant species were identified in the field using a regional wildflower guide (Newcomb 1989), or vouchered and later identified using Gleason and Cronquist’s Manual of Vascular Plants (1991).
Pollinator surveys
On each sample date, pollinators were simultaneously surveyed using passive (pan traps) and active (timed netting sessions) methods. Upon each site visit I deployed thirty 3.25 oz Solobrand P325 white plastic soufflé cups that had been painted either UV-bright blue or yellow, or left white (10 of each color) (Leong and Thorp 1999; Toler et al. 2005; Wilson et al. 2008; Droege 2015). Cups were filled with a solution of 5 mL original blue Dawn dish soap per 4 L of tap water (Droege et al. 2010), and placed on cleared patches of level ground at 1 – 2 m intervals, alternating bowl colors, along the access road at each reclaimed mine site or along a public walking trail at park sites. Overhanging vegetation was removed in a circular area around each cup. The soap detergent decreases the surface tension of the water such that insects once in the water cannot escape. Nearby flowering-visiting insects (including bees, flies, wasps, beetles, and butterflies) that attempt to land on the artificial “flower,” fall into the water, and drown (Droege et al. 2005). Traps were left in place for at least 4 hours, after which pollinators were strained from the water using a fine mesh sieve and transferred to paper towels in re-sealable plastic zipper bags for later preservation and identification in the lab.
On each sample date, pollinators were actively sampled by aerial netting of pollinating insects for one hour each in the field and edge habitats (total of two hours per site per sample date), not including handling time of transferring bees to vials. All netting was done by the same observer (JL) to reduce the collection bias that can occur between multiple observers. I searched an approximately 100m radius circular area (Goulson and Darvill 2004) and recorded or netted all insects coming into contact with plant reproductive structures, and recorded the flowering species on which they were observed. Sampling was focused on the most abundant and attractive plant species for bees in the area, but each flowering species with more than 10 stems was observed for a minimum of three minutes during the one-hour sample period. This flexible sampling approach allowed me to account for the patchiness of flowers in the habitat and to catch more pollinators than if I had restricted sampling to a [smaller] strictly-defined area, such as a 1m x 1m quadrat, or along a standardized belt transect (Nielsen et al. 2011). Because flowers were scarce in the forest understory for the majority of the study duration, I did not conduct plant or pollinator surveys in the adjacent forest. Net sampling was conducted a minimum of 20m away from pan traps, to lessen competition between the researcher and the bowls. Pollinators that were not identifiable to species on the wing were collected and brought to the lab, pinned, and identified. Bees were sight-identified in the field by a trained observer, or identified later in the lab using Discoverlife, an interactive web-based identification tool (Schuh et al. 2010).
Data analysis
As bees were the most likely pollinators of flowers in this study, and were visually confirmed under light microscopy to be carrying more pollen than other insect flower visitors in this study, analyses were restricted to bee taxa only. All analyses were conducted using the R programming language, version 3.3.1 (R Development Core team 2017).
Because some sites were separated by as much as 74 km and others as little as 1 km, I tested for spatial autocorrelation between site geographic distance and the plant and bee communities using Mantel correlation tests (mantel function, ’vegan’ package, R, Oksanen et al. 2016). The pairwise distances between site centers was calculated using an online tool (Veness 2017). There was no significant relationship between site distance and dissimilarity of pollinator communities (r = 0.05, p = 0.34), or between site distance and dissimilarity in plant communities (r = -0.20, p = 0.84).
I used Kruskal-Wallis rank sum tests with Dunn’s post hoc pairwise group comparisons (with Bonferroni family error correction, ‘dunn.test’ R package, Dinno 2017) to assess overall differences in bee genus Shannon diversity between collection methods. Rather than transforming the data to meet assumptions of normality, I choose to use the Kruskal-Wallis test as an alternative to analysis of variance (ANOVA) to test for differences between the two groups of samples (pan versus net). In order to assess differences in Shannon diversity of bee assemblages collected by these two methods without the effect of abundance, pan trap and net samples were each rarified down to 15 individuals. Twenty-six (out of 69) pan trap samples were excluded because fewer than 15 individuals had been collected at that site on that sample date. Bee assemblages were randomly re-sampled 1000 times using function rrarefy (‘vegan’ package), and the mean Shannon diversity of the 1000 iterations was calculated for all rarefied samples.
I used non-metric multidimensional scaling ordination (NMDS; metaMDS function, vegan package), based on Bray-Curtis site dissimilarity, to visualize differences in bee assemblage by sampling method and site history and management characteristics, for both the rarefied and un-rarefied datasets. I tested for significant differences in bee assemblages between net and pan trap samples using adonis tests (‘vegan’ package), and plotted ellipses around each group of samples (drawn to one SD unit around group centroids, function ordiellipse,‘vegan’ package). To determine which sampling method yielded the most stable representation of bee communities across sites and sample dates, I calculated the mean distance between each the genus assemblage of each pan and net sample (based on pairwise Bray-Curtis dissimilarity in genus composition) and the within group centroid (function betadisper, ‘vegan’ package).
Indicator species analysis was used to determine which bee taxa were more likely to be captured by one sampling method or the other (function indval, R package ‘labdsv’, Roberts 2016).
To estimate asymptotic bee genus richness, permutational sample-based taxon accumulation curves were constructed for pan trap and net samples, separately then combined, using the functions specaccum, specpool, and poolaccum (100 permutations, minimum size of 3 samples) (‘vegan’ package, R).
