SEARCH

SEARCH BY CITATION

Keywords:

  • age at reproduction;
  • lifetime seed output;
  • plant life span;
  • plant size;
  • seed size

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References
  • 1
    The trade-off between seed mass and the number of seeds a plant can make for a given amount of energy underpins our understanding of seed ecology. However, there is little information on the magnitude of the fecundity advantage of small-seeded species over an entire plant lifetime.
  • 2
    We compiled data from the literature to quantify the relationships between: (i) seed mass and plant size (because the photosynthetic area of a plant determines how much energy is available for allocation to seed production); (ii) seed mass and plant reproductive lifetime (the number of years a plant has to produce seeds); and (iii) seed mass and the number of seeds produced per individual per year, and per lifetime.
  • 3
    Seed mass was positively related to all measures of plant size (canopy area, plant height, stem diameter, plant mass and canopy volume). There were also positive correlations between seed mass and time to first reproduction, plant life span, and reproductive life span. Thus, although small-seeded species produce more seeds per unit canopy area per year than large-seeded species, large-seeded species tend to have larger canopies and more reproductive years.
  • 4
    These patterns accord well with independently gathered data on annual and lifetime seed production. The negative relationship between seed mass and the number of seeds produced per year was much shallower on a per individual basis than on a per unit canopy basis. Seed mass was not significantly related to the total number of seeds produced by an individual plant throughout its lifetime.
  • 5
    Our previous understanding of seed mass as a spectrum from production of many small seeds, each with low establishment probability, to a few large seeds each with higher establishment probability, was missing some important elements. To understand the forces shaping the evolution of seed mass, we will need to consider plant size and longevity, as well as seedling survival rates and the number of seeds that can be produced for a given amount of energy.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References

The trade-off between seed mass and the number of seeds a plant can make for a given amount of energy is central to much theory regarding the ecology of seed size and the coexistence of a wide range of seed size strategies (Venable & Brown 1988; Geritz 1995; Rees & Westoby 1997; Geritz et al. 1999; Leishman 2001). This seed size/number trade-off has been demonstrated for a wide range of species in many different ecosystems (Shipley & Dion 1992; Greene & Johnson 1994; Turnbull et al. 1999; Jakobsson & Eriksson 2000). The available evidence (and logic) suggest that among coexisting species, the number of seeds produced per unit canopy per year decreases isometrically with the species’ mean seed mass (Aarssen & Jordan 2001; Henery & Westoby 2001).

The fact that small-seeded species can produce more seeds per unit canopy area than large-seeded species gives them a clear advantage during seed production. However, species with large and small seeds must be equally capable of replacing themselves, or the wide range of seed-mass strategies commonly observed within plant communities (Harper et al. 1970; Leishman et al. 2000) would not coexist. It has generally been thought that the fecundity advantages of small-seeded species would be balanced by the superior establishment ability of seedlings from large-seeded species. This idea has been supported by abundant evidence showing that seedlings from large-seeded species have higher rates of survival than seedlings from small-seeded species (reviewed in Leishman et al. 2000; Westoby et al. 2002). However, there is some evidence that the advantage of large-seeded species during seedling establishment is not sufficient to counterbalance the greater number of small seeds that can be produced for a given amount of reproductive effort (Coomes & Grubb 2003; Moles & Westoby 2004). So the greater number of seeds produced by small-seeded species (per unit canopy area per year) may be balanced partly by the higher rate of seedling survival associated with large seeds, and partly at some other stage in the life cycle.

The greater number of seeds produced by small-seeded species per square metre of canopy outline per year might be partly offset by the larger canopy areas and longer reproductive life spans associated with large-seeded species (Levin 1974; Leishman et al. 1995; Rees 1996). However, the potential influence of plant size and life span on the relationship between seed mass and lifetime seed production has not previously been evaluated. Therefore, the present study quantified relationships between seed size and plant size, plant life span, annual seed production and lifetime seed production, in order to assess the overall advantage of small-seeded species during seed production.

seed mass and plant size

The area (or mass) of photosynthetic tissue deployed by a plant is an important determinant of the amount of energy available for reproduction (Bazzaz et al. 2000). Niklas & Enquist (2002, 2003) demonstrated a tight correlation between net annual reproductive biomass (including structures other than seeds) and total leaf mass, and several studies have shown that plant size affects the total number (and/or mass) of propagules produced (Shipley & Dion 1992; Greene & Johnson 1994; Jakobsson & Eriksson 2000; Aarssen & Jordan 2001; Rodríguez-Gironés et al. 2003). If big plants produce a greater total mass of seeds than small plants, then a relationship between plant size and seed mass would influence the relationship between individual seed mass and lifetime seed output.

Most previous data on the relationship between seed mass and plant size are about seed mass and plant height or growth form. Levin (1974) showed that seed mass decreased from trees to shrubby trees to shrubs to herbs, across 1059 angiosperm species, and across 263 leguminous species from around the world. Leishman et al. (1995) found increases in mean seed mass with plant height across species in five floras: semiarid woodlands in western New South Wales, Australia; arid woodlands in central Australia; mixed vegetation in Sydney, Australia; the Indiana Dunes, USA; and Sheffield, UK. The Sheffield data were analysed in more detail by Rees (1996), who found that seed mass was significantly related to adult height across species with wind-dispersed seeds and across species with no specialized dispersal mechanism, but not across animal-dispersed species. Grime et al. (1997) found no significant correlation between seed mass and plant height across 43 common British species, while Thompson & Rabinowitz (1989) found significant relationships between seed mass and plant height within some plant families around Sheffield.

