Page  57 ï~~2004 THE MICHIGAN BOTANIST 57 DOES STEEPLEBUSH (SPIRAEA TOMENTOSA) FACILITATE POLLINATION OF VIRGINIA MEADOW BEAUTY (RHEXIA VIRGINICA)? Raffica J. LaRosa, David A. Rogers, Thomas P. Rooney*, and Donald M. Waller Department of Botany University of Wisconsin-Madison 430 Lincoln Drive Madison WI 53706 ABSTRACT Facilitation through pollinators is a rarely investigated topic. Rhexia virginica (Melastomataceae) is a bumblebee-pollinated, Atlantic coastal plain disjunct herb that is rare in Wisconsin. Spiraea tomentosa (Rosaceae), a native shrub often growing in the same location, blooms at approximately the same time of year, and bumblebees often fly between these species. We tested the hypothesis that the presence of Spiraea increases pollinator visits to Rhexia. We observed sites that varied in their densities of Spiraea inflorescences and Rhexia flowers and recorded bumblebee visits to Rhexia across these sites. Visits to Rhexia increase significantly in the presence of flowering Spiraea. However, bumblebee visits to Rhexia decline with increasing densities of Spiraea inflorescences, suggesting that facilitation is replaced then by competition for pollinators. Facilitation might occur because Rhexia offers only pollen as a reward whereas Spiraea offers nectar, and these are complementary resources for bees. INTRODUCTION The Virginia meadow beauty, Rhexia virginica L., belongs to the primarily tropical family Melastomataceae. While its primary range is the Atlantic coastal plain, disjunct populations are also found in the Great Lakes region. Along with other coastal plain disjuncts, Rhexia inhabits sites in central and northern Wisconsin where soils are very sandy and moist. Disjunct populations in the Great Lakes arrived in that region either by long distance dispersal or by a series of shorter dispersal events along the shorelines of glacial lakes and drainages present as the glaciers receded (Reznicek 1994). Rhexia is rare and listed as a species of "special concern" in Wisconsin. Disjunct populations are especially vulnerable to population loss because they lack a constant influx of seeds to rescue extirpated populations, and because extant populations are often sparse (Reznicek 1994). Little research has been done on Rhexia in the midwest. Rhexia's habitat in Wisconsin consists of wet, sandy soil, typically along lakeshores or in sedge meadows. It is a perennial herb that stands approximately 30 cm high and is pollinated almost exclusively by bumblebees. Recruitment from seed is important for Rhexia populations because populations are often small, and the probability of immigration of seed from other popula *Author for correspondence: 608. 265. 2191; tprooney@wisc.edu

Page  58 ï~~58 THE MICHIGAN BOTANIST Vol. 43 tions is infinitesimal. Baskin and Baskin (1998) examined germination properties of Rhexia mariana. To germinate, seeds need at least 56 days of cold, followed by light and temperatures of 35/200C. Rhexia virginica might well require similar conditions. As Rhexia habitats are prone to flooding, all plants can be killed in some years. Seeds can remain dormant for many years (Keddy and Reznicek 1982). This seed bank ensures populations can persist following floods. Rhexia lacks nectar and only offers pollen as a reward to pollinators. The anthers have pores and are "buzz pollinated." Anthers on newly opened flowers are long and bright yellow. Individual flowers are only fertile for one day. After that, the anthers turn red and the petals fall off. New flowers are produced throughout late summer. In central Wisconsin, steeplebush, Spiraea tomentosa L. var. rosea (Raf.) Fernald, often grows alongside Rhexia, and they flower at the same time. Bumblebees are often observed flying between Rhexia and Spiraea. Because Spiraea is rich in nectar, it might increase the chances of Rhexia being visited by bumblebees, as it offers a complementary food source. If Spiraea facilitates Rhexia pollination, conservation efforts might be directed toward maintaining or restoring mixed populations. MATERIALS AND METHODS We surveyed pollinator visitation rates at two sites in Marquette County, Wisconsin, chosen for their relatively large Rhexia populations. Both populations were located along privately owned lakeshores. Site A was on a small lake (T16N, R10E, Sec 7 SW 4) and site B was on a pond several km away (T17N, R10E, Sec 30 SW 4). To assess local effects of flowers on pollinator visitation, we established circular plots of 10m2. We selected plots of this size because they were both large enough to have a high probability of a bee entering the area and small enough to allow tracking all the bees in a plot. The plots were centered around an individual Rhexia plant. There were 35 plots, 19 at site A and 16 at site B. The plots were established in June, before there were any flowers on either Rhexia or Spiraea. We collected data through July, August, and into September, visiting site A nine times and site B ten times. During each visit, we counted the number of fertile Rhexia flowers and the number of fertile Spiraea inflorescences within each 10 m2 plot. We then recorded the number of bumblebee visits to Rhexia flowers and Spiraea inflorescences within the plot during a 5-minute period. Plots were observed between 7 am- 1pm and 5 pm-8 pm. All plots were surveyed during each visit, and as each site was visited repeatedly, each plot was observed for a total of 45-50 minutes. The visits on 16 days resulted in 47 hours of observation. Although we also recorded weather conditions, these did not appear to affect bumblebee foraging behavior. Larson and Barrett (1999) also observed that visitors to Rhexia are insensitive to weather. As bumblebees are able to forage under harsh conditions, it is not surprising to find little variation in activity in July and August in Wisconsin (Heinrich 1979). It was not necessary to modify the densities of the two species as the density of fertile flowers changed from day to day. The fluctuations in flower densities also acted to make the results from the same plot on different days largely independent from one another. We applied two tests to examine how Spiraea flowering affected patterns of bumblebee visitation to Rhexia. First, we examined how site and the presence or absence of Spiraea affected Rhexia visitation rates, using a two-way ANOVA. We then reduced the data to include only plots that had at least one fertile Rhexia flower and one Spiraea inflorescence and examined how Spiraea density affected Rhexia visits as a covariate in a one-way analysis of covariance, with site as the factor and Spiraea density as the covariate. Because densities of Rhexia and Spiraea are positively correlated, we needed further tests to confirm that any correlation between Spiraea density and visits per Rhexia flower was not actually due

