Effect of Ovule Position within the Pod on the Probability of Seed Production in Bauhinia ungulata (Fabaceae)

Abstract

• Background and Aims It has been claimed that ovules linearly ordered within a fruit differ in their probabilities of reaching maturity. This was investigated by studying the effect the position of an ovule within the pod has on seed abortion and seed production in Bauhinia ungulata.

• Methods Fruits collected during the dry seasons of 1999, 2000 and 2001 were opened, and the number, position and status of each ovule within the fruit were recorded. A GLM model was used to assess the effects of population, tree identity and ovule position within the pod on ovule fertilization, seed abortion, seed damage and seed maturation in two populations of B. ungulata.

• Key Results Nearly 30 % of the ovules were not fertilized in 1999; this percentage dropped to 5 % the following two years. Seed abortion (50 %) and seed damage (15 %) were the same every year during the study period. Only 15 % of the initial ovules developed into mature seeds in 1999; this value increased to 35 % in 2000 and 2001. However, seed survivorship was dependent on the position of the ovule within the pod; non-fertilized and early aborted ovules were found more often near the basal end of the ovary. The frequency of seed damage was not affected by position. Mature seeds were found mainly in the stylar half of fruits, where ovules are likely to be fertilized by fast pollen tubes.

• Conclusions The pattern of seed production in B. ungulata is non-random but is dependent upon the position of the ovule within the pod. The results suggest that the seeds produced within a fruit might differ in their vigour.

INTRODUCTION

Typically, only a small fraction of the initial number of flowers pollinated in a given season develops mature fruits. This initial excess establishes the conditions for resource competition among flowers and their ovaries. Such competition has been proposed as a mechanism through which plants can control the vigour of their offspring (Stephenson and Bertin, 1983; Davis et al., 1987; Torres et al., 2002).

In some species, fruit set seems to be limited by pollen availability (Bierzychudek, 1981; Johnson and Nilsson, 1999). When pollen loads are variable in composition and/or the number of grains per load increases beyond ovule number, the competition among pollen grains for access to ovules becomes intense. Under such conditions, the vigour of the offspring tends to increase (Aizen and Searcy, 1998; Brown and Kephart, 1999; Niesembaum, 1999). Furthermore, in these cases it has been shown that a subset of the competing donors tends to be over-represented among the offspring (Marshall and Ellstrand, 1986).

It is well known that most flowering plants produce more ovules than the number of seeds they mature (Stephenson, 1981). Sometimes, ovules fail to mature seeds because most of their flowers do not receive enough pollen to fertilize all of their ovules and are eventually aborted (Calvo, 1993; Murphy and Vasseur, 1995; Johnson and Nilsson, 1999). However, in the majority of species, seed maturation does not seem to be limited by pollen availability and ovule fertilization (Stephenson, 1981; Rocha and Stephenson, 1991a). On the contrary, most flowering plants receive enough pollen grains deposited on their stigmas to fertilize all of their ovules, and still many are aborted (Sutherland, 1987; Lee, 1988; Corbet, 1998).

Several hypotheses have been proposed to explain the high levels of seed abortion present in most plant species. Some argue that many seeds might be aborted because of the expression of lethal or deleterious alleles. They propose that high levels of homozygosity due to inbreeding may lead to high rates of seed abortion (Wiens et al., 1987). In outcrossing species, seed abortion increases as the level of inbreeding increases, as a result of the genetic load.

Others argue that abortion depends on the supply of resources for seed development (Lee, 1988). Assuming that all the ovules are fertilized, the zygotes that are closer to the source of resources will receive them first and/or in larger quantities than zygotes further from the source. As a consequence, the latter will have a higher probability of being aborted (Lee and Bazzaz, 1982a, b). This could lead to a non-random pattern of seed abortion, with regard to ovule position.

Lastly, other authors have proposed that seed abortion might be due to differences in the ability of the developing seeds to gather maternal resources (Rocha and Stephenson, 1991b; Melser and Klinkhamer, 2001). All seeds within a fruit must compete for resources allocated by the maternal plant. Stronger seeds may more efficiently garner resources than less competent seeds and will mature fully, while the others will starve. Winsor et al. (2000) showed that offspring vigour is related to the vigour of the paternal plant. Therefore stronger, faster-growing pollen tubes will give rise to stronger seeds. Under these conditions, seed abortion is claimed to be a consequence of the gametophytic competition for the access to ovules. From this, it is expected that ovules located closer to the point of entry of pollen tubes will present a lower probability of abortion, while those found farther from the entry point will present a higher probability of abortion (Bawa and Webb, 1984; Rocha and Stephenson, 1990, 1991a; Mohan Raju et al., 1996).

To test these hypotheses, it is necessary first to determine the patterns of seed abortion that a species exhibits under natural conditions. In many legumes, seeds are arranged linearly within the pod, making it relatively simple to study the effect of position of the ovule on seed abortion and seed set. Once these patterns have been described, experiments can be designed to test for the possible effects of abortion on the fitness of the offspring. The purpose of this study was to determine the effect of ovule position within the pod on the patterns of seed abortion and seed set in Bauhinia ungulata, a tropical dry forest tree, as a first step in evaluating the effect of seed abortion on the fitness of the offspring.

MATERIALS AND METHODS

Study site

This study was conducted at Santa Rosa National Park, located in north-western Costa Rica (10°45′ to 11°00′N and 85°30′ to 85°45′W). This park is part of the Guanacaste Conservation Area (ACG). This study was conducted in an area of the ACG classified as moist transition dry forest (Holdridge, 1967; Hartshorn, 1983). The park includes a mosaic of forests of different ages and abandoned pastures (Hartshorn, 1983; Janzen, 1983; Gerhardt, 1993). The climate is highly seasonal, with a well-defined dry season from late November to mid-May. Annual rainfall ranges between 800 and 2600 mm, with an annual mean of 1423·4 mm. The annual mean temperature is 25·7 °C, and the mean annual relative humidity is 81 % (Rojas Jiménez, 2001).

Study species

Bauhinia ungulata L. (Fabaceae: Caesalpinioideae) is a small tree, 5–10 m high, typical of semi-deciduous and deciduous forests. It is distributed from southern Mexico to the dry areas of Venezuela and Colombia (Holdridge et al., 1997). In Costa Rica, this species is found only in the dry parts of the North Pacific region.

Flowering occurs during the dry season, from December to March. Bauhinia ungulata has a floral dimorphism based on the length of the pistil. Most of the flowers have a pistil longer than the anthers. However, some flowers have a very short pistil, the stigma being found right next to the anthers (Wunderlin, 1983). Flower anthesis is crepuscular; the flowers open at around 1700 h and release pollen soon afterwards. Nectar is secreted within 1 h of flower opening and seems to be secreted throughout the night (Mena-Alí, 2003; J. I. Mena-Alí, unpubl. data). Bauhinia ungulata is pollinated by bats. The main pollinator is Glossophaga soricina (Heithaus et al., 1974, 1982), although it is commonly visited by other species of bats, including Phyllostomus discolor (Ramirez et al., 1984), Carollia perspicillata and Anoura sp. (Fischer, 1992). Fruit set occurs from January to May, at the time when the species is flushing its new foliage. The pod is dehiscent and the seeds are ballistically dispersed by the sudden split of the two parts of the pod (Holdridge et al., 1997) throwing the seeds away from the mother tree. The mean number of seeds is 19, although it ranges from 10 to 30 seeds per pod (Mena-Alí, 2003).

Likelihood of seed maturation

To determine the likelihood of seed maturation within fruits of B. ungulata, fruits were collected from 19 randomly selected trees during the fruiting season of 1999. Those trees were located in two populations, 3 km apart, at Santa Rosa National Park. To avoid the effect of age and/or size on the patterns of seed set, all the trees considered in this study had diameters at breast height between 8 and 10 cm. This range represents 90 % of all the trees present in the study area.

At least 50 mature fruits from each tree were collected. The fruits were opened and the number of ovules was recorded. Furthermore, each ovule/seed was classified according to its condition as (a) unfertilized ovules, (b) fertilized but aborted seeds, (c) seed developed but damaged by insects, and (d) mature seeds. Aborted seeds were easily identified because they exhibited signs of growth and were always necrotic, even at early stages of abortion; unfertilized ovules were very small, always green in appearance and never showed signs of growth or necrosis.

In 2000 and 2001 the study was repeated to determine the patterns of seed production within fruits of B. ungulata. As described above, the number of ovules and their condition were recorded; this time, aborted seeds were classified according to the time of abortion—early or late. Furthermore, the position of each ovule or seed within the fruit was recorded. Ovule position was determined by counting each ovule from the stylar end (closest to the style) to the basal end (closest to the peduncle) of the ovary. Ten trees (with diameter at breast height of 8–10 cm) were randomly selected from each population. Fifty fruits per tree were collected throughout the fruiting season. Some trees produced only a small number of mature fruits (<20) and were therefore removed from the analysis; in addition, during 2001 most of the second population was removed from the ACG. This reduced our final population sizes to 15 trees for the year 2000 (eight trees from population 1; seven trees from population 2) and 11 trees for the year 2001 (eight trees from population 1; three trees from population 2).

To examine the effects of ovule position, maternal tree and population were used on the probability of ovule fertilization, seed abortion, seed predation and seed maturation, weighted least-squares mixed-model analyses of variance (GLM, option weight; SAS, 1985), where the maternal tree was nested in population and treated as a random variable and ovule position was treated as a fixed variable. In these analyses, the proportion p of successful ovules/seed in each test, i.e. fertilized, aborted, damaged or viable seeds, was weighted by the inverse of its variance, Var = pq/n, where q = 1 − p (Rocha, 1996).

RESULTS

The results of this study indicate that most ovules within the fruits failed to produce seeds. In 1999, only 10 % of the ovules developed into a seed. In 2001 and 2002 the likelihood of seed production increased to nearly 30 %.

Most ovules were fertilized; about 70 % of the ovules were fertilized in 1999 and close to 95 % in 2000 and 2001. Seed abortion and seed damage appear to be fairly similar among the three years (50 % and 10 %, respectively; Fig. 1). However, seed maturation varied between individual trees; while some trees had many mature seeds, some other trees did not have any mature seeds at all (Fig. 2 and Table 1).

Fig. 1.

Fate of ovules/developing seeds within the pods of Bauhinia ungulata during the study period.

Fig. 2.

Variation in the percentage fate of ovules/developing seeds among individual trees of Bauhinia ungulata during the study period. Trees with same numerals represent the same individual over the 3-year period: (A) for the year 1999; (B) for the year 2000; (C) for the year 2001.

Table 1.

Analysis of variance for the effect of ovule position on the rate of ovule fertilization, early and late seed abortion, seed damage by insects, and seed maturation in Bauhinia ungulata

	F-value (significance level)
Source	Ovule fertilization	Early seed abortion	Late seed abortion	Seed damage by insects	Seed maturation
(A) Fruits collected in 2000
Population	0·24 (0·622)	9·50 (0·003)	45·31 (<0·0001)	26·38 (<0·0001)	4·10 (0·064)
Tree (population)	0·12 (1·000)	6·59 (<0·0001)	22·62 (<0·0001)	21·69 (<0·0001)	53·77 (<0·0001)
Ovule position	22·78 (<0·0001)	75·36 (<0·0001)	1·22 (0·210)	1·01 (0·451)	22·71 (<0·0001)
(B) Fruits collected in 2001
Population	0·01 (0·944)	2·14 (0·150)	6·75 (0·028)	16·48 (0·002)	16·16 (0·003)
Tree (population)	0·08 (1·000)	3·68 (<0·0001)	19·17 (<0·0001)	69·42 (<0·0001)	28·65 (<0·0001)
Ovule position	8·85 (<0·0001)	67·30 (<0·001)	18·82 (<0·0001)	1·43 (0·094)	21·03 (<0·0001)

Significant values are shown in bold.

When the effect of ovule position on the probability of seed set was examined, several trends were found. The results of the weighted least-squares analyses of variance revealed that there is a significant effect of ovule position on the probability of fertilization in 2000 and 2001 (Table 1A and B, respectively). Overall, the ovules in the most peduncular/basal positions within the ovary are less likely to be fertilized than those in the most stylar/distal positions (Fig. 3A). A similar pattern is also seen for early abortions, where ovules in basal positions are more likely to be aborted early in seed development than ovules in stylar positions (Fig. 3B). However, there was no significant effect of ovule position within the fruit on the likelihood of late abortion in the year 2000, but there was a significant effect of ovule position on the probability of late seed abortion in fruits produced in 2001. In 2001, ovules close to the style tend to be aborted more often during the final stages of seed development (Fig. 3C).

Fig. 3.

Likelihood of the different fates of ovules as a function of ovule position within the fruit in Bauhinia ungulata. Position 1 refers to the ovule closest to the style. Bars represent standard deviations. (A) Probability of an ovule being unfertilized. (B) Probability of an initial ovule being fertilized but aborted at an early stage. (C) Probability of a seed being fertilized but aborted at a late stage. (D) Probability of damage by insects. (E) Probability of an initial ovule developing into a mature seed.

Developing and mature seeds were damaged mainly by larvae of a species of pyralid moth and by the beetle Gibbobruchus guanacaste (Bruchidae). It was found that the patterns of damage caused by these insects did not depend on the position of the seed within the fruit: the probability of seed damage seems to be nearly constant among ovule positions (Fig. 3D).

Finally, the probability of a seed developing to maturity is not independent of the position of the ovule within the fruit. In general, ovules closer to the stylar end of the ovary showed a higher probability of reaching maturity than ovules at the basal end. Moreover, ovules in the most basal positions almost never matured into seeds (Fig. 3E).

DISCUSSION

The results presented in this study show that nearly 70 % of the ovules within fruits of Bauhinia ungulata failed to produce mature seeds. Significant effects of population and tree were also found for most of the stages of seed development; this result is not unexpected. Population effects may be due to environmental variation between sites; differences among trees suggest that maternal effects may be involved. Variation in breeding capabilities, resource allocation to seed development and defense responses to attack by herbivores can cause differences among trees. Regardless of the population or the individual tree, the present data revealed that the patterns of seed maturation within the fruits of Bauhinia ungulata are not random, and that the likelihood of seed maturation for each ovule depends on the position it occupies in the pod. Overall, ovule positions located at the basal end of the ovary showed the lowest probability of ovule fertilization/seed maturation.

The present data show that most of the fertilized ovules failed to produce seeds. Even though the amount of pollen deposited on the stigmas of B. ungulata is fairly large, it is likely to include a high proportion of self-pollen (Mena-Alí, 2003). Bauhinia ungulata is self-incompatible (Ramirez et al., 1984); although pollen is able to germinate on stigmas following self-pollination, the growth of these pollen tubes is arrested soon after they enter the style (Mena-Alí, 2003). Self-pollen could therefore interfere with the growth of outcross pollen tubes, and hence result in insufficient fertilization. The rate of ovule fertilization was lower in 1999 (70 %) than in the following two years (95 % for 2000 and 2001). This might be due to differences in the time when fruits were collected. In 1999, fruits were collected in March, close to the end of the flowering season, whereas fruit collection for the following two years was carried out during late January and early February. The low rate of ovule fertilization observed in 1999 could be due to a limited rate of pollination: the reduced number of flowers available at the end of the season might have attracted fewer pollinators, and the rate of geitonogamy might have increased.

Among fertilized ovules, those in the most basal positions within the ovary had a higher probability of being aborted during an early stage of development. This pattern suggests that the basal ovules are not as capable of garnering maternal resources as stylar ovules. It has been argued that ovules farther away from the point of entry of pollen tubes into the ovary are more likely to be fertilized by less vigorous, slow-growing pollen tubes (Davis et al., 1987; Rocha and Stephenson, 1991a, b). Davis et al. (1987) showed an association between the vigour of the male gametophyte and the sporophytic progeny in wild zucchini, where ovules fertilized by less-vigorous pollen tubes gave rise to a less-fit progeny. Rocha and Stephenson (1991a, b) showed that, within a given fruit, ovules that are fertilized by the fastest-growing pollen tubes are more likely to mature into seeds, and also produce more vigorous progeny than ovules fertilized by slower-growing pollen tubes.

The pattern of seed maturation found in B. ungulata suggests that early abortion might be the consequence of gametophytic competition. Hossaert and Valéro (1988) compared the patterns of seed abortion in two species of the genus Lathyrus (Fabaceae). In the autogamous L. sylvestris, they found a random pattern of seed abortion within the fruit, whereas in the outcrossing L. latifolius, the probability of abortion was lower in the stylar end of the ovary. They argued that the difference in the patterns of seed abortion between self- and outcrossing species suggests a relationship between seed abortion and the genetic quality of the pollen that sired the seeds, and hence result in more vigorous offspring.

The fact that abortion at a late stage is not dependent on ovule position might indicate either random segregation of rare deleterious alleles that are not expressed during the gametophytic phase or differences in resource allocation by the maternal plant. When resources are limited, only some of the developing seeds will complete their process of maturation. Since the weak seeds were aborted at an early stage, late abortion could take place in a position-independent fashion. Such interactions are not rare: in Lupinus texensis, the production of seeds seems to depend on both the amount of nutrients available and the magnitude of the genetic load in the offspring (Helenurm and Schaal, 1996).

The present results indicate that seed damage caused by Gibbobruchus guanacaste (Bruchidae) and the larvae of pyralid moths is independent of the position of the seed within the fruit (Fig. 3D). This could be explained by the patterns of seed predation by these insects. It is known that females of G. guanacaste lay only one egg per developing seed, where the larva hatches and feeds on the endosperm (Janzen, 1980). The number of seeds damaged per fruit by these beetles normally ranged from 0–10. On the other hand, the pyralid moth larva initially feeds on the developing seed in which the egg was laid, but it later crawls inside the pod, eating several developing seeds within the fruit. In some cases, the larvae can eat all developing seeds in the fruit (Mena-Alí, 2003).

Bawa and Buckley (1989) found the same patterns of seed abortion in several species of dry forest plants, and found that seeds with a higher probability of abortion are located away from the point of entry of pollen tubes into the ovary. This pattern seems to be fairly common in legumes. Bawa and Webb (1984) found similar patterns in Caesalpinia eriostachys, Pterocarpus rohrii, Dalbergia retusa and Myrospermum frutescens. In these species, there is variation in the proportion of seed abortion within the fruit, and the pattern of seed abortion might be the consequence of selective mechanisms to increase progeny vigour (gametophytic selection) or to increase the dispersal ability of winged fruits.

In summary, the pattern of seed maturation within the fruits of Bahuinia ungulata is not independent of ovule position, i.e. ovules closer to the style have a higher probability of setting seed than ovules closer to the peduncle. The patterns of seed abortion and seed set described in this study suggest that microgametophytic competition might contribute to the mechanisms that ultimately determine the production of offspring in B. ungulata. If this hypothesis is true, it is expected that relaxed competition among ovules within a pod would lead to a less-fit offspring (Rocha and Stephenson, 1991b).