Estimating cost per bee of pan trap versus netted specimens
An argument often used in favor of pan trapping is that it is more cost-effective and less time/labor intensive than net surveys. In order to compare the cost between the two sampling methods, I calculated the price of obtaining a single bee specimen through either pan trapping or netting separately or combined in this study. I took into consideration wages, travel to and from field sites, and field and lab materials. Calculations for labor are based on minimum processing time per bee in the field and lab, based on five years of researcher experience (JL) with handling bee specimens. I did not include time spent identifying bees, entering data, or analyzing data, only the minimum time needed to perform primary lab and field work (i.e. to collect and preserve specimens). See supplemental material (Tables S1 and S2) for details.
Results
In total 12,795 bees were documented from 32 genera (Table 2). Bee genus richness was comparable with, or higher, than in other published studies from northeastern and central North America (Table 3).
Question 1. Which method yielded greater bee genus diversity and had a greater rate of taxon accumulation?
Through netting I recorded 10,524 bees in 30 genera, visiting 147 species of plants. I collected 2,271 total bees from 24 genera in pan traps. Any one 3.25oz bee bowl (left in place for 4 hours) caught approximately the same number of bees (1.10 ± 1.32 SD), as one minute of netting time (1.27 ± 0.63 SD). When honey bees and bumble bees were removed from the dataset (the two most abundant taxa in net samples that were largely absent in pan traps), significant differences remained in the numbers of bees documented by netting versus pan trapping on the average sample date (net > pan; H = 32.21, df = 1, p < 0.01).
Regardless of the differences in yield between the two methods, when pan and net samples were each rarefied down to a standardized number (in this case 15 individuals per sample), net samples had significantly higher bee genus diversity than pan samples (H = 42.14, df = 1, p < 0.01).
The bootstrap estimate of asymptotic bee genus richness in net samples (31.01 ± 0.97 SE) was similar to the observed richness (30 genera across the 69 total samples). The bootstrap estimate of asymptotic bee richness in pan traps (26.43 ± 1.25 SE) was slightly higher than the number of genera actually observed (24 genera across the 62 total pan trap samples). The total estimated bee genus richness for net plus pan samples combined (31.92 ± 0.84 SE) did not differ from the extrapolated net sample richness (Fig. 1). Individual sites ranged in estimated bee genus richness from 2.00 ± 0.00 to 17.54 ± 1.74 SE genera in pan trap samples, and from 17.15 ± 1.41 to 23.93 ± 2.02 in net samples (Table S3).
Whereas genera accumulated more slowly with successive pan trap samples, netting succeeded at documenting the majority of the bee genera that were ultimately collected in a fraction of the samples (Table 4). When considering the total yield of the two methods combined, netting and pan trapping together accumulated genera slightly more quickly than netting alone.
See supplemental material for differences in bee genus diversity and composition by bowl color in pan trap samples.
Question 2. How did pan trap and net samples differ in bee genus composition?
Netting versus pan trapping yielded significantly different bee assemblages (NMDS ordination with adonis test: F = 57.04, df = 1, 129, p < 0.01). These differences in bee composition by sampling method remained when ordination was performed on rarefied 15-bee samples (Fig. 2; adonis test F = 82.16, df = 1, 110, p < 0.01), and when honey bees and bumble bees were removed from the data set (adonis test F = 28.16, df = 1, 129, p < 0.01).
Anthidiellum, Coelioxys, Eucera, Pseudopanurgus, Stelis, Triepeolus, and Xylocopa were only caught by netting, and did not appear in any bowl samples. Peponapis pruinosa and Agapostemon were collected in pan traps, but not netted. Pan traps tended to collect more Lasioglossum (iv = 0.62, p < 0.01) and Calliopsis (iv = 0.41, p < 0.01), while netting favored Anthidium, Apis, Augochloropsis, Bombus, Ceratina, Hylaeus, Megachile, Melissodes, Osmia, Triepeolus and others (Table 2). 80.0% of bees collected in pan traps were from the family Halictidae, compared to 18.6% Halictid bees in net samples.
Question 3. Which method yielded a more stable bee assemblage across sample dates?
There was much greater variability in pan trap yield than in net collections across sample dates, in terms of bee abundance, diversity, and genus composition (Table 5). When pan and net data were combined, there was slightly lower variability between samples in bee abundance and diversity than in net samples alone. Pan traps were more effective at capturing a higher diversity of bees in the spring compared to mid- and late-summer samples (rarefied 15-individual samples; H = 10.19, df = 2, p = 0.01), although rarefied diversity of netted bees remained similar between sample months.
In an NMDS ordination of differences in bee genus assemblage by method, pan trap samples had higher within group dispersion (or higher average distance to group centroid) than did netted bee assemblages.
Question 4. Which method was more sensitive to flower resource availability?
Rarefied bee Shannon diversity of pan trap samples was negatively correlated with flower diversity in transect surveys (r = -0.42, t = -2.95, df = 41, p < 0.01, Fig. 3). Before rarefying down to 15 individuals, pan trap bee diversity was also negatively affected by flower abundance (r = -0.31, t = -2.53, df = 60, p = 0.01), but after rarefaction the relationship disappeared (r = -0.21, t = -1.36, df = 41, p = 0.18, Fig. 4). Rarefied bee diversity of net samples was not significantly influenced by flower diversity (r = -0.17, t = -1.37, df = 67, p = 0.17, Fig. 3), but was negatively correlated with flower abundance (r = -0.42, t = -3.76, df = 67, p < 0.01, Fig. 4).
Regardless of sample method, Bray-Curtis dissimilarity in bee catch was significantly correlated with flower community dissimilarity between samples (mantel.test, vegan, 999 permutations). However, bee net samples more closely tracked differences in flower community between sites and samples (Mantel r = 0.45, p < 0.01) than did pan trap samples (r = 0.25, p < 0.01).