Page  59 ï~~2004 THE MICHIGAN BOTANIST 59 TABLE 1. Two-way ANOVA showing the affects of site and Spiraea inflorescence presence on visitation rates of bumblebees on Rhexia flowers (n=267). Variable df SS F P Spiraea presence/absence 1 0.80 5.98 0.015 Site 1 0.00 0.00 0.97 Spiraea presence X site 1 0.14 1.04 0.21 Error 263 34.99 to a correlation between Rhexia density and visits/flower. Using the method of standardized partial regression coefficients, we calculated path coefficients between: Rhexia density and visits per Rhexia flower, Spiraea density and visits/flower, and Rhexia density and Spiraea density. RESULTS Across both sites, there were no difference in bumblebee visits to Rhexia, but there was a significant increase in visits to Rhexia in plots when at least one Spiraea inflorescence present (Table 1; Fig. 1). At site A, the average visits per 150 125 - >,100 -0 c'75 -LL 50 - E Spiraea absent SSpiraea present Jili r--i,m 25 - U f L Iffffffff _ 1 1" t _ f 1 i 1 1 1 0 N 0 0 0 C4 0 r N c0 0 0 o d co 0 (0 0 0 0 Co 0; + 0 Q~ Mean no. visits/5 min. FIGURE 1. Number of bumblebee visits and visits per 5-minute observation interval to Rhexia flowers when Spiraea is absent (dark bars) and present (white bars).

Page  60 ï~~60 THE MICHIGAN BOTANIST Vol. 43 TABLE 2. One-way ANCOVA showing the effects of site and Spiraea inflorescence density on visitation rates of bumblebees to Rhexia flowers at each of two sites and plots (n = 14; total r2 = 0.41). Variable df SS F P Site 1 0.07 1.50 0.25 Spiraea inflorescence density 1 0.30 6.85 0.024 Error 11 0.49 a, 0 U) a) CL Spiraea patch density FIGURE 2. The relationship between the average bumblebee visits per Rhexia flower and Spiraea inflorescence density per square meter when Spiraea was present (df = 1; r = -0.58; P = 0.024). Rhexia flower increased over twofold per 5 minutes and at site B, it increased over fourfold. Once Spiraea is present, however, bumblebee visits to Rhexia appear to decline with increases in Spiraea inflorescence density (r = -0.58; Table 2, Fig. 2). Again, there were no differences in visits per Rhexia flower due to site. Using path coefficients, the direct effect of Spiraea density on visits per Rhexia flower was -0.577 (Fig. 3). The indirect effect is the product of all coefficients in the pathway. The indirect effect of Spiraea on visits/flower was -0.013. These results show that the stronger and only statistically significant influence on visitation rate to Rhexia flowers is Spiraea density.

