Search

The dispersion and capture of differently shaped particles within a Zostera marina L. (eelgrass; Zosteraceae) bed were examined to understand submarine pollination and other dispersals. During periods of moderate flow in the canopy, the capture rate of 'spherical' (the shape of ancestral pollen) and 'filamentous' (the shape of eelgrass pollen) particles was greater for particles released at the top of the canopy (3.07 and 4.53% x [10.sup.-5] [cm.sup.-2] of collector; i.e., percentage of particles captured normalized to collector area) and greater for filamentous than for spherical particles (4.51% x [10.sup.-5] [cm.sup.-2] vs. 2.01% x [10.sup.-5] [cm.sup.-2]). Estimates of the horizontal P (Joseph-Sendner diffusion velocity) and the vertical diffusivity (Gaussian K) of filamentous particles were small (P [approximately equal to] 4 x [10.sup.-4] m/s; K [approximately equal to] [10.sup.-4] [m.sup.2]/s) compared to theoretical values that do not consider plant canopies. These findings support the concept that eelgrass canopies modify the fluid dynamics (i.e., reduced turbulent mixing) within their canopies. These results indicate that 1000-10000 Z. marina pollen are required to pollinate a single flower. Similarly, it was estimated that under some conditions, the probability of particle impaction on eelgrass vegetation approaches certainty. These results provide insight into the evolution of filamentous pollen and submarine pollination, as well as dispersal and other mass transport phenomena within macrophyte canopies. Key words: canopy flow; dispersion; mixing; particle capture; particle transport; seagrass; submarine pollination; Zosteraceae; Zostera marina.

Penstemon digitalis, a prairie species whose flowers possess a large bristle staminode, is visited by eight bee species of varying size. Two sets of field experiments involving staminode removal were performed to test pollination efficiency in relation to bee size. Our data indicate that bristle staminode presence and function are influenced by size-dependent selection on bee body size and associated pollen transporting attributes. The first experiment compared staminode presence and removal in open-pollinated flowers. Staminode removal significantly reduced total pollen deposition but had no effect on total pollen removal. The second experiment utilized single bee visits to assess the interaction between pollinator size and staminode presence on the rate of pollen deposition and removal. This experiment indicated that staminode removal resulted in fewer pollen grains deposited on stigmas and less pollen removed from anthers for both large and small bees (the contrary was true for medium bees). Although the number of pollen grains deposited was greatly reduced for large bees, staminode removal reduced deposition efficiency most notably for small bees by 41.3%, reducing female reproductive success. Staminode removal increased pollen removal efficiency most notably for medium bees by 22.7%, reducing male reproductive success. Mechanisms of staminode function are discussed. Key words: Apoidaea; Penstemon; Scrophulariaceae; size-dependent selection; staminode.

The distance of pollen movement is an important determinant of the neighborhood area of plant populations. In earlier studies, we designed a method for estimating the distance of pollen dispersal, on the basis of the analysis of the differentiation among the pollen clouds of a sample of females, spaced across the landscape. The method was based solely on an estimate of the global level of differentiation among the pollen clouds of the total array of sampled females. Here, we develop novel estimators, on the basis of the divergence of pollen clouds for all pairs of females, assuming that an independent estimate of adult population density is available. A simulation study shows that the estimators are all slightly biased, but that most have enough precision to be useful, at least with adequate sample sizes. We show that one of the novel pairwise methods provides estimates that are slightly better than the best global estimate, especially when the markers used have low exclusion probability. The new method can also be generalized to the case where there is no prior information on the density of reproductive adults. In that case, we can jointly estimate the density itself and the pollen dispersal distance, given sufficient sample sizes. The bias of this last estimator is larger and the precision is lower than for those estimates based on independent estimates of density, but the estimate is of some interest, because a meaningful independent estimate of the density of reproducing individuals is difficult to obtain in most cases.

Many trees in mature forests show intermittent reproduction. Intensive flowering and seed production occur only once in several years, often synchronized over a long distance. In a previous paper, we showed that the limitation of fruit production by the outcross pollen availability can bring about synchronized reproduction of trees in a constant environment, assuming that pollen availability depends on the mean flowering intensity of the trees. However, pollen exchange normally occurs within a distance much shorter than the extent of the whole forest. We studied a coupled map lattice, in which each tree engages in the chaotic dynamics of energy reserve level, but different trees are coupled by pollen exchange with neighbors. We first derived a relation between two statistics of spatial covariance (dynamic spatial covariance and snapshot spatial covariance). A strong synchronization of tree reproduction can develop over the whole forest that may be orders of magnitude larger than the distance of direct pollen exchange between trees. The fluctuation is close to a cycle with a period of two years. The model of local coupling always generates non-uniform spatial patterns, but the enhanced spatial covariance caused by the spatial heterogeneity is restricted to a short range, only a few times larger than the spatial range of direct interaction. When pollen exchange occurs beyond the nearest neighbors, the local spatial pattern becomes proportionally larger, but the condition for synchronization of the whole forest and its magnitude are the same as for the case with the nearest neighbor pollen exchange. When a fraction of the seeds are sired by globally dispersed pollen and the rest are sired by local pollen, long-range synchronization can occur for a wide-parameter region, and trees may engage in a fluctuation with a masting interval longer than two years. We discuss alternative explanations for the long-range synchronization of beech forests. Key words: beech forest; coupled map lattice; dynamic spatial covariance; Japan; masting; pollen coupling of trees; synchronization, long-range.

TwoGener, a novel two-generation (parent-offspring) approach to the study of genetic structure is presented. It makes possible quantification of heterogeneity among the male gamete pools sampled by scattered maternal trees and to estimate mean pollination distance and effective neighborhood size. TwoGener is illustrated where the male gamete can be categorically or ambiguously determined and applied to an empirical study of Quercus alba in Misouri.

Distance and pollen competition and their effects on gene flow in the wind-pollinated meadow grass Festuca pratensis Huds. are discussed. Gene flow was seen to be affected mainly by the distance between the donor and acceptor plants and to decrease rapidly with distance to the donor field up to 75 m. Beyond that distance, gene flow is much slower.

We investigated pollen dispersal in an obligate pollination mutualism between Yucca filamentosa and Tegeticula yuccasella. Yucca moths are the only documented pollinator of yuccas, and moth larvae feed solely on developing yucca seeds. The quality of pollination by a female moth affects larval survival because flowers receiving small amounts of pollen or self-pollen have a high abscission probability, and larvae die in abscised flowers. We tested the prediction that yucca moths primarily perform outcross pollinations by using fluorescent dye to track pollen dispersal in five populations of Y. filamentosa. Dye transfers within plants were common in all populations (mean +/- 1 SE, 55 +/- 3.0%), indicating that moths frequently deposit self-pollen. Distance of dye transfers ranged from 0 to 50 m, and the mean number of flowering plants between the pollen donor and recipient was 5 (median = 0), suggesting that most pollen was transferred among near neighbors. A multilocus genetic estimate of outcrossing based on seedlings matured from open-pollinated fruits at one site was 94 +/- 6% (mean +/- 1 SD). We discuss why moths frequently deposit self-pollen to the detriment of their offspring and compare the yucca-yucca moth interaction with other obligate pollinator mutualisms in which neither pollinator nor plant benefit from self-pollination. Key words: Agavaceae; fluorescent dyes; mutualism; plant-pollinator interactions; pollinator behavior; Prodoxidae; self-pollination; Tegeticula; Yucca.

Optimal sampling designs for studies of gene flow are discussed. The sampling design has been studied theoretically and an algorithm suggested, but the authors in question reached intuitive conclusions dependent on two quite restrictive assumptions. Another algorithm has been set up for optimizing sampling design relative to the mean integrated squared error (MISE). Denser sampling where the function decreases rapidly and strong effort where pi(1- pi) is high would be desirable. Sampling efforts control variance. Constraints bring a trade-off between the two effects. Strong directional winds across a wind-pollinated crop or where foraging behavior of insect pollinators is affected are discussed.

Multiple paternity in the bird-pollinated mallee tree, Eucalyptus rameliana (Myrtaceae), has been studied using six locations. Estimates of the level of multiple paternity/correlated outcrossing in fruits were made in the predominantly outbreeding population. Behavior of the pollinators promotes pollen dispersal and multiple paternity. When the female contribution to the gene pool is concentrated in few plants, this can be quite important.

Pollen dispersal in a stand of 62 adult bur oaks in northern Illinois was characterized using microsatellite analysis. The analysis showed that 57% of the acorns were pollinated by trees outside the stand. Pollen donors for two of the three maternal trees were randomly dispersed throughout the stand. The average pollination distance for within-stand pollinations was 75 m. The results showed direct evidences for high levels of long-distance gene flow in a wind-pollinated species.

1 Records of Empetrum nigrum pollen from blanket peat profiles at 18 sites in the southern Pennines, covering a variety of topographic situations, were collated. Inferences were drawn from the pollen database about climatic change over the last 3000 years and the time of onset of gully erosion of the peats. 2 Interpretations of the pollen record were based on observations of the present-day distribution of E. nigrum in British blanket mires, and on studies by surface-sample analysis of its pollen production and dispersal at a North Wales bog. Empetrum favours better-drained situations such as hummocks and gully sides, and its pollen is dispersed only over short distances. Pollen was found in quantity only at one hummock site. 3 Two episodes of high Empetrum pollen were found consistently in the peat profiles: prior to about 860 BC, and between c. AD 1100 and 1250. These episodes were interpreted as the product of extended periods of drier climate. At sites exhibiting severe gully erosion at the present day, high Empetrum pollen values have been maintained subsequent to AD 1250; sites along the middle and upper reaches of erosion gullies, however, show high values only within the last 200-250 years. 4 Because of the sensitivity of the southern Pennine bog vegetation to climatic change, the stratigraphic record of Sphagnum and Racomitrium macrofossils, in combination with the Empetrum pollen record, can be used to provide a framework of climatic change in the region over the last 3000 years. Keywords: climate change, Empetrum nigrum, erosion, peat stratigraphy, pollen analysis, southern Pennines

In Sequoiadendron ovules are borne inside the ovulate cone, and pollination drops secreted from these ovules collect pollen. We examined: (1) the relation between ovular position and pollen capture; (2) pollen behavior when in contact with a pollination drop; and (3) ultrastructure of ovules during pollination drop secretion. During wet periods a water sheet forms on the surface of the cone due to bract shape and wettability. Pollination drops persist inside the wetted cone, and pollen capture resumes immediately after drying. Pollen landing on a pollination drop is taken inside the drop and carried into the micropyle when the drop contracts. Several notable ultrastructural features appear in the nucellus, integument, chalaza, and bract lamina during pollination-drop secretion. The abaxial surface of the lamina is covered by a membrane that may contribute to the wettable nature of the surface. Key words: ovulate cone; pollen; pollination drop; Sequoiadendron; Taxodiaceae.

In this paper we study aspects of the breeding system of Hormathophylla spinosa (Cruciferae) to identify the factors responsible for seed production in the absence of insect pollinators. The pollinator-exclusion experiments show that H. spinosa, under natural conditions, does not produce seed by apomixis or spontaneous autogamy. H. spinosa appears to be self-incompatible but slightly geitonogamous. Thus, this plant species needs pollen vectors for reproduction. The results of the wind-exclusion experiments performed during two different years in two populations of H. spinosa support the hypothesis that the wind acts as a pollen vector; flowers excluded from the wind had a lower fruit set and female fertility than flowers excluded from all pollinator insects (winged and wingless). This generalist pollination system (insects and wind) permits this species to colonize and sustain viable populations in high mountains, where this species is the only woody shrub living above 3000 m above sea level in the Sierra Nevada. Key words: Cruciferae; high-mountain ecology; Hormathophylla spinosa; Sierra Nevada; wind pollination.

I studied the effect of independent variation in pollen supply and resources (plant size and light availability) on fruit production by Calyptrogyne ghiesbreghtiana, assessing fruit initiation on a per flowering episode basis. Since this species shows no abortion of fruit between initiation and ripening (although many fruit are eaten by insects), any effect of pollen supply or resource availability limiting fruit production will be most apparent at fruit initiation. 2 Variation in visitation by bats, the pollinators of this species, was correlated with fruit initiation. A survey of pollen loads on 642 bat-visited female flowers revealed, however, that pollen transfer by bats was frequently ineffective. Even after bat visitation, 46% of flowers remain unpollinated. 3 Controlling the supply of pollen to inflorescences at two levels (155 flowers vs. 50 flowers per inflorescence) revealed that the number of fruit initiated per inflorescence is significantly affected by pollen supply. At the high level of controlled hand-pollination the median number of fruit initiated per infructescence was almost five times greater than in the naturally pollinated population. 4 Variation in plant size, light availability, and the number of flowering episodes in the previous year, failed to explain any of the variation in fruit initiation by the unmanipulated field population, or the hand-pollinated population. The resource environment is, however, important to lifetime reproductive success because of its influence on flowering frequency. 5 To my knowledge, there are no other field studies of reproduction by non-herbaceous perennial plants that examine the effect of independent variation in both pollen supply and the resource environment. Keywords: Arecaceae, fruit set, plant size, pollen limitation, resource availability

High potential for outcrossing is the primary reproductive character that mediates the representation of taxa in palynological records. Pollen present in lake sediments is primarily monoecious or dioecious taxa pollinated by anemophily or by insects. Flower morphology is a significant factor in determining the pollinator and the release of pollen.

Bumblebee exhibits preference for flower symmetry in Epilobium angustifolium, as explained by large nectar production in symmetrical flowers. Thus, floral symmetry indicates nectar production, thereby improving pollen transport. Floral asymmetry reflects developmental homeostasis, explaining developmental selection in the plants.

Stamens of Berberis thunbergii respond to a tactile stimulus by snapping toward the stigma. When filaments are stimulated ('tripped'), anthers that strike an object (e.g., a glass cover slip or floral visitor) deposit some of the sticky pollen that is held together by viscin threads. When stamens were stimulated experimentally, 53% of the total pollen was deposited on the first trip, and 20% on the second, resulting in a pattern of diminishing proportional removal. Pollen removal by bees resulted in a similar pattern. However, small bees (e.g., Andrenids) tripped fewer stamens per visit than large bees (i.e., Bombus spp.). Our results do not support either the 'numerical' or 'proportional' removal patterns of Harder and Thomson (1989) but do conform to the diminishing proportional removal model proposed by Harder (1990) for plant/pollinator associations where there is a high variance in the number of pollinator visits.

1 The hypothesis that most of the pollen collected by forest-floor moss polsters originates from local sources (i.e. trees growing within a 20-m radius) was tested by comparing pollen and forest inventory data from 30 plots in three extensively forested regions. Within each region, plots were located in forest stands similar in composition to regional forests (matrix sites) as well as in patches differing substantially from regional forests (patch sites). 2 Scatter plots and regression analyses indicated that between 25 and 90% of the tree pollen in the assemblages originated from trees growing beyond the 20-m forest-sampling radius. Most of the extraneous pollen was from taxa that were (i) regionally abundant in forests, and (ii) highly productive of well-dispersed pollen. 3 Despite dominance of the assemblages by nonlocal pollen, the pollen assemblages recorded subregional patterns in forest composition. 4 Moss polsters may be especially subject to annual or seasonal variations in pollen production and dispersal. Therefore, application of pollen source areas and pollen-vegetation calibration parameters estimated from moss-polster studies to interpretation of fossil-pollen sequences from closed-canopy sediments (humus, small hollows) is uncertain. 5 Studies of modern pollen assemblages in which pollen deposition is integrated over several years are needed in patchy forests to assess pollen source area of closed-canopy basins and to determine the effects of vegetation patterns on pollen representation.

1 The allocation of resources among various reproductive structures and functions can suggest the relative importance of past selective pressures in moulding reproductive patterns. 2 To determine how male and female plants of the dioecious shrub Oemleria cerasiformis differ in allocation to reproductive functions, biomass was partitioned among reproductive structures for 20 plants of each sex. 3 Structures contributing to pollinator attraction (petals and hypanthium) were heavier in males than in females. In males, these structures constituted 63% of total reproductive biomass; inflorescence stems and bracts constituted 28%, and the androecium only 9%. In females at flowering, petals and hypanthium constituted 50% of reproductive biomass, inflorescence stems and bracts 37%, and the gynoecium 13%. 4 In females, fruits constituted 87% of the total reproductive biomass at average fruit set (13.7% of pistils). Even with the lowest observed fruit set (4% of pistils), fruit comprised 75% of reproductive biomass overall, and pollinator attraction no more than 4%. Fruit biomass was distributed about equally between pulp (offspring dispersal) and stone (offspring provisioning and protection). A flower can produce up to 150 times its own weight in fruit, thus fruit set is the main factor determining how biomass is distributed among reproductive structures in females. 5 These data support the view that the primary factors influencing the evolution of reproductive allocation are, in males, pollen limitation, and in females, the necessity of provisioning and dispersing offspring.

Many plants produce more flowers than will set fruit. In some species, a subset of flowers are actually female-sterile and morphologically distinct; these flowers are presumed to enhance fitness primarily through pollen donation. Individual plants of the hummingbird-pollinated shrub Besleria triflora produce both long-styled and short-styled flower morphs, characteristic of these 'andromonoecious' sexual systems. Short-styled flowers set fruit rarely when hand-pollinated and are therefore functionally staminate. To test the hypothesis that selection for increased male function maintains flower dimorphism, I measured relative pollen dispersal per unit investment by staminate vs. perfect (long-styled) flowers. In the field, anthers of perfect flowers dispersed pollen for a significantly longer time than did those of staminate flowers. In laboratory experiments, hummingbirds transferred more pollen per visit from perfect flowers than from staminate flowers, in part due to differences in flower-morph pollen production. As a result, perfect flowers dispersed substantially more pollen than staminate flowers over an average flower lifetime. The two flower morphs did not differ in the amount of biomass invested through the time of corolla abscission, suggesting that staminate flowers do not conserve resources substantially. These results fail to support the 'male-function' hypothesis for the evolution of andromonoecy. Staminate flowers contribute relatively little to plant reproductive success directly through male function, suggesting they are maintained for reproductive functions other than pollen donation.

Size-related patterns of floral sex allocation were examined in 14 anemophilous and eight entomophilous monoecious plants. Among the anemophilous species, eight increased male floral allocation and six increased female floral allocation with increased plant size. An increase in female floral allocation with increased plant size occurred in all eight entomophilous monoecious species examined. Pollination vector has a significant effect on size-related floral sex ratio patterns. We present a heuristic model that considers pollen dispersal of anemophilous monoecious plants as a function of plant geometry. The surface area to volume ratio of plants approximating geometric extremes (cylindrical or spherical forms) should affect pollen dispersal and floral sex ratio due to aerodynamic effects on the whole plant. Plants approximating cylinders should increase male floral allocation with increasing height. Plants that approximate a more spherical form should increase in female floral allocation with increasing size. All entomophilous plants conformed to a resource limitation model; however, anemophilous plants increased in either male or female floral allocation with increased size. Results for 13 anemophilous species support the predictions of our geometric model. One species does not approximate an extreme geometry and does not appear to conform to the geometrical model's predictions.

Conditions just prior to field abandonment and the early stages of forest regeneration have residual effects on patterns of forest regeneration, but are often not available for study. The potential of pollen preserved in acidic soils to reconstruct land-use immediately before field abandonment was tested here by comparing known stand histories with analysis of pollen which had washed downward into the soil. Large amounts of Ambrosia in the soil pollen, accompanied by Asteraceae and Poaceae, indicated that a site was a plowed field prior to the start of forest regeneration. Abrupt shifts in proportions of Pinus and Quercus pollen correlated with known logging histories. Sites of unknown history included similar diversity in the proportions of nonarboreal pollen, and no evidence of logging. Proportions of tree taxa in soil pollen and vegetation surveys correlated only weakly. Soil pollen analysis, under appropriate conditions of pollen preservation, can provide evidence of major differences in site histories for which there is otherwise little evidence.

The distance that pollen is dispersed has important ecological and evolutionary implications. Nevertheless, our understanding of the mechanism that generates spatial patterns of pollen dispersal in animal-pollinated plants, namely the combined processes of pollinator movement and pollen deposition, is at present qualitative. I show how easily measured characteristics of pollinator movement and pollen deposition can be used to build models that generate quantitative predictions about pollen dispersal distance. A field experiment in which I followed honey bees (Apis mellifera) foraging on one-dimensional arrays of mustard plants (Brassica campestris) indicated that individual honeybees have a biased direction of movement and alter their movement patterns in response to plant spacing. I used the model to explore the potential consequences of biased movement and plant spacing for pollen dispersal. Biased movement results in an increase in the expected pollen dispersal distance. In addition, the predicted dispersal distance increased with increasing interplant distance. These results demonstrate that models of pollen spatial dynamics based on individual pollinator behavior can be used to explore the consequences of plant spatial arrangement for gene flow in plant populations.

Little is known about the reproductive biology of the Crassulaceae. We studied a population of Echeveria gibbiflora in the Pedregal de San Angel ecological preserve in Mexico City, Mexico. Each flower is open and producing nectar 7 to 8 days. On the days of maximum nectar production (flowers 4-6 days old) an average of 14.5 microns l accumulates in a flower per day. The maximum rate of nectar production is between 0700 and 0900 hours. The average sugar concentration in the nectar is 43.7%. In a given flower, pollen is exposed and the stigmas are receptive at the same time. The average natural fruit-set and seed-set are 56.6% and 35.5%, respectively. The pollen-ovule ratio is 124, and the plants are fully self-compatible. The flowers are pollinated by only one species of hummingbird (Cynanthus latirostris) and are never visited by insects. Pollen movement is very limited (mean of pollinator flight distances = 0.72 m, mean distance fluorescent dyes = 0.92 m). Seed dispersion is by gravity and wind, and also is very limited (an average of 1.07 m). The total genetic neighborhood area is 15 m squared, to 17 m squared, producing a neighborhood effective population size (N(sub b)) of 5.01 to 39.7 individuals. This is a very small N(sub b), indicating that genetic drift may be a dominant force in the evolution of this species.

The reliability of paternity analysis in estimating mean effective pollen dispersal was evaluated. Pollen parents were inferred based on the degree of genetic relationship with the offspring alone or combined with data on the probability of mating with mother plants. Computer simulation showed that the mean distance between inferred males and mother plants could give reliable estimates of mean effective pollen dispersal under specified circumstances. Some conditions that affect the reliability of paternity analysis were discussed. The results have applications in studies on the role of gene dispersal through pollen on plant population genetic structure.

The hypothesis that genetic correlations among animal-pollinated plants decrease with increasing distance was investigated in Ipomopsis aggregata by determining its inbreeding coefficient, or F(sub ST), analysis with spatial autocorrelation and electrophoretic analysis of leaves and flower buds. The spatial scale of genetic differentiation did not compare conclusively with models of isolation-by-distance. Some reasons for the discrepancy between theory and observations are discussed.

1. The effect of male tree density on pollen flow and fruit set in females was examined for three species of dioecious tree (Neolitsea dealbata, Litsea leefeana and Diospyros pentamera) in complex notophyll rain forest in tropical, northeastern Australia. 2. The spatial distribution for all reproductive individuals combined was significantly clumped in Diospyros but not in either Litsea or Neolitsea. Separate male and female groups were significantly clumped in all species. More than 80% of nearest male neighbours occurred within 10m of Neolitsea females and within 20m of Litsea and Diopsyros females. The ratio of males to females was 1-6:1 (Neolitsea), 1.5:1 (Diopsyros) and 1.3:1 (Litsea). 3. No species produced fruit asexually when flowers were isolated in mesh bags. All species populations flowered synchronously but Diopsyros flowered for a longer period (15 weeks), than either Neolitsea (3-4 weeks) or Litsea (4-5 weeks). All species had small, unspecialized flowers and were pollinated by small, generalist insects. Male trees produced more flowers per reproductive shoot than females and flower production was positively correlated with tree size in males but not in females. Fruit set was not correlated with tree size in female trees. 4. The ratio of fruit set to flowers in females was negatively related to two measurements of male to female distance (distance to the nearest male tree (all species) and mean distance to the nearest 10 (Neolitsea and Diospyros) or three (Litsea) males, termed |local male distance'). The rate of reduction in fruit set with increasing nearest male distance was the same for all species, although Diospyros produced a higher proportion of fruit relative to flowers. The ranges of fruit set for Neolitsea, Litsea and Diospyros were 1-30%, 2-20% and 20-40%, respectively. Fruit set in Neolitsea females was similar in two successive seasons, indicating a consistent effect of local male distance on fruit set. 5. The amount of pollen trapped at female trees stress over the whole flowering season was unrelated to fruit set or to local male distance in all species. Amounts of pollen trapped at females after the first 6 (Neolitsea) and 15 (Diospyros) days of flowering were negatively correlated with local male distance. In Neolitsea females, the amount of pollen trapped at females after 6 days of flowering was positively related to fruit set, indicating that effective fertilzations occur early in the flowering period. 6. Although the mechanism for density-dependent pollen movement is the same for all three species, the longer flowering period and more gradual production of flowers found in Diospyros allowed more time for pollen to move through the population. It is concluded that rain-forest trees which have obligate outcrossing breeding systems, brief, synchronous flowering periods and are pollinated by a non-specialized fauna risk reduced rates of fruit set when relatively isolated from conspecifics. Key-words: dioecy, pollen limitation, tree density Journal of Ecology 1992, 80, 57-69

Pollen-mediated gene flow along a linear series of patches of the functionally gynodioecious, bee-pollinated Cucurbita foetidissima was assessed using electrophoretic analysis of seed allozymes. Gene flow was documented at distances up to 0.7 km. For the 13 patches examined, interpatch pollen must have sired from 0% to 48.3% (X = 8.5%) of the seeds of monoecious plants (hermaphrodites). Rates of interpatch siring of seeds of pistillate plants (females) averaged 20.4% (range 8.6%-40%) for the three patches examined. Heterogeneity among fruits in seeds sired by interpatch pollen indicates that the arrival of interpatch pollen is clumped with respect to stigmas. Within patches, plants of the same sex type usually shared identical five-locus genotypes, suggesting that clonal propagation predominates. Since approximately 90% of seeds are sired by intrapatch pollen, seeds of monoecious plants appear to result primarily from geitonogamous (self-) fertilization. This may help explain the existence of female plants in natural populations, since self-fertilization has been shown to severely reduce the survival of seedlings in this species.

The anthers of Torenia fournieri were found to shed pollen forcibly by lever action. Anther structure was modified for this function by a flangelike outgrowth of the lateral pollen sac wall forming a lever. When pressed, this lever causes an infolding of the thinner, subadjacent, pollen sac wall forcing pollen from the stomium. A force of 1-1.5 g pressing against the four levers of an anther pair resulted in the forcible shedding of 2,000-3,000 pollen grains in two parallel rows. During the 2-day anthesis the flowers shift from functioning as males to hermaphroditic outcrossers, and yet only have a pollen: ovule ratio of 98.6, a ratio more indicative of facultative autogamy. The outermost pair of anthers functions on the first day of anthesis, while the second, inner pair functions on both days. In each flower, the 2-day anther pair produces approximately twice as many pollen grains as the 1-day anther pair, a pollen production highly correlated with the length of their functional lives. This difference in pollen production is apparent in the larger size of the day2-day anthers, a size difference that first appears with the initiation of the anther primordia.

A technique using micronized metal powders was developed for both general labeling of pollen and marking of individual pollen grains. After labeling, pollen flow is analyzed by the use of backscatter scanning electron microscopy and X-ray microanalysis. To test the efficiency and efficacy of the technique, we assessed differences in pollen distribution in Solanum rostratum, an enantiostylous species with dimorphic anthers which are putatively feeding and pollinating anthers. Pollen from each set of anthers was labeled using different micronized metal powders. We could not confirm the differentiation of functional anthers in S. rostratum. This technique provides an efficient and convenient method for tracking pollen movement within and between flowers, and anthers within a single blossom can be differentially marked.

We compared success and speed of pollinium germination and putative ovule penetration in self- and cross-pollinated flowers from three clones of common milkweed (Asclepias syriaca). Event hough common milkweed is largely or entirely self-incompatible, success of selfed flowers equalled or exceeded that of crossed ones (87% vs. 74% germination, 78% vs. 63% putative penetration). This result is consistent with the postfertilization incompatibility mechanism proposed for A. syriaca. Clones different in success, and the recipient clone with the highest combined selfed and crossed success (97% vs. 74% germination for the others, 95% vs. 56% putative penetration for the others) also developed most rapidly (90% vs. 45% germination for the others 1 day after pollination; 90% vs. 13% putative penetration after 1 day). The significant among-clone variation suggests that factors intrinsic to the clones strongly affect rates of fertilization and putative penetration.

In Cleistes divaricata, a hinged anther dispenses a sequence of loosely aggregated pollen tetrad masses onto the dorsal thorax of Bombus and Megachile workers. This mechanism allows more than one bee to receive pollen from a single flower. To determine the effectiveness of this strategy, plants in bud were caged and hand pollinated using pollen from sequential releases. Capsules produced from a flower's first three pollen dumps were significantly larger and heavier and contained more seeds than those produced by later pollen deposits with fewer tetrads. Decreasing pollen dosage did not affect fruit set, capsule development, individual seed weight, or percentage of healthy looking embryos per fruit. Fruit set for artificial pollinations was 93% for pollen dumps 1-3 and 97% for dumps 4-18. As set of naturally pollinated marked plants was only 38%, low frequency of pollinator visits apparently limited seed production. Efficacy of infrequent visits is maximized, however, because a flower's first pollen releases contain the most pollen, but later low-dosage dumps can produce some seeds. An unexpected cost was a higher investment of pericarp per seed in low-dosage capsules (130 ng pericarp [seed.sub.-1] fresh weight) than in high-dosage pollinations (only 27 ng pericarp [seed.sub.-1]). Flower fading in plants receiving low pollen dosage was as slow as that in caged, unpollinated controls, thus increasing opportunity for additional pollination.

The role of gene establishment in gene flow was investigated in a population of the annual legume Chamaecrista fasciculata by determining the effect of interparent distance on progeny fitness throughout the entire life history. A decelerating gain in progeny fitness with increasing interparent distance was observed. Selfed progeny suffered a 2-fold fitness disadvantage compared to progeny derived from mating events between individuals in the same neighborhood. Progency derived from within neighborhood crosses had lower fitness than progeny from crosses between neighborhoods. Coupling the effect of interparent distance on gene establishment with information on gene dispersal resulted in a considerable increase in estimated gene flow. However, gene flow was still limited, as the average neighborhood area corresponded to a circle with radius of approximately 3.0 m. Yearly fluctuations in population size and variation in reproductive output lowered the estimate of [N.sub.e] below the census estimate to approximately 100 individuals. The role of a seed bank in increasing the estimate of [N.sub.e] was found to be insignificant. It is likely that genetic drift plays a major role in determining the distribution of genetic variation in this population.

Bryophytic polsters and surface samples were collected from four sites in Shannon and Carter counties in the SE Missouri Ozarks to determine modern pollen rain. Regional pollen rain and variation in the local pollen rain are reflected by modern pollen spectra. Regional pollen rain was calculated by averaging the percentages of the various taxa in the pollen spectra from the 10 samples collected. In this area the average regional pollen rain is dominated by Pinus (18.5), Quercus (51.5), Carya (4.4) and Ambrosia (8.7). The data are consistent with the mosaic of pine-oak and oak-hickory-pine forests characteristic of this region.

SUMMARY (1) Pollen diagrams are presented for five sites in lowland Craven in north-west Yorkshire, an area (about 200 [KM.sup.2] of varied relief and soils. (2) Associated radio-carbon dates show that the vegetational history extends from the end of the Devensian late-glacial to the present day. (3) The early immigration and spread of Pinus and Alnus are discussed in relation to changes in climate, soil and human interference. (4) It is suggested that up to the Ulmus decline human interference, although probably present, was not a major influence on the vegetation of this lowland area. (5) After the Ulmus decline human interference was of prime importance in determining vegetation history, but with greatly differing intensity and progress on different soils. On high-quality limestone soils, major clearance occurred in the Bronze Age, but on the heavy clay soils in the western part of the region, intensive clearance did not take place until Anglo-Saxon times.