Positive relationships between seed mass and plant height, combined with the fact that taller plants generally have larger canopy areas (both within species, White 1981, and across species, Rees 1996) are likely to lead to a positive correlation between seed mass and canopy area. However, the only quantitative evidence known to us for the relationship between species’ seed mass and canopy area is that of Rees (1996), who showed a positive relationship between seed mass and lateral spread across 169 species from the Sheffield area. Further, although plant size is one of the strongest correlates of seed mass (Leishman et al. 1995), little is known about the slope of the relationship between seed mass and plant size, or even its shape. Our aim was to determine the slope and shape of this correlation, in order to quantify the extent to which a positive relationship between seed mass and canopy area might affect the relationship between seed mass and the number of seeds produced per plant.

seed mass and plant life span

There is some evidence that large seeds are associated with long plant life spans. Rees (1996) found a significant relationship between seed mass and adult life span across species with abiotic seed dispersal, but not across species with animal dispersal, nor across all species combined. Leishman et al. (1995) found that annual species had smaller seeds than perennial species across 248 species from western New South Wales, Australia, 196 species from central Australia and 266 species from Sheffield, UK, but not across 641 species from the Indiana Dunes, USA. Baker (1972) showed that the average seed mass of annual herbs was smaller than the average seed mass of perennial herbs, across 1604 species from California. Similarly, Silvertown (1981) showed that the average seed mass of annuals was smaller than that of biennials and perennials across 54 species from calcareous grassland in Britain. However, some studies have shown no relationship between seed mass and plant life span (e.g. across 229 species in the Australian arid zone, Jurado et al. 1991, or between 82 species of annuals and 177 species of perennials from herbaceous vegetation in Britain, & Thompson 1984).

Here, we aimed to gather information on the relationships between seed mass and total plant life span, time to first reproduction and plant reproductive lifetime across as many species as possible from around the world, in order to assess how much a relationship between seed mass and plant life span might affect the overall relationship between seed mass and lifetime seed output.

seed mass and seed output

Several studies have quantified the relationship between seed mass and annual seed production. Shipley & Dion (1992) found a strong negative relationship between seed mass and the number of seeds produced in one year across 57 herbaceous angiosperms from North America. Greene & Johnson (1994) found a strong negative relationship between seed mass and annual seed production per plant across 17 species of canopy trees from North America. Turnbull et al. (1999) found a strong negative relationship between seed mass and annual seed production per plant across seven species of grassland annuals in South Wales, and Jakobsson & Eriksson (2000) found a negative relationship between seed mass and seed output per plant across 72 grassland species in Sweden. However, few data have previously been compiled on the relationship between seed mass and lifetime seed production. The only cross-species study we are aware of is Turnbull et al. (1999), and that included only seven species of annuals.

We aimed to quantify the relationship between seed mass and the total number of seeds produced per plant (genet) per year, and per lifetime. We also compiled additional data on the relationship between seed mass and the number of seeds produced per square metre of canopy per year, in order to assess the generality of previous findings (Aarssen & Jordan 2001; Henery & Westoby 2001).

within-habitat and within-taxa relationships

Our major questions are about the slopes of the relationships between seed mass and plant size, plant life span and seed production. All of these slopes are predicted to remain constant across different habitats and different taxa (though the elevation of the relationship between seed mass and seed production might shift as a result of differences in photosynthate acquired per square metre in different habitats). However, we investigated the possibility that cross-habitat relationships might behave differently to relationships within habitats. Patterns of significant interest were also investigated within major taxa, to determine whether similar relationships have emerged within multiple evolutionary lineages and to exclude the possibility that cross-species relationships were obscured, or artificially created, through aggregation of within-taxa relationships. For example, if the range of seed masses differed among taxa, and there were shifts in elevation between different taxa, several strong within-taxa relationships could combine to appear as no relationship (Fig. 1a), or even as a positive relationship across species (Fig. 1b).

image

Figure 1. Schematic diagram illustrating the confounding that could result from juxtaposition of data from different taxa (or habitats). If differences in intercept between taxa are associated with shifts in seed mass, it is possible for strong negative relationships within taxa to appear as a lack of relationship across all species (Fig. 1a), or even as a positive relationship (Fig. 1b).

Download figure to PowerPoint

In summary, the questions addressed in this paper are:

  • 1
    What is the slope (and shape) of the relationship between seed mass and plant size?
  • 2
    What is the slope of the relationship between seed mass and (a) time to first reproduction, (b) mean and maximum life span, and (c) reproductive life span of plants?
  • 3
    What is the slope of the relationship between seed mass and the number of seeds made (a) per m2 of canopy outline, per year, (b) per whole individual per year, and (c) per whole individual per lifetime?
  • 4
    Are there differences in the relationships described above: (a) within habitats, and (b) within taxa?

