Improving apple and peach pollination by advancing knowledge of how forest management affects wild bee functional diversity

(1) Determine how forest management practices, through their influence on forest structural complexity, affect wild bee functional diversity on forest lands

How does forest management and related environmental variables affect the distribution of bee functional traits? 

Results

Spring bee functional assemblages 

We performed RLQ analysis to assess how bee trait assemblages respond to forest management and environmental variables. RLQ is an ordination technique that allowed us to visualize trait-management relationships across multiple dimensions. In the first dimension (RLQ axis 1), we found that unmanaged plots were most different from thin+burn plots. We found that burn-only plots were most similar to unmanaged plots, and thin+burn plots fell in between burn-only and thin-only plots (Figure 2; Table 1). Thin-only plots were positively correlated with the amount of litter, holes and the total number of floral resources (Table 1), opposite to unmanaged plots. Additionally, in the second dimension (RLQ axis 2), we found that burn-only plots were most different from unmanaged plots (Figure 3; Table 1). There was a positive correlation between burn-only plots and bulk density and slope (Table 1). In addition, unmanaged plots had significantly higher canopy cover and increased dead wood and holes (Table 1).

Body size, nesting strategy, parasitism, sociality, scopa number, tongue length and flight phenology were significantly associated with the first RLQ axis (Table 1). Larger bees with longer tongues were more prevalent in unmanaged plots, as well as cavity renting bees. (Table 1; Figure 2). The community of bees most associated with thin-only plots was smaller, non-cleptoparasitic, cavity-excavators, primitively eusocial bees, bees with more scopa locations and bees with longer activity periods (Table 1; Figure 2).

Table 1: Relationships between site scores and environmental variables and species scores and bee traits for RLQ axis 1 and 2 in spring. Correlation coefficients are based on Spearman rank correlations (ρ) for continuous variables and Kruskal-Wallis chi-squared statistics (χ² ) for categorical variables. p < 0.1*; p < 0.05**; p < 0.01***

Summer bee functional assemblages 

Unlike spring, we did not find that treatment significantly associated with RLQ axes and affected environmental variables (Table 2). Rather, RLQ axis 1 separated plots by nesting resources and floral resource availability. Plots on the negative side of the axis were characterized by increased bare ground and more wood and holes within sites, while on the positive side they were characterized by increased litter, pithy stems, total number of blooms and steeper slopes (Table 2; Figure 4).

In summer, environmental variables were associated with different bee functional traits. Nesting strategy, sociality and scopa number were significantly associated with RLQ axis 1. Ground nesting bees, primitively eusocial bees and bees with more scopa locations on their body were more abundant in plots with increased floral resource availability and steeper slopes (i.e. litter, pithy stems and total number of blooms) (Table 2; Figure 4). Plots with more nesting resources (bare ground, wood and holes) had more cavity nesting (excavators and renters) and diverse nesting bees and solitary bees (Table 2; Figure 4).

Table 2: Relationships between site scores and environmental variables and species scores and bee traits for RLQ axis 1 and 2 in summer. Correlation coefficients are based on Spearman rank correlations (ρ) for continuous variables and Kruskal-Wallis chi-squared statistics (χ² ) categorical variables. p < 0.1*; p < 0.05**; p < 0.01***

Discussion

Our findings demonstrate that specific bee functional traits align with different management strategies, including no management, and this relationship changes throughout the year. Both managed and unmanaged habitat provide vital nesting and floral resources that alter bee functional trait composition. Collectively, our results suggest that retaining both unmanaged and sustainably managed stands within a temperate forested landscape can promote a functionally diverse bee community and maintain crucial pollination services for the plant community.

In spring, bee traits aligned with specific management strategies depending on availability of resources. Thinned plots with increased floral resources supported small bees that are more dispersal limited. This is important as bee body size is linked to many behavioral and morphological traits that affect fitness and potentially pollination services, including but not limited to foraging flight distance, overwintering survival and offspring survival. 

