Printer Friendly

Number of seeds per pod in three species of perennial legumes: a compromise between ecological and physiological constraints.

Abstract. Most pods (70%) of Cercidium floridum (Leguminosae) contain only one seed even though ovaries contain 8.2 [+ or -] 0.18 (mean [+ or -] SE) ovules. We asked if: (1) Few seeds per pod resulted from balancing selection between costs of pod biomass and seed predation; (2) insect pollinators deposit too few pollen grains to pollinate all ovules within an ovary; (3) limitation of water and nutrient resources prevents maturing of all seeds within a pod.

We compared the pattern of seed packaging in Cercidium floridum with packaging patterns in similar species (C. microphyllum and Olneya tesota) where seed predation does not vary with the number of seeds per pod but where costs of pod biomass remain. The frequency distribution of pods by seed number did not differ among the three species; thus we rejected hypothesis (1). Flowers did not produce significantly more seeds per pod when pollen loads were augmented by, hand compared to insect-only pollinated controls; thus hypothesis (2) was also rejected. In a drought year, C. microphyllum plants with greater reproductive output at a site that received water run-off did not produce more seeds per pod compared to plants with lower reproductive output at a dry site. Instead, greater reproductive output was achieved by increasing the density of pods. We believe there may be energy and anatomical constraints to producing too many seeds at one site (pod) at the same time. Thus, we suggest that resource limitation has been an important ultimate factor influencing the number of seeds produced per pod in these species. However, alternative hypotheses are discussed.


The number of seeds in a fruit like a legume can vary widely among plant species (e.g., Janzen, 1975), among individual plants within a species (e.g., Janzen, 1975; Stephenson, 1984), and even among fruits within a plant (Mitchell, 1977). Greater insights into factors and processes affecting seed production might be achieved with studies that incorporate variation at all these levels. For instance, to avoid idiosyncratic hypotheses associated with studies of single species, several species ought to be compared (Harvey and Pagel, 1991). Thus, it is important that variation at all of these levels is examined by evolutionary ecologists concerned with factors that determine the number of offspring produced by an organism during a reproductive bout (Willson, 1983; Lee, 1988; Lessells, 1991).

Intra- and interindividual variation in seed production has been observed in the desert palo verde tree Cercidium floridum (Leguminosae) (Mitchell, 1977). In California populations, C. floridum produced 71% one-seeded, 23% two-seeded, 4% three-seeded, 1% four-seeded and 0.5% five-seeded fruits. As many as eight seeds per pod were observed. Biomass of pod material per seed increases with fewer seeds in a legume pod (Janzen, 1975), as was found for C. floridum (Mitchell, 1977). Thus, if resources for reproduction are limited, the added cost of pod biomass in pods with few seeds should select for increased frequency of multiseeded pods.

Seed predation, however, may select for increased frequency of one-seeded pods in C. floridum (Mitchell, 1977). The bruchid beetle Mimosestes amicus lays eggs on the indehiscent pods of Cercidium floridum, larvae burrow into a seed, and after successful development an emerging adult cuts an exit through the pod. A second bruchid, Stator limbatus, only lays eggs on seed coats and can gain access to seeds only through the M. amicus exit holes. Once inside a pod, all the seeds in a pod are then accessible to S. limbatus because there is enough space in the pod for a female to crawl over and among seeds. Thus, destruction by M. amicus of one seed in a pod with many seeds puts all the remaining seeds at risk to the second bruchid, S. limbatus. One-seeded pods of C. floridum were much better protected from S. limbatus than multiseeded pods because all seeds, regardless of the number of seeds per pod, are attacked with the same frequency by M. amicus (Mitchell, 1977). Mitchell (1977) estimated a 48% increase in seeds surviving S. limbatus if all seeds were packaged in one-seeded pods.