Question 5. Which method was preferable for detecting differences in bee genus diversity and assemblage by site history and management characteristics?
Analysis of netted bees or pan trap bees alone, compared to net plus pan combined, resulted in different conclusions about how bee diversity was influenced by site history and management characteristics. Even though it was not apparent from the pan trap data, sites that had been seeded with native wildflowers within the past five years (three park sites plus one reclaimed mine site) had significantly higher netted bee diversity than reclaimed mine sites planted with the standard reclamation mix of non-native grasses and legumes (H = 8.98, df = 1, p < 0.01). Sites supplemented with native wildflowers also differed somewhat in netted bee assemblages from those planted with the base seed mix, but not significantly so based on NMDS ordination with adonis tests (F = 2.04, df = 1, 67, p = 0.08). Based on NMDS ordination of pan trap data only, pan trap bee genus assemblages differed significantly between reclaimed mine sites and park sites (F = 3.04, df = 1, 60, p = 0.01), but not by site flower type (standard reclamation mix versus supplemented wildflower plantings; F = 1.08, df = 1, 60, p = 0.34). When NMDS ordination was performed on net plus pan data combined, it accentuated differences in bee assemblages by site and flower community types (site type, reclaimed mine versus park: F = 3.01, df = 1, 67, p = 0.01; flower community type, standard versus supplemented: F = 2.20, df = 1, 67, p = 0.05).
Question 6. Which method was more cost effective?
If I had used either one sampling method or the other alone, the cost per bee for pan trap specimens would have been $2.12 (Table S1) and the cost of netted specimens would have been $0.84 per bee (Table S2) in this study. Since I left pan traps in place while conducting net surveys at the same site, the actual cost per bee in this study was $0.84, the same per unit as if I had done netting alone. The additional cost of field labor and materials needed for pan trapping was balanced out by the increased number of specimens generated.
Discussion
Regardless of the difference in bee abundance between methods, netting collected a higher diversity of bee genera than pan trapping and accumulated new taxa faster with successive samples. Netting, or both netting plus pan trapping combined, also had a more consistent yield in terms of diversity and genus assemblage across sample dates than pan trapping alone. Pan traps were more effective in the spring than in mid- and late summer samples, and at sites where flower diversity was low. As the bee assemblages collected in pan traps were largely a subset of those collected by netting, either net data, or considering net plus pan yield together, was more informative for assessing differences in bee communities between sites based on management and site history. The most cost-effective strategy for surveying bees in this study system, where sites were large meadows with patchily-distributed resources and were geographically far apart, was to net primarily and to supplement with pan traps, because the cost per bee was the same whether I had done netting alone or both simultaneously.
Netting succeeded at documenting the core suite of bee genera commonly found in the study region, as I observed 30 out of 53 genera reported to occur in the state. Through net surveys, I also documented some of the less frequently observed genera in Ohio including Eucera, Heriades, Perdita, Pseudopanurgus, Ptilothrix, and Stelis (Prajzner 2017; Spring 2017; Spring et al. 2017). Surprisingly, given the success of pan traps in other regional surveys (e.g. Spring et al. 2017), pans added only two genera, Peponapis and Agapostemon, to my inventory. There were fifteen genera that have appeared in other recent bee surveys in the region that I failed to detect using either method: Chelostoma, Colletes, Dianthidium, Dieumonia, Dufourea, Epeolus, Habropoda, Hesperapis, Macropis, Melecta, Melitoma, Protandrena, Pseudoanthidium, Svastra, and Xenoglossa (Table 3). Among other regional bee surveys (Grixti and Packer 2006; Richards et al. 2011; Burkle et al. 2013; McLeod 2013; Tonietto et al. 2017; Griffin et al. 2017; Spring 2017, Spring et al. 2017), trained researchers have identified >75,000 individual bees and detected a total of 48 genera of the 56 reported in Ohio and surrounding states (Schuh et al. 2010). The maximum reported in a single bee survey in our study region has been 36 (Grundel et al. 2011), but that study was conducted in a specialized habitat type that is not widely distributed (the Indiana Dunes National Lakeshore park; Porter County, Indiana). This suggests that 30 – 35 genera in a given survey may be a suitable benchmark when considering the number of genera a survey needs to accumulate to adequately represent the core bee fauna in central or northeastern North America.
After comparing bee genus assemblage in pan traps to those from net collections, in this study pan trap samples did not, by themselves, adequately represent the regional bee fauna. While others have found pan trapping to broaden the scope of their survey when used in addition to netting (Nielsen et al. 2011; Richards et al. 2011; Williams 2011), in this case pans collected a lower diversity of bees compared to netting, had more variable yield between sample dates, and did not lessen the researcher workload or project cost. My results agree with those of Cane et al. (2000), Roulston et al. (2007), Kwaiser and Hendrix (2008), and Richards et al. (2011) that pan trapping alone is of limited use for inventorying bee diversity and documenting the most common and important flower visitors, but is a useful supplement to net surveys when and where flowers are scarce or patchily distributed in space and time (Wilson et al. 2008; Popic et al. 2013). In an analysis of 11 multi-year studies that employed a variety of pollinator sampling methods, LeBuhn et al. (2013) found that bee species richness was less variable between sites and samples when measured using both pan and netting surveys rather than one method or the other alone. In this study, considering net and pan data together provided a slight advantage over netting alone in interpreting how bee communities responded to land management, flower community, and site history on grassland-reclaimed former coal mines and managed park meadows planted with native wildflowers.
