Temperature-dependent pollinator-mediated selection on floral thermoregulation

Abstract

The thermal environment is one of the most pervasive agents of selection. Most plants cannot choose their microclimate, so understanding how they cope with thermal variability is of critical concern. Several floral traits can modify the floral thermal microenvironment, which may alleviate negative impacts of thermal extremes on gametophytes and plant-pollinator interactions. While estimates of selection on traits associated with thermoregulation exist, selection on floral thermoregulation itself (i.e. the deviance of floral temperature from air) has not been quantified. We quantified pollinator-mediated and viability selection on floral thermoregulation in an alpine and lower elevation population of a widespread plant, Argentina anserina (Rosaceae), over two seasons using three fitness components. At high elevation, pollinators favored floral warming-selection via seed number and pollen export was more positive than at low elevation. At low elevation however, selection favored cooling via pollen viability in one season. Across populations and years, selection favored floral warming under cooler conditions but cooling under warmer conditions, and this pattern was driven by pollinator-mediated selection. These results provide the first direct evidence of temperature-dependent selection on floral temperature modulation. Consistent geographic differences in selection should drive local adaptation of thermoregulatory mechanisms.

Keywords: floral evolution; local adaptation; phenotypic selection; pollen viability; pollination; thermoregulation.

Fig. 1

Predicted ways that flower thermoregulation (∆T, Flower temperature – air temperature) impacts plant reproductive fitness through the pollination process. Floral thermoregulation should influence pollinator‐mediated pollen transfer (solid lines) and have temperature‐mediated effects on gametophyte performance and zygote development (dashed lines).

Fig. 2

Pollinator‐mediated selection favored floral warming via seed number at high elevation. (a–h) depict standardized total selection gradients on floral ΔT (flower temperature – air temperature) for open‐pollinated and hand‐pollinated flowers in low elevation (a–d) and high elevation (e–h) populations of Argentina anserina. Selection was measured during peak solar radiation at midday (10:01 h–14:00 h) and in the later afternoon (14:01 h–18:00 h) across 2 yr. (i) depicts total pollinator‐mediated selection on ΔT at low elevation calculated as the univariate selection gradient from hand‐pollinated flowers subtracted from the univariate selection gradient in open‐pollinated flowers with associated SE. When zero‐inflated mixed‐effect ANCOVA revealed that selection differed between open‐ and pollen‐supplemented treatments, it is denoted by an asterisk (P < 0.05). The order from left to right is 2022 midday ΔT, 2022 afternoon ΔT, 2023 midday ΔT, 2023 afternoon ΔT. (j) depicts total pollinator‐mediated selection on ΔT at high elevation. Inset photographs show an image of a small solitary bee foraging on A. anserina at low elevation and a larger Bombyliid fly at high elevation. (k) shows the average air temperature outside of flowers at flower height within population, time window, and year corresponding with (i, j).

Fig. 3

Selection favored floral warming via pollen export at high elevation but not low elevation. Univariate selection gradients on midday (a, c; 10:01 h–14:00 h) and afternoon (b, d; 14:01 h–18:00 h) ΔT (flower temperature – air temperature) in a low‐ and high‐elevation population of Argentina anserina in 2022 (a, b) and 2023 (c, d) using pollen export as the component of fitness. P‐values for interactions between population and ΔT from linear mixed‐effect models are provided when significant.

Fig. 4

Selection favored floral cooling via pollen viability at low elevation in one season. Univariate selection gradients on midday (a, c; 10:01 h–14:00 h) and afternoon (b, d; 14:01 h–18:00 h) ΔT (flower temperature – air temperature) in a low‐ and high‐elevation population of Argentina anserina in 2022 (a, b) and 2023 (c, d) using pollen viability as the component of fitness. P‐values for significant interactions between population and ΔT from linear mixed‐effect models are provided in (a).

Fig. 5

Air temperature affected the magnitude and direction of selection on floral thermoregulation. (a) Total and (b) direct selection on floral ΔT (flower temperature – air temperature) plotted against average flower‐level air temperature at the time of selection in Argentina anserina. Linear fits to total selection gradients for each fitness component are depicted in solid lines. Temperature × Fitness Component impacted total (P = 0.001) and direct selection (P = 0.02) in linear models.