Hurricane‐induced pollinator shifts in a tightly coadapted plant–hummingbird mutualism

Citations Over TimeTop 10% of 2024 papers

Abstract

The majority of flowering plants are pollinated by animals, especially in warm and wet tropical regions where > 90% of all plant species rely on animal pollination (Rech et al., 2016). As a rule, on oceanic islands, pollinators and their plants are thought to be predominantly generalized, likely because of depauperate pollinator faunas and reduced interspecific competition, resulting in ecological release and generalized feeding niches (Olesen et al., 2002; Castro-Urgal & Traveset, 2014; Traveset et al., 2016; Dalsgaard et al., 2021). Colonization history may also matter for level of generalization, as mainland generalists should more easily colonize and establish on islands than mainland specialists (Olesen et al., 2002; Castro-Urgal & Traveset, 2014; Traveset et al., 2016; Dalsgaard et al., 2021). Finally, combined with characteristics of species on oceanic islands, that is, small and isolated populations at high risk of extinction, specialized mutualists on oceanic islands should be particularly rare and vulnerable due to natural perturbations, such as hurricanes (Waser et al., 1996; Clavel et al., 2011; Leimberger et al., 2022; Gonçalves et al., 2024). As such, it may appear unexpected that the mutualistic associations between two species of Heliconia plants and their hummingbird pollinator, the purple-throated carib (Eulampis jugularis), have been reported to be highly specialized and tightly coadapted on small Lesser Antillean islands frequently struck by hurricanes (Temeles et al., 2000; Temeles & Kress, 2003). Heliconia bihai and H. caribaea have flowers shown to reflect the size and curvature of the bill of the sexually dimorphic E. jugularis occurring across the Lesser Antillean islands (Temeles et al., 2000; Temeles & Kress, 2003). Females of E. jugularis have long, curved bills and have been shown to be the primary pollinator of H. bihai, while H. caribaea is pollinated by both the females and the shorter, straighter-billed males of E. jugularis (Temeles et al., 2000; Temeles & Kress, 2003). At many locations on the Lesser Antillean island of Dominica, H. caribaea has two colour morphs: a red morph with significantly longer and more curved flowers visited more frequently by females of E. jugularis, and a yellow morph with shorter, less curved flowers preferred by males of E. jugularis (Temeles et al., 2000; Temeles & Kress, 2003). As on most of the Caribbean islands, Dominica periodically experiences hurricanes (Temeles & Bishop, 2019; Fairbairn et al., 2022). Between 1851 and 2017, Dominica experienced 33 hurricanes, many of them weaker ones probably with little impact on biodiversity. However, it was also hit by two category four hurricanes and three category five hurricanes (Knapp et al., 2010), which may have had devastating impacts on biodiversity (Gonçalves et al., 2024). Notably, in 2017, Dominica was hit by Hurricane Maria, a catastrophic category five hurricane with island-wide devastating effects on bird populations, especially hummingbirds including the purple-throated carib (Temeles & Bishop, 2019; Fairbairn et al., 2022). After the hurricane passed over Dominica, dead and dying hummingbirds were observed on the ground. Hundreds of hummingbirds were observed feeding on open fruits, and only c. 25% of the purple-throated carib population is thought to have survived (Temeles & Bishop, 2019; Fairbairn et al., 2022; Leimberger et al., 2022). Hummingbirds are also thought to be more vulnerable to hurricanes than passerine birds with a more generalized feeding habit because the narrow feeding niches and high metabolic rates of hummingbirds make them especially vulnerable to lack of floral resources in the immediate aftermath of hurricanes (Wiley & Wunderle, 1993). As hurricanes may not only influence bird populations but also disrupt pollination systems (Rathcke, 2000; Rivera-Marchand & Ackerman, 2006; Temeles & Bishop, 2019), we wanted to examine the consequence of Hurricane Maria on the pollination system of H. bihai and H. caribaea. To do so, in 2022, we implemented a multi-facetted approach integrating information on: (1) bird visitation rates on the flowers of H. bihai and H. caribaea; (2) pollen depositions on the stigmas of H. bihai and H. caribaea; and (3) Heliconia pollen transport on mist-netted birds from one locality with H. bihai (see Materials and Methods section). We found avian visitation to H. bihai and H. caribaea to be fundamentally different from studies conducted in the decades before Hurricane Maria (Temeles et al., 2000; Temeles & Kress, 2003; Dalsgaard et al., 2009). Notably, flowers of H. bihai were visited by other hummingbird species and the bananaquit (Coereba flaveola), a generalized passerine pollinator, at similar frequencies as the purple-throated carib (Fig. 1). The same pattern was observed for the red morph of H. caribaea, but with the bananaquit having even higher visitation rates than the purple-throated carib; the yellow morph of H. caribaea was visited by even higher frequencies of both the bananaquit and the other hummingbird species on Dominica. With respect to pollen deposition, all visiting species of nectar-feeding birds deposited pollen on Heliconia plants (Supporting Information Table S1), with no overall association between bird visitation rate and pollen deposition (Fig. 2a). Given the visitation rate, the purple-throated carib was shown to deposit significantly more pollen than the other birds, possibly