Glycol traps used to document spring bee communities in forest and a demonstration of their general utility
Below is the text from a little report I put together for a study of 7 woodlands using 9 trap glycol array last spring. I will try and send the full report with figures separately with the group, but it may not make it out due to size limitations of the listserv...if that is the case then feel free to email me for the report.
Because of the reasons and results listed below, more and more I think that some version of glycol trapping makes sense for inventory, community, and monitoring purposes for bees.
Investigation of the Spring Bee Fauna of Seven Woodland Sites in the Coastal Plain of Maryland using Continuously Trapping Arrays of Propylene Glycol Bee Cup Traps
Free flying adult bees are present in closed canopy forests in Eastern North America only in the spring. Within these woodlands, particularly deciduous ones, there is usually a compliment of canopy, understory, and shrub blooming woody plants while the forest floor can contain and often dense herbaceous layer of vernal flowers. At times, these resources can be abundant but the composition varies greatly among forest types, ranging from carpets of blooming forbs in the flood plains of large bottomland forests to sparse ericaceous shrub layers in oak-hickory forests. In early successional woods, weedy non-native plants at times dominate and spring blooming plants may be absent or in low numbers. Soils with extreme physical or chemical characteristics of deep sand, serpentine, acid, or basic materials also impact that resulting set of blooming plants. All such factors together would, in theory, impact the abundance and composition of bees inhabiting each woodland site.
Bee communities have been studied in a number of forested systems in Eastern North America often as part of a larger faunal study. Published studies have used malaise traps, netting, trap nests, and bowl traps.
While woodland environments are certainly studied by bee biologists, a general problem in comparing bee studies of any sort is the lack of a comparable protocol and the problem of how to adjust for when and how often samples were taken. The problem is often one of phenology. You can set your sampling dates any way you like using fixed dates or calculations of degree days or even vegetative presentation, but it is always unclear as to how to correct for major differences in bee phenology changes within and among sites or weather problems (see Figure 3 below). Better is to have a sampling regime that continuously traps bees. Maliase traps do that, but few of us have the budget to afford and replace such traps in numbers sufficient to satisfy the statisticians. However, recent work using propylene glycol based colored bowl or cup traps does provide an alternative.
In this study I demonstrate the use of inexpensive arrays of plastic beer cups filled with propylene glycol to capture and characterize the bee community of a set of 7 woodlands in the coastal plain of Maryland.
Materials and Methods
Seven sites in the coastal plain of Maryland, U.S.A. were chosen on the properties of the USDA Beltsville Agriculture Research Center and the USFWS Patuxent Wildlife Research Refuge as exemplars of different forest types present in the region (Table 1, Figures 1 and 2). Sites were chosen non-randomly for their convenience to roads, to each other, and to the author’s office. As such, specimens from traps at all sites could be collected in 1.5 hours if one didn’t dawdle.
Nine traps were located at each site and were spaced 5 meters apart. Trap arrays began approximately 15m from the edge of the secondary road and continued to the interior of the woods. Each trap consisted of a short length of plastic electrical conduit with a thin ring of 3 inch (7.6cm) PVC drain pipe screwed to the side a short distance from the top. The conduit was pushed or pounded into the ground so that the ring would hold a 12 ounce (0.35L) plastic beer cup upright with the bottom of the cup touching the ground. Each beer cup had 3 small weep holes drilled into it just above the lip of the cup to release liquid during heavy rain events. The beer cups were an opaque white color and one third of them were painted fluorescent blue and one third fluorescent yellow on their interior walls. Fluorescent paints were created using a white latex silicon flat base paint mixed with yellow and blue fluorescent pigment from Guerra Paint and Pigment. Traps were filled with propylene glycol that had been dyed blue by the plumbing company where it was purchased and was diluted by approximately half with tap water. A small amount of Blue Dawn Dishwashing liquid was added to the glycol to decrease surface tension and each cup was filled to approximately 7/8ths full.
