The plot thickens: does low density affect visitation and reproductive success in a perennial herb, and are these effects altered in the presence of a co-flowering species?

Abstract

Plants may experience reduced reproductive success at low densities, due to lower numbers of pollinator visits or reduced visit quality. Co-occurring plant species that share pollinators have the potential to facilitate pollination by either increasing numbers of pollinator visits or increasing the quality of visits, but also have the potential to reduce plant reproductive success through competition for pollination. I used a field experiment with a common distylous perennial (Piriqueta caroliniana) in the presence and absence of a co-flowering species (Coreopsis leavenworthii) in plots with one of four different distances between conspecific plants. I found strong negative effects of increasing interplant distance (related to conspecific density) on several components of P. caroliniana reproductive success: pollinator visits to plants per plot visit, visits received by individual plants, conspecific pollen grains on stigmas, outcross pollen grains on stigmas, and probability of fruit production. Although P. caroliniana and C. leavenworthii share pollinators, the co-flowering species did not affect visitation, pollen receipt or reproductive effort in P. caroliniana. Pollinators moved very infrequently between species in this experiment, so floral constancy might explain the lack of effect of the co-flowering species on P. caroliniana reproductive success at low densities. In co-occurring self-incompatible plants with floral rewards, reproductive success at low density may depend more on conspecific densities than on the presence of other species.

Appendix

I tested whether distance and co-flowering treatment affect the number of outcross pollen grains on stigmas. To determine the ranges of pollen grain sizes produced by anthers of P. caroliniana plants of the two floral morphs, I measured diameters of pollen grains from anthers of five long- and five short-styled P. caroliniana plants. I used digital photographs of pollen from anthers to calculate pollen diameters [in pixels; 75.40 ± 0.46 pixels (mean ± SE)] using Adobe Photoshop^® (Adobe 1999–2002), converting these numbers to actual pollen diameters using the number of pixels per micrometer from digital photographs of a stage micrometer taken at the beginning and end of each photographic session.

Pollen grains of short-styled plants have larger diameters on average than pollen grains of long-styled plants [4.22 ± 0.30 vs. 3.74 ± 0.09 μm, respectively (mean ± SE); seven short-styled anthers; nine long-styled anthers; df = 10; t = 4.57; P < 0.01; 10–30 pollen grains were measured per anther]. However, pollen grain sizes of the two style morphs overlap (Fig. 5). Knowing both the style morph of each plant and the size range of pollen grains produced by each morph, I was able to estimate the probability that a given pollen grain was from a plant of the opposite style morph, using the equations:

$$ P_{L} (X) = \frac{{S_{X} }} {{S_{X} + \alpha L_{X} }} $$

(1)

$$ P_{S} (X) = \frac{{\alpha L_{X} }} {{S_{X} + \alpha L_{X} }} $$

(2)

Fig. 5

A frequency histogram of pollen grain diameters (in tenths of millimeters) measured from pollen grains from P. caroliniana anthers. Dark bars indicate pollen from long-styled plants, and open bars indicate pollen from short-styled plants. Downward-pointing arrows indicate the lower (for short-styled plants) or upper (for long-styled plants) 95% confidence limits of the distribution

where X indexes the diameter class of a pollen grain, S _X and L _X are the fractions of pollen grains produced by short-styled and long-styled flowers, respectively, that fall into size class X, P _L (X) and P _S (X) are the probabilities that a pollen grain in size class X found on the stigma of a long- or short-styled flower, respectively, is a compatible outcross pollen grain, and α is the estimated ratio of the number of pollen grains produced in anthers of long-styled plants versus short-styled plants. In some distylous plants, one style morph produces more pollen than the other (Li and Johnston 2001). If anthers of long-styled flowers produce more pollen than those of short-styled plants (α > 1), then P _L (X) is smaller and P _S (X) is larger than if anthers of both style morphs produce equal amounts of pollen.

I measured diameters of pollen grains in digital photographs of mounted stigmas collected during the field experiment (using methods described above). I used maximum likelihood methods to estimate α by fitting the fraction of pollen grains produced by long-styled anthers (f) [f/(1 − f) = α in Eqs. 1 and 2], using data on diameters of pollen grains found on stigmas as the representation of the overall pollen pool. Specifically, I used the equation:

$$ P(X,f) = fL_{X} + (1 - f)S_{X} $$

(3)

where P(X, f) is the probability that a pollen grain randomly chosen from the pollen pool is in size class X. I assumed that the fractions of pollen grains among size classes followed a multinomial distribution. The log likelihood function of f is:

$$ \log L = {\sum\limits_{i = 1}^n {[X(i)\log (P(X,f))]} } $$

(4)

Using these methods, the estimated fraction of pollen produced by long-styled morphs (f) = 0.5893 and α [=f/(1 − f)] = 1.4346.

I then determined, for each pollen grain on a given stigma, the probability that it was produced by a plant of the opposite style morph (using the estimate of α from Eqs. 3 and 4 in Eqs. 1 and 2), and summed these probabilities for the subset of pollen grains measured on that stigma (I measured diameters only for pollen grains in the focal plane of the photograph). I estimated the number of outcross pollen grains per stigma, by dividing the total proportion of outcross pollen by the number of pollen grains measured, and multiplying by the number of pollen grains per stigma lobe.