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References

We used three Biosis searches to compile data from the literature. The first search used the terms ‘plant size’ or ‘canopy area’. The second search used the terms (fecundity and seed), ‘seed output’ or ‘seed production’, or (‘reproductive output’ and seed). The third search used the terms ‘longevity’ or ‘lifespan’, limited to ‘plants’. All searches were for journal articles published in English before 2002. We tried to include only studies of natural plants growing under natural conditions. Additional terms were used to exclude crops, genetically altered plants, cultivars, invasive species, and species in highly modified or disturbed environments. Studies in which plants were raised in pots, within exclosures, under shelters, with extra watering, with pesticide, supplementary fertilization or weeding, and studies in which plants were grown in glasshouses or transplanted were excluded, as were populations growing at unnatural densities through thinning or seed addition. Parasitic and semiparasitic species were excluded from the compilation on plant size, as plant nutrition depends on processes other than photosynthesis in these species. Plants grown in seed orchards were excluded from the seed output compilation, as these species are managed for high seed output. Studies presenting data for fewer than five individuals per species were also excluded.

We recorded any information given on plant size, seed output, plant life span, or time to first reproduction, whether means or maxima. Means and maxima each have limitations. Maxima are vulnerable to outliers, and may give non-representative results. Means are in many ways more appropriate to our questions. However, fewer means were available than maxima, especially for plant height. It was not always possible to tell whether published studies were reporting mean values for reproductively mature individuals only, or for all individuals of a species, but where possible, means of plant size, life span and seed output were limited to individuals that had reached reproductive maturity.

Annual and lifetime seed output were not calculated by multiplying seed production per unit of canopy by canopy area and/or reproductive life span for any species. That is, data on annual and lifetime seed production are completely independent of data on plant size and reproductive longevity.

Seed mass data were taken from the same papers as the size, life span or seed output data wherever possible, or compiled from other published sources.

A full list of species, data and data sources is provided in Appendix S1 of the Supplementary Material.

within-habitat and within-taxa relationships

Selected relationships were investigated within habitats and within the nine best-represented taxa (Amaranthaceae, Asteraceae, Brassicaceae, Fabaceae, Gymnosperms, Monocotyledons, Myrtaceae, Proteaceae, Rosaceae). Habitat types were assigned based on the vegetation classification scheme provided by Udvardy (1975). Where data for one species were available from more than one habitat, the species was assigned to the category ‘other’.

data analyses

Where the direction of causation between x and y variables was unclear, relationships were characterized using model II analyses (also known as standardized major axis analyses; McArdle 1988; Sokal & Rohlf 1995) on log scaled variables. In these cases, model II techniques provide a more appropriate estimate of the line summarizing the relationship between two variables than model I regression, because residual variance is minimized in both x and y dimensions, rather than the y dimension only (McArdle 1988; Sokal & Rohlf 1995). Where we wished to predict the value of one variable from another (i.e. one variable was clearly the x variable), model I regression was employed. Thus, we used model II analyses to describe the relationships between seed mass and plant size, and between seed mass and measures of plant longevity (neither plant longevity nor plant size are dependent on seed mass, though the traits are correlated). We used model I regression to describe the relationships between seed mass and seed output, because the number of seeds produced is dependent on seed mass. We also used model I slopes to investigate the extent to which the canopy area and reproductive life span associated with a given seed mass might influence annual and lifetime seed output.

Model I statistics were calculated using SPSS v11.0. Model II analyses were performed in (S)MATR (Falster et al. 2003), which provides for the equivalent of analysis of covariance. The program first fits slopes across species within each group, with confidence intervals calculated following Pitman (1939). Then it tests for statistical differences in slope between groups, using methods outlined by Warton & Weber (2002). Where within-group slopes were not significantly heterogeneous, shifts in elevation were of interest. Data were transformed to a common slope of 0: y′ = y − βx and x′ = y + βx, where β was the common slope of the untransformed data. Group means of y′ and x′ were then tested for significant differences.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References

seed mass and plant size

There were positive relationships between seed mass and all measures of plant size (canopy area, canopy volume, mean and maximum plant height, dry mass and stem diameter; Table 1, Fig. 2). Scatter was moderate in each relationship (R2 < 0.5 in most cases). However, the data clouds were mostly elliptical. This indicates a linear relationship between seed mass and plant size (small plants make small seeds and large plants make large seeds) rather than a triangular relationship in which just one corner of the space is unoccupied (the pattern that would be seen if small plants only made small seeds, but large plants could make either large or small seeds).