Traps were initially deployed on March 16, 2011 and were run for 9 weeks ending May 19, 2011 (Table 2).
Specimens were collected from traps each week. If trap liquids were low then more was added. There were only 2 instances in which cups were spilled and contents lost. Trap liquids quickly became darkened and the blue color not noticeable. All traps from an individual site were pooled during each sampling period and stored in a freezer until processing (7 Sites X 9 Weeks = 63 samples). Specimens were washed, dried, pinned, and labeled prior to identification by the author. Trapping finished when captures rates became very low in May after the canopy had formed and blooming had finished in woodland environments. Weather was normal for the season with an exceptionally cool and rainy period occurring during the second trapping week, resulting in few captures (Table 3. Figure 3).
Paleontological Statistical Software Package version 2.14 was used in all analyses.
A total of 1439 bees were collected representing at least 58 species captured over the 9 week sampling period (Appendix 1 lists captures by species by site and Appendix 2 by date). Bees from the genera Andrena, Osmia, and Lasioglossum dominated the captures, comprising 91% of all captures from the 16 genera represented (Table 4).
At the species level, Andrena erigeniae (479 captures, 33%) and the recently established non-native Osmia taurus (291 captures, 20%) were captured the most frequently. Captures rates for all species combined other than A. erigeniae were high for the first 5 weeks and tapered off rapidly during the last 4 (Figure 3). A. erigeniae counts remained relatively steady throughout the time period except during week 7 when 242 were captured, surpassing the next highest capture period during week 1 (63) by several times (Table 5 and Figure 4).
There were clear differences in species composition among sites (Appendix 1). When sites were subjected to Correspondence Analysis the first axis was found to explain 58% of the variation and the second 15% (Table 6).
When plotted out (Figure 5) sites 2, 3, and 4 clustered closely together along the X-axis as did sites 6, 7 and 1 (all with high numbers of A. erigeniae) and 5 located between the others.
A possible interpretation of Axis 1 is that it represents a transect of sites from dry Forests to the bottomlands though this seems more interpretable when A. erigeniae is not included in the analysis (Table 7 and Figure 6). The second Axis accounts in both analyses accounts for much less variation and seems uninterpretable.
A date by species correspondence analysis explained less of the variance in the system than site by species with only 26% of the variance explained by axis 1 without A. erigeniae and 46% with A. erigeniae included. Plots of the scatter grams indicated that each point’s nearest neighbors were likely to be adjacent dates (Figure 7).
Summary of Interesting Results for Select Genera
Agapostemon, Augochlora, Augochloropsis, Augochlorella – Only 1 specimen of Agapostemon, 3 of Augochlorella, and 2 of Augochloropsis occurred at 3 of the sites. This is not unsurprising giving this group’s proclivity for fields. Of the metallic green halictids 11 individuals were captured of Augochlora pura across most of the sites mirroring most people’s experiences with this species being associated with woodlands and the edges of woodlands.
Andrena – 15 species present, of those A. erigeniae accounted for 479 of the 618 individuals captured. Not unsurprisingly this pollen specialist on Claytonia virginica was very abundant at the sites with obvious patches of this flower (Sites 1,6,7) and only in low numbers elsewhere. A. carlini occurred at all sites but reached its maximum count (26) at an oak-hickory site (4). 19 individuals of the relatively uncommon A. pruni were captured with small numbers found at most of the sites. Small numbers of regionally uncommon Andrena such as A. bradleyi, A. hilaris, A. rugosa, and A. tridens occurred among the plots and may or may not indicate a preference for wooded environments.
Anthophora –4 specimens of A. plumipes occurred across 4 plots. The site introduction for this introduced species was within but a kilometer of Site 1 and its presence here indicates both its establishment within the region’s woodlands and an indication that its numbers may not become overwhelming in native habitats. It is now of regular occurrence throughout the Washington D.C. area.
Apis – Only 3 honeybees were captured in this study despite the proximity of many hives associated with the USDA’s honeybee research lab. All captures occurred at Site 1 which was closest to those hives. Apis tends not to occur regularly in colored bowl type traps despite being regionally abundant.
Bombus – 13 of the 15 individuals were the early spring B. bimaculatus with the remaining 2 individuals B. griseocollis. Despite no Bombus captures coming from the bottomland sites they certainly occur there but, similar to Apis, do not often go into bowl traps.
Ceratina – Of the 3 very common species present in the region C. calcarata and C. strenua were present, but in small numbers. This follows observations that these species will inhabit spring woodlands while C. dupla is more associated with dry open sites.
Habropoda - 2 specimens of this regionally uncommon blueberry specialist were captured in the 2 oak-hickory sites which have a strong ericaceous shrub understory.
Halictus – Only 2 captures of Halictus species were made. One each of H. poeyi/ligatus and one of H. rubicundus. Many more would have been expected if these traps were located in fields.
Lasioglossum – This was perhaps the most interesting group. Normally, I think we think of Lasioglossum as being associated with open field situations and as being pollen and often habitat generalists with the exception of some sand loving species. However, it seems clear from these data that there is a strong woodland group of Lasioglossum species that at least I often overlook. Lasioglossum coeruleum occurred across all sites and is known to nest in decaying wood. Lasioglossum cressonii is a species that occurs in many habitats, but here occurs in large numbers, particularly in the drier sites. Lasioglossum gotham is a newly described species that appears to be associated with woodlands and has been found nesting in upturned tree root masses. Lasioglossum nigroviride is a very uncommon species that occurred in this study only once, but perhaps is to be found more often if woodland situations are more thoroughly checked. Lasioglossum quebecense is the most abundant species of Lasioglossum captured and clearly avoids the pine/oak-hickory sites and favors bottomlands and streamside locations. Lasioglossum subviridatum is a species that is almost never found in fields or open areas and the large numbers found here was perhaps the biggest surprise.
Nomada – As the nest parasites of Andrena and as common as Andrena are in this study it is not surprising to find that there were good numbers of Nomada also. No particular pattern or associations leap out here. While it would be tempting to associate the two most common Nomada (N. pygmaea and N. luteoloides) with the two most common Andrena (A. erigeniae and A. carlini) the distribution of their captures seems to bear no relationship.
Osmia – A classic spring bee, this group contains a mix of native and introduced species. Unfortunately the non-native O. taurus and O. cornifrons dominate the captures. Of the two, O. taurus is by far the most common with captures 6 times those of all the native species combined. O. lignaria, which is in the same subgenus as O. taurus and O. cornifrons is completely absent from this study and perhaps is suffering from competition with this group as it has been captured regularly in these woodlands in the past. O. virga is an ericaceous specialist and fittingly occurs only in the pine/oak-hickory sites with ericaceous understories.
Surveys using glycol traps hold promise as inexpensive long-term trapping techniques for native bee species. They appear to diminish the impact of phenological change in bee communities across years as well as the choice of sampling date within years. They also have the advantage that traps are set only once and servicing these traps can be done on a schedule without regard to weather. They are a natural addition to long-term weather stations and monitoring sites in general.
Examples of how to set up a glycol trap array and how to process specimens can be found at the following internet channels
Acknowledgements: Thanks to Sue Boo and Jelle Devalez for collecting specimens from traps while I was away. I stand on the shoulders of all those who have tried and failed and tried and succeeded in developing better ways of Silver bark of beech, and sallow surveying bees; my thanks and encouragement to continue on.
Bark of yellow birch and yellow
Twig of willow.
Stripe of green in moosewood maple,
Color seen in leaf of apple,
Bark of popple.
Wood of popple pale as moonbeam,
Wood of oak for yoke and barn-beam,
Wood of hornbeam.
Silver bark of beech, and hollow
Stem of elder, tall and yellow
Twig of willow.
- Edna St. Vincent Millay