Seeds of another palo verde species, Cercidium microphyllum, are attacked by Stator limbatus and Mimosestes amicus in nearly the same way (Siemens et al., 1992). One major difference is that pods of C. microphyllum are narrow and closed between seeds so that S. limbatus cannot travel between seeds in multiseeded pods (Fig. 1). Stator limbatus is small enough that larvae can feed on the tissue left in a seed that has already been fed upon by M. amicus. Thus, to S. limbatus all seeds of C. microphyllum are effectively in one-seeded pods. There is then no selection by S. limbatus on C. microphyllum to increase the frequency of one-seeded pods. In addition, most mature pods of C. microphyllum abscise and drop from canopies before M. amicus emerge and lay more eggs, presumably as a defense (Siemens et al., 1992), which further limits accessibility of seeds to S. limbatus because seeds on the ground are not attacked.

Desert ironwood (Olneya tesota; Leguminosae) occurs sympatrically with the two palo verde species in the Southwest (Elias, 1980). Ironwood trees are not attacked by any species of bruchid beetles, although in some populations the bruchid Stator pruininus has been reared from a few seeds of some trees (C.D. Johnson, pers. observ.; Johnson and Kingsolver, 1976). Without attack, there is no selection by bruchids to reduce the number of seeds per pod. The cost in pod biomass remains and therefore we expect ironwood to produce a higher frequency of multiseeded pods compared to Cercidium floridum. The palo verdes are in the subfamily Caesalpinioideae, and desert ironwood is in the subfamily Papilionoideae. Thus, any similarities in seed packaging between ironwood and C. floridum would probably not be the result of close phylogenetic relationships, as might be the case in comparisons between the palo verde species.

The objective of this study was to test the hypothesis that few seeds per pod is the result of balancing selection between the cost of pod biomass and seed predation. Based on varying patterns of seed predation among species and the increase in pod biomass per seed with fewer seeds per pod, we expected frequencies of multiseeded pods to be much greater for Cercidium microphyllum and Olneya tesota compared to C floridum. We also assessed two alternative hypotheses for high frequencies of pods with few seeds; that insect pollinators deposit too few pollen grains to pollinate all ovules within an ovary, and that the limitation of water and nutrient resources prevents trees from maturing a complete complement of seeds within a pod.


Seed production. -- The study site was a 500 [m.sup.2] area located in Paradise Valley 19 km S of Carefree, Maricopa County, Arizona. The predominant vegetation consisted of the palo verdes and ironwood (site details in Siemens and Johnson, 1990). To determine the frequency of one- and multiseeded pods produced by trees of Cercidium floridum, C microphyllum and Olneya tesota, we systematically sampled pods from plants of each species in late June of 1986. Sampling was two-stage, simple random sampling (Cochran, 1977). We first randomly selected six plants of each species within the area (trees with fruit were tagged and numbered, then selected with random numbers from a table). Four main branches on each plant with fruit were similarly randomly selected and all pods (ca. 200) on each branch were collected. Because the number of seeds per pod varies more within plants (among pods) than between plants (Mitchell, 1977), our sampling effort was focused within plants. We counted the number of seeds in each pod. To determine if pod biomass per seed decreased with increasing numbers of seeds per pod in C. microphyllum, as reported for C. floridum and other species of legumes, we compared pod biomass of 21 one-seeded pods with 21 two-seeded pods of C. microphyllum.

Variation in seed number was analyzed with regression analysis (SYSTAT; Wilkinson, 1990). The frequency data were arcsine transformed to satisfy assumptions of normality (Zar, 1984). The regression model contained two factors, plant species (a categorical variable) and number of seeds per pod. The best model contained higher order terms of seeds per pod. Residuals were examined for homoscedasticity.

Bruchid egg production. -- We counted the number of Mimosestes eggs on 65 pods per tree to determine if egg numbers per seed varied with the number of seeds per pod. Because M. amicus emerged after our collections, too few M. amicus emergence holes were observed to count eggs laid by Stator limbatus. The early collection was necessary because pods of Cercidium microphyllum abscise and drop from canopies in early July, before most M. amicus emerge from seeds (Siemens et al., 1992). The number of M. amicus eggs on pods represents the potential risk of attack by S. limbatus. For each palo verde species, the effect of seeds per pod on egg production was analyzed with one-way ANOVA.

Ovule production. -- To determine if seed production was limited by the number of ovules per ovary, we counted ovule number in ovaries of 52 flowers of Cercidium floridum and 21 flowers of C. microphyllum. Flowers were collected during peak flowering time from six plants of each species at or near the study site where pods were collected. We collected 3-10 flowers from each plant. Flowers were preserved in 70% ethanol and dissected under magnification in the laboratory.