Page  61 ï~~2004 THE MICHIGAN BOTANIST 61 2004 THE MICHIGAN BOTANIST 61 0. 6 VisitsiFkower Inflorescence flensitv FIGURE 3. Diagram showing the factors that influence bumblebee visits to Rhexia and the possible paths and coefficients of influence. (Direct effect of Spiraea = -0.577; Indirect effect of Spiraea = -0.013). DISCUSSION In central Wisconsin, Rhexia has higher visitation rates per flower when Spiraea inflorescences are present. Yet visitation rates to Rhexia when Spiraea is present decline with increasing Spiraea inflorescence density. Thus, small densities of Spiraea inflorescences appear to facilitate Rhexia visitation, whereas Spiraea appears to compete with Rhexia for pollinator visits when Spiraea inflorescences become dense. This pattern may arise because of resource complementarity. Rhexia only offers pollen as a pollinator reward; while bumblebees need pollen as a protein source, they also need nectar as an energy source (Heinrich 1979). Spiraea offers both pollen and nectar as pollinator rewards. Therefore, bumblebees might be attracted to Rhexia by its copious pollen but remain in the area because Spiraea provides them with plentiful nectar. Rathcke (1983) argues resource complementation could result in higher pollinator visitation rates as these resource patches are highly attractive to pollinators. Resource complementarity has been experimentally demonstrated with frugivorous birds. Whelan et al. (1998) found that birds reared on complementary fruit resources had higher fitness than those reared on a single fruit resource. The potential for resource complementarity has important implications for species coexistence. Callaway (1995) observed that co-flowering species experience facilitation when one plant species is more attractive to pollinators than a neighboring species. This is consistent with what we observed with Rhexia and Spiraea. The presence of Spiraea appears to provide an advantage to Rhexia by attracting bumblebees to the area. While bumblebees are there, they find a complementary food source (pollen) in Rhexia. However, visits to Rhexia decreased in plots with the most Spiraea inflorescences, suggesting the bees are "majoring" in collecting nectar (Heinrich 1979). The increased number of Spiraea inflorescences might also visually overwhelm the Rhexia display, distracting bees from the Rhexia flowers. Spiraea and Rhexia often co-occur in the same habitat types, and their ranges overlap (Keddy and Reznicek 1982; Larson and Barrett 1999). Because Rhexia is consistently found near Spiraea, it might be more persistent when Spiraea is

Page  62 ï~~62 THE MICHIGAN BOTANIST Vol. 43 present. If Spiraea can increase pollinator visits to Rhexia, it might also increase seed set. This hypothesis should be tested by comparing seed set in Rhexia populations with and without Spiraea present. These data did not indicate what densities of Rhexia and Spiraea might be optimal for enhancing bee visitation or seed set in Rhexia. We also do not know whether the ratio of Spiraea to Rhexia or the absolute density of Spiraea affects bumblebee visitation the most. Further studies might clarify this issue as well by manipulating Spiraea densities to see what Spiraea inflorescence density or ratio to Rhexia flower density optimizes bee visits and/or seed set. ACKNOWLEDGMENTS This study was financially supported by a Holstrom Environmental Scholarship to R. La Rosa. LITERATURE CITED Baskin, C. C., & J. M. Baskin. 1998. Seeds: Ecology, Biogeography and Evolution of Dormancy and Germination. Academic Press, New York. xiv + 666 pp. Callaway, R. M. 1995. Positive interactions among plants. Bot. Rev. 61: 306-349. Heinrich, B. 1979. Bumblebee Economics. Harvard University Press, Cambridge. viii + 245 pp. Keddy, P. A., & A. A. Reznicek. 1982. The role of seed banks in the persistence of Ontario's coastal plain flora. American Journal of Botany 69: 13-22. Larson, B. M. H., & S. C. Barrett. 1999. The ecology of pollen limitation in buzz-pollinated Rhexia virginica (Melastomataceae). Journal of Ecology 87: 371-381. Rathcke, B. 1983. Competition and facilitation among plants for pollination. Pages 305-329 in L. Real, ed. Pollination Biology. Academic Press, Orlando. Reznicek, A. A. 1994. The disjunct coastal plain flora in the Great Lakes region. Biological Conservation 68: 203-215. Whelan, C. J. K. A. Schmidt, B.B. Steele, W.J. Quinn, & S. Dilger. 1998. Are bird-consumed fruits complementary resources? Oikos 83: 195-205