Available online at www.sciencedirect.com South African Journal of Botany 75 (2009) 720–725 www.elsevier.com/locate/sajb Rodent pollination in Protea nana ⁎ A. Biccard , J.J. Midgley Department Botany, University of Cape Town, Private Bag X3, Rondebosch 7701, South Africa Received 19 March 2009; received in revised form 3 August 2009; accepted 4 August 2009 Abstract Floral morphology, late winter flowering time and a distinct yeasty odour in Protea nana suggest that it is adapted for rodent pollination. To test this hypothesis, rodents were trapped in a stand of P. nana, examined for presence of pollen (on their noses and in their scats) and then observed in the laboratory for pollination behaviour. Inflorescences were placed in wire and shade-cloth exclosures and seed-set compared to controls in the field. Otomys irroratus, Aethomys namaquensis, Rhabdomys pumilio and Myomyscus (Praomys) verreauxi were captured and all tested positive for the presence of protea pollen. All, except O. irroratus, displayed legitimate pollination of P. nana. Myomyscus verreauxi was regarded as the most competent pollinator as it displayed superior climbing ability in comparison to the other rodent species. R. pumilio sometimes displayed highly destructive behaviour of P. nana inflorescences and is the likely explanation for the observed 20% reduction in average number of inflorescences per plant over a two month period. Seeds per inflorescence were lowest in shade-cloth covered inflorescences (total pollinator exclusion), but not significantly lower in wire exclosures which provided access for insects. A high percentage of sugar (29.4% by weight) in the nectar of P. nana is similar to known rodent pollinated species. We conclude that P. nana is at least partially rodent pollinated. © 2009 SAAB. Published by Elsevier B.V. All rights reserved. Keywords: Aethomys namaquensis; Myomyscus verreauxi; Protea nana; Rhabdomys pumilio; Rodent pollination 1. Introduction Protea nana could be rodent pollinated as it has inflexed styles, nectar with a high sugar content (Nicolson and Van Wyk, Small mammals have long been known as pollinators of South 1998), a characteristic yeasty smell and a winter–spring flowering African Proteaceae (Wiens and Rourke, 1978). However, of the season which coincides with that of known rodent pollinated many (N35) species suggested to be small mammal pollinated Proteas (Wiens et al., 1983; Fleming and Nicolson, 2002). Rourke (Rourke, 1980, Rebelo and Breytenbach, 1987), only a few and Wiens (1977) suggested the Cape striped field mouse, species have been investigated. The basic floral features of Proteas Rhabdomys pumilio, is the most plausible mammal pollinator of used to suggest pollination by small mammals include: (1) bowl P. nana. shaped heads borne on short (3–4 mm) stout peduncles, often with P. nana could also be insect pollinated (Rebelo and the outsides of bracts darkly coloured, (2) copious, sugar-rich Breytenbach, 1987). It is an erect, rounded shrub up to a metre nectar production with a high (ca. 36%) total carbohydrate in height, rather than being shorter and more geoflorous as is composition, (3) often inflexed, wiry styles ca. 30–40 mm long, typical of rodent pollinated species. Also, the inflorescences are (4) cryptic, geoflorous, axillary positioning of the heads, (5) a terminal, pendulous and held up to a metre off the ground on distinctive “yeasty odour,” (6) and a stigma–nectar distance of slender branches, rather than being cauliflorous on stout stems, approximately 10 mm to “fit” between the rostrum of a pollinating vertically presented and close to the ground. Our objective was to rodent and the stigma and (7) a winter–spring flowering period determine if small mammals are legitimate pollinators of P. nana. (Wiens et al., 1983, Rebelo and Breytenbach, 1987). 2. Materials and methods 2.1. Study site 2 ⁎ Corresponding author. Tel.: +27 21 650 2438. This study was carried out within a small (~1 km ), dense E-mail address: [email protected] (A. Biccard). population of P. nana located along Bain's Kloof Pass outside 0254-6299/$ - see front matter © 2009 SAAB. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.sajb.2009.08.003 A. Biccard, J.J. Midgley / South African Journal of Botany 75 (2009) 720–725 721 Wellington in the Western Cape (33°37′44.5″S, 19°05′58.3″E, Furthermore, viable and sterile achenes do not differ morpho- 514 m altitude). One of our main sources of evidence was the logically, however, viable endoderm-containing seeds can be presence of typically triangular protea pollen in animal faeces. readily identified by their soft, milky-white texture; whereas Therefore, on Bain's Kloof Pass where P. nana can co-occur seeds interpreted as sterile have a dull-white, dry, fibrous content with Protea amplexicaulis, a well known rodent pollinated (Wiens et al., 1983; see Fig. 1C). Hence ovules with swollen species, we chose a site