Hand-pollinations. -- To determine if pollination was limiting seed production in the palo verdes, we augmented natural pollen loads by hand-pollinating 150 flowers from each species. After hand-pollination, flowers were left open to natural insect pollinators. We compared seed production in the hand-pollinated flowers (that were also naturally pollinated) to control flowers that only received pollen from natural pollinators. If pollination limited seed production, then flowers with augmented pollen loads should produce more seeds per pod. We pollinated 50 flowers on each of three Cercidium floridum plants and 25 flowers on each of six C. microphyllum plants. Cercidium floridum and C. microphyllum are self-incompatible and have protandrous flowers (Jones, 1978); stigmatic surfaces are receptive about a day after anthesis. Pollen was collected from several trees within 100 m. We used pollen from different plants to avoid adding incompatible pollen. Stigmas were covered with an excess of mixed-tree pollen using a glass rod. For each species of palo verde we tested the difference between treatments and controls with the nonparametric two-sample Mann-Whitney U test (MINITAB; Ryan et aL, 1985) because the data did not appear normally distributed.

Resource limitation. -- To determine if water run-off had an effect on the number of seeds per pod, we compared trees from a "dry" site to those at a "wet" site during a drought year (1988). In 1988 there was a marked contrast in foliage and reproductive output between plants next to surfaces such as roads where water run-off occurred and plants at dryer sites. Trees at the edge of Filthy Five Park (FFP) in Black Canyon City, Arizona, caught water run-off from the flat, sloping park grounds and the bordering wash, while trees at a dryer location were found a few kilometers to the S off the Table Mesa Road (TMR) exit on Interstate 17, Ariz. The plants in these wet sites produced many more flowers and pods than plants at dry sites [e.g., at the FFP wet site, the density of racemes (22.4 [+ or -] 2.7 racemes per 30 cm of stem) was 34 times greater than at the TMR dry site (0.58 [+ or -] 0.18), although the number of flower buds per raceme only differed by about a factor of two (Siemens, 1994)]. Thus, the sites differed dramatically in their reproductive output, probably because of water run-off, but nutrients and other factors associated with the different sites may also, have been important. We determined the number of seeds produced per pod from 100 pods collected from each of six plants at the wet site (FFP), and from about five pods from each of six plants at the dry site (TMR). Very few pods were produced by each plant at the dry site.


Seed production. -- Frequencies of pods with different numbers of seeds were remarkably similar for the palo verdes and ironwood (Fig. 2) in spite of the dramatic differences in patterns of attack by bruchid beetles. There were no statistically significant differences in pod frequencies among species (Table 1). No interaction was detected suggesting differences in pod frequencies among species did not change with the number of seeds per pod (Table 1). Most pods (ca. 70%) were one-seeded in Cerridium floridum, C. microphyllum and Olneya tesota at the Paradise Valley site. The frequency of multiseeded pods decreased sharply with increasing seeds per pod in a pattern that was the same FOF each species examined.

Table 1. -- Regression analysis of seed production. The best regression model was y = constant + x + [x.sup.2] + [x.sup.3] + species + species.x + species.[x.sup.2] + species.[x.sup.3] where y is the arcsine square root transformed pod frequencies and x is the number of seeds per pod. [r.sup.2] = 96%

Source df MS F P
Seeds per pod     3   1.970   487.073   0.000
Species           2   0.003     0.809   0.449
Interaction       6   0.005     1.114   0.363
Error            70   0.004

As expected, pod biomass per seed in Cercidium microphyllum also increased with fewer seeds in a pod. The weight of two-seeded pods, 0.100 [+ or -] 0.003 g, was significantly less than twice the weight of one-seeded pods, 0.122 [+ or -] 0.004 g (two-sample t-test, t = 4.64, df = 38, P[is less than] 0.001).

Bruchid egg production. -- The number of eggs per seed laid by Mimosestes amicus on pods of Cercidium floridum and C. microphyllum did not vary significantly with the number of seeds per pod (Table 2). Sample sizes of pods with more than three seeds per pod were too small to be included in the analysis because of inadequate replication.