In this study pan traps were lower-yielding than expected based on large published surveys (reviewed in Westphal et al. 2008), but on par with two other recent bee surveys in Ohio (Prajzner 2017: ~2600 bees in 23 genera; Spring 2017: 3004 bees in 28 genera). Even though I was careful to place pans on patches of bare ground that were highly visible to flying insects and to remove overhanging vegetation (Droege et al. 2015), pan trap yield is subject to stochastic forces and relies on chance discovery by passing bees. For example, two small-bodied soil-nesting genera associated with pan traps in my study nest in either social familial colonies (Lasioglossum) or aggregations of solitary females (Calliopsis), so their catch may have been “hit and miss” based on bowl proximity to favorable nest sites. I also found evidence that natural forage plants interfered with pan traps, which are artificial flower mimics. Diversity of bees caught in pans decreased significantly with flower diversity, and to a lesser extent flower abundance. Meanwhile, there was no effect of flower diversity on netted bee diversity. This result suggests that bowls are outcompeted by natural flowers when they are available, and most effective when the abundance and diversity of bee forage is low (Wilson et al. 2008), especially early or late in the season (Grundel et al. 2011). Baum and Wallen (2011) similarly found that pan traps were more effective in plots where flowers had been experimentally removed. Where flower resources are plentiful and spread out evenly over a large study area, it is more productive for researchers to actively seek bees out than to rely on passive collection techniques (Cane et al. 2000, Roulston et al. 2007; Kwaiser and Hendrix 2008; Richards et al. 2011, Popic et al. 2013).
Given the high degree of variation globally in bee catch between sample dates, sites, and across years, LeBuhn et al. (2013) argue that high intensity sampling (>200 sites sampled at least twice within a five-year period) is needed to thoroughly document the bee fauna in a region. Westphal et al. (2008) and LeBuhn et al. (2013) advocate for high-intensity sampling on the grounds that there is ultimately greater risk to the stability of pollination services if bee declines go undetected for long periods of time due to inadequate sampling than of the surveys themselves bringing about declines by over-harvesting bees. However, there is some concern that repeated pan trap sampling may deplete local populations and negatively impact bees’ reproduction (Gibbs et al. 2017). Pan trapping removes individuals from the population indiscriminately, unlike netting surveys, in which the researcher may choose to spare reproductive individuals in the interest of bee conservation. Gibbs et al. (2017) found evidence that blue vane traps could lead to small scale depletion of certain wild bee species in study orchards if they were deployed at the time of year when only foundress bees of social taxa were active. On the other hand, Gezon et al. (2015) found no evidence that pan trapping every other week for three consecutive years diminished bee community diversity.
Others have tried to estimate the minimum number of specimens needed to document a bee fauna (Grundel et al. 2011) or to an answer an ecological question, when considering the accumulation of functionally-important species as sampling effort increases and the integrity of plant-pollinator networks as a whole (Hegland et al. 2010). If the intent is to assess the likely quality of pollination services at a site, a relatively small amount of sampling is needed to capture the basic functional groups of bees (Roulston et al. 2007; Droege et al. 2010), that is, the core group of common species that have a disproportionately large effect on pollination services (Kleijn et al. 2015; Winfree et al. 2015). In this study, I could still have documented 95% of the bee genera ultimately collected with only half the sampling effort (35 out of 69 total samples) and lessened my impact on bee communities. If identification to genus or grouping by some functional guild (e.g. Hoehn et al. 2008) is sufficient for assessing pollinator community function, I could have further minimized the impact of net surveying on bee populations by immobilizing bees with compressed air from office keyboard dusters (which chills and temporarily anesthetizes them), identifying them to genus in the field, and re-releasing them, rather than collecting them for later pinning and identification in the lab. Bee bodies could be swabbed for pollen in the field and later identified in the lab for additional insight into bee resource use (as in Alarcón et al. 2010).
In place of high-mortality bee surveys, a conservative approach to rapidly assessing bee community habitat use and response to land use change would be to net non-lethally by identifying to genus in the field to document the most important and abundant flower visitors, and to pan trap only selectively as a supplement when the goal is to reach a complete species inventory. Alternatively, for groups of observers of varying levels of experience, a combination of pan trapping and netting could be done in which the researcher netted or conducted timed observations to record visitation by bumble bees and honey bees, or other focal taxa that are integral in plant-pollinator networks, and deployed pan traps simultaneously to survey for broader diversity of the bee community.
