3.1. Bee community summary
We collected 2,121 bee specimens representing six families, 27 genera, and 125 species (Table S1). In 2013, we collected 225 bees representing 52 species in the Downeast growing
region. Sites near lowbush blueberry fields contained 147 bees representing 39 species, and isolated sites had 78 bees representing 27 species. In 2014, we collected 847 bees from 92 species across both the Midcoast and Downeast growing regions, and in 2015, we collected 1,049 bees from 95 species across both regions. Across all three sampling years, we collected 1,177 bees from 107 species in sampling sites near lowbush blueberry fields and 944 bees from 98 species in sites isolated from lowbush blueberry fields.
The most abundant bee species collected was the sweat bee Lasioglossum cressonii (Robertson) (535 individuals collected). Other common bees included the bumble bees Bombus ternarius (Say) (249) and B. vagans (Smith) (159) and the sweat bee Augochlorella aurata (Smith) (193). Uncommon bee genera included Colletes and Osmia, both of which contain species associated with lowbush blueberry. We collected two specimens of Macropis nuda (Provancher), a rare solitary bee that specializes on oils produced by Lysimachia spp., a plant group we found throughout our sites in both growing regions. We also collected one new bee species state record for Maine, represented by one specimen of Melitta americana (Smith). This species is a rare specialist of Vaccinium spp., particularly cranberry; our specimen was collected Downeast, where wild cranberry (Vaccinium macrocarpon Aiton) is abundant (Jones et al. 2014). A complete list of bee species, their size class, and the site types and growing regions in which they were collected is presented in Table S1.
We found annual differences in bee communities between 2013 and 2014 (Dunn’s test abundance Z=-2.14, p=0.01) and 2013 and 2015 (abundance Z=3.44, p<0.001, richness Z=-2.40, p=0.001), with no differences between 2014 and 2015. Additionally, bee abundance was temporally autocorrelated when 2013 data were included in analyses (Durbin-Watson test statistic=1.36, p=0.03); however, species richness was not (D-W=1.77, p=0.28). Owing to our different sampling scheme in 2013, we conducted all bee community data analyses on specimens collected in 2014 and 2015. Bee abundance and species richness changed throughout the sampling season (Kruskal-Wallis abundance χ2 = 19.19, df=3, p<0.001, richness χ2 = 23.30, df=3, p<0.001). Bees were less abundant (538 individuals) and less species rich (62 species) in the late season than early (595 individuals, 63 species) or mid-season (976 individuals, 88 species), with no significant differences in abundance or species richness between early and mid-season. We found no temporal autocorrelation in our data for bee abundance (D-W=2.07, p=0.72) or species richness (D-W=2.16, p=0.98) in 2014-2015. Further, we found no spatial autocorrelation within growing regions across sampling years (Mantel test: Midcoast abundance r=-0.12, p=0.79, species richness r=0.14, p=0.15; Downeast abundance r=-0.19, p=0.98, species richness r=-0.04, p=0.59).
3.2. Floral resource availability
We observed 62 plant species in flower while sampling, 10 of which were non-native, in our corridor sites. Floral species richness was greater in the Midcoast region, where we recorded 51 species, than in the Downeast region, where we recorded 31 species (F(1,22)=13.11, p=0.001). There was no significant difference in floral abundance between growing regions, and no difference in floral abundance or species richness between corridor sites near to or isolated from lowbush blueberry fields. Eleven flowering plant species were recorded exclusively Downeast, whereas 31 occurred exclusively Midcoast. Downeast-exclusive species tended to be ericaceous and bog associated, including sheep laurel (Kalmia angustifolia L.), Labrador tea (Ledum groenlandicum Oeder), and Rhodora (Rhododendron canadense (L.) Torr.). In contrast, Midcoast-exclusive species were often disturbance-associated, including boneset (Eupatorium perfoliatum L.), chickweed (Stellaria/Cerastium sp.), yellow clover (Trifolium arvense Pollich), hawkweed (Hieracium spp., 2 species), and evening primrose (Oenothera sp.). Common species in both growing regions were meadowsweet (Spiraea alba Du Roi), creeping raspberry (Rubus spp.), bunchberry (Chamaepericlymenum canadense L.), and fall asters (Solidago spp., 3 species; Euthamia graminifolia (L.) Nutt. var. graminifolia). We also frequently observed two species of native loosestrife, Lysimachia quadrifolia L. and L. terrestris (L.) Britton, Sterns, & Poggenb, in both growing regions. These species produce oils that attract rare bees in the genus Macropis. A full list of floral species and the growing region(s) in which they were observed is provided in Table S2.
3.3. Landscape composition
We found more coniferous forest and wetland land cover at all spatial scales surrounding our corridor sites in the Downeast growing region than the Midcoast region (Table S3). In contrast, we found more deciduous/mixed forest at 100, 500, and 1000 m, more urban land cover at 500 and 1000 m, and more deciduous/mixed forest edge and agriculture/pasture at all spatial scales around our corridor sites in the Midcoast growing region than the Downeast region (Table S3). Our corridor sites near lowbush blueberry fields were surrounded by more lowbush blueberry land cover at all spatial scales, and we also found more coniferous forest around our sites isolated from lowbush blueberry fields at 500 and 1000 m (Table S3). We did not find a difference in the amount of lowbush blueberry surrounding our sites between growing regions.
