Climate-induced phenological shifts in a Batesian mimicry complex

Abstract

Climate-induced changes in spatial and temporal occurrence of species, as well as species traits such as body size, each have the potential to decouple symbiotic relationships. Past work has focused primarily on direct interactions, particularly those between predators and prey and between plants and pollinators, but studies have rarely demonstrated significant fitness costs to the interacting, coevolving organisms. Here, we demonstrate that changing phenological synchrony in the latter part of the 20th century has different fitness outcomes for the actors within a Batesian mimicry complex, where predators learn to differentiate harmful "model" organisms (stinging Hymenoptera) from harmless "mimics" (hoverflies, Diptera: Syrphidae). We define the mimetic relationships between 2,352 pairs of stinging Hymenoptera and their Syrphidae mimics based on a large-scale citizen science project and demonstrate that there is no relationship between the phenological shifts of models and their mimics. Using computer game-based experiments, we confirm that the fitness of models, mimics, and predators differs among phenological scenarios, creating a phenologically antagonistic system. Finally, we show that climate change is increasing the proportion of mimetic interactions in which models occur first and reducing mimic-first and random patterns of occurrence, potentially leading to complex fitness costs and benefits across all three actors. Our results provide strong evidence for an overlooked example of fitness consequences from changing phenological synchrony.

Keywords: Batesian mimicry; climate change; hover flies; mismatch; phenology.

Fig. 1.

(Top) Heat map of mimetic ratings between 56 Syrphidae and 42 Hymenoptera. Colors indicate the mean similarity rating for each pair. The heat map has been ordered to place the Syrphidae with the highest mean similarity on the left of the plot and the Hymenoptera with the highest mean similarity at the bottom of the plot. See SI Appendix, Fig. S2 for an enlarged version with species names. (Bottom) Representative pairs of Hymenoptera (Top row) and Syrphidae mimics (Bottom row) that were rated as high similarity (A = Anthophora plumipes, E = A. superbiens, rating = 7.5), medium-high similarity (B = V. vulgaris, F = Chrysotoxum festivum, rating = 4.9), low-medium similarity (C = A. mellifera, G = Cheilosia impressa, rating = 3.1), and low similarity (D = Bombus pascuorum, H = Baccha elongata, rating = 1.0). A, B, and E–H courtesy of Steven Falk (photographer). D courtesy of Arnstein Staverløkk (Norwegian Institute for Nature Research, Trondheim, Norway).

Fig. 2.

Fitness consequences of phenological asynchrony in Batesian mimics, models, and predators. Dotted horizontal line at y = 1 in A and B shows model-first odds ratio, against which the other two phenological patterns are compared: (A) Batesian mimics are predated more under mimic-first scenarios, benefitting most from model-first scenarios. (B) Models are predated more often under mimic-first scenarios, benefitting more from random scenarios. (C) Predator fitness is greatest under model-first scenarios, then mimic-first, and lowest under random scenarios. Error bars are 95% confidence intervals.

Fig. 3.

Trends in the number of high-fidelity model-mimic pairs in which (A) the model emerges first, (B) the mimic emerges first, and (C) emergence is random. Mimetic fidelity is derived from a large citizen science study, phenological trends are derived from >1 million biological records, and the three categories of emergence are defined using known fitness consequences from behavioral experiments. Shaded areas are 95% confidence intervals.