References
Alarcón R (2010) Congruence between visitation and pollen-transport networks in a California plant-pollinator community. Oikos 119:35-44
Bartomeus I, Ascher JS, Gibbs J, Danforth BN, Wagner DL, Hedtke SM, Winfree R (2013) Historical changes in northeastern US bee pollinators related to shared ecological traits. PNAS 110(12):4656-4660
Baum KA, Wallen KE (2011) Potential bias in pan trapping as a function of floral abundance. J Kansas Entomol Soc 84(2):155-159
Brown MJF, Paxton RJ (2009) The conservation of bees: a global perspective. Apidologie 40:410-416
Burkle L, Marlin JC, Knight TM (2013) Plant-pollinator interactions over 120 years: Loss of species, co-occurrence, and function. Science 339:1611-1615
Cane JH, Minckley RL, Kervin LJ (2000) Sampling bees (Hymenoptera: Apiformes) for pollinator community studies: Pitfalls of pan-trapping. J Kansas Entomol Soc 73:225-231
Cane JH, Tepedino VJ (2001) Causes and extent of declines among North American invertebrate pollinators: Detection, evidence, and consequences. Ecol Soc 5. http://www.consecol.org/vol5/iss1/art1/
Dinno A (2017) Dunn.test: Dunn’s test of multiple comparisons using rank sums. R package. version 1.3.3. https://cran.r-project.org/web/packages/dunn.test/index.html. Accessed 1 February 2017
Droege S, LeBuhn G, Williams N, Minckley R, Griswold T, Kremen C, Messinger O, Cane J, Roulston T, Parker F, Tepedino V, Buchmann S (2005) The bee inventory plot. http://online.sfsu.edu/beeplot/. Accessed 1 May 2014
Droege S (2015) The very handy manual: How to catch and identify bees and manage a collection. United States Geological Survey Bee Inventory and Monitoring Lab, Patuxent Wildlife Research Center, Beltsville, Maryland. https://www.pwrc.usgs.gov/nativebees/Handy%20Bee%20Manual/The%20Very%20Handy%20Manual%20-%202015.pdf. Accessed 1 May 2015
Droege S, Tepedino VJ, LeBuhn G, Link W, Minckley RL, Chen Q, Conrad C (2010) Spatial patterns of bee captures in North American bowl trapping surveys. Insect Conserv Diver 3:15-23
ESRI (2016) ArcGIS Desktop: Release 10.3.0.4322 Environmental Systems Research Institute. Redlands, California
Fry J, Xian G, Jin S, Dewitz J, Homer C, Yang L, Barnes C, Herold N, Wickham J (2011) Completion of the 2006 National Land Cover Database for the conterminous United States. Photogramm Eng Rem S 77:858-864
Gezon ZJ, Wyman ES, Ascher JS, Inouye DW, Irwin RE (2015) The effect of repeated, lethal sampling on wild bee abundance and diversity. Methods Ecol Evol 6:1044-1054
Gibbs J, Joshi NK, Wilson JK, Rothwell NL, Powers K, Haas M, Gut L, Biddinger DJ, Isaacs R (2017) Does passive sampling accurately reflect the bee (Apoidea: Anthophila) communities pollinating apple and sour cherry orchards? Environ Entomol 46(3):579-588
Gleason HA, Cronquist A (1991) Manual of vascular plants of northeastern United States and adjacent Canada. 2nd Ed. The New York Botanical Garden Press: New York
Goulson D, Darvill B (2004) Niche overlap and diet breadth in bumblebees; are rare species more specialized in their choice of flowers? Apidologie 35:55-63
Griffin SR, Bruninga-Socolar B, Kerr MA, Gibbs J, Winfree R (2017) Wild bee community change over a 26-year chronosequence of restored tallgrass prairie. Restor Ecol. https://doi.org/10.1111/rec.12481
Grixti JC, Packer L (2006) Changes in the bee fauna (Hymenoptera: Apoidea) of an old field site in southern Ontario revisited after 34 years. Can Entomol 138:147-164
Grundel R, Frohnapple KJ, Jean RP, Pavlovic NB (2011) Effectiveness of bowl trapping and netting for inventory of a bee community. Environ Entomol 40(2):374-380
Hegland SJ, Dunne J, Nielsen A, Memmott J (2010) How to monitor ecological communities cost-efficiently: The example of plant-pollinator networks. Biol Conserv 143:2092-2101
Hoehn P, Tscharntke T, Tylianakis J, Steffan-Dewenter I (2008) Functional group diversity of bee pollinators increases crop yield. P R Soc B 275:2283-2291
Hopwood J (2008) The contribution of roadside grassland restorations to native bee conservation. Biol Conserv 141:2632-2640
Kimoto C, DeBano SJ, Thorp RW, Rao S, Stephen WP (2012) Investigating temporal patterns of a native bee community in a remnant North American bunchgrass prairie using blue vane traps. J Insect Sci 12:108
Kleijn D, Winfree R, Bartomeus I, Carvalheiro LG, Henry M, Isaacs R, Klein AM, Kremen C, M’Gonigle LK, Rader R, Ricketts TH, Williams NM, Adamson NL, Ascher JS, Baldi A, Batary P, Benjamin F, Biesmeijer JC, Blitzer EJ, Bommarco R, Brand MR, Bretagnolle V, Button L, Carivaeu DP, Chifflet R, Colville JF, Danforth BN, Elle E, Garratt MPD, Herzog F, Holzschuh A, Howlett BG, Jauker F, Jha S, Knop E, Krewenka KM, Le Feon V, Mandelik Y, May EA, Park MG, Pisanty G, Reemer M, Riedinger V, Rollin O, Rundlof M, Sardinas HS, Scheper J, Sciligo AR, Smith HG, Steffan-Dewenter I, Thorp R, Tscharntke T, Verhulst J, Viana BF, Vaissiere BE, Veldtman R, Westphal C, Potts SG (2015) Delivery of crop pollination services is an insufficient argument for wild pollinator conservation. Nat Commun 6: 7414