Table 1.  Results of model II analyses of relationships between seed mass and aspects of plant size, plant life span and seed production. All variables were log10 transformed prior to analysis. Lower and upper b represent 95% confidence intervals for the model II (standardized major axis) slopes
 Number of speciesPR2Model II slopeLower bUpper b
Plant size
 Maximum height (m)2113< 0.0010.35 0.59 0.57 0.61
 Mean height (m) 129< 0.0010.33 0.67 0.58 0.77
 Mean canopy volume (m3)  12< 0.0010.80 2.33 1.71 3.19
 Mean plant mass (mg) 224< 0.0010.43 1.40 1.27 1.55
 Maximum plant mass (mg)  49< 0.0010.22 1.63 1.26 2.11
 Maximum stem diameter (mm) 179< 0.0010.06 0.42 0.36 0.48
 Mean stem diameter (mm)  64   0.0990.04 0.62 0.48 0.79
 Maximum canopy area (m2)  73< 0.0010.37 0.99 0.82 1.20
 Mean canopy area (m2) 153< 0.0010.63 1.41 1.28 1.56
Plant life span
 Mean life span (year)  67< 0.0010.18 0.60 0.48 0.75
 Maximum life span (year) 616< 0.0010.19 0.62 0.58 0.66
 Minimum juvenile period (year) 253< 0.0010.26 0.34 0.31 0.38
 Maximum reproductive life span (year) 210< 0.0010.19 0.49 0.44 0.56
Seed production
 Maximum annual seed production (no of seeds adult plant−1 year−1) 122< 0.0010.21−1.23−1.45−1.05
 Mean annual seed production (no of seeds adult plant−1 year−1) 395< 0.0010.08−1.07−1.18−0.97
 Maximum lifetime seed production (no of seeds adult plant−1)  25   0.3860.03−0.94−1.42−0.62
 Mean lifetime seed production (no of seeds adult plant−1) 135   0.77140.00 1.07 0.90 1.27
 Mean annual mass of seeds produced (mass of seeds adult plant−1 year−1; mg) 182< 0.0010.29 1.42 1.25 1.60
 Mean lifetime mass of seeds produced (mass of seeds adult plant−1; mg) 122< 0.0010.32 1.35 1.16 1.57
 Maximum annual mass of seeds produced (mass of seeds adult plant−1 year−1; mg)  53< 0.0010.15 1.51 1.17 1.95
 Mean seeds per m2 of canopy per year (no of seeds m−2 year−1 on adult plants)  60< 0.0010.55−1.24−1.47−1.04
image

Figure 2. Relationships between seed mass and (a) maximum plant height, (b) mean plant height, (c) maximum and mean canopy area, (d) maximum and mean canopy volume, (e) maximum and mean plant dry mass, and (f) maximum and mean stem diameter. Slopes and statistics represent the results of model II regressions (details in Table 1). Mean values represent means for reproductively mature individuals. Each point represents one species. White circles and solid lines represent relationships across means, black circles and dashed lines represent relationships across maxima.

Download figure to PowerPoint

The model II slope of the relationship between log10 seed mass (x) and log10 maximum canopy area (y) was 0.99 (95% CI 0.82–1.20, Fig. 2c), while the model II slope of the relationship between log10 seed mass and log10 mean canopy area was 1.41 (95% CI 1.28–1.56, Fig. 2c; significantly steeper than with maximum canopy area, P < 0.001). That is, a 10-fold increase in seed mass was associated with a 9.8-fold increase in maximum canopy area, but with a 26-fold increase in mean canopy area.

The relationship between seed mass and stem diameter was much weaker than the other relationships. This was surprising, especially in light of a strong positive relationship between maximum plant height (y) and maximum stem diameter (x, P < 0.001, model II slope = 0.84 (95% confidence intervals: 0.78–0.92), R2 = 0.80, n = 117 species). We investigated the possibility that variation in the height at which stem diameter was measured (typically breast height for trees, and near ground level for herbs) weakened cross-species relationships with seed mass. The slope of the relationship between seed mass and stem diameter (mean and maximum) differed significantly (P < 0.01) between species measured at breast height and species measured at the base. The relationship across species measured at the base appeared similar to that seen with other measures of plant size, while the relationship across species measured at breast height was surprisingly flat.

seed mass and plant life span

There were positive relationships between seed mass and plant life span (mean and maximum), time to first reproduction and maximum reproductive life span (Table 1, Fig. 3). That is, plants with big seeds tended to take longer to reach reproductive maturity, but enjoyed longer reproductive life spans. There was also a strong positive relationship between maximum plant life span (y) and maximum plant height (x, P < 0.001, model II slope = 0.96 (95% confidence intervals 0.89–1.03), R2 = 0.60, n = 283 species).

image

Figure 3. Relationships between seed mass and (a) maximum plant life span, (b) mean plant life span, (c) maximum reproductive life span, and (d) minimum duration of the juvenile period. Slopes and statistics represent the results of model II regressions (details in Table 1). All variables are on log10-axes. Mean life span values represent means for reproductively mature individuals. Maximum reproductive life span is the difference between the maximum life span and the minimum duration of the juvenile period. Each point represents one species. White circles and solid lines represent relationships across means, black circles and dashed lines represent relationships across maxima.

Download figure to PowerPoint

Some life span data from Benson and McDougal's compilations (references in Appendix 1) were estimates made by experienced field biologists. It seemed possible that these less accurate age estimates might influence our results. However, the relationship between seed mass and maximum life span excluding data from Benson and McDougal (slope = 0.57, n = 380 species) was not significantly different (P = 0.10) from the relationship across the full data set (slope = 0.62, n = 616 species).

seed mass and seed production

Henery & Westoby (2001) found a relationship between seed mass and seed production per unit canopy area across 47 species from sclerophyll shrubland in Sydney with a model I slope (on log-log axes) not significantly different from −1 (Fig. 4a, white points). Here, we added data for 13 new species, from a variety of environments around the world and of a range of growth forms (Fig. 4a, black points). The new data points were closely aligned with the old. Only two species (Senecio jacobaea (high) and Protea lepidocarpodendron (low)) lay apart from the previously compiled data. The model I slope was −0.92 (95% CI: −1.14 to −0.71, R2 = 0.55), not significantly different from −1.

image

Figure 4. (a) The seed size/number trade-off. White points represent data from Henery & Westoby (2001), grey points represent data from other studies. (b) Seed mass vs. annual seed production (per adult plant). (c) Seed mass vs. lifetime seed production (per adult plant). (d) Seed mass vs. total mass of seeds produced per year, and per lifetime. Each point represents one species. All slopes were fitted using model I regression. In 4b,c,d, white circles and solid lines represent relationships across means, black circles represent relationships across maxima. In 4a, grey circles and line represent mean lifetime data.

Download figure to PowerPoint

There was a significant negative relationship between seed mass and the number of seeds made per adult plant per year (mean and maximum; Table 1, Fig. 4b).

A crucial finding was that there was no significant relationship between seed mass and the total number of seeds produced per adult plant per lifetime (Fig. 4c). This was the case across both mean data (P = 0.77) and maximum data (P = 0.39).

The model I slope of the relationship between seed mass and the number of seeds produced per square metre of canopy outline per year was −0.92. This slope had slackened to −0.31 when taken per individual and per year (incorporating the effect of canopy area), and was not significantly different from 0 when taken on a per individual per lifetime basis (incorporating the effects of both canopy area and reproductive life span, though recall that seed output data were collected independently of canopy area and reproductive life span data). We therefore asked whether shifts in the relationship between seed mass and seed output between Fig. 4(a,c) were consistent with relationships between seed mass and canopy area, and seed mass and reproductive lifetime.

Combining the slope of the relationship between seed mass and the annual number of seeds produced per square metre of canopy (model I slope = −0.92) with the slope of the relationship between seed mass and canopy area (model I slope = 1.12) actually leads to the prediction of a slight positive relationship between seed mass and the number of seeds produced per individual plant per year, not the negative (−0.31) slope actually observed. However, annual seed output per individual would only be expected to be a product of the number of seeds made per canopy area per year and canopy area if the percentage of a plant's energy devoted to reproduction remained constant across the range of canopy areas (Harper et al. 1970). Thus, the negative relationship between plant size and reproductive allocation (Bazzaz et al. 2000) might explain the disjunction between the observed negative relationship between seed mass and annual seed output, and the positive relationship predicted above.

If both large- and small-seeded species make approximately equal numbers of seeds in their lifetimes, large-seeded species must produce a greater total mass of seeds. This relationship is illustrated in Fig. 4(d). However, the statistics associated with this regression should be treated with care. For most species the average seed mass (x) was multiplied by average seed output to obtain the average total mass of seed produced per individual (y), thus violating the assumption of independence of variables. However, the positive relationships between seed mass and plant size, and individual seed mass and total mass of seeds produced, found here were consistent with strong positive relationships between plant size and total mass of reproductive structures reported by Niklas (1993) and Niklas & Enquist (2003).

within-habitat and within-taxa relationships

Almost all major relationships within habitats and taxa were broadly similar to cross-species relationships (Figs 5 and 6; Appendix S2 in Supplementary Material). Although there were some statistically significant differences between slopes, the actual magnitude of the difference was small in most cases. Only one relationship was dramatically different on a within-taxa or within-habitat basis. The positive relationship between seed mass and plant height found across species and within all angiosperm groups was completely reversed across 33 gymnosperm species (Fig. 5a; see Appendix S3 for discussion of this result).

image

Figure 5. Within-taxa relationships between seed mass and (a) maximum plant height, (b) mean adult canopy area, (c) mean annual seed production, (d) maximum life span, and (e) mean lifetime seed production. Analyses were performed using model II techniques. Analysis of covariance was used to test for heterogeneity of slopes between taxa. Only relationships significant at P = 0.01, that included > 6 species were plotted. Species from poorly represented families were omitted. Mean values represent averages across reproductively mature individuals. Each point represents one species.

Download figure to PowerPoint

image

Figure 6. Within-habitat relationships between seed mass and (a) maximum plant height, (b) mean adult canopy area, (c) mean annual seed production, (d) maximum life span, and (e) mean lifetime seed production. Analyses and presentation as for Fig. 5. Species that could not be assigned to one of the major habitat groups were omitted from these plots.

Download figure to PowerPoint

There was no significant relationship between lifetime reproductive output and seed mass within any habitat or any taxon. This excludes the possibility that relationships within habitats or taxa were obscured when species from different habitats or taxa were pooled.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References

The major finding from this compilation was that although small-seeded species made more seeds per square metre of canopy outline per year (Fig. 4a), large-seeded species tended to have larger canopies (Fig. 2c), and longer reproductive life spans (Fig. 3c). Consequently, there was no overall relationship between seed mass and the total number of seeds made by an individual throughout its life (Fig. 4c). This finding shifts our perspective on the ecology of seed size. Most previous work has emphasized that small-seeded species can produce more offspring than large-seeded species (reviewed in Leishman et al. 2000). This remains true among individuals having similar size and life histories. But questions about the consequences of shifts in seed mass that do not consider coordinated shifts in other life-history traits are only seeing part of the picture.

The lack of a relationship between seed mass and lifetime seed output is clearly of importance to theory on the evolution of seed mass (Geritz 1995; Rees & Westoby 1997). However, this finding does not invalidate the seed size/number trade-off hypothesis, even at multispecies level, because this theory assumes that small-seeded species produce a greater number of propagules per patch occupied per year, rather than per individual per lifetime.

why is there a positive relationship between seed mass and plant size?

Although many studies have found positive relationships between seed mass and plant size (Levin 1974; Thompson & Rabinowitz 1989; Leishman et al. 1995; Rees 1996), the slope and shape of this relationship had not previously been quantified. This lack of information made it difficult to assess the likely importance of the many proposed mechanisms through which plant size and seed mass might be coordinated. Here, we attempt such an assessment.

Corner (1949) suggested that a positive correlation between seed mass and plant size might arise because simple hydraulic and mechanical constraints impose an upper limit on the mass of seed that could be supported by a given size of plant. Logically, this must be true. However, the actual sizes of seeds may not be close enough to this theoretical limit for physical constraints to play a major role in shaping the correlation between seed mass and plant size (Thompson & Rabinowitz 1989). Further, if seeds are borne on terminal branches, the diameter of branches supporting seeds might not be different between tall trees and small shrubs. In any case, the ‘physical constraint’ idea cannot explain the scarcity of large plants that produce really small seeds. Thus, although the positive relationship between seed mass and plant size observed in this study does not preclude the operation of this mechanism, it does indicate that physical constraints are not the only factor shaping this relationship.

The idea that plant height and seed mass might be constrained by the inability of small plants to effectively disperse large seeds (Thompson & Rabinowitz 1989) does not explain why large plants do not produce very small seeds either. Further, Leishman & Westoby (1994) found that the association between plant size and seed size was strongest when species with unassisted dispersal and wind-dispersed seeds were excluded from analyses. Seed dispersal therefore seems unlikely to be an important determinant of the relationship between plant size and seed mass.

Seed mass and plant height might be coordinated across habitats by light environment (Salisbury 1974; Rees 1996), because competition for light increases plant height (Givnish 1982; Iwasa et al. 1985; King 1990; Falster & Westoby 2003) and shaded environments favour species with larger seeds (Salisbury 1974; Thompson & Hodkinson 1998). The successional stage of habitats might act similarly (Salisbury 1974). Early successional sites tend to be dominated by small plants (perhaps because small plants take less time to reach maturity; Fig. 7) with relatively small seeds (which might be associated with a higher colonizing ability; Turnbull et al. 1999). Both of these hypotheses are intuitively appealing. However, the relationship between seed mass and plant height operates within habitats similarly to across habitats (Fig. 6a), suggesting that differences in habitat are not the major factor shaping the correlation of seed mass and plant height either.

image

Figure 7. Minimum time to reproductive maturity vs. maximum plant height.

Download figure to PowerPoint

Under an extension of the Smith-Fretwell model developed by Venable (1992), optimal seed mass is predicted to increase with increasing maternal resource status (and therefore with plant size) if there are diminishing returns on producing larger numbers of seedlings. Returns might diminish because of density-dependent predation or pathogen attack, seedling–seedling competition, or simply as a result of an increased number of offspring per available adult site. These circumstances can occur in natural vegetation (Augspurger & Kitajima 1992; Hulme & Borelli 1999; Webb & Peart 1999), and the available data fit the predictions of Venable's theory better than they fit the predictions of theories based on physical constraints, seed dispersal, habitat type or successional status. However, under Venable's model, optimal seed mass should increase with plant size within species just as strongly as across species. This does not seem to be the case. There is some evidence for positive relationships between seed mass and plant size within species (Venable 1992). However, within-species relationships between seed mass and plant size are often non-significant, and even significant relationships generally have slopes much shallower than the cross-species slope (Appendix S4). This difference in slope suggests that a different mechanism operates at within- vs. cross-species levels.

Explanations for the relationship between seed mass and plant longevity are similarly unconvincing. It has been suggested that the positive relationship between seed mass and plant life span might arise through constraint. Large seeds take longer to complete development than small seeds (Moles & Westoby 2003), so very short-lived plants may be unable to produce very large seeds. However, this cannot explain the absence of long-lived small-seeded species (Fig. 3a). Further, most plants live much longer than the time required to develop even the largest seeds. Thus, although this constraint might place an upper limit on the seed mass attainable for a given longevity, it is unlikely to be a major determinant of the relationship observed.

In summary, none of the above theories provides an entirely satisfactory explanation of the observed relationships between seed mass and plant size or plant longevity. Here, we consider a further possibility: that the correlations arise through evolutionary co-ordination among life-history traits.

evolutionary coordination among life-history traits

This study has shown a positive relationship between seed mass and the time taken to reach reproductive maturity (Fig. 3d). This relationship might be partly driven by a positive relationship between maximum plant size and the time taken for seedlings to reach maturity (P < 0.001, R2 = 0.46; model II slope = 0.60; n = 122 species; Fig. 7). That is, plants generally require a long juvenile period to become large adults. Further, it is well known that there is a positive correlation between seed mass and seedling survival through a given amount of time (Bakker 1989; Eriksson 1997; Dalling & Hubbell 2002; Moles & Westoby 2004). Although the higher survival rates of large-seeded species are likely to be transitory (Leishman et al. 2000), the slower initial decline of populations of seedlings of large-seeded species is expected to have a lasting effect on the proportion of seedlings remaining alive after a given amount of time. Thus, species require a long juvenile period to become large adults, and to survive a long juvenile period requires high juvenile survivorship, which is associated with large seeds. Large plants would not be able to make small seeds, because seedling survivorship would be too low for sufficient seedlings to reach maturity.

