Wind drives temporal variation in pollinator visitation in a fragmented tropical forest

1. Introduction

Animal pollinators such as bees provide critical ecosystem services that support biodiversity [1] and global crop yields [2]. The composition and abundance of pollinator communities can vary substantially in both space [3] and time [4], likely as a result of both stochastic fluctuations and small-scale variation in the biotic and abiotic environment [5].

Wind strongly affects flying insects and may be an important factor in spatio-temporal heterogeneity of pollinator visitation. Wind affects macroecological patterns of insect dispersal and migration [6,7]. Mean wind flow [8] and fluctuations (i.e. turbulence) pose biomechanical challenges [9–13] that push manoeuvrability limits in flying insects [14] and can impose energetic costs on flight [15]. The mechanical and physiological challenges posed by wind may have important effects on insects' interactions with plants, including herbivory [16,17], and pollinator visitation and landing [18]. Wind also indirectly impacts flying insects by inducing plant movements [19], which can impose additional manoeuvrability challenges [14]

In addition to biophysical challenges, wind disperses chemical cues and signals critical for interactions between insects (e.g. attracting mates [20], or locating prey [21,22]), as well as between insects and plants (e.g. pollinator attraction [23–25] and herbivory [26]). Turbulence (i.e. fluctuations in wind on top of mean flow speed and direction) is a critical factor in olfactory navigation, as it disperses odours into complex plumes [27,28].

While wind is known to have many effects on pollinator behaviour and ecology, our understanding of its influence on bee abundance and plant visitation rates is limited. While extreme wind speeds restrict bee flight and foraging (reviewed in [29]), recent studies have found that wind can have either positive [30] or negligible [10,31] effects on bee abundance and activity. Some of the variable effects of wind could be explained by differences in sensory ecology, particularly the importance of olfactory cues for locating floral resources.

Orchid bees (Apidae: Euglossini) are a key group of pollinators for which wind-borne odours likely play a uniquely critical role. The approximately 200 species (across four genera) of orchid bees are important neotropical pollinators of orchids and several other plant families [32], in addition to pollinating important crop plants such as tomato [33,34]. In this group, foraging behaviour intersects with mating strategies, as male orchid bees gather species-specific combinations of volatile compounds from a wide variety of flower species [35]. These fragrance ‘bouquets' are thought to play a key role in attracting mates [36]. Orchid bees are primarily forest associated, and they collect fragrances from flowers and other sources that are often sparsely distributed within tropical rainforests. Accordingly, male orchid bees show strong patterns of long-distance movement and dispersal across the landscape [37–40] and are thought to locate floral scent resources in dense vegetation using olfactory cues.

Despite its likely importance in dispersing scents, the role of wind in orchid bee ecology is not well understood, although previous observers have noted temporary increases in orchid bee arrivals at baits after wind gusts [41]. Recent work has also shown that male orchid bees performing mating displays strongly prefer to orient on the downwind side of trees, presumably to maximize the dispersal of odour plumes [42]. However, to our knowledge, quantitative studies of the impacts of wind (or turbulence) on spatial or temporal variation in orchid bee abundance and visitation rates at scent sources have not been made.

Local wind patterns can be strongly affected by forest fragmentation [43], which is increasingly characteristic of the Neotropical range of orchid bees. Previous work has shown that forest fragmentation significantly alters orchid bee communities (compared to continuous forest [44]), and some studies have found that fragment traits (such as size and shape) significantly affect orchid bee abundance and community composition [45–47] (although other studies have not found evidence for such a relationship [48,49]). In addition to variation across fragments, abundance and composition of orchid bee communities can vary locally within fragments, for example as a result of microhabitat variation [47,50,51] and edge effects [52,53]. Recent work has also shown that local landscape structure (e.g. forested area within 1000 m) can have significant effects on bee community structure and abundance [54].

Here, we explore the effects of local landscape structure, wind speed and turbulence on male orchid bee visitation rates to scent sources within and adjacent to a large tropical forest fragment. In addition to a positive association between visitation and local forest cover, we predict that wind speed positively correlates with orchid bee visitation, because higher wind speeds will further disperse scents and attract bees from a wider area. Conversely, we hypothesize that turbulence is negatively correlated with visitation, because stronger turbulence will result in higher costs of flight and make odour plumes more challenging for bees to track.

2. Material and methods

(a) Sampling sites and orchid bee collection

We collected male orchid bees from nine different sites within or adjacent to a large forest fragment at the Las Cruces Biological Station (8.79°, −82.96°) in Coto Brus, Puntarenas Province, Costa Rica (figure 1, electronic supplementary material, table S1). The forested areas in this region are characterized by premontane forest, with a high abundance and diversity of euglossine bees [45].

• Download figure
• Open in new tab
• Download PowerPoint

Male orchid bees were collected from each site multiple (5–9) times between 1 October and 16 November 2014 (electronic supplementary material, table S2). All collections occurred between 08.30 and 11.30, roughly corresponding to peak daily abundance. Bees were sampled by saturating tissue paper with one of two compounds (cineole or methyl salicylate) and suspending this scent bait approximately 1.5 m above the ground in a permeable metal tea infuser. Male orchid bees arriving at the scent bait were collected by hand netting for 20 min. Bees were identified (electronic supplementary material, table S1) independently by two authors (J.B. and J.D.C.) using an established taxonomic key [55]. Identifications of example specimens for each species were reviewed and corrected by Dr Santiago Ramírez (UC Davis).