Kwaiser KS, Hendrix SD (2008) Diversity and abundance of bees (Hymenoptera: Apiformes) in native and ruderal grasslands of agriculturally dominated landscapes. Agr Ecosyst Environ 124:200-204
LeBuhn G, Droege S, Griswold R, Minckley RL, Roulston T, Cane J, Parker F, Buchmann S, Tepedino V, Williams N, Kremen C, Messenger O (2003) A standardized method for monitoring bee populations—the bee inventory (BI) plot. http://online.sfsu.edu/beeplot/pdfs/Bee%20Plot%202003.pdf
LeBuhn G, Droege S, Connor EF, Gemmill-Herren B, Potts SG, Minckley RL, Griswold R, Jean R, Kula E, Roubik DW, Cane J, Wright KW, Frankie G, Parker F (2013) Detecting insect pollinator declines on regional and global scales. Conserv Biol 27(1):113-120
Leong JM, Thorp RW (1999) Colour-coded sampling: the pan trap colour preferences of oligolectic and nonoligolectic bees associated with a vernal pool plant. Ecol Entomol 4:329-335
McCravy KW, Geroff RK, Gibbs J (2016) Malaise trap sampling efficiency for bees (Hymenoptera: Apoidea) in a restored tallgrass prairie. Fla Entomol 99(2):321-323
McLeod K (2013) Native bee (Hymenoptera: Apiformes) response to ecological restoration in southern Ontario. Master’s Thesis, University of Waterloo
Newcomb L (1989) Newcomb’s wildflower guide. Little, Brown and Company, Boston
Nielsen A, Steffan-Dewenter I, Westphal C, Messinger O, Potts SG, Roberts SPM, Settele J, Szentgyörgyi H, Vaissière BE, Vaitis M, Woyciechowski M, Bazos I, Biesmeijer JC, Bommarco R, Kunin WE, Tscheulin T, Lamborn E, Petanidou T (2011) Assessing bee species richness in two Mediterranean communities: importance of habitat type and sampling techniques. Ecol Restor 26:969-983
Office of Surface Mining Reclamation and Enforcement, US Department of the Interior (2014) Abandoned Mine Land Inventory System (AMLIS). https://amlis.osmre.gov. Accessed 1 May 2014
Office of Surface Mining Reclamation and Enforcement, US Department of the Interior (2017) Laws and regulations. https://www.osmre.gov/lrg.shtm. Accessed 1 May 2017
Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E, Wagner H (2016) Vegan: Community ecology package. R package version 2.4-1. https://CRAN.R-project.org/package=vegan. Accessed 1 Oct 2016
Popic TJ, Davila YC, Wardle GM (2013) Evaluation of common methods for sampling invertebrate pollinator assemblages: Net sampling outperform pan traps. PLoS ONE 8(6):e66665
Potts SG, Biesmeijer JC, Kremen C, Neumann P, Schweiger O, Kunin WE (2010) Global pollinator declines: trends, impacts, and drivers. Trends Ecol Evol 25:345-353
Prajzner SP (2017) Effects of land use change on bee (Anthophila) community structure and function. Electronic Dissertation, the Ohio State University. https://etd.ohiolink.edu/. Accessed 1 July 2017
R Development Core Team (2017) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. v3.3.2. https://www.r-project.org. Accessed 1 Oct 2017
Richards MH, Rutgers-Kelly A, Gibbs J, Vickruck JL, Rehan SM, Sheffield CS (2011) Bee diversity in naturalizing patches of Carolinian grasslands in southern Ontario Canada. Can Entomol 143:279-299
Roberts DW (2016) Labdsv: Ordination and multivariate analysis for ecology. R package version 1.8-0. https://CRAN.R-project.org/package=labdsv
Roulston TH, Smith SA, Brewster AL (2007) A Comparison of pan trap and intensive net sampling techniques for documenting a bee (Hymenoptera: Apiformes) fauna. J Kansas Entomol Soc 80(2):179-181
Roulston T, Goodell K (2011) The role of resources and risks in regulating wild bee populations. Annu Rev Entomol 56:293-312
Saville NM (1993) Bumblebee ecology in woodlands and arable farmland. Dissertation, University of Cambridge
Schuh RT, Hewson-Smith S, Ascher JS (2010) Specimen databases: A case study in entomology using web-based software. Am Entomol 56:206-216
Spring M (2017) Impacts of urban greenspace management on beneficial insect communities. Master’s Thesis, The Ohio State University
Spring M, Lustofin KS, Lin C, Gardiner MM, McShaffrey D (2017) Quantifying bee diversity and resource use in the Appalachian foothills near Marietta, Ohio. Ohio Biological Survey Notes 7:1-13
Surface Mining Control and Reclamation Act (SMCRA), Pub.L. 95-87 (1977) United States Code, Title 30, Ch 25. http://uscode.house.gov. Accessed 1 Oct 2017
Toler TR, Evans EW, Tepedino VJ (2005) Pan-trapping for bees (Hymenoptera: Apiformes) in Utah’s West desert: the importance of color diversity. Pan-Pac Entomol 81:103–113
Tonietto RK, Ascher JS, Larkin DJ (2017) Bee communities along a prairie restoration chronosequence: Similar abundance and diversity, distinct composition. Ecol Appl 27(3):705-717
