Discussion
For a long time, the pollination capacity of non-bee insects visiting flowers has remained largely unexplored (Solomon and Kendall 1970, Boyle and Philogène 1983, Currah and Ockendon 1983, Kumar et al. 1985). Only recently developed molecular methods, such as metabarcoding, seem to have the potential to change this (Keller et al. 2015, Lucas et al. 2018) as confirmed by the much greater resolution and diversity of pollen present on the bodies of flower visiting insects collected from strawberry flowers in our study.
We observed a high diversity of non-bee visitors on day-neutral strawberry flowers at three sites in Southern Ontario and used metabarcoding to identify the pollen they carried. More than half of the non-bee genera were found to transport similar amounts of pollen to native bees from the genus Halictus (Figure 1). In fact, two members of the syrphid genus Eristalis had the largest overall pollen loads per individual (Figure 1, Table 3).
When assessing pollen loads at a coarse level, Syrphidae had the most available pollen, contributing more than four-times as much pollen as the species of Halictidae (Figure 1, Table 2). This was primarily due to the high pollen loads found on Eristalis tenax andEristalis similis (Figure 1b; Table 2), as well as the high abundance of Toxomerus marginatus (n = 114), which carried much less pollen per individual, but considering total pollen available in the field, all three species contributed meaningfully (Figure 1b; Table 2). Our observations during sampling indicated that syrphids were not stationary on flowers, they took flight at the slightest disturbance and alighted on neighbouring flowers. This observation is consistent with previous findings in studies on effective syrphid pollination, showing large pollen loads and rapid flower-flower movement (Bohart and Nye 1970, Solomon and Kendall 1970, Kendall and Solomon 1973, Nye and Anderson 1974, Kumar et al. 1985, Hodgkiss et al. 2018). Syrphid abundance has also been correlated with an increase in pollination and fruit set with a decrease in malformation of strawberry fruits (Stewart et al. 2017).
The fly families Polleniidae and Anthomyiidae were also found with larger amounts of pollen (Figure 2A). Polleniid flies are already known to be efficient pollinators of strawberry, imparting services equivalent to honey bees and have been used for stocking greenhouses (Free 1966, Carden and Emmett 1973, Clements 1982). Anthomyiids, also known as root-maggot flies, are a crop pest to strawberries, and thus their role as potential pollinators needs to be weighed against the consequence of their pest status. Interestingly, two of the three recorded ant species (Prenolepis imparis and Formica subsericea ) are confirmed pollinators (Ashman and King 2005) but the proportion of strawberry pollen on them varied substantially (30% to 100% respectively) (Appendix 1). The third ant species, Tetramorium caespitum(Formicidae), mostly carried strawberry pollen (92%; Appendix 1). The exclusion of Lepidoptera is unlikely to affect the assessment of non-bee flower visitors as their abundances were low (Figure 4B). The exclusion of Drosophila was necessary given the resources and collection methods. However due to their high abundance, even with small pollen loads, it is possible that they collectively provide substantial pollination services that we were unable to track. However, while observing them in the field they often did not move from flower to flower, but rather stayed clustered together and stationary on a single flower.
Most of the pollen found on non-bee flower visitors collected on strawberry flowers was indeed strawberry pollen, with an average of 69% across all measured pollen loads (Figure 3A). The species with the largest pollen loads carried over 70% strawberry pollen:Eristalis tenax (N= 350,017, 85% strawberry), Eristalis similus (N = 216,117, 70% strawberry), Toxomerus marginatus (N = 141,783, 76% strawberry), and Pollenia rudis (N = 105,320, 87% strawberry). Thus, these flower visiting species are likely contributing significantly to pollination and should be investigated further to rule out the possibility that they are stationary on a particular flower and therefore cannot be classified a pollinator.
Interestingly, the most generalist families (counted by plant genera) were also those that carried the largest amount of pollen (Figure 2, Supplementary Table 3), Syrphidae (58 plant genera), Polleniidae (53 plant genera), and Anthomyiidae (35 plant genera). Anthomyiid flies have been recorded as a largely generalist family of flower visitors (Larson et al. 2001). Within the Syrphidae, we found two generalist and one specialist species within Toxomerus (Figure 3). The genusToxomerus included both a generalist (T. marginatus ) and a specialist (T. geminatus ) which could be the result of speciation due to differing food exploitation strategies (Schluter et al. 1985). However, all plant (pollen) diversity estimates should be treated cautiously when considering which plants these insects visit, as many of the genera identified with metabarcoding were grasses (Poaceae) with 15 genera identified, and other wind-pollinated plants (Rabinowitz et al. 1981). The presence of wind-pollinated plants in the samples could be incidental, found on these insect bodies as the result of contact with windborne pollen when flying, rather than a confirmed visit to the plant itself (although this also cannot be excluded as a possibility).
The RDA analysis demonstrated that a number of environmental variables were poor predictors of insect community visitation to strawberry flowers (Figure 4A). A strong explanatory variable in the model was sampling date and to a lesser degree time of day (during sampling days). This suggests that the flower visitor community was quite different on each day of sampling. As such, this model could be detecting phenological patterns of the non-bee visitors; insects that emerge and are abundant for a short time and not recorded outside of their biological timeline. This is supported by observations (see Figure 4B), where large peaks of activity can be found for some taxa, particularly Syrphidae and Formicidae. Many insects are restricted to narrow ranges of temperature for flight, as endothermy is a rare trait in insects, requiring a rise in ambient temperature or basking in sunlight to warm their flight muscles (Inouye et al. 2015). Most syrphid species, however, do have endothermic capabilities which allows them to forage in cloudy and cool weather (Morgan and Heinrich 1987). Other dipteran families also forage when bees and butterflies do not (Hooper 1932, Inouye et al. 2015). Indeed, during field sampling, syrphids and other flies were foraging on cool, overcast days and even in light rain. Low abundance of solitary bees, particularly Dialictus , were out visiting strawberry flowers during these less-than-ideal weather conditions but they were stationary, and not apparently actively engaged in pollinating during this time. This range in degree of specialization(s) could reduce the effect of the environmental variables in the model.