Thin-only management and unmanaged plots increased nesting resources for bees in spring. Snags and stumps that were not removed from thin-only plots may have provided additional nesting resources for cavity-excavators. Unmanaged plots specifically provided both deadwood and holes for bees. These results suggest that management practices that retain dead wood in the forest post-management (e.g., retention forestry) should be used to increase nesting resources for bees. This may also indicate that burning by itself or burning in combination with thinning may eliminate key nesting resources that bees rely on in forests.

Unmanaged plots also supported several of the most vulnerable bee groups (i.e., cleptoparasites and specialists). Both of these bee groups are more likely to face future declines and may be more sensitive to disturbance compared to other bee groups. Cleptoparasitic bees stabilize and regulate the bee community by preferentially targeting dominant bee species and are therefore imperative for maintaining a healthy bee community. Given this, it is imperative that areas of unmanaged forest habitat remain on the landscape intermixed with managed areas. 

Overall, findings from this portion of the project emphasize the importance of using an array of forest management strategies to support a functionally diverse bee community. 

How does structural complexity affect wild bee communities? 

Results 

Spring and summer bee diversity

Spring bee diversity was negatively related to overall vertical height variability (Fig. 5A), and positively related to the mean vegetation density between 0 - 2 m (Fig. 5B). Summer bee diversity was affected by different LiDAR-derived structural metrics. Summer bee diversity was negatively related to percent canopy gap (Fig. 6A), mean vegetation density between 2 - 5 m (Fig. 6B), and the variability of vegetation density between 0 - 2 m (Fig. 6C).

Spring and summer bee abundance 

Spring bee abundance was negatively related to mean overall vegetation density (Fig. 7A) and positively related to bloom abundance (Fig. 7B). Again, summer bee abundance was affected by different LiDAR-derived structural metrics. Summer bee abundance was positively related to the variability of overall vegetation density (Fig. 8A), and negatively related to mean vegetation density between 2 -5 m (Fig. 8B) and mean vegetation density between 5 – 15 m (Fig. 8C). 

Discussion

Many studies have found that understory vegetation density is an important predictor of forest bee communities. The majority have studied this in the context of increased floral resources provided by herbaceous plants on the forest floor. In temperate forests in spring, ephemeral wildflowers provide some of the earliest floral resources and support a taxonomically and functionally diverse bee community. My results support this, as bloom abundance outcompeted all LiDAR metrics in predicting spring bee abundance. However, increased vegetation density between 0 - 2 m also positively affected bee abundance and diversity in spring. This may reflect increased floral resources from woody shrubs and small flowering trees (e.g., serviceberry, dogwood, and American plum).

In summer, vegetation density between 2 - 5 m had a negative effect on bee diversity and abundance. This mirrors other work that has found that subcanopy openness and not just canopy openness is a strong predictor of bee community structure. This relationship is likely more apparent in summer once the canopy has leafed out. The effect was especially pronounced in mesophytic forest plots, where we observed an abundance of mid-story beech trees and dense shade. Densely shaded stands can lead to a reduction in floral resource availability, which can have negative effects on bee communities.

(2) Determine which wild bee functional groups are present in apple orchards; compare those wild bee functional groups to managed forest lands

Results and Discussion: 

We found that bee species composition in apple orchards was significantly different from all forest management types with the first two PCoA aces explaining 31% of total variation (P = 0.001; R² = 0.61; Fig. 9).

We sampled apple orchards in 2021 but were unable to gather data in 2022 due to poor weather conditions that hindered our ability to sample bees. During the apple flowering period in 2022, conditions were rainy and windy. On days with clear skies, temperatures were too low and there were very few bees seen foraging from apple flowers. As a result, we only have results from 2021 apple sampling to share. This limits our ability to provide recommendation to growers and relay information regarding wild bee functional composition between apple orchards and managed forest lands. As a result, we instead compared how bee species composition differed between apple orchards and forest management strategies. While there were significant differences in bee composition between apple orchards and forest management treatments, we were unable to detect if a specific management treatment had more similar bee composition to apple orchards over others. Given this, we cannot say which forest management strategy may be best for apple pollination. However, other papers have found that increased bee functional diversity can increase pollination services to apples. Given that we found different bee functional traits aligned with different forest management strategies, this may mean a diversity of management strategies surrounding apple orchards is best to increase pollination services. We hope to continue this work in the future.