Tscharntke T, Gathmann A, Steffan-Dewenter I (1998) Bioindiciation using trap-nesting bees and wasps and their natural enemies: community structure and interactions. J Appl Ecol 35(5):708-719
Veness C (2017) Movable Type Scripts: Calculate distance, bearing, and more between latitude/longitude points. http://www.movable-type.co.uk/scripts/latlong.html. Accessed 1 Oct 2017
Westphal C, Bommarco R, Carre G, Lamborn E, Morison N, Petanidou T, Potts SG, Roberts SPM, Szentgyorgyi H, Tscheulin T, Vaissiere BE, Woyciechowski M, Biesmeijer JC, Kunin WE, Settele J, Steffan-Dewenter I (2008) Measuring bee diversity in different European habitats and biogeographical regions. Ecol Monogr 78(4):653–671
Williams NM (2011) Restoration of nontarget species: Bee communities and pollination function in riparian forests. Restor Ecol 19(4):450-459
Wilson JS, Griswold T, Messinger O (2008) Sampling bee communities (Hymenoptera: Apiformes) in a desert landscape: Are pan traps sufficient? J Kansas Entomol Soc 81(3):288-300
Winfree, R, Fox J, Williams N, Reilly J, Cariveau D (2015) Abundance of common species, not species richness, drives delivery of a real-world ecosystem service. Ecol Lett 18:626-635
Tables and Figures
Site | Type | Year | Latitude | Longitude (-) | H pan | n pan | H net | n net |
HG | mine | 2015-2016 | 39°24’5.44″N | 82°27’56.06″W | 0.74 | 230 | 1.84 | 839 |
Orland | mine | 2015-2016 | 39°22’47.44″N | 82°25’44.97″W | 1.20 | 182 | 2.02 | 560 |
Cannelville | mine | 2015-2016 | 39°48’14.59″N | 82° 2’0.60″W | 0.69 | 318 | 1.71 | 823 |
Rock Run | mine | 2015-2016 | 39°35’5.51″N | 82°13’13.80″W | 1.23 | 180 | 2.19 | 823 |
Lost Run | mine | 2015-2016 | 39°33’17.78″N | 82°14’50.76″W | 1.70 | 99 | 1.57 | 1227 |
Pine Run | mine | 2015-2016 | 39°37’22.25″N | 82°10’36.61″W | 0.65 | 262 | 1.84 | 799 |
SSL | mine | 2015-2016 | 39°37’3.98″N | 82°13’46.38″W | 0.87 | 274 | 1.65 | 976 |
Spencer | mine | 2015-2016 | 39°31’16.85″N | 82°10’26.68″W | 1.27 | 120 | 1.89 | 1145 |
Superior Fibers | mine | 2015-2016 | 39°36’36.24″N | 82°12’39.51″W | 1.33 | 407 | 2.14 | 887 |
Moores Junction | mine | 2015-2016 | 39°43’13.32″N | 82° 7’9.02″W | 1.60 | 104 | 1.84 | 878 |
Chestnut Ridge | park | 2016 | 39°48’0.65″N | 82°45’12.50″W | 0.35 | 9 | 1.79 | 616 |
Clear Creek | park | 2016 | 39°35’48.52″N | 82°33’7.12″W | 1.62 | 65 | 2.14 | 501 |
Three Creeks | park | 2016 | 39°52’47.49″N | 82°53’48.25″W | 1.55 | 22 | 2.06 | 464 |
Table 1. List of sites, site type (reclaimed mine or park meadow), years sampled, and location. H is the Shannon diversity index of total catch in pan trap or net/observation samples, while n is the total number of individuals caught, by sampling method.
27
Taxon | Total | Net | Pan | No. Plant Spp. | Association with Method | iv | p |
Agapostemon | 6 | 0 | 6 | – | – | – | – |
Andrena | 342 | 294 | 48 | 36 | net | 0.49 | < 0.01 |
Anthidiellum | 26 | 26 | 0 | 2 | net | 0.16 | < 0.01 |
Anthidium | 198 | 168 | 30 | 5 | net | 0.35 | < 0.01 |
Anthophora | 6 | 5 | 1 | 2 | – | – | – |
Apis mellifera | 4364 | 4349 | 15 | 56 | net | 0.98 | < 0.01 |
Augochlora | 205 | 198 | 7 | 48 | net | 0.66 | < 0.01 |
Augochlorella | 157 | 116 | 41 | 38 | net | 0.40 | < 0.01 |
Augochlorini undet. | 309 | 309 | 0 | 62 | net | 0.80 | < 0.01 |
Augochloropsis | 36 | 35 | 1 | 17 | net | 0.30 | < 0.01 |
Bombus | 1683 | 1675 | 8 | 73 | net | 0.95 | < 0.01 |
Calliopsis andreniformis | 127 | 14 | 113 | 4 | pan | 0.41 | < 0.01 |
Ceratina | 1004 | 896 | 108 | 81 | net | 0.82 | < 0.01 |
Coelioxys | 24 | 24 | 0 | 12 | net | 0.25 | < 0.01 |
Eucera | 5 | 5 | 0 | 2 | – | – | – |
Halictus | 486 | 382 | 104 | 50 | net | 0.67 | < 0.01 |
Heriades | 7 | 6 | 1 | 4 | – | – | – |
Holcopasites | 17 | 5 | 12 | 3 | – | – | – |
Hoplitis | 64 | 40 | 24 | 10 | – | – | – |
Hylaeus | 362 | 344 | 18 | 34 | net | 0.74 | < 0.01 |
Lasioglossum | 2556 | 901 | 1655 | 81 | pan | 0.62 | 0.01 |
Megachile | 384 | 380 | 4 | 37 | net | 0.79 | < 0.01 |
Melissodes | 83 | 80 | 3 | 8 | net | 0.22 | < 0.01 |
Nomada | 35 | 22 | 13 | 14 | – | – | – |
Osmia | 188 | 134 | 54 | 23 | net | 0.32 | 0.03 |
Peponapis pruinosa | 1 | 0 | 1 | – | – | – | – |
Perdita | 6 | 5 | 1 | 3 | – | – | – |
Pseudopanurgus | 5 | 5 | 0 | 2 | – | – | – |
Ptilothrix bombiformis | 1 | 1 | 0 | 1 | – | – | – |
Sphecodes | 14 | 11 | 3 | 5 | – | – | – |
Stelis | 2 | 2 | 0 | 2 | – | – | – |
Triepeolus | 7 | 7 | 0 | 3 | net | 0.09 | 0.03 |
Xylocopa virginica | 85 | 85 | 0 | 20 | net | 0.35 | < 0.01 |
Total | 12795 | 10524 | 2271 | 147 |
Table 2. Number of bees observed of each genus, in total and by sampling method. The number of flowering plant species on which each taxon was netted is given in the column “No. plant spp.” Significant associations between a given genus and a particular sampling method (based on indicator species analysis) are listed, along with the indicator value (iv) and p value.
