Matching floral and pollinator traits through guild convergence and pollinator ecotype formation

Abstract

Background and aims: Pollinator landscapes, as determined by pollinator morphology/behaviour, can vary inter- or intraspecifically, imposing divergent selective pressures and leading to geographically divergent floral ecotypes. Assemblages of plants pollinated by the same pollinator (pollinator guilds) should exhibit convergence of floral traits because they are exposed to similar selective pressures. Both convergence and the formation of pollination ecotypes should lead to matching of traits among plants and their pollinators.

Methods: We examined 17 floral guild members pollinated in all or part of their range by Prosoeca longipennis, a long-proboscid fly with geographic variation in tongue length. Attractive floral traits such as colour, and nectar properties were recorded in populations across the range of each species. The length of floral reproductive parts, a mechanical fit trait, was recorded in each population to assess possible correlation with the mouthparts of the local pollinator. A multiple regression analysis was used to determine whether pollinators or abiotic factors provided the best explanation for variation in floral traits, and pollinator shifts were recorded in extralimital guild member populations.

Key results: Nine of the 17 species were visited by alternative pollinator species in other parts of their ranges, and these displayed differences in mechanical fit and attractive traits, suggesting putative pollination ecotypes. Plants pollinated by P. longipennis were similar in colour throughout the pollinator range. Tube length of floral guild members co-varied with the proboscis length of P. longipennis.

Conclusions: Pollinator shifts have resulted in geographically divergent pollinator ecotypes across the ranges of several guild members. However, within sites, unrelated plants pollinated by P. longipennis are similar in the length of their floral parts, most probably as a result of convergent evolution in response to pollinator morphology. Both of these lines of evidence suggest that pollinators play an important role in selecting for certain floral traits.

Keywords: Coevolution; Prosoeca longipennis; ecotype; floral guild; geographic variation; pollination; proboscis length; specialization; speciation; trait matching; tube length.

Fig. 1.

(A) The long proboscid fly Prosoeca longipennis visiting Pelargonium carneum (Geraniaceae) and (B) Cyrtanthus leptosiphon (Amaryllidaceae). Scale bar = 10 mm.

Fig. 2.

Geographic variation in tube and proboscis length of P. longipennis and the flowers that it pollinates across 14 different localities. Black columns represent mean fly proboscis length and white columns represent grand means of all floral guild members within a site. Numbers refer to different populations, and these correspond to Fig. 6 as well as to the tables throughout the manuscript. The phytogeographical regions of the Cape Floristic Region (CFR), including the Extra-Cape (EC) area of South Africa, are indicated on the map, namely the North West (NW), Roggeveld (RV), South West (SW), Agulhas plain (AP), Langeberg (LB), South East (SE) and Karoo Mountain (KM).

Fig. 3.

Flower colour of floral guild members present at each site as seen by human vision. Crosses in the centre of cells refer to the presence of nectar guides and their colour as perceived by human vision.

Fig. 4.

A sub-set of floral guild members pollinated by P. longipennis, namely (A) Pelargonium pinnatum (Geraniaceae), (B) Gladiolus oppositiflorus (Iridaceae). Photo: Petra Wester. (C) Pelargonium dipetalum (Geraniaceae), (D) Nerine humilis (Amaryllidaceae), (E) Pelargonium carneum (Geraniaceae), (F) Geissorhiza fourcadei (Iridaceae), (G) Wahlenbergia guthrie (Campanulaceae), (H) Tritoniopsis antholyza (Iridaceae) and (I) Gladiolus engysiphon (Iridaceae).

Fig. 5.

A representative sample of reflectance spectra measured from plants within the P. longipennis pollination guild. The different colours on the graph are approximations of colours as perceived by the human eye.

Fig. 6.

Study sites ranked in order of increasing P. longipennis proboscis length, showing the variation in the floral traits associated with proboscis length for each plant species within and among study sites.

Fig. 7.

Scatterplot with an ordinary least-squares regression, indicating the relationship between the grand means of all guild members per study site and P. longipennis proboscis length across those study sites. Each symbol indicates the mean trait value and s.e. per population.