Table 2. -- Mimosestes egg production as a function of seeds per pod of two species of palo verdes. Values represent the means [+ or -] 1 SE
                                 Number of seeds per pod
Host [species.sup.a]              1                    2

C. [microphyllum.sup.b]   0.33 [+ or -] 0.05   0.35 [+ or -] 0.06
C. [floridum.sup.c]       0.60 [+ or -] 0.09   0.56 [+ or -] 0.91

Host [species.sup.a]              3

C. [microphyllum.sup.b]   0.45 [+ or -] 0.51
C. [floridum.sup.c]       0.54 [+ or -] 0.09

[a] The host 0. tesota was not attacked by bruchids [b] ANOVA: F = 0.59; df = 2,14; P = 0.57 [c] ANOVA: F = 0.08; df = 2,23; P = 0.92

Ovule production. -- Based on the mean number of ovules in an ovary, each pod of Cercidium floridum is capable of producing 8.2 [+ or -] 0.18 (mean [+ or -] SE) seeds. Cercidium microphyllum flowers also have the potential to produce many seeds with a mean of 8.0 [+ or -] 0.13 ovules per ovary. Pods of Olneya tesota can have at least six seeds in them, and similar to the palo verde species, ovaries also contain many ovules (D.H. Siemens, pers. observ.).

Hand-pollinations. -- There were no statistically significant differences in the numbers of seeds per pod between flowers which we augmented pollen loads by hand and insect-pollinated controls as determined by Mann-Whitney U tests (Table 3). Of 150 hand-pollinated flowers for each species, only 23 Cercidium floridum and 20 C. microphyllum flowers produced pods, as occurs naturally (Siemens, 1994). We compared these pods to the same number of insect-pollinated pods sampled haphazardly from the same trees. Despite excess of pollen from different trees, and eight ovules per ovary, hand-pollinated flowers contained only ca. 1-2 seeds as did insect-only pollinated flowers (Table 3). Thus, pollen limitation, as tested in these hand-pollination experiments cannot account for the low numbers of seeds produced in palo verde pods.

Table 3. -- Seed production in hand plus insect and insect-only pollinated flowers for each palo verde species. Each value represents the mean [+ or -] 1 SE
                        C. [floridum.sup.a]  C. [microphyllum.sup.b]

Hand-+insect-pollinated   2.26 [+ or -] 0.24    1.35 [+ or -] 0.11
Insect-pollinated         2.04 [+ or -] 0.28    1.15 [+ or -] 0.08

[a] U = 593, n = 23, P = 0.23 [b] U = 450, n = 20, P = 0.15

Resource limitation. -- There was no statistically significant difference in the mean number of seeds per pod between Cercidium microphyllum trees at the TMR dry site (1.35 [+ or -] 0.08) and the FFP wet site (1.47 [+ or -] 0.17), as determined by a Mann-Whitney U test (U = 40.5, n = 6 trees each site, P = 0.87).


Our main question was whether the number of seeds per pod varied among species in accordance with dramatically different patterns of selection by bruchid beetle seed predators. In the absence of selection by bruchids for one-seeded pods in Cercidium microphyllum and Olneya tesota, the assumption was that costs in pod biomass would select for increased numbers of seeds per pod in these species relative to C. floridum.

We also considered some alternative hypotheses for patterns of seed packaging. Proximate factors that may influence the number of seeds per fruit within a plant species include pollen limitation (e.g., Bierzychudek, 1981; Campbell and Halama, 1993), and resource limitation (Lee and Bazzaz, 1982; Campbell and Halama, 1993). Other hypotheses for low seed:ovule ratios exist (Haig and Westoby, 1988; Lee, 1988), but appropriate data were either not observed in this system (eg., seed abortion) or were not assessed directly (e.g., parent-offspring conflict, predominant male function of some hermaphroditic flowers, excess flowers for pollinator attraction resulting in many fruits with few seeds). If ecologists are to understand what factors are important in the evolution of life history characteristics, studies that consider both ultimate and proximate factors simultaneously, as in the present study, are needed.

We found no differences among plant species in the frequencies of pods with varying numbers of seeds. Clearly, current patterns of seed predation cannot explain the number of seeds produced per pod in Cercidium floridum, C. microphyllum and Olneya tesota. Although seed predators like bruchids have direct effects on host plant fitness, seed packaging, especially in perennials, may be better understood phylogenetically.

The results of this study do not suggest that the production of mostly one-seeded pods by Cercidium floridum is an evolved defense against the bruchid Stator limbatus (Mitchell, 1977). Instead, early abscission of mature pods by C. microphyllum which reduces attack by bruchids, and chemically resistant seed coats on C. floridum may function as alternative seed defense mechanisms against bruchid beetles and other seed predators (Siemens et al., 1992).

Although nutrients and especially water are probably limited for many desert plants, developing pods of these desert legumes are green and probably photosynthesize as they develop. Thus, at least in terms of energy requirements, the green pods may be largely self-sustaining (Bazzaz et al., 1979; Watson and Casper, 1984). For example, Bazzaz et al. (1979) found that photosynthesis by fruits of Acer platanoides contributed 64.5% of photosynthate to fruit development. The cost of pod biomass would be lower if the developing pods were photosynthetic.

Pollen limitation did not explain the high frequency of pods with few seeds. This conclusion is based on hand-pollinations in the field with presumed outcrossed pollen which augmented natural pollen loads. Information on the genetic structure of the palo verde population is needed to eliminate the possibility that pollen used was from closely related individuals. Ovules fertilized with pollen from closely related plants may abort due to deleterious effects of inbreeding (Endler, 1992). Although pollination intensity is usually positively correlated with the number of seeds per fruit (Lee, 1988), other factors such as pollen source or limited resources may be more important.

Hermaphroditic flowers may function primarily as pollen donors (e.g., Sutherland, 1987; Sutherland and Delph, 1984). This might explain the low fruit: flower ratios observed in our experiments. Flowers that have a predominant male function may also produce few seeds per fruit. Thus, excess flowers may function in pollen donation or pollinator attraction (Willson and Price, 1977), which may result in the production of more fruits with fewer seeds per fruit as observed in the two palo verde species.

In a drought year, palo verdes with increased reproductive output at a site that received water run-off did not produce more seeds per pod compared to plants with low reproductive output at a dry site. The additional resources were apparently not allocated to more seeds per pod. Instead, plants with additional resources had increased densities of inflorescences and pods (Siemens, 1994). Spreading resources to mature more pods with fewer seeds per pod may be a mechanism to utilize limited resources more efficiently. There may be anatomical or energetic constraints that prohibit the maturation of many seeds within a pod (Lee, 1988). Similarly, Stephenson (1984) argued that to produce more fruits per plant, it may be more efficient for Lotus corniculatus to produce few fruits per inflorescence and more inflorescences. This explanation is based on clutch theory (Burley, 1980), which was developed to explain the combination of small clutch size, many clutches, and clutch overlap in birds. This effect of resource allocation may overshadow any effects of pollen limitation, but studies are needed which investigate the interaction of these factors (Campbell and Halama, 1993). Producing few seeds per pod as a resource allocation strategy would resolve any parent-offspring conflict within a pod in favor of the offspring (Haig and Westoby, 1988).

Acknowledgments. -- We thank Rodger Mitchell, and anonymous reviewers for many helpful comments on an earlier draft of the manuscript. Partial support was provided by NSF Grants BSR82-11763 and BSR88-05861 to CDJ.

Literature Cited

Bazzaz, F. A., R. W. Camon and J. L. Harper. 1979. Contribution to reproductive effort by photosynthesis of flowers and fruits. Nature, 279:554-555.

Bierzychudek, P. 1981. Pollinator limitation of plant reproductive effort. Am. Nat., 117:838-840.

Burley, N. 1980. Clutch overlap and clutch size: alternative and complementary reproductive tactics. Am. Nat., 115:223-246.

Campbell, D. R. and K. J. Halama. 1993. Resource and pollen limitations to lifetime seed production in a natural plant population. Ecology, 74:1043-1051.

Cochran, W. G. 1977. Sampling techniques. Wiley, New York.