Study | Method(s) | Year(s) | Location | No. Bees | No. Genera | No. Species | Genera collected that were not observed in our study |
this study | pan, aerial net | 2015-2016 | Ohio, USA | 12722 | 32 | n/a | |
pan | 2015-2016 | Ohio, USA | 2272 | 24 | n/a | ||
aerial net | 2015-2016 | Ohio, USA | 10450 | 30 | n/a | ||
Spring 2017 | pan | 2014-2016 | Ohio, USA | 3004 | 28 | 81 | Pseudoanthidium |
Spring et al. 2017 | pan* | 2013 | Ohio, USA | 2753 | 35 | 130 | Colletes, Melecta, Melitoma, Panurginus |
Griffin et al 2017 | pan, blue vane | 2014 | Illinois, USA | 2097 | 23 | 85 | Colletes, Dufourea, Svastra, Xenoglossa |
Tonietto et al. 2017 | pan, aerial net | 2010-2012 | Illinois, USA | 6561 | 32 | 115 | Chelostoma, Colletes, Dufourea, Epeolus, Svastra, |
McLeod 2013 | pan, sweep net | 2012 | Ontario, CAN | 10602 | 30 | n/a | Chelostoma |
Richards et al. 2011 | pan, sweep net, aerial net | 2003 | Ontario, CAN | 15733 | 29 | 124 | Colletes, Dufourea, Protandrena |
Grundel et al. 2011 | pan, aerial net | 2003-2004 | Indiana, USA | n/a | 36 | 175 | Chelostoma, Dianthidium, Dieunomia, Epeolus, Hesperapis, Trachusa |
Grixti & Packer 2006 | sweep net | 1968-1969 | Ontario, CAN | 9784 | 26 | 105 | Colletes, Macropis |
Grixti & Packer 2006 | sweep net | 2002-2003 | Ontario, CAN | 10437 | 27 | 150 | Colletes, Epeolus, Macropis |
Lanterman unpub. data | pan | 2012-2015 | Ohio, USA | 2602 | 24 | n/a | Colletes, Melecta, Melitoma |
Table 3. Comparison of total bee yield in this study versus in ten other regional bee surveys, with a list of genera captured in each other study that were not observed in this study and details of study design.
29
Table 4. Estimate of the number of samples, out 69 total, needed to reach 75%, 90%, or 95% of the genera ultimately observed, by collection method. Taxon accumulation curves for each collection method (net, pan, or both combined) were calculated based on presence/absence of bee genera in net samples (estimated from 500 permutations, with samples added in a random order; R, vegan package, function specaccum).
Method | Total No. Genera | Number of samples needed to accumulate
__ % of the total genera |
||
75% | 90% | 95% | ||
net | 30 | 10 | 26 | 38 |
pan | 24 | 21 | 44 | 55 |
net + pan | 32 | 9 | 23 | 35 |
Table 5. Variability between sample dates by method in mean bee abundance and genus-level Shannon diversity per sample. Also given isthe mean distance to group centroid on an NMDS ordination plot (calculated with function betadispers, ‘vegan’ package, R), by sampling method.
Method | Mean Bee Abundance | % CV Bee Abundance | Mean Bee Diversity | % CV Bee Diversity | Mean Distance to Group Centroid |
net | 152.81 | 49.52 | 1.66 | 19.22 | 0.40 |
pan | 32.93 | 120.58 | 0.85 | 60.39 | 0.46 |
net + pan | 185.74 | 48.49 | 1.72 | 16.90 | 0.40 |
Fig. 1 Bee genus accumulation curves by sampling method (a pan trap samples; b net samples; c net plus pan data combined),createdusing function specaccum (‘vegan’ package, R) with samples added in random order and 500 permutations. See Oksanen et al. 2016 for details.
Fig. 2 Non-metric multidimensional scaling plot of bee genus assemblage by sampling method (net and pan), rarefied down to 15 individuals per sample. Red circles represent bee genus composition in rarefied pan trap samples (n = 43), and black circles represent bee composition in rarefied net samples (n = 69). Filled gray ellipses represent 1 SD around group centroid for each pollinator data collection method.
Fig. 3 Plot of rarefied bee genus Shannon diversity in pan trap (gray circles) and net samples (black circles) by Shannon diversity of flowering species. Each point represents one sample date at a single site (pan samples, n = 43; net samples n = 69). Bee data were rarefied down to a random 15 individuals per sample. The gray dashed trend line with correlation coefficient shows the significant negative correlation between rarefied bee diversity in pan trap samples and the diversity of flowering species.
Fig. 4 Plot of rarefied bee genus Shannon diversity in pan trap (gray circles) and net samples (black circles) by transect flower abundance score. See methods for description of how flowering species were scored on a 1-5 abundance scale on each sample date. Each point represents one sample date at a single site (pan samples, n = 43; net samples n = 69). The trend line with correlation coefficient (black) shows the significant negative correlation between netted bee diversity and site flower abundance on a given sample date.
Cite This Work
To export a reference to this article please select a referencing stye below:
Related Services
View allRelated Content
All TagsContent relating to: "Environmental Science"
Environmental science is an interdisciplinary field focused on the study of the physical, chemical, and biological conditions of the environment and environmental effects on organisms, and solutions to environmental issues.
Related Articles
DMCA / Removal Request
If you are the original writer of this dissertation and no longer wish to have your work published on the UKDiss.com website then please: