Pollination and Floral Ecology pp. 179-180 (2011)
Pat Willmer

Adaptive Features?

Roulston and Cane (2000), in common with earlier authors, concluded that there was no good evidence for specifically adaptive relations between pollen food value and the type of pollinator: and there was equally little or no evidence that flower visitors could assess or respond to the pollens nutritional quality except perhaps in the crude sense of grain size.

In search of possible adaptive nutritional features. the focus has inevitably been on bees, whose main dietary input is pollen. Franchi et al. (1996) presented evidence that bees preferentially took pollens with low or negligible starch content, selecting floral species in which the stored starch had already been hydrolyzed to sugars: but this apparent preference may just reflect pollen longevity (see section 7, The Longevity and Viability Pollen, below). There is only limited evidence in the older literature that bees can, or do deliberately, select more protein-rich pollens (Schmidt and Johnson 1984; van der Moezel et al. 1987). For example, honeybees with artificially reduced pollen quality in their hives could not make a switch to gathering more protein-rich pollens (Pernal and Currie 2001). However, S. Cook et al. (2003) showed that Apis preferentially took higher protein pollens when given artificial choices, and Robertson et al. (1999) recorded Bombus as choosing between patches of Mimulus flowers on the basis of pollen protein quality. Hanley et al. (2008) reported that in British herbaceous plants, the obligate insect-pollinated flowers, and especially bumblebee flowers, had a significantly higher pollen protein content; curiously, they found no relation between protein content and grain size, although their samples did not include the very large or very small pollens.

Bee growth and performance are certainly affected by the quality of their pollen diet, however, For Osmia eggs grown on ten different pollen types, only those pollens richest in protein supported development through to adulthood (M. Levin and Haydak 1957). Likewise, for bumblebees growth was faster on higher-protein pollens (Regali and Rastoont 1995). while for honeybees, longevity increased with pollen protein content (Schmidt et al. 1987). There is also evidence that bees may produce smaller offspring when feeding on protein-poor pollen (Roulston and Cane 2000).

Aside from bees, some cases are known where pollen content appears to be specifically coadapted with a particular pollinator. Most strikingly, there can be very different overall protein contents within genera that have different vectors for different species. For example, in the genus Agave, A. americana (pollinated nocturnally by bats) has 43% protein in pollen, whereas A. schottiana and A. parviflorum (both insect pollinated) have 8% and 10% protein, respectively. Taking this a stage further, in the bat-pollinated saguaro cactus (Carnegiea). protein is abundant in the pollen and is particularly rich in the two amino acids tyrosine and proline (Howell 1974). Bats not only have high energy demands per se (chapter 10) but also need unusually large amounts of tyrosine to make the collagen that forms their wing membranes (collagen is 80% tyrosine repeats), and proline is an important component of hat milk. At first sight, this observation indicates adaptive nutrition provided by pollen. However, proline has important functions for the plant too, being crucial in cell-wall formation during pollen tube growth thus high proline and high overall protein levels are needed in larger flowers with longer styles, and for this reason alone these features can seem (spuriously) to be associated with large bat-pollinated flowers. In fact, some anemophilous pollens such as Zea, which have particularly rapid pollen tube growth, are substantially richer in proline than any of the Agate pollens (Solberg and Remedios 1980).

Does pollen ever have specific nutritional attractants for specific visitors? Although early examples have been discounted, the answer may still be yes; there is some evidence that bees like particular fatty acids found in some pollens (C.Y. Hopkins et al. 1969), some having an antibacterial effect and others improving larval nutrition (R. Manning 2001). Adaptive nutritive traits in pollen might be more likely where pollen is offered specifically for feeding, as in heteranthous plants. In some Lecythidaceae (fig. 2.5), the food pollen grains are known to be larger than the reproductive pollen (Mori et al. 1980). However, in Commelina some of the "food anthers" are deceptive and have little or no food reward, and some contain a milky fluid instead of pollen. Curiously, sterile pollen is reasonably common even in nonheteranthous but functionally female flowers (Cane 1993a), for example, Actinidia, Rosa setigera, and some Solanum spp. This case has usually been regarded as one of deception, but the sterile pollen can still be nutritionally useful to bees (especially its lipid-rich pollenkitt layer; e.g., Cane l993b) and may even have as much amino acid content as the fertile pollen (Davies and Turner 2004).