Pollinators at high elevations face multiple threats from climate change including heat stress, failure to phenological match advancing flower resources and competitive pressure from range-expanding species of lower elevations. We conducted long-term multi-site surveys of alpine bumble bees to determine how phenology of range-stable and range-expanding species is responding to climate change. We ask whether bumble bee responses generate mismatches with floral resources, and whether these mismatches in turn promote community disruption and potential species replacement.

In alpine environments of the central Rocky Mountains, range-stable and range-expanding bumble bees exhibit phenological mismatches with flowering host plants due to earlier flowering of preferred resources under warmer spring temperatures. However, workers of range-stable species are more canalised in their foraging schedules, exploiting a relatively narrow portion of the flowering season. Specifically, range-stable species show less variance in phenology in response to temporally and spatially changing conditions than range-expanding ones. Because flowering duration drives the seasonal abundance of floral resources at the landscape scale, we hypothesize that canalisation of phenology in alpine bumble bees could reduce their access to earlier or later season flowers.

Warmer conditions are decreasing abundances of range-stable alpine bumble bees above the timberline, increasing abundance of range-expanding species, and facilitating a novel and more species-diverse bumble bee community. However, this trend is not explained by greater phenological mismatch of range-stable bees. Results suggest that conversion of historic habitats for cold-adapted alpine bumble bee species into refugia for more heat-tolerant congeners is disrupting bumble bee communities at high elevations, though the precise mechanisms accounting for these changes are not yet known. If warming continues, we predict that the transient increase in diversity due to colonization by historically low-elevation species will likely give way to declines of alpine bumble bees in the central Rocky Mountains.

Study system

In our alpine (above treeline) study sites (Colorado, USA), we classified bumble bees as resident or range-stable if pre-1980 records showed them completing their entire life cycle above timberline (B. kirbiellus and B. lapponicus sylvicola, formerly B. balteatus and B. sylvicola, respectively; Byron, 1980). Observations of B. lapponicus likely included the recently discovered cryptic taxon “incognitus” as it overlaps in distribution and is morphologically indistinguishable (Christmas et al., 2021). Species were deemed colonizing or range-expanding if populations before 1980 were commonly found in lower subalpine and montane habitats and nesting was not observed above treeline (B. bifarius, B. flavifrons, B. frigidus, B. melanopygus, B. mixtus, and B. nevadensis; Byron, 1980; Macior, 1974). Flowering phenology and floral density were monitored for nine plant species that together accounted for ~ 90% of floral resources used historically by alpine bumble bees: Trifolium dasyphyllum, T. parryi and T. nanum, Penstemon whippleanus, Pedicularis groenlandica, Castilleja occidentalis, Mertensia lanceolata, Polemonium viscosum, and Rhodiola rhodantha (Byron, 1980; Macior, 1974; Miller-Struttmann et al., 2015). 

Floral resources were surveyed as part of a long-term phenological study (13 yrs. between 1977 and 2019; Table 1) on Pennsylvania Mountain, Colorado (39°15.803′N, 106°8.564′W; 3600-3970 m). Timing and abundances of bumble bee foragers were monitored from 2012-2014 in elevationally-stratified zones of two Front Range sites [Niwot Ridge (low alpine: 40.0567°N, 105.5916°W, 3564m and mid alpine: 40.0585°N, 105.6119°W, 3700) and Mount Evans (low alpine: 39.6454°N, 105.5925°W, 3550m; mid alpine: 39.6334°N, 105.6046°W, 3700m; and high alpine: 39.5964°N, 105.6275°W, 3850m)] where L. W. Macior documented the historic ranges (1966-1969) and flowering hosts of species sampled in this study. Host plants and bumble bees were sampled weekly throughout the flowering season (June-August) in alpine habitats including krummholz vegetation, rocky fellfields, exposed ridges, mesic tundra meadows, and swale or marsh vegetation.

(1) Climate cues affecting flowering time in the host plant-community

Twenty-three 2 m × 10 m plots were established on Pennsylvania Mountain by P.G. Kevan and surveyed weekly from the onset to the end of the flowering season in 1977-1981. Plots were located nonrandomly to sample prevalent alpine habitat types (Byron, 1980). Using aerial photographs and maps in 2012, we found at least one edge of each original plot and established a larger survey area that included the original plot in its entirety (Miller-Struttman et al. 2015) . To account for larger sampling scale from 2012-2019, flower density was based on the average flower density in all possible independent 2 × 10 m subplots within the larger plots. Open flowers or inflorescences were counted weekly to provide comparable data to surveys from 1977-81. Inflorescence counts at randomly chosen points were used to estimate the average number of open flowers per m2. From flower counts we extracted day of first flowering, day of peak flowering, and cumulative flower density (total flowers produced per m2 over the entire season) for each of the nine host-plant species.

