Ovule bet-hedging at high elevation in the South American Andes: Evidence from a phylogenetically controlled multispecies study

How animal-pollinated plants support low and stochastic pollination in the high alpine is a key question in plant ecology. The ovule bet-hedging hypothesis proposes compensation for stochastic pollination via ovule oversupply in flowers allowing the benefits of windfall pollination events to be reaped. Under this hypothesis, ovule number is expected to increase from tree line upward on high mountains characterized by steep declines in flower visitation rates and increasingly more variable pollination. Ovule/floret number was investigated for a total of 174 simple-flowered and pseudanthial species in the central Chilean Andes (2,100–3,650 m.a.s.l.). Phylogenetic reconstruction was undertaken using ITS sequences and a constrained ordinal-level backbone reflecting the APG-IV topology. Ovule/floret number was modelled with ordinary least squares regression (OLS) and phylogenetic generalized least squares regression (PGLS) with elevation, floral biomass, life history, pollinator efficiency, pollination generalization, and seasonal flowering period as explanatory variables. The best performing OLS and PGLS models for simple-flowered species consistently included vegetation belt and floral biomass, and with PGLS, pollination efficiency and flowering period. For pseudanthial species, explanatory variables were always floral biomass and its interaction with elevation. Effects of life history and generalized pollination was not found. Ovule/floret number showed high phylogenetic signal, increased with floral biomass and was generally higher in the upper alpine belt in both floral categories. Simple-flowered species with efficient pollination and flowering early, respectively, had larger ovule numbers. Synthesis. Ovule number increases with elevation in the central Chilean alpine in two separate floral groups independently of some effects of flowering period and pollinator efficiency. Greater disparity in pollen deposition on stigmas than with inefficient pollination under low visitation rates might explain the association between efficient pollination and higher ovule numbers. Our study provides the first empirical evidence for ovule bet-hedging in the alpine environment. Future studies on the ovule bet-hedging hypothesis should include a measure of flower size