where we could be sure protea pollen endoderm were used as a proxy for seed-set. Such “viable seeds” was from P. nana inflorescences. To ensure this we chose a site were identified in the laboratory by sequentially cutting where P. amplexicaulis was rare. To demonstrate differences in transverse sections through all developing achenes and observ- density, a tally was made of all individuals of these two species ing them under a dissecting microscope. in 6 plots, each 5 m×5 m that were laid out in the study area. The mean density of P. nana was 28.1 per plot and a mean of only 2.3. Sugar concentration of nectar 0.66 P. amplexicaulis individuals per plot was recorded. Another factor which makes it unlikely that P. amplexicaulis Nectar was randomly sampled shortly after sunrise at the pollen would be found in faeces at our study site is the beginning of November from ten newly opened, unvisited observation that the phenology of these two species is different. P. nana inflorescences using capillary tubes. Each sample was P. amplexicaulis flowers earlier in the year. For example, a optically analysed for sugar content using an Eclipse handheld survey of flowering state in late July indicated that only 1 out refractometer (Bellingham & Stanley Ltd.) which had been of 30 P. nana individuals had open flowers whereas all 30 calibrated with distilled water at 20 °C. Readings were corrected P. amplexicaulis individuals had open flowers. Finally, P. nana according to the International Temperature Correction Table for produces more than 10 times the mean number of inflorescences Brix (% Sugar) scale. per individual than does P. amplexicaulis (40 versus 3.5), although the inflorescences are smaller in the former (Fig. 1D). 2.4. Mammal trapping 2.2. Exclosure experiments Mammal nomenclature follows Skinner and Chimimba (2005). Trapping was conducted on 15, 16, 22 and 23 August A pollinator exclusion experiment was setup at this site in 2007 (using Sherman traps). Traps were baited with a mixture mid-August during the start of the 2008 flowering season. In an of peanut butter and rolled oats and spaced at approximately attempt to exclude small mammals, but also to allow insects 10 m intervals within the P. nana stand. Traps were set in the access, 20 wire cages constructed out of 13 mm “chicken mesh” early evening before sunset and then inspected the following and were placed around P. nana inflorescence buds on different morning. A few traps were left open during the day for the bushes. In order to avoid weighing the plants down, or damaging capture of diurnal species. Captured rodents were identified, inflorescences, the wire cages were secured with a cable-tie to a sexed, weighed, and ventrally marked using black hair dye. dowel stick (Fig. 1A). An additional 20 wire cages were covered Breeding status for each individual was recorded. in shade-cloth and placed around P. nana inflorescences in a Each trapped animal was tested for the presence of pollen similar fashion. These served to exclude all potential pollinators grains on their nose to assess which species had visited P. nana. (including insects) and tested for self-pollination — the shade- Pollen was collected from the rostra of captured rodents by cloth was secured such that it could not be removed by rodents rubbing a cube of gelatin (impregnated with basic Fuchsin dye — and the mesh size excluded all potential insect pollinators Beattie, 1971) mounted on a dissecting needle against the fur from (Fig. 1B). Throughout the experiment, inflorescences were the eye-level down to the nose and around the mouth. The size of randomly selected at various heights and distances from the the gelatin cubes and the effort applied when collecting pollen centre of each bush. Care was taken when selecting undamaged from the rostra of rodents was not standardised. Gelatin cubes inflorescence buds for this experiment. Only immature (closed) were immediately placed into plastic vials in order to avoid inflorescences were selected and occasionally, multiple inflor- contamination. In the laboratory the cubes were melted inside the escences on the same branch were placed in a single exclosure. plastic vials by placing them in a bath of boiling water. Two drops To quantify predation of flowers, ten P. nana bushes were of the molten Fuchsin Gelatin solution were taken from each randomly chosen. Each bush was then marked and the total sample and mounted on separate microscope slides. Pollen grains number of fresh inflorescences recorded. were counted for the full field of view at 40×magnification After three months and five days, all inflorescences within the (0.45 mm diameter) over three different scans of the entire exclosures were removed from the bushes for counting of coverslip length (22 mm). This count yielded an approximate 3% numbers of seeds per inflorescence (averages were calculated for sample of the area of the coverslip.