Elias, T. S. 1980. The complete guide to North American trees. Times Mirror Magazines, Inc., New York.

Endler, J. A. 1992. Genetic heterogeneity and ecology. In: R. J. Berry, T J. Crawford and G. M. Hewitt (eds.). Genes in ecology. Blackwell, London.

Haig, D. and M. Westoby. 1988. Inclusive fitness, seed resources, and maternal care, p. 60-79. In:J. Lovett Doust and L. Lovett Doust (eds.). Plant reproductive ecology: patterns and strategies. Oxford Univ. Press.

Harvey, P. H. and M. D. Pagel. 1991. The comparative method in evolutionary biology. Oxford Univ. Press, New York.

Janzen, D. H. 1975. Interactions of seeds and their insect predators/parasitoids in a tropical deciduous forest. In: P. W. Price (ed.). Evolutionary strategies of parasitic insects and mites. Plenum Publishing Corp., New York.

Johnson, C. D. and J. M. Kingsolver. 1976. Systematics of Stator of North and Central America (Coleoptera: Bruchidae). U.S. Dep. Agric. Tech. Bull. 1537.

Jones, C. E. 1978. Pollinator constancy as a pre-pollination isolating mechanism between sympatric species of (Cercidium microphyllum; Fabaceae). Evolution, 32:189-198.

Lee, T. D. 1988. Patterns of fruit and seed production. In: J. Lovett Doust and L. Lovett Doust (eds.). Plant reproductive ecology: patterns and strategies. Oxford Univ. Press.

_____ and F. A. Bazzaz. 1982. Regulation of fruit and seed production in an annual legume, Cassia fasciculata. Ecology, 63:1363-1373.

Lessells, C. M. 1991. The evolution of life histories. In: J. R. Krebs and N. B. Davies (eds.). Behavioral ecology. Blackwell, Oxford, England.

Mitchell, R. 1977. Bruchid beetles and seed packaging by Palo Verdes. Ecology, 58:644-651.

Ryan, B. F., B. L. Joiner and T. A. Ryan, Jr. 1985. Minitab handbook, 2nd ed. Duxbury, Boston.

Siemens, D. H. 1994. Factors affecting regulation of maternal investment in an indeterminate flowering plant (Cercidium microphyllum; Fabaceae). Am. J. Bot., 81:1403-1409.

_____ and C. D. Johnson. 1990. Host-associated differences in fitness within and between populations of a seed beetle (Bruchidae): effects of plant variability. Oecologia, 82:408-413.

_____, _____ and K. J. Ribardo. 1992. Alternative seed defense mechanisms in congeneric plants. Ecology, 73:2152-2166.

Stephenson, A. G. 1984. The regulation of maternal investment in an indeterminate flowering plant (Lotus corniculatus). Ecology, 65:113-121.

Sutherland, S. 1987. Why hermaphroditic plants produce many more flowers than fruits: experimental tests with Agave Mckelveyana. Evolution, 41:750-759.

_____ and L. F. Delph. 1984. On the importance of male fitness in plants: patterns of fruit set. Ecology, 65:1093-1104.

Watson, M. A. and B. B. Casper. 1984. Morphogenic constraints on patterns of carbon distribution in plants. Annu. Rev. Ecol Syst., 15:233-258.

Wilkinson, L. 1990. SYSTAT: the system for statistics. SYSTAT, Inc., Evanston, Ill.

Willson, M. F. 1983. Plant reproductive ecology. Wiley, New York.

_____ and P. W. Price. 1977. The evolution of inflorescence size in Asclepias (Asclepiadaceae). Evolution, 31:495-511.

Zar, J. H. 1984. Biostatistical analysis, 2nd ed. Prentice-Hall, Englewood Cliffs, NJ.
COPYRIGHT 1995 University of Notre Dame, Department of Biological Sciences
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Siemens, David H.; Johnson, Clarence Dan
Publication:The American Midland Naturalist
Date:Apr 1, 1995
Previous Article:Rapid decomposition of summer-input leaves in a northern Michigan stream.
Next Article:Growth response of baldcypress (Taxodium distichum (L.) Rich.) to an altered hydrologic regime.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |