The costs of reproduction in plants


  • José Ramón Obeso

    Corresponding author
    1. Ecology unit, Departamento Biología Organismos & Sistemas, Universidad de Oviedo, E-33071 Oviedo, Spain
      Author for correspondence: José Ramón Obeso Tel: +34 98 510 4789 Fax: +34 98 510 4866 Email:
    Search for more papers by this author

Author for correspondence: José Ramón Obeso Tel: +34 98 510 4789 Fax: +34 98 510 4866 Email:



  • Summary321

  • I.   Introduction321
  • II.     Theory on costs of reproduction322
  • III. Methodological aspects324
  • IV. Empirical evidence328
  • V. Plant size and costs of reproduction330
  • VI.  Costs of reproduction in sexually dimorphic plants331
  • VII. Compensation of the costs333
  • VIII. Concluding comments and future perspectives336
  • Acknowledgements337

  • References337


This review reports on the processes associated with costs of reproduction, including some theoretical considerations, definitions and methodological aspects, followed by a list of the situations where costs are difficult to find. Despite some exceptions, case studies, examined by trade-offs between reproduction and other life-history traits, generally support the predictions of the cost of reproduction hypothesis. The cost of reproduction as an evolutionary determinant of sexual dimorphism in life history traits in dioecious species was specifically tested, considering that the higher cost of reproduction in females has driven the life history traits related to sexual dimorphism. Females of woody dioecious species were consistently smaller than males supporting the costs of reproduction hypothesis. By contrast, females of herbaceous perennials were generally the larger sex, which did not fit the expectations of the hypothesis. Finally, the mechanisms that enable the compensation of the reproductive costs are detailed, including the plastic responses of photosynthesis and growth, the effects of the timing of investment, plant architecture and plant physiological integration.

I. Introduction

Life history traits, such as growth, survival and reproduction, are linked through constraining relationships, implying that to be the best in all ecological situations is not biologically possible and to be well fitted to even one situation requires a compromise. We refer to these compromises as trade-offs or costs between life history variables (Reznick, 1985). The concept of costs has taken on an important role in the development of many fields in evolutionary biology since the introduction of cost-benefit analyses into biology and it has been closely linked to the notion of evolutionary trade-offs and constraints (Williams, 1966a,b; Levins, 1968; Roff, 1992). Specifically, the costs of reproduction are defined in terms of losses in the potential future reproductive success caused by current investments in reproduction (Jönsson, 2000).

In a historical perspective, the idea that reproduction competes with other functions is as old as the study of natural history itself (Jönsson & Tuomi, 1994). The historical development of this idea was mainly due to animal ecologists but a number of works by plant ecologists were compiled by Harper (1977) who concluded that perennial polycarpic plants often show an inverse correlation between vegetative growth and the production of fruit and seed which suggests that fecundity and vegetative activity are not wholly compatible. If reproduction carries no costs either in terms of future survival or fecundity an organism should begin reproducing at the earliest possible age (Roff, 1992). The fact that this does not seem to happen (most plants need to achieve a minimum size for reproduction) suggests that reproduction costs exist. The question of whether this threshold minimum size for reproduction mainly reflects the cost of reproduction or, alternatively, is a reflection of some allometric relationships may be a subject of controversy. However, a recent review by Bender et al. (2000) supported the conclusions of Harper & White (1974) for polycarpic perennials: plants with a brief juvenile period have a short life span, and plants with a long juvenile period have a long life span with a long reproductive life.

However, what appeared to be crystal clear in theory later became less evident when field biologists attempted to evaluate the cost of reproduction hypothesis. A number of field and laboratory experiments have been conducted during the past decades, yielding ambiguous and contradictory results. This has lead to a controversy as to the appropriate methods for measuring reproductive costs (Jönsson & Tuomi, 1994). On the other hand, Tuomi et al. (1983) postulated that reproductive costs might not necessarily be observed, since reproducing individuals may increase their resource intake and develop some compensatory mechanisms. Nevertheless, despite the methodological difficulties and the fact that the concept of costs of reproduction is mainly descriptive and is not a predictive tool, the importance of its study resides in its possibilities in linking physiology to demography and life history evolution.

II. Theory on costs of reproduction

1. Concepts on costs of reproduction hypothesis

The three most important components of fitness for a plant might be: the maintenance of the soma and the struggle for existence with competitors, where vegetative size may determine most of the success; the avoidance of predators, which may be achieved by investing in defences; and the investment in reproduction which is closely related to fitness and may be interpreted as a measure of the success of the other two components. Each of these activities requires the expenditure of energy, which is in limited supply, and current investments in each one of these activities result in losses in the potential investments in the other. If resources are limited, the organisms can only acquire a limited amount of resources and energy for which two more processes compete directly, then an increase in energy allocated to one must result in a decrease in energy allocated to the other(s). This is known as the Principle of Allocation (Levins, 1968). We can therefore define cost of defence as the losses in the potential future reproductive success caused by current investments in defence and cost of reproduction as the losses in the potential future reproductive success caused by current investments in reproduction.

Williams (1966a,b ) formalized the demographic theory of optimal reproductive tactics, that can be summarized as three postulates:

  • 1When both reproductive and nonreproductive organisms have the same energy available for investment, an increase in reproductive effort inevitably results in both an increase in current reproductive output and a reduction in somatic investment.
  • 2When reproduction takes place at the expense of somatic investment, the somatic cost reduces the probability of survival and future fecundity.
  • 3If reproduction results in survival and/or fecundity costs, there is a trade-off between current reproductive output and future reproduction. This trade-off is optimised by natural selection which will favour individuals exhibiting higher lifetime reproductive values. The reproductive value of an individual ( Fisher, 1930 ) is its total lifetime expectation of reproductive output, that is the sum of expectations of reproduction in all ages or reproductive states, which is a combination of age-specific fecundity and survival rates. The reproductive value is the sum of the value of the current breeding season and the residual reproductive value (RRV) ( Williams, 1966b ). Then, the theory predicts a negative correlation between the intensity of the current reproductive effort and RRV.

According to the simple algebraic definition presented by Begon et al. (1986) the importance of the residual reproductive value varies with population changes. If the population is expanding, the relative value of current reproductive output increases in relation to residual reproductive output. Conversely, in declining populations future reproduction (RRV) can be of greater value than current reproduction because of its greater proportionate contribution. This suggests that there should be an advantage of early intense reproductive output in expanding populations (i.e. colonising species). Contrarily, individuals from populations threatened by extinction would take advantage of delaying reproduction.

The demographic effect of costs of reproduction and its evolutionary implications were usually studied considering that populations are in equilibrium but will be different in expanding or declining populations; however, this link between demography and analysis of cost of reproduction has rarely been established.

2. Trade-offs, constraints and costs

The major premise in the theory on costs of reproduction is that there is a trade-off at the physiological level of resource allocation to either vegetative growth or reproduction. From the genetic and evolutionary perspective, trade-offs have been defined as negative, genetically–based, associations between traits (Reznick, 1985). However, the trade-off reflects the consequences of constraints that are the ultimate causal factors underlying trade-off phenomena (Gould, 1989; Roff, 1992; Stearns, 1992). For instance, the trade-off is a reflection of the allocation of resources to two different traits, made within the constraints of the conservation laws (Mole, 1994). The processes involved in making the division of resources may be under genetic control, but at the physiological level the key object of concern is the fate of limiting resources (Chapin, 1989; Mole, 1994). The constraints underlying the trade-offs may be formal constraints (which reflect physical laws, i.e. conservation laws), historical constraints (reflecting long-term evolutionary history, i.e. ontogenetic constraints) and functional constraints (Gould, 1989). If we consider the constraints imposed by the conservation laws, the trade-offs patterns can be expressed as a Y-shaped model (de Jong & van Noordwijk, 1992), where the stem of the Y reflects the input of resources available to be divided between two traits. If a single resource is divided over many traits, a trade-off pattern can be modelled as a ‘tree’ where each allocation decision is a dichotomous ‘branching point’ (de Jong, 1993).

The foremost plant functions (growth, defence and reproduction) may be connected by trade-offs. Herms & Mattson (1992) have framed a plant defence theory in terms of a trade-off between growth and defence, proposed as a plant dilemma. The underlying constraint is that at the cellular level the processes of growth and differentiation are incompatible functions. In fact, the fitness costs of defence have been widely documented (Zangerl & Bazzaz, 1992; Bergelson & Purrington, 1996; Koricheva, 2002). Nevertheless, rather than a dilemma, plant allocation may be defined by a triangle of competing functions: growth, reproduction and defence. This may be interpreted as a ‘tree’ where the resources of the stem are divided into three interconnected branches. The three branches compete for limited resources, but at the same time, the resources allocated to one branch may be partially reallocated to other branches (some structures may share two or more functions). For instance, green fruits are reproductive structures that exhibit photosynthetic activity and defence functions (i.e. the stone, chemical defences).

Otherwise, there are considerable structural parallels between the plant defense theory and plant responses to costs of reproduction. Mole (1994) proposed strong parallels between the plant defense theory (avoidance strategies or ‘resistant’ responses) and life-history and I specifically propose the parallels between the plant defense theory and the trade-offs between reproduction and growth, which can be supported by considering that plant responses to herbivory (tolerance strategies or ‘resilient’ responses) and reproduction share certain similarities (see section VII below).

3. Definitions related to costs of reproduction

Contributions to science need a common language, but sometimes different authors use different terms to refer the same concept. A summary of the most commonly used definitions in relation to costs of reproduction is included in Table 1. The assessment of costs of reproduction may be focused from two complementary points of view: resource allocation at the time of reproduction, also termed direct costs of reproduction, and indirect costs, referred to as indirect, delayed or demographic costs of reproduction (Newell, 1991; Nicotra, 1999), which may be interpreted as the demographic consequences of resource investment in the reproductive process. These indirect costs may, in turn, be examined in the short term (reduced growth or reproduction in the subsequent season (Newell, 1991; Ashman, 1992) or in the long term (lifetime costs). Long-term, delayed costs of reproduction are often inferred from the reduced and more evenly spaced future reproductive events, higher mortality rates, lower growth rates and lower vegetative propagation of individuals allocating higher amounts of resources to current reproduction. Direct manipulations to test for delayed cost in dioecious woody species were performed in Lindera benzoin, Salix alaxensis and Siparuna grandiflora revealing a trade-off between current reproduction and subsequent reproduction in females (Fox & Stevens, 1991; Cipollini & Whigham, 1994; Nicotra, 1999).

Table 1.  Summary of definitions commonly used in relation to costs of reproduction. Sometimes there are several denominations for the same concept or the same definition is used for different concepts
Direct costs:Somatic costs of reproduction during current reproductive season
Delayed costs (Newell, 1991), indirect (Ashman, 1992) or demographic costs of reproduction (Karlsson et al., 1990; Nicotra, 1999):The demographic consequences of resource investment in the reproductive process, such as mortality rate and frequency of future reproduction
Short-term costs:Reduced growth or reproduction in the subsequent season (Newell, 1991; Ashman, 1992)
Long-term costs:Lifetime costs or fitness costs
Physiological costs (Karlsson, 1994):Physiological trade-offs derived from investment in reproduction (i.e. photosynthetic and respiration rates, phenology of growth, nectar production)
Relative somatic costs (RSC) (Tuomi et al., 1983):Differences in vegetative growth rate associated with differences in reproductive investment
Static measures (Ashman, 1994):Estimates of current reproductive investment neglect the physiological processes that influence actual reproductive expenditure
Dynamic measures (Ashman, 1994):Accurate assessment of reproductive investment considering all physiological process involved (see physiological costs)
Direct fecundity costs (Reznick, 1985): Reduction in energy available for future reproduction (long-term energy stores)
Indirect fecundity cost (Reznick, 1985): Approximately equivalent to Relative Somatic Cost (Karlsson et al., 1990)

However, the term direct costs may be used in another sense as in the case of Bell (1980) who considered that the quantity of resources devoted to reproduction are direct estimates of the costs (construction costs, Chapin, 1989). Reznick (1985, 1992) divided the costs into two main categories, the first being survival costs, where current reproductive effort influences the probability of future survival. The second is fecundity costs, where current reproductive effort influences future reproduction. These fecundity costs may be direct if they reduce the energy available for future reproduction by reducing long-term energy stores that are used for reproduction. Indirect fecundity costs reduce growth rates, indirectly influencing future reproduction whenever clutch size is proportional to body size.

Lloyd (1984 ) distinguished three different types of costs of reproduction. Production costs include the construction costs of structures related to reproduction or the acquisition and storage of energy for reproduction. Fixed costs reflect the costs that are incurred only once before any reproduction is possible and that are not added subsequently (e.g. the production of peduncles and pedicels). Fixed costs consider preparation for offspring production, are more important in animals than in plants and have also been referred to as prebreeding costs ( Jönsson et al., 1995a,b ). When the reproductive structures are repeated more than once (i.e. flowers, fruits) Lloyd (1984 ) used the term recurrent costs that are more important in plants than in animals.

The application of the ‘economic analogy’ to plant allocation (Bloom et al., 1985) has led to the use of terms such as ‘opportunity costs’ (Chapin, 1989), ‘fixed costs’ (Lloyd, 1984; Jönsson, 2000) and ‘future discounts’ that allows a calculation of future costs and benefits in terms of their present value, considering that both have a higher absolute value in the present than in the future (Lerdau, 1992). For instance, the ‘opportunity costs’ are incurred through losses of growth and competitive status as a result of allocation to reproduction instead of growth early in the ontogeny. Consequently, these costs are predicted to be higher in presence of competitors. Finally, the ‘ecological costs’ of reproduction are incurred through interaction with biotic and abiotic environment. For instance, reproductive individuals may divert resources from defenses to reproduction and then increase their susceptibility to herbivory.

III. Methodological aspects

1. Reproductive allocation, reproductive effort and costs of reproduction

To examine the correlation between costs of reproduction and the intensity of the current reproductive effort we first need to consider the methods for estimating an index of reproductive effort (RE). The concepts of reproductive resource allocation (RA), RE and cost of reproduction are not synonymous; however, sometimes these concepts are intermingled (Bazzaz & Ackerly, 1992). RA may be defined as the proportion of resources allocated to reproduction and is usually estimated as a static measure. The notion of RE is directly connected to the concept of reproductive costs, since it can be defined either in terms of the supply of resources invested in reproduction which is derived from the vegetative plant (Bazzaz & Ackerly, 1992) or by examining changes in vegetative biomass resulting from reproduction (Tuomi et al., 1983; Reekie & Bazzaz, 1987a,c). As Stearns (1992) points out, two individuals could devote the same amount of energy to reproduction at equivalent body sizes, but differ greatly in the absolute amount of energy gathered or the time at which it was gathered. In such cases, the true proportional allocation to reproduction (RE) would be unequal, whereas the RA would be identical.

Nevertheless, reproductive effort (RE) may be calculated as the proportion of total resource pool devoted to reproduction (Thompson & Stewart, 1981; Tuomi et al., 1983):

RE  =  Ir/(Ir  +  Is)

the sum of resources allocated to reproductive (Ir) and somatic (Is) structures (above- and below-ground plant parts). Additionally, RE may be calculated as RE = Ir/Is in order to avoid autocorrelations (Samson & Werk, 1986; Klinkhamer et al., 1992). However, these calculations measure reproductive allocation (RA) more than RE, because they are static measures that do not represent the proportion of energy flowing through the organism that is devoted to reproduction.

The relative somatic cost of reproduction, RSC (sensuTuomi et al., 1983) can be calculated, as the relative difference in somatic investment (I) between nonreproductive (In) plants (or plants prevented from reproduction or seed production by removing reproductive parts in an early state: debudding experiments, flower removal) and naturally reproducing plants (Is):

RSC = (In − Is)/In

 If the entire investment in reproduction (Ir) occurs at the expense of somatic growth (Is) then Ir + Is = In, and in such a case RE = RSC. Nevertheless, the relationships between RE and RSC were examined in several species and RSC were frequently lower than RE indicating that all resources invested in reproduction were not directly derived from the somatic resource pool but additional resources can be obtained by other mechanisms (Karlsson et al., 1990; Thorén et al., 1996; Hemborg & Karlsson, 1998a). This estimate of RSC reflects a mean population measure rather than calculating individual plant somatic costs and in this sense individual variations cannot be detected. Both RE and RSC may be estimated in different currencies: dry mass, carbon, nitrogen, and phosphorous.

Cost of reproduction may also be determined in terms of lost growth per unit of reproductive structure (i.e. fruits or seeds) production (Reekie & Bazzaz, 1992): (Bn – Br)/R, where Bn is the biomass (or other currency) of nonreproductive plants, Br is the biomass of reproductive plants and R is the biomass of the seeds (or reproductive structures).

The above measures of reproductive investment are static estimates that may not reflect the true cost to the plant because they ignore the dynamic processes associated with reproduction and its physiological costs (Ashman, 1994). In fact, measures of respiration rates seem to provide the most complete estimate of the reproductive effort and costs of reproduction when energy or carbon is the limiting resource since they include growth, maintenance and composition of the reproductive structures. But respiration varies considerably with many factors, and thus poses difficulties on its own (Bazzaz et al., 1979; Goldman & Willson, 1986). An accurate method for calculating RE that takes into account the atmospheric losses was formulated by Bazzaz & Reekie (1985) and Bazzaz & Ackerly (1992) but its use in empirical work has considerable difficulties.

In woody plants, long-term costs of reproduction may be estimated by examining tree-ring growth (Obeso, 1997; Obeso et al., 1998). Silvertown & Dodd (1999) described methods for overcoming the problem of measuring costs of reproduction, illustrated by a field study of Abies balsamea where using serial correlation and a permutation test on chronosequences, they reconstructed the age- and size-specific dynamics and detected growth and demographic costs of reproduction.

2. Currency of allocation

If resources are not limited, trade-offs derived from resource allocation make no sense and this is why the relevant limiting resource should be identified in order to index all costs in that currency. However, there exist problems in identifying the relevant limiting resource in plants, which limits both reproductive and vegetative growth (Goldman & Willson, 1986; Bazzaz & Ackerly, 1992). Most published studies of costs of reproduction are based on measurements of the biomass. If carbon is the resource in limited supply, or when the estimates of proportional resource allocation based on biomass or on caloric content do not differ from each other, or if all resources vary in parallel, the use of biomass seems reasonable (Goldman & Willson, 1986; Chapin, 1989). However, Ashman (1994) supports the contention that biomass or carbon may not be a good currency for reproductive investment in plants because reproductive structures may be self-supporting in terms of carbon.

Furthermore, the proportion of nitrogen allocated to reproductive structures is often higher than the proportion of biomass or carbon (Reekie & Bazzaz, 1987b) and plant growth is not always limited by a single resource but tends to be limited by different resources simultaneously (Chapin et al., 1987). The identification of the nutrient(s) truly limiting growth and reproduction may be complicated (Harper, 1977; Lovett-Doust, 1989). However, Bazzaz & Ackerly (1992) stated that any of the various limiting resources might be appropriate as a currency of allocation and that problem has its origin in the interpretation of the allocation patterns rather than their assessment. Conversely, Ashman (1994) proposed that it might be more important to measure nutrient investment dynamically than to measure many nutrients or to determine which nutrient is limiting.

Empirical results clearly demonstrated that conclusions about costs of reproduction depend on the currency. For example, Hemborg & Karlsson (1998a) when examining the costs of reproduction in eight subarctic plant species found that the relative costs were generally of similar magnitude in terms of nitrogen and phosphorous but lower in terms of biomass. In fact, the costs in terms of carbon may be at least partially compensated by increasing photosynthetic rates or by the photosynthesis of the reproductive structure itself. In this sense, over one growing season, male cones of Pseudotsuga menziesii var. glauca required only 8% of all carbon allocated to reproduction, with female ones consuming the remaining 92%. However, when nitrogen was used as currency, male cones received 22% of all nitrogen allocated to reproduction and female cones received the remaining 78% (McDowell et al., 2000).

3. The experimental manipulation of reproductive effort in plants

Since costs of reproduction may be difficult to find in natural populations sometimes it is necessary to use experimental manipulation to evidence the effects of reproductive allocation in other functions. In fact, trade-offs are best verified through manipulative experiments, where the level of expression of a trait is modified (Stearns, 1989; Partridge, 1992). If individuals are allowed to reproduce at their chosen rate, individual variation due to genetic, recent history and size differences will produce an important variance in reproductive traits and negative correlations with other life history traits will be difficult to find (Partridge, 1989). Furthermore, experimental manipulation of current reproductive allocation represents a fixed environmental effect that is independent of genetic background (Reznick, 1992), so it can be useful in evaluating the cost of reproduction.

Some of the experimental possibilities of manipulation of reproductive investment and environmental control are summarized in Table 2. The extent of reproduction may also be controlled experimentally using photoperiod manipulations (Reekie & Bazzaz, 1992). Studies that artificially increase the fruit set using hand pollination may illustrate mechanisms of costs of reproduction by creating extreme cases of resource-demand (Primack & Hall, 1990; Delph & Meagher, 1995; Ramsey, 1997). Similarly, considering that in perennial plants seed production is expected to be resource-limited more than pollen-limited (Willson & Burley, 1983), the experimental reduction of available resources is expected to result in higher costs of reproduction. Resources may be manipulated in a number of ways: shading, avoiding prey captures in carnivorous plants, and nutrient addition. Defoliation experiments may be included in this category but the timing of leaf removal should be controlled (Obeso & Grubb, 1994). One way to assess ecological costs is to compare the magnitude of the examined trade-offs in the presence and absence of herbivores and at different densities of plants (Horvitz & Schemske, 2002).