The correlation between plant size and seed mass (Fig. 2) is similar to the relationship between adult body size and offspring mass at independence observed in mammals (Charnov 1993). Further, the mechanism outlined above strongly parallels Charnov's (1993) life-history theory for mammals. Charnov's theory predicts that larger adult body sizes will be associated with larger offspring mass at weaning because the long period of growth required for large adults to develop exposes individuals to a long period of juvenile mortality. This disadvantage can be offset either by decreasing the total amount of growth required to reach adult size, or by increasing juvenile survival rates. Both of these outcomes are achieved by increasing offspring mass at independence (Charnov 1993). Charnov's theory also predicts a positive relationship between offspring mass and reproductive life span, because the reduced birth rate associated with production of large offspring is offset by the lower replacement rate required in long-lived species.

A full test of these ideas will require quantification of the relationship between seed mass and survival from seed production to reproductive maturity. However, Charnov's theory is well supported by the data we do have, and provides a coherent explanation for the positive associations between seed mass, plant size and plant life span, as well as suggesting a way to reconcile previous demonstrations of a positive correlation between seed mass and rates of seedling survival with the present finding of no relationship between seed mass and lifetime seed production.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References

We are very grateful to John Tweddle and John Dickie from Kew Gardens, who allowed us to use data from the Seed Information Database, and to Ian Wright and Sandra Diaz, who allowed us to use some of their unpublished data. Thanks to Glenda Wardle for helpful discussion of results and for help in deriving lifetime seed output from transition matrices. Supported by an Australian Postgraduate Award to A.T.M. and by Australian Research Council funding to M.W. and M.R.L. This work was finalised while A.T.M. was a postdoctoral student at the National Center for Ecological Analysis and Synthesis.

Supplementary material

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References

Appendix S1 Full list of species, data and data sources.

Appendix S2 Within-habitat and within-taxa analyses.

Appendix S3 Why the relationship between seed mass and plant height might have been different in gymnosperms.

Appendix S4 Within-species relationships between seed mass and plant size.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Supplementary material
  9. References
  • Aarssen, L.W. & Jordan, C.Y. (2001) Between-species patterns of covariation in plant size, seed size and fecundity in monocarpic herbs. Ecoscience, 8, 471477.
  • Augspurger, C.K. & Kitajima, K. (1992) Experimental studies of seedling recruitment from contrasting seed distributions. Ecology, 73, 12701284.
  • Baker, H.G. (1972) Seed weight in relation to environmental conditions in California. Ecology, 53, 9971010.
  • Bakker, J.P. (1989) Nature Management by Grazing and Cutting: On the Ecological Significance of Grazing and Cutting Regimes Applied to Restore Former Species-Rich Grassland Communities in the Netherlands. Kluwer Academic, London.
  • Bazzaz, F.A., Ackerly, D.D. & Reekie, E.G. (2000) Reproductive allocation in plants. Seeds: the Ecology of Regeneration in Plant Communities (ed. M.Fenner), pp. 129. CAB International, Wallingford.
  • Charnov, E.L. (1993) Life History Invariants: Some Explorations of Symmetry in Evolutionary Ecology. Oxford University Press, Oxford.
  • Coomes, D.A. & Grubb, P.J. (2003) Colonization, tolerance, competition and seed-size variation within functional groups. Trends in Ecology and Evolution, 18, 283291.
  • Corner, E.J.H. (1949) The Durian theory or the origin of the modern tree. Annals of Botany, 8, 966952.
  • Dalling, J.W. & Hubbell, S.P. (2002) Seed size, growth rate and gap microsite conditions as determinants of recruitment success for pioneer species. Journal of Ecology, 90, 557568.
  • Eriksson, O. (1997) Clonal life histories and the evolution of seed recruitment. The Ecology and Evolution of Clonal Plants (eds H.De Kroon & J.Van Groenendael), pp. 211226. Backhuys, Leiden.
  • Falster, D.S., Warton, D.I. & Wright, I.J. (2003) (S) MATR: Standardised Major Axis Tests and Routines, Version 1.0. http://www.bio.mq.edu.au/ecology/SMATR.
  • Falster, D.S. & Westoby, M. (2003) Plant height and evolutionary games. Trends in Ecology and Evolution, 18, 337343.
  • Geritz, S.A.H. (1995) Evolutionarily stable seed polymorphism and small-scale spatial variation in seedling density. American Naturalist, 146, 685707.
  • Geritz, S.A.H., Van Der Meijden, E. & Metz, J.A.J. (1999) Evolutionary dynamics of seed size and seedling competitive ability. Theoretical Population Biology, 55, 324343.
  • Givnish, T.J. (1982) Adaptive significance of leaf height in forest herbs. American Naturalist, 120, 353381.
  • Greene, D.F. & Johnson, E.A. (1994) Estimating the mean annual seed production of trees. Ecology, 75, 642647.
  • Grime, J.P., Thompson, K., Hunt, R., Hodgson, J.G., Cornelissen, J.H.C., Rorison, I.H. et al. (1997) Integrated screening validates primary axes of specialisation in plants. Oikos, 79, 259281.
  • Harper, J.L., Lovell, P.H. & Moore, K.G. (1970) The shapes and sizes of seeds. Annual Review of Ecology and Systematics, 1, 327356.
  • Henery, M.L. & Westoby, M. (2001) Seed mass and seed nutrient content as predictors of seed output variation between species. Oikos, 92, 479490.
  • Hulme, P.E. & Borelli, T. (1999) Variability in post-dispersal seed predation in deciduous woodland: relative importance of location, seed species, burial and density. Plant Ecology, 145, 149156.
  • Iwasa, Y., Cohen, D. & Leon, J.A. (1985) Tree height and crown shape, as results of competitive games. Journal of Theoretical Biology, 112, 279298.
  • Jakobsson, A. & Eriksson, O. (2000) A comparative study of seed number, seed size, seedling size and recruitment in grassland plants. Oikos, 88, 494502.
  • Jurado, E., Westoby, M. & Nelson, D. (1991) Diaspore weight, dispersal, growth form and perenniality of central Australian plants. Journal of Ecology, 79, 811830.
  • King, D.A. (1990) The adaptive significance of tree height. American Naturalist, 135, 809828.
  • Leishman, M.R. (2001) Does the seed size/number trade-off model determine plant community structure? An assessment of the model mechanisms and their generality. Oikos, 93, 294302.
  • Leishman, M.R. & Westoby, M. (1994) Hypotheses on seed size-tests using the semiarid flora of Western New South Wales, Australia. American Naturalist, 143, 890906.
  • Leishman, M.R., Westoby, M. & Jurado, E. (1995) Correlates of seed size variation – a comparison among five temperate floras. Journal of Ecology, 83, 517529.
  • Leishman, M.R., Wright, I.J., Moles, A.T. & Westoby, M. (2000) The evolutionary ecology of seed size. Seeds – the Ecology of Regeneration in Plant Communities (ed. M.Fenner), pp. 3157. CAB International, Wallingford.
  • Levin, D.A. (1974) The oil content of seeds: an ecological perspective. American Naturalist, 108, 193206.
  • McArdle, B.H. (1988) The structural relationship – regression in biology. Canadian Journal of Zoology, 66, 23292339.
  • Moles, A.T. & Westoby, M. (2003) Latitude, seed predation and seed mass. Journal of Biogeography, 30, 105128.
  • Moles, A.T. & Westoby, M. (2004) Seedling survival and seed size: a synthesis of the literature. Journal of Ecology, 92, 372383 .
  • Niklas, K.J. (1993) The allometry of plant reproductive biomass and stem diameter. American Journal of Botany, 80, 461467.
  • Niklas, K.J. & Enquist, B.J. (2002) On the vegetative biomass partitioning of seed plant leaves, stems, and roots. American Naturalist, 159, 482497.
  • Niklas, K.J. & Enquist, B.J. (2003) An allometric model for seed plant reproduction. Evolutionary Ecology Research, 5, 7988.
  • Pitman, E.T.G. (1939) A note on normal correlation. Biometrika, 31, 912.
  • Rees, M. (1996) Evolutionary ecology of seed dormancy and seed size. Philosophical Transactions of the Royal Society of London – Series B: Biology Sciences, 351, 12991308.
  • Rees, M. & Westoby, M. (1997) Game theoretical evolution of seed mass in multi-species ecological models. Oikos, 78, 116126.
  • Rodríguez-Gironés, M.A., Sandsten, H. & Santamaría, L. (2003) Asymmetric competition and the evolution of propagule size. Journal of Ecology, 91, 554562.
  • Salisbury, E. (1974) Seed size and mass in relation to environment. Proceedings of the Royal Society of London B. Biology Sciences, 186, 8388.
  • Shipley, B. & Dion, J. (1992) The allometry of seed production in herbaceous angiosperms. American Naturalist, 139, 467483.
  • Silvertown, J.W. (1981) Seed size, lifespan & germination date as co-adapted features of plant life history. American Naturalist, 118, 860864.
  • Sokal, R.R. & Rohlf, F.J. (1995) Biometry: the Principles and Practice of Statistics in Biology Research, 3rd edn. W.H. Freeman, New York.
  • Thompson, K. (1984) Why biennials are not as few as they ought to be. American Naturalist, 123, 854861.
  • Thompson, K. & Hodkinson, D.J. (1998) Seed mass, habitat and life history – a re-analysis of Salisbury (1942, 1974). New Phytologist, 138, 163166.
  • Thompson, K. & Rabinowitz, D. (1989) Do big plants have big seeds? American Naturalist, 133, 722728.
  • Turnbull, L.A., Rees, M. & Crawley, M.J. (1999) Seed mass and the competition/colonization trade-off: a sowing experiment. Journal of Ecology, 87, 899912.
  • Udvardy, M.D.F. (1975) A Classification of the Biogeographic Provinces of the World. International Union for Conservation of Nature and Natural Resources, Gland, Switzerland.
  • Venable, D.L. (1992) Size-number trade-offs and the variation of seed size with plant resource status. American Naturalist, 140, 287304.
  • Venable, D.L. & Brown, J.S. (1988) The selective interactions of dispersal, dormancy and seed size as adaptations for reducing risks in variable environments. American Naturalist, 131, 360384.
  • Warton, D.I. & Weber, N.C. (2002) Common slope tests for bivariate structural relationships. Biometrical Journal, 44, 161174.
  • Webb, C.O. & Peart, D.R. (1999) Seedling density dependence promotes coexistence of Bornean rain forest trees. Ecology, 80, 20062017.
  • Westoby, M., Falster, D.S., Moles, A.T., Vesk, P.A. & Wright, I.J. (2002) Plant ecological strategies: some leading dimensions of variation between species. Annual Review of Ecology and Systematics, 33, 125159.
  • White, J. (1981) The allometric interpretation of the self-thinning rule. Theoretical Biology, 89, 475500.