(2) Cues driving foraging phenology of colonizing vs. resident bumble bee species

In the Front Range (Niwot Ridge and Mount Evans), we collected and identified to species and caste all bumble bees observed foraging on flowers without regard to plant taxon, during weekly, one-hour walking surveys in each elevational zone (median area of 4.22 ha, and range of 2.13-6.57 ha; Miller-Struttmann et al., 2015). Individuals were anesthetized, photographed, marked with non-toxic paint, and released. In the rare event that a marked bee was recaptured, it was released without additional data collection and was not included in the analyses presented here. A small subset of individuals were lethally collected for archival purposes and deposited in the Enns Entomology Museum at the University of Missouri.

Samples were pooled elevationally to characterize bumble bee communities within the average flight distance of alpine bumble bees (<1km; Geib et al., 2015) for a total of five elevational zones in the Front Range (Niwot Ridge and Mount Evans). Species abundances were weighted by collection effort in person hours for each elevational zone (median = 663; range: 198-840 person hours). For queen and worker bumble bees of each species encountered, we estimated the timing of emergence and peak foraging activity. Respectively, emergence was operationally recognized as the day of year (DOY) of the first observation of that species/caste, and peak activity as the DOY when the maximum number of foraging individuals were observed in a given elevational zone. 

Queens of three colonizing species (B. flavifrons, B. melanopygus, and B. mixtus) were infrequently collected (fewer than five populations sampled over three years) and therefore excluded from these analyses.

To address whether seasonal patterns of forager activity are more canalised for resident than colonizing bumble bees, we compared curves for temporal variation in forager activity over the season. The number of workers of each species on each date was calculated at the regional level by pooling samples from all locations and elevational bands. The activity curves for each species were tested for kurtosis, which quantifies the degree to which the peak of a curve is narrower or broader than expected for a normal distribution. We calculated the mean elevation at which bumble bees of each species were collected along the 2700 m elevational gradient (mean number of individuals per species = 1603).

(3) Impacts of climate cues on synchrony of colonizers and residents with floral resources

We surveyed flowering weekly in the nine host-plant species using 12-15, 1m x 2m plots placed within each elevational zone on Niwot Ridge and Mount Evans. In 2012, we established an initial set of ten 1m x 2m plots before the onset of flowering in each elevational zone. As the flowering season progressed, an additional 2-5 plots were established to survey areas of later-blooming. Open flowers or inflorescences for each of the nine host plants were counted weekly from onset to end of the flowering season and converted to number of flowers using the mean flowers per inflorescence for a sub-sample (3-30) of inflorescences per plot. For two species (T. dasyphyllum and T. parryi), we used published flower counts per inflorescence from Pennsylvania Mountain (Geib, 2010) and for one (C. occidentalis) flower counts from individuals outside of plots on Niwot Ridge and Mount Evans (2012-2013). Days of first and peak flowering and cumulative flower density were calculated for each elevational zone.

For workers of each bumble bee species, we estimated the yearly phenological mismatch between foraging activity and flowering of the host-plant community in each elevational zone. To test for mismatch between worker foraging activity and flowering, we calculated the difference between day of peak worker foraging and day of peak flowering. Negative numbers indicate that worker activity peaked before flowering.

(4) Relationship of phenological mismatch to abundances of resident and colonizing species

Cumulative abundance of each bumble bee species was estimated as the sum of the weekly counts of all workers of that species weighted by collection effort over the season within an elevational zone. Relative abundance was calculated as the proportional contribution of a species to the bumble bee forager community.

References

Byron, P. A. (1980). On the ecology and systematics of Coloradan bumblebees. In Dissertation. University of Colorado, Boulder.

Christmas, M. J., Jones, J. C., Olsson, A., Wallerman, O., Bunikis, I., Kierczak, M., Peona, V., Whitley, K. M., Larva, T., Suh, A., Miller-Struttmann, N. E., Geib, J. C., & Webster, M. T. (2021). Genetic Barriers to Historical Gene Flow between Cryptic Species of Alpine Bumblebees Revealed by Comparative Population Genomics. Molecular Biology and Evolution, 38(8), 3126–3143. https://doi.org/10.1093/molbev/msab086

Geib, J. C. (2010). The Impacts of Pollinator Abundance on Benefits from Facultative Pollination Mutualism. In Ph. D. Dissertation, University of Missouri (Issue December). University of Missouri.

Geib, J. C., Strange, J. P., & Galen, C. (2015). Bumble bee nest abundance, foraging distance, and host-plant reproduction: implications for management and conservation. Ecological Applications, 25(3), 768–778.

Macior, L. W. (1974). Pollination ecology of the front range of the Colorado Rocky Mountains. Melanderia, 15, 1–59.

Miller-Struttmann, N. E., Geib, J. C., Franklin, J. D., Kevan, P. G., Holdo, R. M., Ebert-may, D., Lynn, A. M., Kettenbach, J. A., Hedrick, E., & Galen, C. (2015). Functional mismatch in a bumble bee pollination mutualism under climate change. Science, 349(6255), 75–78.

R markdown, R, and the follow R packages: lme4, AICcmodavg, car, r2glmm, nlme, PerformanceAnalytics