due to the better morphological match between their bills and Heliconia flowers. Out of 210 mist-netted birds belonging to 15 species, 52 individuals (25%) from five species (33%) carried Heliconia pollen (Table S2). The bananaquit presented the highest number of individuals carrying Heliconia pollen (46 individuals out of 102; 45%), followed by the hummingbirds E. jugularis (three individuals out of 6; 50%), Riccordia bicolor (1 individual out of 2; 50%), Orthorhyncus cristatus (one individual out of 7; 14%), and the omnivorous Loxigilla noctis (one individual out of 28; 3.5%), creating a pollen transport network dominated more or less equally by E. jugularis, R. bicolor and C. flaveola (Fig. 2b). Taken together, these results are in sharp contrast to previous reports of a tightly coadapted pollination system with the purple-throated carib as the sole pollinator of H. bihai and H. caribaea (Temeles et al., 2000; Temeles & Kress, 2003). We suggest that the hurricane-induced population crash of the purple-throated carib led to a breakdown in competitive exclusion (Temeles & Pyle, 2024), allowing the smaller species of hummingbirds and, notably, the opportunistic bananaquit, to visit these nectar-rich Heliconia plants. In further support of this, mist-netting data shows that the bananaquit increased in both population size and body weight after Hurricane Maria, suggesting increased resource availability for the bananaquit (Fairbairn et al., 2022), possibly due to Heliconia plants not being visited frequently by purple-throated caribs. This breakdown provides rare empirical evidence that tight coevolution between mutualists on islands may periodically be disrupted by natural disturbances such as hurricanes (Rathcke, 2000; Rivera-Marchand & Ackerman, 2006; Spiller & Schoener, 2007; Temeles & Bishop, 2019), which could lead to new evolutionary trajectories in plants and pollinators, and suggests that more generalized pollinators can provide resilience by sustaining plant sexual reproduction in the aftermath of hurricanes. These results also suggest that even highly specialized pollination systems, such as Heliconia pollination in the Lesser Antilles, may not be evolutionary dead-ends, consistent with phylogenetic studies of floral and hummingbird evolution (Tripp & Manos, 2008; Rombaut et al., 2022). Contrarily, a more generalized pollination system may develop fast if influenced by a high-impact disturbance. Previous studies (Temeles et al., 2000; Temeles & Kress, 2003) on Heliconia pollination were conducted two decades after the previous major hurricane, Hurricane David, which struck Dominica in 1979 as a category four hurricane. This indicates that the Lesser Antillean Heliconia pollination system entails an innate ‘recovery period’, which is somewhere between 6 and 20 yr, before the tight relationship between the Heliconias and E. jugularis is restored after a major hurricane. Hurricanes may thus be seen as pulses of disturbance, much like fires in fire-prone environments (Geerts et al., 2012). Given that hurricanes are predicted to increase in frequency and intensity due to global warming (Wehner & Kossin, 2024), whether such tight mutualisms will recover from future more powerful, more frequent hurricanes are a legitimate conservation concern, especially for species on small oceanic islands. Fieldwork was conducted in the southern part of Dominica (15°25′N, 61°20′W) during the Heliconia's flowering peak from 6 April to 17 July 2022. On Dominica, the two species occur allopatrically along an altitudinal gradient, with H. caribaea predominately found at lower elevations (c. 100–600 m) and H. bihai occupying higher elevations (c. 600–900 m). Therefore, we sampled bird–flower interactions at two different elevations: mid- (c. 530 m) and high-elevation (c. 800 m) covering the distribution of both Heliconia species. The high-elevation H. bihai-prone area was located in the Morne Trois Pitons National Park and consisted of mountainous elfin forests defined by strong winds and heavy precipitation. The mid-elevations H. caribaea-prone area was located close to the Middleham Falls and the village of Laudat, characterized by comparably lower rainfall and situated in a small valley, creating shelter from the wind, resulting in higher rainforest habitat. As for the location of hummingbirds in our study sites, the green-throated carib (Eulampis holosericeus) is primarily found at the mid-elevation study site, while the other three species of hummingbirds (Eulampis jugularis, Orthorhyncus cristatus, and Riccordia bicolor) are found at both mid- and high-elevation study sites. To quantify bird visitation rates and pollen deposition on stigmas, 60 mature Heliconia plants were chosen for data collection, consisting of 20 individuals of H. bihai at high elevation and 40 individuals of H. caribaea at mid-elevation. The H. caribaea samples were divided equally between the two colour morphs, with 20 individuals of both red and yellow morphs at the mid-elevation site. Each individual Heliconia plant was marked with flagging tape and a plant ID to ensure easy identification. The day before observation, one mature flower bud was bagged to exclude flower visitors. On the day of observation, the mesh bag was removed and the flower was carefully emasculated using tweezers. Subsequently, using a hand-lens, the stigma was inspected for excess pollen and, if found, delicately removed using a small piece of tape. For each Heliconia individual, bird visitation was either visually observed, between 06:00 h and 14:00 h, at a distance of c. 7 m, or with time-lapse cameras with zoom lenses at a distance of c. 3 m. The observation periods lasted between 4 and 6 h. Avian visitation rates were calculated as visits per flowers h−1 (VR). After an observation period, the stigma was removed from the flowers, placed on a microscope slide, and preserved in fuchsine gel (Kearns & Inouye, 1993). In the laboratory, we counted all conspecific and heterospecific pollen grains deposited on the stigmas using a light-contrast microscope with a magnification of ×100 or ×400. The pollen counts were used to calculate rates of multiple-visits pollen deposition (PD). For each of the three Heliconia species/morphs, we averaged the amount of pollen grains from the samples without registered bird visits and used these as controls. Differences between mean control and pollen counts were used to calculate deposited pollen by observed birds, with negative differences denoting zero pollen deposition. We evaluated pollen transport on mist-netted birds in the high-elevation H. bihai-dominated area. We captured birds across six consecutive days and different sites using 10 mist nets (2.6 × 12 m) daily. Mist nets were opened under favourable weather conditions for 5 h after sunrise, totalizing 3600 m2-h capture effort (net area multiplied by netting time) and regularly inspected for captured birds. Birds were individually sampled for pollen load by swabbing a small cube of glycerine jelly, stained with fuchsine, on their beak, throat, and forehead feathers. The gelatine cube was then placed on a microscope slide, melted by a weak heat source to produce a single layer of stained pollen grains. Preparations were then sealed with a coverslip and clear nail polish, labelled, and stored. Pollen grains were later identified by means of a reference collection and counted using a light-contrast microscope. Whereas it is difficult to distinguish between H. bihai and H. caribaea pollen under a light-contrast microscope, Heliconia pollen is relatively easy to distinguish from other types of pollen. As we caught the birds in an area with H. bihai only, it is most likely also H. bihai pollen we detect on the birds, although we cannot rule out that they might have picked up the pollen from H. caribaea at lower elevations. Irrespectively of being H. bihai or H. caribaea pollen, we document the ability of birds at picking up pollen from Heliconia plants. For the visitation rate (VR) analysis, a Mann–Whitney U test was utilized to compare visitation rates of E. jugularis with all other hummingbird species (E. holosericeus, O. cristatus, and R. bicolor) combined as a group. Subsequently, a Kruskal–Wallis test with a Dunn–Bonferroni post hoc test examined differences in VR among all visiting bird species (E. jugularis, E. holosericeus, O. cristatus, R. bicolor, C. flaveola, and L. noctis). For the pollen deposition (PD) analysis, a linear mixed-effects model was used to test whether pollen deposition was correlated with general visitation rates, also examining whether each bird species showed a positive effect on pollen deposition. All bird species were separated into presence/absence variables, and all individual bird species and VR set as fixed factors. Heliconia species were set as a random factor in the model, thus examining PD as a function of VR and bird species while accounting for the different species of Heliconias. All observations without visits were removed before analysis, and an ANOVA used to test for significance. All statistical analyses were conducted using R, v.4.2.3 (R Core Team, 2023). TSOS, FG, MGRV, BIS, CNK-B, and BD thank the Independent Research Fund Denmark (grant no. 0135-00333B). FG is supported by the SNSF Swiss Postdoctoral Fellowships (TMPFP2_217531). MGRV is supported by a Marie Curie Postdoctoral Fellowship (Horizon-TMA-MSCA-101149502). We thank the forestry department of Dominica for permitting us to conduct research in the Morne Trois National Park of Dominica, and for a great cooperation. MG is supported by CEPID FAPESP Center for Research on Biodiversity and Climate Change (2021/1063-95) and CNPq (306928/20213). None declared. TSOS, FG, MGRV, EJT and BD were directly involved with conceptualization and designed the project. TSOS, FG, MGRV, RDJ and SK collected the data. TSOS, FG, MGRV, RDJ and BD contributed to the data analysis. All authors (TSOS, FG, MGRV, TZ, RDJ, FLT-T, MG, BIS, CNK-B, EJT and BD) contributed to the writing, thus providing essential feedback to the writing of the manuscript. The data that supports the findings of this study are available in the Supporting Information of this article. Tables S1 and S2 provide information on multiple-visits pollen deposition and information on recorded Heliconia pollen on mist-netted birds, respectively. Table S1 Information on multiple-visits pollen deposition. Table S2 Information on recorded Heliconia pollen on mist-netted birds. Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Related Papers

• → Nectar extraction by hummingbirds: response to different floral characters(1984)64 cited
• → Nectar ecology of the endemic epiphytic hummingbird-pollinated bromeliad Vriesea altodaserrae: secretion dynamics and pollinator visitation pattern(2018)13 cited
• → SECONDARY NECTAR ROBBING BY A VOLCANO HUMMINGBIRD (SELASPHORUS FLAMMULA)(2023)4 cited
• → Antagonists and their effects in a hummingbird–plant interaction: Field experiments(2008)10 cited
• → Nectar ecology of the endemic epiphytic hummingbird-pollinated bromeliad Vriesea altodaserrae: secretion dynamics and pollinator visitation pattern(2018)