3.4. Bee communities by growing region and site type
We collected 820 individual bees from 84 species in the Downeast growing region and 1,076 bees from 97 species in the Midcoast growing region in 2014-2015. We found significant differences in both bee abundance (df=1,21, deviance = 3.94, p=0.05) and species richness (df=1,21, deviance=16.24, p<0.001) between the two growing regions. We collected 1,030 bees from 98 species in corridor sites near lowbush blueberry fields and 866 bees from 92 species in corridor sites isolated from lowbush blueberry fields, though differences in bee communities across these site types were not significant. However, we found a significant type x region interaction for bee species richness (df=1,21, deviance= 5.21, p=0.02). Bee species richness was significantly different between corridor sites near to and isolated from lowbush blueberry in the Downeast growing region (df=1,10, deviance=7.24, p=0.007), while we found no difference in bee species richness between site types in the Midcoast growing region (Table 1).
Table 1
Wild bee abundance and species richness (± standard error) in power line corridor sites near to and isolated from lowbush blueberry fields in Downeast and Midcoast Maine, USA, 2014-2015.
Abundance
|
Near lowbush blueberry
|
Isolated from lowbush blueberry
|
Downeast
|
448 (±12.6)
|
372 (±6.8)
|
Midcoast
|
582 (±12.6)
|
494 (±8.8)
|
Species Richness
|
|
|
Downeast **
|
71 (±2.3)
|
47 (±1.6)
|
Midcoast
|
76 (±2.6)
|
80 (±2.2)
|
** = Significantly different at p<0.01 |
3.5. Bee life-history traits and corridor context across growing regions
We found significant differences in medium (df=1,21, deviance=15.10, p<0.001) and extra-large sized (df=1,21, deviance=14.59, p=0.008) bee species richness by growing region (Fig. 2). Medium sized bee species richness is greater in the Midcoast growing region, where it is associated with more deciduous/mixed forest edge (df=1,22, deviance=7.88, p=0.004) and less coniferous forest (df=1,22, deviance=8.65, p=0.003) surrounding corridor sites at 250 m (Fig S1a, b). Extra-large sized bee species richness is also greater Midcoast, where it is influenced by more agriculture/pasture at 1000 m surrounding corridor sites (df=1,22, deviance=4.11, p=0.042) (Fig. S2a). Extra-large sized bee species richness was also positively influenced by floral abundance (df=1,22, deviance=17.79, p=0.042) (Fig. S2b).
We found significant differences in the species richness of social bees (df=1,21, deviance=7.12, p=0.007) and solitary bees (df=1,21, deviance=10.97, p<0.001) between growing regions. Solitary bee species richness was greater Midcoast, where it is associated with less coniferous forest and wetland and more agriculture/pasture and deciduous/mixed forest edge surrounding corridor sites (Fig. 3a; Table S4a). We found a significant site type by growing region interaction for the species richness of solitary bees (df=1,20, deviance= 6.98, p=0.008) owing to an interactive effect of lowbush blueberry land cover surrounding our corridor sites at 250 (df=1,20, deviance= 7.54, p=0.006), 500 (df=1,20, deviance= 6.46, p=0.01), and 1000 m (df=1,20, deviance= 8.68, p=0.003). Lowbush blueberry had a negative association with solitary bee species richness in the Midcoast region and a positive association in the Downeast region (Fig. 4), though as a single variable, the amount of lowbush blueberry surrounding corridor sites did not significantly influence solitary bee species richness (Table S4a). Furthermore, we found that greater floral abundance (m2 per transect) in corridor sites was associated with greater solitary bee species richness (df=1,22, deviance=6.62, p=0.01) (Fig. S3a). Social bee species richness was also greater in the Midcoast region, increasing with less coniferous forest cover and more agriculture/pasture (Fig. 3a; Table S4b) in the surrounding landscape.
Ground nesting (df=1,21, deviance=9.98, p=0.001) and cavity nesting (df=1,21, deviance=4.45, p=0.034) bee species richness was greater in corridors in the Midcoast region (Fig. 3b). Less coniferous forest and more agriculture/pasture influenced ground nesting bee species richness at multiple scales surrounding corridors, whereas more deciduous/mixed forest edge and urban area and less wetland were influential at small scales surrounding corridors (Table S4c). Greater floral abundance in corridor sites led to greater ground nesting bee species richness (df=1,22, deviance=7.93, p=0.004) (Fig. S3b). We found greater cavity nesting bee species richness in corridor sites surrounded by less coniferous forest and more deciduous/mixed forest edge at small scales and more wetland at 500 m (Table S4d).
We found significant differences in bee species richness associated with body size (df=1,10, deviance=13.00, p=0.05), social habit (df=1,10, deviance=4.85, p=0.027; Fig. 5a), and nesting preference (df=1,10, deviance=5.43, p=0.019; Fig. 5b) between corridor sites near to and isolated from lowbush blueberry fields in the Downeast growing region. Large sized (500 m scale; df=1,10, deviance=4.73, p=0.03), solitary (all spatial scales; Table S5a), and ground nesting (all spatial scales; Table S5b) bee species richness were positively associated with surrounding lowbush blueberry land cover at sites near crop fields. Ground nesting bee species richness decreased in both near and isolated sites with greater coniferous forest cover surrounding corridor at 500 and 1000 m, whereas greater emergent wetland land cover at 1000 m led to increased species richness of ground nesting and solitary bee species richness. Finally, more deciduous forest in the 1000 m surrounding corridor sites near lowbush blueberry fields reduced solitary bee species richness, and interestingly, there was no effect of coniferous land cover on bee species richness at any spatial scale surrounding near sites (Table S5a, b). We found no significant differences in bee life-history traits between site types in the Midcoast growing region.