3. Results

We collected a total of 409 male orchid bees, from 23 different species (electronic supplementary material, table S3). Orchid bee visitation varied significantly among sites (Kruskal–Wallis test, χ² = 26.2, d.f. = 8, p = 9.7 × 10⁻⁴) and was positively correlated with the percentage of forested area within 200 m of collection sites (figure 2a, generalized linear mixed model, d.f. = 64, z = 2.56, p = 0.01). In addition, we found that wind speed had a significant, positive effect on orchid bee visitation over time within sites (figure 2b, generalized linear mixed model, d.f. = 64, z = 3.45, p = 0.0005). Turbulence also had a significant positive effect on orchid bee visitation (figure 2c, generalized linear mixed model, d.f. = 64, z = 2.87, p = 0.004).

• Download figure
• Open in new tab
• Download PowerPoint

The same positive effects of wind speed and turbulence on visitation were not found between sites. Both wind speed and turbulence also varied significantly between sites (Kruskal–Wallis test; wind speed, χ² = 31.3, d.f. = 8, p = 0.0001; relative TKE, χ² = 24.1, d.f. = 8, p = 0.02). However, sites with higher average wind speeds did not have higher median visitation (electronic supplementary material, figure S1A). Likewise, sites with higher mean turbulence did not have higher median visitation (electronic supplementary material, figure S1B).

4. Discussion

Our results suggest that both landscape and wind play important roles in driving male orchid bee visitation rates. In particular, our findings support previous observations that orchid bee communities can vary significantly across small spatial scales (i.e. less than 1 km) [47,50,52] and suggest that forest cover in particular has important effects on local variations in orchid bee abundance. This supports previous work showing that local forested area (within 1000 m) can significantly alter the composition and abundance of bee communities in general [54], and orchid bee communities in particular [49]. Our findings show that these effects occur on even shorter spatial scales (i.e. 200 m) and can drive spatial variation within forested fragments.

We also found a significant, positive effect of wind speed on visitation rates over time within sites (figure 2b), despite relatively low wind speeds (median wind speed of 0.122 m s⁻¹, versus a global terrestrial average of approximately 3.1–3.4 m s⁻¹ [62]). The simplest explanation of this pattern is that increased wind speeds disperse attractive scents over greater distances, attracting male orchid bees from a wider area.

The positive association between wind speed and visitation within sites does not hold true between sites, as windier sites did not have higher visitation (electronic supplementary material, figure S1). A possible explanation for this pattern is that spatial variation (i.e. between sites) in orchid bee male abundance is primarily driven by local landscape context, while temporal variation (i.e. within sites) in visitation is more strongly affected by wind. For example, increased forest cover may increase the number of individuals located near a site at any given time, while higher wind speeds may increase the likelihood of those individuals visiting a scent source.

We found no evidence that turbulence decreases orchid bee visitation; instead, higher relative turbulence was positively associated with visitation (figure 2c). While turbulence is often considered a challenge for olfactory navigation, we speculate that a possible explanation for this result is that turbulence also plays a role in dispersing attractive odours via turbulent diffusion, an effect that could be especially important when mean wind speeds are low.

Overall, our results underscore the importance of wind for pollinator ecology, particularly in species that rely strongly on olfactory cues. Previous work has noted that orchid bee abundance can vary significantly across time within sites [45]. Our results are consistent with these observations and suggest that temporal variation within sites is driven, in part, by variation in wind speed and turbulence, consistent with previous anecdotal observations [48]. Future studies investigating a wider range of environmental conditions and a diversity of pollinator species will be critical for understanding the generality of the findings here (e.g. whether the positive effects of wind are restricted to species that rely primarily on olfactory cues).

These results may also have important implications for the timing of volatile release in Euglossine bee-attracting plants. The release of pollinator-attracting fragrances from many flowers shows strong rhythmicity, likely synchronized with the activity of pollinating animals [63]. Our findings suggest that timing and efficacy of scent release by flowering plants could also be shaped by wind conditions.

More broadly, our results underscore the potentially important but understudied role of wind in pollinator behaviour and ecology. In addition, anthropogenic environmental change is altering wind conditions at both the local (e.g. forest fragmentation [43,64]) and global (e.g. planetary shifts in wind and weather patterns [62,65]) scales. However, the effects of these changes on pollinators and the ecosystem services they provide are not well understood and represent an important direction for future study.

Data accessibility

Raw data and custom scripts are available on Zenodo https://doi.org/10.5281/zenodo.3743898 [66].

Authors' contributions

J.D.C., N.E.P. and S.A.C. conceived and designed the study. J.D.C., S.G. and J.B. collected data on orchid bees and wind. C.D.M. generated landscape data. J.D.C. analysed data and performed statistical analyses. All authors contributed to writing and editing the manuscript. All authors agree to be held accountable for the content herein and approve the final version of the manuscript

Competing interests

We declare we have no competing interests

Funding

This work was supported by a NSF GRFP fellowship to J.D.C., an Organization for Tropical Studies Pilot grant to J.D.C., a Company of Biology Traveling Fellowship to J.D.C., a David Rockefeller Center for Latin American Studies grant to J.D.C., NSF IOS-1257543 to N.E.P., and NSF CAREER IOS-1253677 to S.A.C.