Table 2.  Experimental manipulations of reproductive investment and environmental control
(a) Manipulation of reproductive traits maintaining the environment constant
1Nonreproductive individuals (from modules to entire plants) used as control of the control: No cost of reproduction is expected
2Suppression of reproduction in otherwise reproductive individuals: experiments of bud, flower or small fruits removal
3Nonmanipulated reproductive individuals: control
4Individuals with increased reproductive effort RE: maximum cost expected (i.e. supplementation of natural pollination in cases of pollen limitation)
(b) Experimental manipulation of the environment
1Herbivore exclusion vs herbivore presence
2Low density (competition) vs high density (competition)
3Low resource availability vs high resource availability. Experiments of nutrient addition vs limitation, leaf removal (defoliation) and branch girdling experiments

As costs during the current growth season or even during the subsequent year may not be detected, long-term studies are necessary as the only way to detect costs in long-lived species (i.e. Primack et al., 1994; Hemborg, 1998b; Antos & Allen, 1999).

4. Techniques used to detect costs

Despite the central role of costs of reproduction in life-history theory, there is a great deal of controversy about their existence, derived at least in part from differences in the techniques used to detect costs, some of which fail to control major sources of confounding variation (Partridge & Harvey, 1988). The measurement of costs of reproduction in plants is difficult (Goldman & Willson, 1986), and few field studies have measured these costs in terms of fitness or demographic components (but see Svensson et al., 1993), thus creating a hiatus between theory and data.

The surveys of laboratory and field studies of animals showed the widespread occurrence of reproductive costs (Reznick, 1985; Roff, 2000), with negative results and cases of apparent negative costs of reproduction occurring mainly in correlational studies. Reznick (1985) evaluated four different methods being used to measure such costs (a similar classification was proposed by Bell & Koufopanou, 1986):

  • 1Phenotypic correlation studies correlate an index of reproductive effort with the presence of a potential cost, such as parental survival, parental growth, or future fecundity. This method deals with naturally occurring variation in reproductive effort, is the most commonly used method of measuring costs and is often used in the field.
  • 2Experimental studies that manipulate some aspect of reproduction or some environmental variable which affects reproduction and observe a cost-correlated response.
  • 3Genetic correlation studies use quantitative genetic experiments conducted in the laboratory to determine genetic correlations between reproductive expenditure and cost.
  • 4Selection experiments that observe correlated changes in some index of reproductive effort and some potential cost in responses to selection pressures on particular life history traits in the laboratory.

It has been suggested that genetic correlations and selection experiments are the best way to test the assumptions behind models concerned with the evolution of life history traits (Reznick, 1985; Stearns, 1989; Roff, 2000). However, most authors agree that field studies are also required to demonstrate the dynamics of costs in the field, and also to elucidate the mechanisms that organisms use to avoid the costs (Stearns, 1989; Reznick, 1992). Even the use of genetic correlation might not be a valid test. If the genotype for an optimum reproductive effort, leading to a minimum cost, is prevalent in a population, the genetic variation in that trait will be near zero. Consequently, genetically based correlations will be weak (Bailey, 1992). Although phenotypic correlations between current reproductive allocation and other life-history traits have an underlaying genetic basis (Reznick, 1985) environmental factors can potentially alter current investment. Whether such environmental effects obscure the detection of the genetically based costs of reproduction is currently a subject of debate (Sinervo & DeNardo, 1999); however, the effects of environment and the interaction environment × genotype may be experimentally controlled by holding the environment constant. One way to circumvent the objections raised to phenotypic correlations between life-history related traits could be the use of path analysis, which may remove the effects of confounding variables (Sinervo & DeNardo, 1999; Sinervo, 1999). This method may be used to remove the effects of environmental variability on reproduction and the environmental-free residuals should reveal costs of reproduction that are not obscured by environmentally induced reproductive plasticity (Fig. 1).

Figure 1.

The use of the techniques of path analysis estimates the contribution of environmental variables to current reproduction and the effects of this relationship on future reproductive investment. The correlation between current reproductive investment and future reproductive investment (or other life history trait) may be decomposed into a set of correlated environmental effects (environmental variables) and environmental-free residuals that probably reflect variability in life history traits per se . It is expected that these residuals (life history variables) to be negatively correlated.

5. Difficulties in detecting the costs of reproduction

The trade-off between reproduction and growth is not a general phenomenon (Tuomi et al., 1983; Obeso, 1993a; Delph & Meagher, 1995) and below I list several difficulties in detecting the costs of reproduction:

  • 1There are some methodological problems in detecting the existence of costs ( Reznick, 1985, 1992 ). The different empirical approaches used to evaluate trade-offs and/or the difficulties in assessing reproductive effort in plants may lead to inconsistent evidences for the cost of reproduction hypothesis.
  • 2It should be noted that there is considerably more information content in studies that found costs (or even negative costs) than those that found no evidence at all for a cost. The absence of costs is usually a consequence of failing to reject a null hypothesis which is not a proof that the null hypothesis is true. On the other hand, there are substantial differences between the experimental designs examining costs of reproduction and we still lack an appropriate experimental protocol to accurately test this hypothesis regardless of the species considered.
  • 3Selection against individuals that suffer high costs, which involves development of mechanisms of compensation ( Obeso, 1993a ) as well as increased resource inputs, which would uncouple vegetative costs from reproductive costs (compensation hypothesis, Tuomi et al., 1983 ).
  • 4The trade-offs may be weak, and so undetectable, when resources are abundant, then the allocation principle may be not fulfilled ( Horvitz & Schemske, 1988 ). Reproductive costs may be observed only when the investment in reproduction is very large or resource levels exceptionally low (threshold hypothesis, Tuomi et al., 1983 ).
  • 5There may be contradictory results as a consequence of the modular nature of plants, which implies that costs of reproduction should be examined at several levels within the modular hierarchy: from shoot current year growth to integrating growth at the whole plant level ( Lovett-Doust & Lovett-Doust, 1988 ; Obeso, 1997 ).
  • 6Taking into account the fact that juvenile mortality is unpredictable and often high, while adult mortality is low and approximately stable, the bet-hedging hypothesis ( Schaffer, 1974 ; Stearns, 1976 ) predicts that it is not profitable to take risks of reproduction. Long-living organisms such as perennial plants should maintain reproduction at low levels which has minor effects on growth and survival.
  • 7The trade-offs between life history traits may be detected in some environments but not in others, which may be explained by the existence of phenotypic plasticity ( Stearns, 1989 ).
  • 8Inter-individual variation in resource acquisition, allocation and degree of resource depletion after reproduction may lead to nonsignificant or even positive correlation between life history traits. The model by van Noordwijk & de Jong (1986 ) considers that when some individuals (well performed) spend much and others (poor performed) spend little, positive correlations may be observed depending on the relative variation in the availability of resources. In the same way, interindividual variation is not considered when estimates of cost of reproduction are obtained by comparing population means. In these cases, the costs of reproduction may not be detected at population level despite the fact that some individuals clearly exhibit costs. Inter-individual differences in resource acquisition ability may be genetically based but it seems that an enhanced ability to acquire resources is only attained with a cost (‘acquisition cost’, Reznick et al., 2000 ).

IV. Empirical evidence

1. Case studies

Compilations by Harper (1977) and Willson & Burley (1983) have shown that the production of fruits is correlated with reduced vegetative growth or reduced fruiting in the following year(s) as a result of internal allocation shift of resources. They showed no exceptions to the trade-offs expected according to cost the of reproduction hypothesis.

A survey of the recent papers documenting trade-offs between reproduction and other life history traits in nonagricultural nondioecious species is included in Appendix 1 and summarised in Fig. 2. Under natural conditions, there will be strong selection against individuals that suffer high costs (Jönsson & Tuomi, 1994); but in agricultural selected plants, when the fruits are crops, these plants were selected for higher reproductive allocation. Costs should be expected to be higher in these species and then easier to detect. Dioecious species are considered independently in the next section. This is not an exhaustive list but aims to include recent and the most cited works in the literature on costs of reproduction. General trends from Appendix 1 were extracted using a type of vote-counting meta-analysis (Gurevitch & Hedges, 1993; Chalfoun et al., 2002) in which I calculated the frequency (number of tests) of significant negative correlations vs nonsignificant and significant positive correlations. Two-dimensional contingency tables were used to tests the independency of the grouped factors (Fig. 2).

Figure 2.

Number of tests from Appendix 1 in which examined trade-off resulted in negative correlations or other results (nonsignificant, positive correlations), considering separately (a) phenotypic and genotypic studies ( inline image  = 0.725, P > 0.05) (b) observational and experimental designs ( inline image  = 1.411, P > 0.05), and (c) the trade-off examined: current reproduction vs future reproduction (FR), vegetative growth (VG), clonal growth (CG), survival (SV) and somatic resource pool (SRP) ( inline image = 5.409, P > 0.05).

Four points from Appendix 1 and Fig. 2 deserve attention: first, most of the papers reported the existence of a negative correlation between a measure of reproduction and one or more life history traits (somatic resource pool, growth, frequency of reproduction, survival). Most of the studies considered only phenotypic correlations and detected significant negative trade-offs, therefore it seems that the approach of the study has no effect on the result (the results of the tests were independent of the method of analysis selected, phenotypic vs genotypic, Fig. 2a). On the other hand, the results of the tests were independent of the experimental approach (observational vs experimental, Fig. 2b). Nevertheless, Tripsacum dactyloides, Pinguicula alpina and Lathyrus vernus showed no significant costs in natural conditions but exhibited significant costs after manipulative experiments: removal of reproductive structures increased subsequent growth. Salix alaxensis showed the opposite result: under natural conditions it exhibited negative correlations but experimental bud removal had no effect on subsequent growth.

Second, the same species render different results depending on the trade-off examined. Seed production in Agropyron repens showed detrimental effects on growth of leaves, stems and roots but showed no effect on rhizome production (Reekie, 1991). Production of male catkins in Betula pubescens resulted in reduced shoot growth but did not affect future reproduction (Henriksson & Ruohomäki, 2000). In Geranium sylvaticum, seed production reduced subsequent flowering but had no effect on plant survival (Ågren & Willson, 1994).

Third, the same species can exhibit responses along a continuum from positive to negative correlations. This is the case of Pinguicula alpina, Ranunculus nivalis and Trollius europaeus. Significant positive correlations between allocation to reproduction and other plant functions, as an exclusive response, were detected in just two species, Plantago rugellii and Pinus banksiana.

Fourth, the results of the trade-off examined were independent of the life-history trait used (Fig. 2c); however, there are some differences in the availability of data. The effects of current reproduction on somatic resource pool and survival were examined less frequently than its effect on vegetative growth and future reproduction. The trade-offs between sexual reproduction and vegetative propagation have been repeatedly demonstrated (Sohn & Policansky, 1977; Snow & Whigham, 1989; Rameau & Gouyon, 1991; Hossaert-McKey & Jarry, 1992; Méndez & Obeso, 1993; Worley & Harder, 1996). In fact, the cost of successful clonal growth is a reduction of sexual reproduction and the longer the duration of clonal growth, the less likely that sexual reproduction occurs (Klekowski, 1997).

2. Physiological costs of reproduction

The direct costs incurred for maintain reproductively functional flowers include resources invested in pollen, nectar production, maintenance respiration and transpiration by floral structures. On the other hand, reproductive structures may contribute to carbon gain. Thus, the consideration of physiological costs of reproduction entails a perspective of dynamic estimates of the costs, which are more accurate than static ones, particularly when variables such as floral longevity are taken into account (Ashman, 1994). There must also be an indirect cost of floral longevity (Chapin, 1989), which should result in a diversion of resources from other reproductive or vegetative functions. Schoen & Ashman (1995) suggested that a possible indirect cost of floral longevity is reduced flower production.

Nectar may represent a considerable investment for plants (Southwick, 1984; Pyke, 1991; Ashman & Schoen, 1997; Harder & Barret, 1992). By removing nectar production, Pyke (1991) experimentally increased nectar production in Blandfordia nobilis and reduced seed production. He concluded that there should be a trade-off between extra-fertilized seeds resulting from enhanced nectar production and the cost in terms of the seeds of nectar production.

Ashman & Schoen (1997 ) found that flowers of Clarkia tembloriensis experimentally maintained 35% longer invested proportionately more in nectar sugar (30%) and these plants with long-lived flowers showed 12% reduction in seed production, which may be interpreted as a cost of flower maintenance. A similar trade-off between floral longevity and seed production was found in Calochortus leichtlinii ( Holtsford, 1985 ). Ashman & Schoen (1997 ) proposed a model that expresses the costs of flower maintenance as a trade-off between floral longevity and seed production and define an optimal floral longevity determined by the rate of fitness gain and the cost of floral maintenance.

The models of sex allocation in hermaphrodites assume that total reproductive effort is resource-limited and must be divided between male and female functions, and that an increased allocation to one of these should decrease allocation to the other (Goldman & Wilson, 1986; Atlan et al., 1992). There is some evidence that trade-offs between male and female allocation, but there also exist mechanisms to avoid this trade-offs such as phenological separation between both functions. Much of the costs of seed production may begin after the male costs have terminated (Goldman & Wilson, 1986). The costs of male function were considered in a number of studies (Caesar & Macdonald, 1984; Delesalle & Mooreside, 1995; Karlsson et al., 1996; Henriksson & Ruohomäki, 2000; Houle, 2001) and some of them specifically considered the costs of pollen production (Rameau & Gouyon, 1991; Goldman & Wilson, 1986).

Most studies on the costs of reproduction exclusively estimated the costs of female function by examining the reproductive output. This static perspective neglects not only the male function but also the costs of fruit abortion. If dynamic estimates are not considered, this fraction of reproductive investment will be a source of error in the calculation of the trade-off between reproduction and growth, since plants or modules exhibiting high initial fruit set and low final fruit set have allocated an important pool of resources to reproduction that is not taken into account. Despite the fact that fruit abortion generally occurs in earlier phases of fruit development, an important fraction of the resources invested in reproduction may be lost if we consider that the cost of the fruit initiation may be greater than the costs of increasing its size (Walker & Ho, 1977). The possibility of resorption of nutrients from developing fruits before their abortion has not been examined in detail. The overproduction of flowers and subsequent abortion is an extended phenomenon (Lee, 1988), which might be a mechanism for adjusting reproductive output to the level of resources available by an individual (Lloyd, 1980; Stephenson, 1981). However, this phenomenon may also be associated with several other functions leading to an increase in the inclusive fitness of the plant (Lee, 1988). In summary, some of the variability found when examining costs of reproduction might be due to the static perspective, which neglects costs of construction, maintenance, male function and fruit abortion.

3. Habitat and costs of reproduction

Phenotypic correlations between life-history traits may be negative (indicating trade-off) in some environments and no significant or even positive (absence of trade-offs) in others (Stearns, 1989). This may be interpreted as phenotypic plasticity and might explain some of the variability in costs of reproduction found in plants. Within species variability in costs of reproduction due to habitat differences were found in several species (Obeso, 1993b; Syrjänen & Lehtilä, 1993; Biere, 1995; Thorén et al., 1996; Hemborg & Karlsson, 1998b; Hemborg, 1998a). Reznick (1985) suggested that the occurrence of costs would be most apparent in habitats with low resource availability or other stress conditions. Accordingly, in Lychnis flos-cuculi, investment in reproduction reduced next-year fecundity in the less fertile sites, whereas no cost was detected in the most fertile site (Biere, 1995). This kind of data are particularly interesting because they would reveal the role of environmental factors in costs of reproduction expression and whether these are possible crossing reaction norms in the trade-offs (Syrjänen & Lehtilä, 1993; Biere, 1995).

Other authors have proposed that reproductive investment is often negatively related to altitude or short growing season indicating that restrictive climatic conditions probably entail strong selection against individuals expressing reproductive costs (Hemborg, 1998a). Therefore, cost of reproduction will be minimised and not easily detectable. Particularly, long-term costs of reproduction may be more difficult to detect in long-lived plants under restrictive conditions (Hemborg & Karlsson, 1998b; Hemborg, 1998a; Houle, 2001) such as the climate at high altitudes and latitudes, because, as a result of weather anomalies, reproductive failure is frequent and the resources saved by not investing in reproduction may contribute to reducing long-term reproductive costs. In fact, restrictive climatic conditions might explain the positive relationship between reproduction and growth when climatic conditions are favourable (Despl& & Houle, 1997; Houle, 2001). However, Hemborg & Karlsson (1998b) studying two herbaceous perennials, Trollius europaeus and Ranunculus acris, found no effects of altitude on somatic costs.

In clonal plants an interaction between reproduction, growth and clonal propagation is expected; however, there are also conflicting expectations in relation to reproduction and vegetative propagation. In this sense, Williams (1975) and Abrahamson (1975, 1980) predicted that sexual reproduction should increase at high plant densities, and vegetative reproduction should increase at low densities. Conversely, Newell & Tramer (1978) proposed the opposite predictions noting that sexual reproduction declines and vegetative propagation increases with succession toward closed deciduous forests. Loehle (1987) presented a benefit-cost model that predicts changes in reproductive effort in response to change in nutrient, shade, and environmental stress. Changes in partitioning to sexual and vegetative reproduction depend on the detection by the plant of environmental cues and the plant should adjust partitioning to match changes in the estimated success of the two reproductive modes. Some field studies examined by Loehle supported his model but others did not. It has also been suggested that, in clonal species, it may be the morphology of clonal growth which controls the partitioning of energy between sexual reproduction and clonal growth (Armstrom, 1984). We need further experimental analysis considering the relative allocation to reproduction, growth and vegetative propagation at different plant densities and under different environmental conditions. In this respect, the interaction between factors (biotic and abiotic) should deserve our attention because sometimes the effects reproduction are only observed in presence of other factors such as herbivory (natural or experimental) or fire (Primack et al., 1994; Horvitz & Schemske, 2002).

V.Plant size and costs of reproduction

Size is a determinant variable in life history that might explain most of the plant allocation patterns; nevertheless, variations of reproductive allocation and reproductive effort with plant size are not clearly understood and no clear interspecific trend is evident even among closely related species (Niklas, 1994). Weiner (1988) noted that many plant species must reach a minimum size before they initiate reproduction and above this size threshold the intraspecific relationship between reproductive and vegetative biomass is predicted to be linear (see Klinkhamer et al., 1992 for nonlinear relationships). According to Samson & Werk's model (Samson & Werk, 1986), a threshold size for reproduction is frequently associated with a negative Y-intercept, which entails an increase of relative reproductive investment with plant size. If larger plants invest proportionately more in reproduction, costs of reproduction might increase with plant size. One can reach the same conclusion if life expectance decreased with plant size (Horvitz & Schemske, 1988).

Calow (1981 ) used an allometric approach to determine when a plant should switch from vegetative to reproductive growth. Growth and reproduction are fuelled by the energy remaining after considering inputs (energy intake is regulated by 2/3-power law, Niklas, 1994 ) and outputs (respiration is controlled by 3/4-power law, Niklas, 1994 ; Gillooly et al., 2001 ). During vegetative growth, inputs exceed outputs but the greater slope of the respiratory cost leads to an intersection at the point that represents the maximum metabolic limit and the maximum body size. The model considers that switching should occur at the size which maximizes the energy available for reproduction, where the difference between intake and cost of maintenance is maximum and this difference will define the optimum size for reproduction. Subsequent models have considered size and age as is the case of optimal energy allocation models of Kozlowski (1991, 1992 ).

Despite the importance of plant size on absolute and proportional reproductive allocation, few studies have considered size-dependency in relation to costs of reproduction. Cost of reproduction increased with plant size in Plantago major (Reekie & Bazzaz, 1992), Pinguicula vulgaris (Worley & Harder, 1996) and Ranunculus acris (Hemborg & Karlsson, 1998a), but no trends were found in Plantago rugelii (Reekie & Bazzaz, 1992), Arum italicum (Méndez & Obeso, 1993) and Trollius europaeus (Hemborg & Karlsson, 1998b). In Asphodelus albus, the cost of reproduction increased with plant size at one study site but there was no significant effect at other localities (Obeso, 1993b).

On the other hand, size-dependent variation in cost of reproduction precludes the possibility of comparing mean costs for populations with unmatched size distributions (Samson & Werk, 1986; Hemborg & Karlsson, 1998b; Hemborg, 1998a).

Reproductive allocation strongly depends on plant size, but plant size may be dependent on individual history: previous reproduction, interaction with competitors and herbivores. For instance, during one growing season, reproductive individuals of the clonal perennial Asphodelus albus, may either increase or decrease their previous size (Obeso, 1993b). In this sense, the knowledge of the ‘past investment’ (Coleman & Gross, 1991) is crucial in determining an organism's ability to invest in the future. The relationship between size mass, resource gain, competence and age needs to be understood further for a better assessment of the costs of reproduction and plant size should be used as a covariate in the experimental designs.

VI. Costs of reproduction in sexually dimorphic plants

Dioecious plants provide an excellent opportunity for examining the costs of reproduction and their importance in determining the evolution of sexual dimorphism in life history. Sexually dimorphic plants usually exhibit gender-related differences in reproductive investment enabling the comparison of individuals expending a lot in reproduction (generally the females) with individuals investment much less (generally the males), in this sense they constitute a nonmanipulative experiment to study costs of reproduction.

Females of woody dioecious usually expend proportionally more of their resources on reproduction and less on maintenance and growth when compared to males (Lloyd & Webb, 1977; Willson, 1983; Obeso, 1997; Delph, 1999; Nicotra, 1999). As a result, females may incur a higher cost of reproduction which should be measurable in terms of a lower vegetative growth rate (Jing & Coley, 1990; Krischik & Denno, 1990; Ataroff & Schwarzkopf, 1992; Cipollini & Whigham, 1994), delayed flowering, lower flowering frequency (Bullock & Bawa, 1981; Hoffmann & Alliende, 1984; Cavigelli et al., 1986; Lovett-Doust & Lovett-Doust, 1988; Armstrong & Irvine, 1989; Ataroff & Schwarzkopf, 1992; Thomas & LaFrankie, 1993) and/or reduced survival (Waser, 1984; Lovett-Doust & Lovett-Doust, 1988; Allen & Antos, 1993). However, it is difficult to interpret resource allocation in terms of cost of reproduction, because there should be a strong selection against individuals that suffer high costs, which may lead to difficulties in detecting such costs (Jönsson & Tuomi, 1994). Even with the amount of information on costs of reproduction in dioecious plants, we still lack a clear picture of how differences in resource allocation to reproduction are translated into costs of reproduction. Nevertheless, according to the predictions of Lloyd & Webb (1977) and Bell (1980), the review by Delph (1999) presents information supporting the view that differences in the cost of reproduction between sexes are significant evolutionary determinants of sexual dimorphism in the life history of plants. She clearly stated that the higher cost of reproduction in females has driven the life history traits related to sexual dimorphism. If this is true, we should expect that the sex exhibiting greater reproductive investment should also incur in both higher somatic and demographic costs, and as a consequence there should be somatic intersexual dimorphism (SSD) and demographic intersexual dimorphism (DSD).

To examine whether intersexual differences exist in variables related to SSD and DSD, I reviewed studies concerning intersexual differences in one or more variables related to SSD (plant size, biomass, and relative growth rate) and variables related to DSD (mortality rates, frequency of reproduction, and vegetative propagation). Appendix 2 shows 128 intersexual comparisons corresponding to 103 dimorphic species (91 dioecious, 2 subdioecious, 9 gynodioecious and 1 trioecious) with different life histories: 62 woody (40 trees and 22 shrubs) and 41 herbaceous. I believe that this species list provides an extensive and representative sample of available data enabling intersexual comparisons in variables related to the life history of dimorphic plants. Subdioecious species were included due to their functional dioecy and because some individuals of dioecious populations may occasionally bear some flowers of the opposite sex (Lloyd & Bawa, 1984). Gynodioecious species were also considered because they exhibit clear gender specialisation: female plants produce more seeds and invest more in reproduction than hermaphrodites. The androdioecious were not included because of the rarity of this sexual expression as a route to the evolution of dioecy (Charlesworth, 1984). The cases of labile sex expression (diphasy) were considered when the species showed environmental sex determination, but species exhibiting true sex change were not, due to the difficulties for testing previously cited hypothesis in these species.

Each species in Appendix 2 was assigned either to ‘NS’ if there were no significant differences between morphs, to ‘F’ if females outperformed males or hermaphrodites, or to ‘M’ if the males (or hermaphrodites, only one species) outperformed females in one or more variables. In most species the results for different variables related to one type of sexual dimorphism (SSD or DSD) were coherent within the same study. In the same way, different studies on the same species usually provided similar results. However, related variables, different studies or localities sometimes provided different results for the same species. In these cases, if one study found male advantage for one variable and another study and/or site female advantage for the same species (e.g. Acer negundo, Appendix 2) this species was included in the group of species without sexual dimorphism (NS).

The information concerning sex-related demographic variables (DSD) is very scant with the exception of reproductive frequency, and there is also absence of information about relative growth rate, particularly for herbaceous (Table 3). When examined, reproductive allocation was consistently higher for females in most of the species and there were only three references reporting no significant differences between sexes (Table 3).

Table 3.  Number of species from appendix 2 showing different results for variables examined in relation to sexual dimorphism. The variables were grouped as variables related to somatic sexual dimorphism (SSD), demographic sexual dimorphism (DSD) and reproductive allocation. F > M: females outperform the males; M > F: males outperform the females; NS: no significant differences between sexes
 NSM > FF > M
Variables related to SSD
  Woody species2021 4
  Herbaceous species 8 620
  Woody species 5 9 1
  Herbaceous species 1 1 0
Variables related to DSD
 Vegetative propagation
  Woody species 4 0 2
  Herbaceous species 2 3 0
  Woody species 1 9 0
  Herbaceous species 0 4 3
 Frequency of reproduction
  Woody species 221 0
  Herbaceous species 1 7 0
Reproductive allocation
  Woody species 1 011
  Herbaceous species 2 0 9

Most of the species showed significant DSD (inline image = 19.565, P < 0.001; species showing-nonsignificant differences between sexes for demographic variables were scarce) and the case studies demonstrating SSD were also significantly more frequent than those showing no SSD (inline image = 5.500, P < 0.019, Fig. 3(a). When only the species showing sexual dimorphism in DSD were considered, the males usually outperformed females regardless of the growth form (inline image= 1.667, P= 0.197, Fig. 3(b). However, when only species showing significant SSD were considered, the sex that outperformed the other significantly depended on the growth form (inline image = 7.937, P < 0.001, Fig. 3(a). Males of woody species outperformed females in size and/or RGR, but females of herbaceous species were frequently larger than males and/or hermaphrodites (Table 3, Fig. 3a).

Figure 3.

Number of species showing different results for both types of sexual dimorphism (somatic intersexual dimorphism (SSD) and demographic intersexual dimorphism (DSD)) and considering the growth form (woody (open columns), herbaceous (closed columns)) separately. NS, no significant differences between sexes; M, males outperformed females; F, females outperformed males.

Phylogenetic effects might influence the interspecific patterns found in across-species comparisons. However, 12 out of 15 genera including two or more species in the Appendix 2 showed heterogeneous responses, that is, nonsignificant intersexual differences, advantage for males or for females reported within the same genus, which probably means that the phylogenetic is not the most important factor influencing the observed patterns.

These results were unexpected if we consider that the idea of females growing less than males as a consequence of higher investments in reproduction seems to be widely accepted. This might be understood if the most cited references were biased towards woody species. The results are difficult to explain in the context of costs of reproduction hypothesis. In fact, somatic and demographic costs cannot be uncoupled because they reflect a mixture of physiological and demographic processes (Bazzaz, 1997). In both herbaceous and perennials, males exhibited a demographic advantage in accordance with the hypothesis of the costs. In concordance, woody females resulted consistently smaller than males supporting the hypothesis of the costs of reproduction. By contrast, herbaceous perennials do not behave as they should and females were generally the larger sex, which did not fit the expectations of the cost of reproduction hypothesis. Nevertheless, these results may be explained in the context of sexual selection. In the woody species, the advantage of males exhibiting larger size lies on its success when competing for access to females, since larger males may produce more pollen and then increase their sired progeny (Willson, 1994; Emms et al., 1997). On the other hand, in the herbaceous, the female advantage of larger size lies in higher fecundity and, at the same time, smaller size in males might exhibit the advantage of earlier maturation with related demographic advantages of precocious reproduction (Andersson, 1994). If this is true, the evolution of sexual dimorphism in life history traits may be related to different selection pressures in woody and herbaceous plants.

Within related groups of organisms, males tend to be larger than females in larger species and females tend to be the larger sex in smaller species. This has been widely accepted in studies of sexual size dimorphism in animals and is known as Rensch's rule (Rensch, 1960; Abouheif & Fairbairn, 1997; Fairbairn, 1997; Colwell, 2000). If we consider that in general terms woody plants tend to be of larger size than herbaceous ones, Reensch's rule might provide an explanation to the patterns found here for dioecious plants, males tend to be the larger sex in woody plants and females tend to be the larger sex in herbaceous. This idea needs further development and exhaustive tests considering the phylogenetic effects, but unfortunately, data of life history-variables within phylogenetic groups of dioecious is still scant. However, the final conclusion is that observed sexual dimorphism in life history traits in dioecious plants is not fully explained by the hypothesis of costs of reproduction despite the fact that this idea seems to be widely accepted (Delph, 1999) and it remains a challenge to determine the relative importance of sexual selection, cost of reproduction and physiological differentiation hypotheses (Obeso & Retuerto, 2002).

VII. Compensation of the costs

One of the problems of costs of reproduction estimates is to what extent the costs may be compensated. Reproduction is usually less costly than indicated by simple trade-offs (but see Worley & Harder, 1996 for costs more costly than expected), which means that compensatory mechanisms exist (compensation hypothesis, Tuomi et al., 1983). Listed below are the mechanisms enabling compensation of the costs that, at the same time, may hamper their detection. Note that ‘resilient’ responses (tolerance strategies) to herbivory and defoliation (Belsky et al., 1993; Obeso & Grubb, 1993, 1994; Obeso, 1993a) exhibit some common response mechanisms with compensatory responses related to reproduction.

1. Compensatory growth and increased photosynthesis

In some cases, the carbon required for reproduction may be supplied by increased photosynthesis in nearby leaves (Gifford & Evans, 1981; Lehtilä & Syrjänen, 1995). There are many examples where a change in assimilate demand has resulted in a change in the rate of photosynthesis; but probably, the strongest cases of sink limited photosynthesis are provided by the relation between high photosynthetic rates and fruiting (Wardlaw, 1990). Accordingly, females of dioecious species exhibit higher photosynthetic rates than males (Dawson & Bliss, 1989; Dawson & Ehleringer, 1993; Lokker et al., 1994; Laporte & Delph, 1996). Within female Silene latifolia plants, individuals that were setting fruit had higher photosynthetic rates throughout the flowering period than those that were not setting fruit (Laporte & Delph, 1996).

In other cases, fruit-bearing branches exhibited lower photosynthetic rates than nonreproductive ones. In Ilex aquifolium, leaves on nonfruiting branches of female trees exhibited higher photosynthetic efficiency than leaves on fruiting branches (Obeso et al., 1998). Similarly, Karlsson (1994) observed a cost of reproduction in terms of CO2-assimilation capacity of the branch in Rhododendron lapponicum, suggesting that reproduction affects nutrient utilization by nearby leaves, causing a decrease in photosynthetic capacity by lowering the leaf nitrogen content.

Reproductive individuals of Pinguicula alpina showed a higher photosynthetic rate, in relation to nitrogen content, than nonreproductive individuals (Méndez & Karlsson, 1999). However, the presumed compensation is not achieved and there is additional evidence for demographic costs in this species (Svensson et al., 1993).

An increase in photosynthetic activity should be accompanied by an increase in resource uptake in order for a plant to function as a balanced system in terms of resource uptake and use (Bazzaz, 1997). However, much less attention has been given to roots and plant ability to adjust its below-ground activities mainly due to methodological difficulties.

2. Photosynthesis of reproductive structures

The photosynthetic capacity of reproductive structures was reviewed by Watson & Casper (1984) but there is also a number of recent works documenting the contribution of flowers and fruits to their own carbon maintenance (Jurik, 1985; Williams et al., 1985; Koppel et al., 1987; Reekie & Bazzaz, 1987a,b; Dick et al., 1990; Whiley et al., 1992; Galen et al., 1993; Ogawa et al., 1995; Ogawa & Takano, 1997; McDowell et al., 2000). Estimates of how flowers and fruits contribute to their own carbon budget range from 2.3% to 64.5%, but positive net photosynthesis is rare.

When carbon assimilation by green carpels is prevented the resultant achenes show a 17% reduction in mass in Ranunculus adoneus. Then, assimilation by reproductive structures compensates to some extent for energy costs associated with nectar and seed production (Galen et al., 1993). Positive rates of net photosynthesis were documented for reproductive structures at all stages of inflorescence development in the orchid Spiranthes cernua, and their contribution to seasonal carbon assimilation was estimated in 8.5% (Antlfinger & Wendel, 1997).

In Pseudotsuga menziesii var. glauca female cones had maximum instantaneous refixation rates of 54%, which, integrated over the season, offset 6% of their total carbon requirements, while male cones were completely dependent on vegetative tissues for carbon. Foliage near female cones had elevated photosynthesis during the early stages of cone development and consistently lower nitrogen concentration than foliage far from cones (McDowell et al., 2000).

3. Meristem availability and plant architecture

In some species, growth and reproduction may be limited primarily by the availability of meristems rather than carbon or nutrients (Watson, 1984; Geber, 1990; Bazzaz & Ackerly, 1992). Besides nutrient limitation, meristem limitation is a proximate mechanism that may generate trade-offs between life-history traits, then we can use the concept of meristem cost of reproduction (Watson, 1984). Thus, in Polygonum arenastrum, the developmental pattern of axillary meristem commitment to vegetative growth vs reproduction exhibited genetic variation that resulted in differences in the age of first reproduction, in age-specific fecundity and growth (Geber, 1990).

Plant growth and architecture result from the reiteration of modules and can be described as a consequence of the fates of individual buds (Lovett-Doust, 1989). Therefore, the availability of meristems is an important parameter affecting resource allocation and responses to reproductive costs. Meristems can be reproductive, vegetative or can remain quiescent. By developing vegetative and/or quiescent buds plants can increase the number of their carbon-acquiring structures to meet the carbon costs of reproduction. The obvious limitations to this kind of response are the time period in which carbon availability acts as a developmental constraint, the degree to which translocation of assimilate is architecturally constrained and the morphology or basic geometric properties of clonal growth (Watson & Casper, 1984; Lovett-Doust, 1989). Meristem commitment may be interpreted on the basis of modularity and sink/source regulation (Honkanen & Haukioja, 1998). The establishment of the bud hierarchy determines their growth potential and indirectly, the growth patterns of the plant. Within shoots, apical dominance leads to the unequal provisioning of individual sinks. However, reproductive structures are high priority sinks that might reduce the importance of apical dominance and the loss of apical dominance may lead to the activation of dormant lateral buds (Wardlaw, 1990), which would contribute to subsequent fruit filling. The trade-off between apical dominance and reproduction may be interpreted as the cost of reproduction, or alternatively, as the cost of apical dominance. This trade-off was demonstrated in several monocarpic Brassicaceae (Duff et al., 1999), but there was lack of evidence for a cost of apical dominance in Lythrum salicaria (Venecz & Aarssen, 1998).

These plastic responses may also be limited by architectural constraints (Diggle, 1995; Preston, 1999). Individual plant architecture is partly based on the genotype but it also has a strong ontogenetic component which reflects the developmental history of each plant (Tomlinson, 1983), and influences several plant functions, such as the reproduction by constraining the number of meristems available for flowering (Maillete, 1985; Geber, 1990; Schmitt, 1993; Preston, 1999). However, plasticity at other morphological levels, such as flowers per inflorescence or seeds per flower, could compensate for these constraints (Preston, 1999).

4. Timing of investment in growth and reproduction

The timing of reproduction may be responsible for differences between two Plantago species in the effects of reproduction on vegetative growth (Reekie & Bazzaz, 1992). Within the same species, the timing of investment in growth vs reproduction within a season may vary between the sexes (Delph, 1999). In many woody dioecious plants, females invest proportionately more in reproduction than males at fruiting, nevertheless during flowering males allocate more biomass to reproduction than females (Delph, 1990; Obeso, 1997). If there is a trade-off between investment in reproduction and vegetative growth, it should be expected that early in the growing season, during flowering, the vegetative growth of the females outperforms that of the males and contrarily, later in the growing season, at fruiting, the vegetative growth of the males exceeds that of the females (Ågren, 1988; Popp & Reinartz, 1988; Delph, 1990).

The morphs of the subdioecious Hebe subalpina differed in the timing of their growth, with each morph growing less during the phase in which its reproductive costs were highest (Delph, 1990). Nevertheless, despite the greater allocation by females to reproduction, the two morphs grew the same amount over the entire reproductive cycle. This may be a result of the fact that females produced more leaves earlier in the season, which enables them to accumulate more resources than males. The timing of development differed between vegetative (produced in late winter) and reproductive (produced in spring and early summer) ramets in the clonal perennials Lathyrus latifolius and L. sylvestris (Hossaert-McKey & Jarry, 1992).

5. Nutrient resorption from senescing structures

The costs of reproduction may be reduced if a significant proportion of the nutrients invested in reproduction can be recovered and reallocated to other functions once reproduction has occurred (Goldman & Willson, 1986; Chapin, 1989; Ashman, 1994). For instance, a substantial proportion of nitrogen and phosphorous invested in reproductive structures may be recovered in Sidalcea oregana ssp. spicata (Ashman, 1994) and Asclepias speciosa (Goldman & Wilson, 1986). By contrast, little resorptions occur from male flowers of Zea mays (Goldman & Wilson, 1986). In three species of the genus Pinguicula, resorption from senescent leaves was similar in reproductive and nonreproductive individuals (Eckstein & Karlsson, 2001).

6. Module specialization and physiological integration

The importance of considering the modular level has been emphasised in several studies (Lovett-Doust & Lovett-Doust, 1988; Obeso, 1997; Henriksson & Ruohomäki, 2000; Hasegawa & Takeda, 2001).

Resource allocation at lower modular levels might respond to local trade-offs (e.g. fruit production and shoot growth) but are probably integrated at higher modular levels (e.g. tree growth). It may be predicted that this integration probably depends on the degree of branch autonomy; however, there is still some discussion about the degree of autonomy of the different modules within a plant (Honkanen & Haukioja, 1998).

In this sense, if branches are autonomous, the costs might be detected at lower modular levels but might be compensated at higher modular levels by nonreproductive branches. Contrarily, if branches are only partially autonomous, and their resources are easily translocated between modules, the costs of reproduction might not be detected at branch level (Henriksson & Ruohomäki, 2000; Bañuelos, 2001), but should be at the whole plant level. The only way of integrating costs at different modular levels is considering the autonomy of these modules, but this information is scant. The current view of branch autonomy is that, under normal conditions, most branches are autonomous during the growing season; however, when local sources of photosynthate are eliminated, there may be some import of photonsynthate (Sprugel et al., 1991; Obeso, 1998).

Some experiments designed to examine the degree of branch autonomy considered the combined effect of experimental defoliation and branch girdling on the reproductive output and branch growth (Obeso, 1998). The extent of resource importation after defoliation of branches may be trivial (Stephenson, 1980), intermediate (Janzen, 1976) or even the main contribution to fruit maturation (Tuomi et al., 1988; Trueman & Turnbull, 1994; Obeso, 1998). In fact, Watson & Casper (1984) proposed that the degree of modular integration varies among species on a continuum from nearly complete autonomy to high interdependence.

Intra-plant regulation can emerge from simultaneous local and plant-wide regulation by hierarchically organized modules via sink/source competition for resources (Wardlaw, 1990; Wilson, 1990; Cannell & Dewar, 1994; Honkanen & Haukioja, 1998). This system regulates the development, growth and reproduction of hierarchically organized plants. Within branches and shoots, apical dominance leads to unequal provisioning of individual sinks, but different plant structures show differential sink strength, according to the following rank order: seeds > fleshy fruit parts > shoot apices > leaves > woody tissues (Wardlaw, 1990). The establishment of bud hierarchy determines the potential growth of shoots; however, actual shoot growth patterns are ultimately regulated by the existence of higher priority sinks, such as developing fruits. Some experiments considering local defoliations in trees have found a decrease in the mass and number of seeds produced by defoliated unit (Tuomi et al., 1988; Marquis, 1992; Ruohomäki et al., 1997), which demonstrate that they are supplied by local resources and do not depend on resources from the general storage compartments of the tree. Alternatively, experiments of defoliation in Ligustrum vulgare found no effects on fruit abortion (Obeso & Grubb, 1993) and experiments of defoliation and girdling in Ilex aquifolium (Obeso, 1998) and Rhamnus alpinus (Bañuelos, 2001) showed that fruit developing on defoliated branches were able to translocate resources from other branches or general storage compartments; whereas girdled branches were able to mature fruits when they were undefoliated. Branches behave as autonomous for fruit maturation in both species; however, when local sources are not available fruits are able to act as sinks from distant sources. Contrarily, shoot growth seems to depend exclusively on local resources. These results show certain branch autonomy but in order for the whole plant to function as a co-ordinated individual, the investment patterns seem to be regulated according to the priority of the sinks, and resources may be transferred between branches. Similar results were obtained in other woody dioecious, such as Macadamia integrifolia (Trueman & Turnbull, 1994).

This interpretation of simultaneous autonomy and dependence to obtain whole plant integration may be extended to patterns of shoot growth. If some of the resources invested in fruits are not from the shoot bearing the fruits but from other nonreproductive shoots, the cost of reproduction could not be detected at shoot level. However, this compensation is not possible at whole plant level.

7. Other mechanisms

In dioecious plants some mechanisms that may reduce the detrimental effects of the costs of reproduction have been reported. For instance, physiological differentiation between sexes may state the basis for habitat partitioning between the sexes (Freeman et al., 1976; Crawford & Balfour, 1983; Bierzychudek & Eckhart, 1988; Dawson & Ehleringer, 1993; Freeman et al., 1997). When males and females of dioecious species are spatially segregated, the more costly reproduction of females restricts them to habitats where they can best meet those costs (Dawson & Ehleringer, 1993).

Pre-reproductive females of Sipanura grandifolia (Nicotra, 1999) and Ilex aquifolium (Retuerto et al., 2000) grow more quickly than males. In general, females delay the initiation of reproduction up to a higher threshold of available resources than males (Delph, 1999). Both life-history traits may allow females to meet the specific demands associated with reproduction. During the reproductive phase, the females reduce their frequency of flowering compared to males (Delph, 1999), defer the negative effects to subsequent years (Obeso et al., 1998; Nicotra, 1999) and use their remaining resources more efficiently than males (Dawson & Ehleringer, 1993; Kohorn, 1994; Delph & Meagher, 1995, but see Retuerto et al., 2000).

VIII. Concluding comments and future perspectives

The cost of reproduction hypothesis has the potential to link physiology, life history and demography but we still lack a clear picture of how resource allocation to reproduction is reflected in other life history traits. The responses to investment in reproduction are extremely variable due to individual variation, plant size and previous history, plant architecture, modular integration, timing of reproduction habitat effects, stress level and resource availability, among others. In the same way, the life-history trait negatively affected by reproduction varies between and within species and we need further careful analysis in order to understand why a particular life history trait is affected but not others.

The use of dynamic estimates is absolutely necessary to accurately measure the costs of reproduction. The consideration of nutrient resorption and flower and fruit abortion not only leads to accurate estimates but also avoids the problem of determining the appropriate currency (Ashman, 1994). An experimental approach is necessary to modify the level of expression of the traits that at natural levels probably exhibit low variances due to selective pressures against their expression. The use of genetically controlled individuals (half-sibs, full-sibs, individuals from vegetative propagation) in different habitats is particularly recommended in order to examine the phenotypic plasticity of the life history traits.

Nevertheless, despite the fact that there is often considerable debate about methods, results based on phenotypic correlations, using both naturally occurring variation and experimentally modified reproductive investment, provide support for the expected trade-offs between reproductive investment and other life-history traits. In fact, the results presented here do not support the idea of Reznick (1985) considering that cases not detecting costs occur mainly in correlational studies. Furthermore, the use of path analysis may overcome some of the problems derived from phenotypical correlations and will be particularly useful in plant species not amenable to genetic studies.

While the effects of current reproduction on growth and future reproduction have been frequently studied, there is relative scarcity of data on trade-offs between current reproduction and somatic resource pool and survival. Particularly, there is lack of information about direct estimates of plant fitness, which is seldom assessed as life-time production of viable seeds (most of the plant fitness estimations measured some life-history parameters related to fitness).

In many plants, overall size and associated allometry may be sufficient to explain much of their reproductive behaviour and competitive ability. Unfortunately, our understanding of variations in patterns with plant size is far from adequate; however, the progress in this field will be particularly relevant due to the predictive capacity of the regression models derived from allometric hypotheses.

Compensatory growth and increased photosynthesis after defoliation or herbivory may compensate or even over-compensate for loss of tissue (i.e. Paige, 1992, 1999; Obeso & Grubb, 1993; Obeso, 1993a; Domínguez & Dirzo, 1994; Wegener & Odasz, 1997; Bakker & Loonen, 1998; Lennartsson et al., 1998; Meyer, 1998). In the same way, responses to herbivory are also limited by architectural constraints and limited availability of meristems (Maschinski & Whitham, 1989; Tuomi et al., 1994). The parallels between both responses seems to be related to the maintenance of the plant as a balanced system and tolerance strategies as responses to herbivore might be by-products of intraplant regulation rather than herbivore selected traits (Honkanen & Haukioja, 1998).

Ecological costs are usually neglected, and we should determine to what extent plants investing in reproduction increase their susceptibility to herbivory or decrease their competitive ability. Some studies need to be made to determine these ecological effects, which imply long-term monitoring of individuals in field conditions. Finally, to obtain significant progress in the development of the cost of reproduction hypothesis, we first need to understand the allometry of plant allocation patterns and its interactions with life-history traits in different habitats.


I thank R. Zamora, J.M. Gómez, A.G. Nicieza and two anonymous referees for their valuable comments. M.J. Bañuelos, M.B. García and J. Ehrlén provided unpublished data. While writing this review I was founded by the DGICYT, project MCYT-BOS2000-0451.


Table Appendix 1.  Trade-off among reproduction and several life history traits related to vegetative growth and propagation, somatic resource pool, survival and future reproduction. Dioecious species not included (see appendix 2). Expt: Experimental manipulation; NS: not significant
SpeciesExptReproductionTrade-off examinedPhenotypic correlationGenotypic correlationReference
Aesculus californica Fruit productionSubsequent growth and reproductionNegative Newell (1991 )
Agropyron repensYesReproductive structuresGrowthWeakWeakReekie & Bazzaz (1987c )
Agropyron repensYesSeed productionRhizome productionNSNSReekie (1991 )
Agropyron repensYesSeed productionGrowth of leaves, stems and rootsNegative Reekie (1991 )
Alnus hirsuta var. sibirica Fruiting shootsShoot productionNegative Hasegawa & Takeda (2001 )
Alnus viridis ssp. crispaYesReproductive structuresGrowth and future reproductionNegative Houle (2001 )
Arum italicum Reproductive structuresGrowth and clonal propagationNegative Méndez & Obeso (1993 )
Asarum canadense Reproductive structuresGrowth, storageNegative Muir (1995 )
Aspasia principissa Fruit productionShoot and inflorescence productionNegative Zimmerman & Aide (1989 )
Asphodelus albus Reproductive structuresClonal growthNegative Obeso (1993b )
Astrocaryum mexicanum Reproductive structuresFuture reproduction and survivalNegative Piñero et al. (1982 )
Betula papyrifera Male catkinsSubsequent shoot growthNegative Caesar & Macdonald (1984 )
Betula pubescens ssp. czerepanovii Male functionShoot growthNegative Henriksson & Ruohomäki (2000)
Betula pubescens ssp. czerepanovii Male functionFuture reproductionNS Henriksson & Ruohomäki (2000)
Betula pubescens ssp. tortuosa Male catkinsSubseq shoot and catkin productionNegative Karlsson et al. (1996 )
Bidens pilosaYesDeflowering, defruitingGrowthNegative Zobolo & van Staden (1999 )
Blandfordia grandiflora Fruit productionGrowth, survivalNS Ramsey (1997 )
Calathea ovandensisYesReproductive structuresGrowth, survival, future reproductionNS Horvitz & Schemske (1988 )
Clintonia borealis Reproductive structuresCLonal growthNS Pitelka et al. (1985 )
Cucumis meloYesFruit removalGrowthNegative El-Keblawy & Lovett-Doust (1996a )
Cucurbita pepoYesFruit removalGrowthNegative El-Keblawy & Lovett-Doust (1996b )
Cyclopogon cranichoidesYesFruit productionSubsequent growth and reproductionNS Calvo (1990 )
Cypripedium acauleYesReproductive structuresGrowth, flowering frequencyNegative Primack & Hall (1990 ); Primack et al . (1994)
Drosera rotundifolia Reproductive statusSomatic resource poolNegative Hemborg & Karlsson (1998a )
Eichhornia crassipes Inflorescence productionCLonal growthNegative Watson (1984 )
Empetrum nigrumYesFlower removalSubsequent flower bud productionNegative Mutikainen & Ojala (1993 )
Epidendrum ciliareYesFruit productionFuture reproductionNegative Ackerman & Montalvo (1990 )
Eurya japonica Flowering intensityShoot growthNegative Suzuki (2000 )
Geranium maculatumYesSeed productionSubsequent flowering, survivalNS Ågren & Willson (1994 )
Geranium sylvaticumYesSeed productionSubsequent floweringNegative Ågren & Willson (1994 )
Geranium sylvaticumYesSeed productionSurvivalNS Ågren & Willson (1994 )
Gladiolus sp.  Female functionSurvival, vegetative propagationNegativeNegativeRameau & Gouyon (1991 )
Gladiolus sp.  Male functionSurvival, vegetative propagationNegativeNegativeRameau & Gouyon (1991 )
Gymnadenia conopsea Reproductive statusSomatic resource poolNegative Hemborg & Karlsson (1998a )
Helianthus tuberosusYesBud removalVegetative propagationNegative Westley (1993 )
Lagenaria sicerariaYesMale functionBiomass, node numberNegative Delesalle & Mooreside (1995 )
Lagenaria sicerariaYesFruit productionBiomass, node numberNegative Delesalle & Mooreside (1995 )
Lathyrus vernus Reproductive structuresGrowth, survival, future reproductionNS Ehrlén & Groenendael (2001 )
Lathyrus vernusYesFlower removalGrowth, subsequent fruit setNegative Ehrlén & Groenendael (2001 )
Lotus corniculatus Fruit productionShoot growthNegative Briggs & Schultz (1990 )
Lychnis flos-cuculi Reproductive structuresGrowthNegative/NSNegat/NSBiere (1995 )
Lythrum salicariaYesInflorescence removalSubsequent reproductionNS Venecz & Aarssen (1998 )
Oenothera biennisYesEarly reproductionGrowthNegative Reekie & Reekie (1991 )
Pinguicula alpinaYesReproductive structuresSomatic resource poolNegative Thorén et al. (1996 )
Pinguicula alpina Reproductive structuresResource pool, future reproductionNS/positive Karlsson et al. (1990 )
Pinguicula villosaYesReproductive structuresSomatic resource poolNegative Karlsson et al. (1990 ); Thorén et al. (1996)
Pinguicula vulgarisYesReproductive structuresSomatic resource poolNegative Karlsson et al. (1990 ); Thorén et al. (1996)
Pinguicula vulgaris Reproductive structuresGrowth, vegetative propagationNegative Worley & Harder (1996 )
Plantago majorYesSeed productionGrowth, subsequent reproductionNegative Reekie (1998 )
Plantago majorYesSeed productionGrowthNegativeNegat/positReekie & Bazzaz (1992 )
Plantago rugeliiYesSeed productionGrowthPositiveNegat/positReekie & Bazzaz (1992 )
Plantago lanceolataYesReproductive structuresGrowthNegative Antonovics (1980 )
Poa annua Reproductive structuresGrowth, survivalNegative Law (1979 )
Podophyllum peltatum Fruit productionFuture reproduction and survivalNegative Sohn & Policansky (1977 )
Polemonium viscosum Early reproductionSurvivalNegativeNegativeGalen (1996 )
Polygonum arenastrum Seed productionGrowthNegativeNegativeGeber (1990 )
Pinus banksiana Cone productionGrowthPositive Despland & Houle (1997 )
Primula verisYesFlower removalSubsequent reproductionNegative/NS Syrjänen & Lehtilä (1993 )
Primula verisYesSeed productionGrowth, survival, future reproductionNS García & Ehrlén, unpublished data
Pseudotsuga menziesii Seed-cone productionAnnual ring widthNegativeNegativeEl-Kassaby & Barclay (1992 )
Ranunculus acrisYesIncreased reproductionFuture reproductionNegative Hemborg (1998b )
Ranunculus nivalis Reproductive statusSomatic resource poolNeg/posit Hemborg & Karlsson (1998a )
Rhododendron lapponicumYesReproductive statusPhotosynthetic capacityNegative Karlsson et al. (1996 )
Rubus caesius Inflorescence productionVegetative propagationNegative Abrahamson (1975 )
Rubus vestitus Inflorescence productionVegetative propagationNegative Kirby (1980 )
Rumex acetosellaYesReproductive structuresVegetative propagationNegativeNegativeHoussard & Escarré (1995 )
Salix alaxensis Reproductive structuresSubsequent growthNegative Fox & Stevens (1991 )
Salix alaxensisYesBud removalGrowthNS Fox (1995 )
Scrophularia nodosaYesReproductive structuresRhizome productionNegative Baalen et al. (1990 )
Senecio keniodendron Reproductive structuresFuture reproduction and survivalNegative Smith & Young (1982 )
Sidalcea oregana ssp. spicataYesReproductive structuresSubsequent reproductionNegative Ashman (1994 )
Silene virginicaYesSeed productionGrowth, survival, future reproductionNS Dudash & Fenster (1997 )
Spiranthes cernua Flower productionSubsequent reproductionNS Antlfinger & Wendel (1997 )
Tipularia discolor Reproductive structuresCorm size, leaf areaNegative Snow & Whigham (1989 )
Toefieldia pusilla Reproductive statusSomatic resource poolNegative Hemborg & Karlsson (1998a )
Tolumnia variegataYesFruit productionGrowth and future reproductionWeak Calvo (1993 )
Trillium grandiflorumYesReproductive structuresStorage, survivalNegative Lubbers & Lechowicz (1989 )
Tripsacum dactyloides Seed productionSubsequent growth rateNSNSJackson & Dewald (1994 )
Tripsacum dactyloidesYesStalk removalSubsequent growth rateNegative Jackson & Dewald (1994 )
Trollius europaeus Reproductive statusSomatic resource poolNeg/posit Hemborg & Karlsson (1998a )
Viscaria vulgarisYesIncreased reproductionSubsequent reproductionNS Jennersten (1991 )
  Flower removal    
Table Appendix 2.  Species included in the analyses. BS = breeding system (D = dioecious; SD = subdioecious; GD = gynodioecious; T = trioecious); GF = growth form (T = tree, S = shrub, HP = herbaceous perennial, A = annual); SR = sex ratio; RA = reproductive allocation; S = size; B = biomass, RGR = relative growth rate, RP = ramet production; SV = survival, RF = reproductive frequency. M = males outperform females; F = females outperform males; H = hermaphrodites outperform females; ns = nonsignificant intersexual differences
Acer negundoDT    0   Willson (1986 )
Acer negundoDT    M   Jing & Coley (1990 )
Acer negundoDT    F   Ramp & Stephenson (1988 )
Acer negundoDT        Dawson & Ehleringer (1993 )
 Dry site  MF  M    
 Wet site  FF  F    
Acer rufinerveDTM M   M Matsui (1995 )
Aporusia microstachyaDTM      MThomas & LaFrankie (1993 )
Aporusia symplocoidesDT?      0Thomas & LaFrankie (1993 )
Baccaurea parvifloraDTM      MThomas & LaFrankie (1993 )
Baccaurea racemosaDT?      0Thomas & LaFrankie (1993 )
Componeura spruceiDT       MBullock (1982 )
Chamaedorea bartlingianaDT    M  MAtaroff & Schwarzkopf (1992 )
Chamaedorea tepejiloteDT0F  M  MOyama (1990 )
Cordia collococcaDTM ?   MMOpler & Bawa (1978 )
Fraxinus lanceolataDT  M M   Cited in Sakai & Burris (1985)
Fraxinus pennsylvanicaDT  F  F  Davidson & Remphrey (1990 )
Gingko bilobaDT  M     Snow (1942 )
Guarea rhopalocarpaDT       MBullock et al. (1983 )
Ilex aquifoliumDT  M M   Obeso (1997 )
Ilex montanaDT  0    MCavigelli et al. (1986 )
Ilex opacaDT       MClark & Orton (1967 )
Jacaratia dolichaulaDT       MBullock & Bawa (1981 )
Juniperus communisDT-S      M Cited in Lloyd & Webb (1977)
Juniperus communisDT-S0 0 0   Marion & Houle (1996 )
Juniperus excelsaDT  M     Fisher & Gardner (1995 )
Juniperus virginianaDT  M 0   Vasiliauskas & Aarssen (1992 )
Lagarostrobos frankliniiD,ST0 0     Shapcott et al. (1995 )
Lithrea causticaDT-S  M     Hoffman & Alliende (1984 )
Lodoicea maldivicaDT  M     Silvertown (1987 )
Macrozamia communisDT  0     Ornduff (1990 )
Myristica insipidaDT  M     Cited in Armstrong & Irvine (1989)
Myristica insipidaDTMF0    MArmstrong & Irvine (1989 )
Nyssa aquaticaDT  MF    Shea et al. (1993 )
Nyssa sylvaticaDTMF0    MCipollini & Stiles (1991 )
Peumus boldusDT-S  M    MHoffman & Alliende (1984 )
Pisonia macranthocarpaDT0 0     Opler & Bawa (1978 )
Populus deltoidesDT  M     Cited in Lloyd & Webb (1977)
Populus grandidentataDT  0  0  Sakai & Sharik (1988 )
Populus nigraDT  M     Chailakhyan & Khrianin (1987 )
Populus tremuloidesDT    0   Mitton & Grant (1980 )
Populus tremuloidesDT  F,0 0F  Sakai & Burris (1985 )
Populus tremuloidesDT  0     Cited in Lloyd & Webb (1977)
Populus tremuloidesDT    F   Grant & Mitton (1979 )
Randia spinosaDTM ?   MMOpler & Bawa (1978 )
Randia subcordataDTM ?   MMOpler & Bawa (1978 )
Taxus baccataDT  M     Snow (1942 )
Trichila cuneataDT0 ?    MOpler & Bawa (1978 )
Triplaris americanaDTF 0     Opler & Bawa (1978 )
Triplaris americanaDTF 0     Melampy & Howe (1977 )
Zanthoxylum setulosumDTM 0     Opler & Bawa (1978 )
Baccharis halimifoliaDS  M M   Krischik & Denno (1990 )
Ceratiola ericoidesDS0 0     Gibson & Menges (1994 )
Corema conradiiDSMF  M   Rocheleau & Houle (2001 )
Gardenia actinocarpaDSM 0    MOsunkoya (1999 )
Hebe subalpinaGDS    0   Delph (1990 )
Laretia acaulisDS  M    MHoffman & Alliende (1984 )
Lindera benzoinDS F  M   Cipollini & Whigham (1994 )
 DSM   M M Cipollini et al. (1994 )
Ochrademus baccatusGDS0 0     Wolfe & Shmida (1997 )
Oemleria cerasiformisDSMFM M MMAllen & Antos (1988, 1993 )
Osyris quadripartitaDS  M,F     Herrera (1988 )
Pistacia lentiscusDS0 0     Verdú & García-Fayos (1998 )
Rhamnus alpinusDSMF0 0   Bañuelos (2001 )
Rhus typhinaDS  0 0 MMLovett-Doust & Lovett-Doust (1988 )
Rhus typhinaDS    M   Luken (1987 )
Salix arcticaDS        Dawson & Bliss (1989 )
 Wet site      F    
 Dry site      M    
Salix viminalisDSF  0    Ahman (1997 )
Salix cinereaDS  0     Alliende & Harper (1989 )
Schriedea globosaSDS     00 Sakai & Weller (1991 )
Simmondsia chinensisDS F    M Waser (1984 )
Thymelaea hirsutaTS        Ramadan et al. (1994 )
Favourable habitats   F       
Unfavourable habitats   M       
Xanthoxylum americanumDS        Popp & Reinartz (1988 )
 Open field   F M   M 
 Woods   0 0   0 
Zamia skinneriDS  F     Clark & Clark (1987 )
Zamia integrifoliaDS0FM    MOrnduff (1996 )
Aciphylla aureaDHP       MCited in Lloyd & Webb (1977)
Aciphylla scott-thomsoniiDHP  M     Cited in Lloyd & Webb (1977)
Aciphylla subflabellataDHP       MCited in Lloyd & Webb (1977)
Antennaria dioicaDHP  F     Cited in Lloyd & Webb (1977)
Aralia nudicaulisDHP     M  Barret (1981 )
Aralia nudicaulisDHP       MCited in Lloyd & Webb (1977)
Asparagus acutifoliusDHP       MCited in Lloyd & Webb (1977)
Asparagus officinalisDHP   F MMMCited in Lloyd & Webb (1977)
Borderea pyrenaicaDHP  M     García & Antor (1995 )
Buchloe dactyloidesDHP00 0    Quinn (1991 )
Carex pictaDHP00 0 0 0Delph et al. (1993 )
Catasetum viridiflavumSDHP0FFF    Zimmerman (1991 )
Chamaelirium luteumDHP  F   M Meagher & Antonovics (1982 )
Chamaelirium luteumDHPMF     MMeagher (1984 )
Chionographis japonicaGDHP  0     Maki (1996 )
Cucurbita foetidissimaDHP  F     Kohn (1989 )
Ecballium elaterium dioicumDHP  0   0,M Costich (1995 )
Fragaria chiloensisDHP  MMM   Hancock & Bringhurst (1980 )
Glecoma hederaceaGDHP  H   H Slade & Hutchings (1989 )
Limnanthes douglasiiGDHP   F    Kesseli & Jain (1984 )
Mercurialis perennisDHP   F    Cited in Lloyd & Webb (1977)
Pachycereus pringleyGDHP  0     Fleming et al. (1994 )
Petasites niveusDHP   F    Cited in Lloyd & Webb (1977)
Phoradendron juniperinumDPMFF   F Dawson et al. (1990 )
Phyllospadix torreyiDHPF  F    Williams (1995 )
Rubus chamaemorusDHP      M Ågren (1987 )
Rumex acetosaDHP FM     Putwain & Harper (1972 )
Rumex acetosellaDHPF    M  Lovett-Doust & Lovett-Doust (1987 )
Rumex hastatulusDHP0FF     Conn & Blum (1981 )
Sidalcea oreganaGDHP   00   Ashman (1994 )
Silene acaulisGDHP  0     Hermanutz & Innes (1994 )
Silene albaDHPFFFF    Armstrong & Irvine (1989 )
Silene dioicaDHP?FF    MElmqvist & Gardfjell (1988 )
Silene latifoliaDHPFF F    Gehring & Linhart (1993 )
Silene latifoliaDHP F F  F Delph & Meagher (1995 )
Silene roemeriDHP   F    Cited in Lloyd & Webb (1977)
Spinifex sericeusDHP  0     Connor (1984 )
Urtica dioicaDHP   F    Mutikainen et al. (1994 )
Vallismeria americanaDHP  MF 0  Lovett Doust & Laporte (1991 )
Amaranthus cannabinusDAM M     Bram & Quinn (1995 )
Cannabis sativaDAF F     van der Werf & van der Berg (1995)
Mercurialis annuaDA  F     Cited in Lloyd & Webb (1977)
Phacelia linearisGDA   F    Eckart (1992 )
Spinacia oleraceaDAM,0  F    Onyekwelu & Harper (1979 )