The advantage of early flowering in the spring ephemeral annual plant Floerkea proserpinacoides


  • Gilles Houle

    Corresponding author
    1. Département de Biologie, Université Laval, Sainte-Foy, Québec, Canada G1K 7P4
      Author for correspondence: Gilles HouleTel: +1 418 656 2131×2691Fax: +1 418 656 2043Email:
    Search for more papers by this author

Author for correspondence: Gilles HouleTel: +1 418 656 2131×2691Fax: +1 418 656 2043Email:


• Internal conflicts in the allocation of resources lead to trade-offs or compromises that can influence fitness and the evolution of specific life-history traits.

• Here, the reproductive schedule of Floerkea proserpinacoides plants, a spring ephemeral annual of the deciduous forests of eastern North America, was manipulated to determine the presence of potential conflicts between growth and reproduction, and the advantage of early reproduction.

• Plants on which flowering was delayed had a greater relative growth rate (RGR) and vegetative biomass than control plants after 8 wks (when control plants started to senesce). Delayed-flowering plants took 1 week longer to senesce, which allowed them to mature almost as many seeds as did control plants in 8 wks.

• However, late flowering in F. proserpinacoides may be maladaptive. With reduced light availability through canopy closure, early flowering plants can rapidly divert most of the resources accumulated in the stem and leaves to seed maturation. Thus, the advantage of maintaining a higher RGR by delaying flowering is associated with the greater disadvantage of necessitating more time to mature seeds.


The resources available to an organism are limited and must be allocated to different, competing functions such as maintenance, growth and reproduction (Cody, 1966; Williams, 1966). Internal conflicts in the allocation of resources lead to trade-offs or compromises that are susceptible to influence fitness and the evolution of specific life-history traits (Gadgil & Bossert, 1970). For example, current reproduction may reduce growth and survival and hence reduce the probability of future reproduction. If this was the case, one solution could be to decrease current reproduction so as to increase survival and growth and thereby optimize life-long fitness. Natural selection should favor those genotypes with a solution to their conflicts that optimizes their fitness in the context of their habitat (Schaffer, 1974; Stearns, 1976).

In a large number of annual plants, reproduction is induced only after a given period of vegetative growth (Cohen, 1971; Paltridge & Denholm, 1974). Conflicts between functions are thus reduced. However, growth and reproduction co-occur to various degrees in some annuals (King & Roughgarden, 1982a,b; Fox, 1992b). One of the most extreme situations exists in Floerkea proserpinacoides, an ephemeral annual plant of deciduous forests, which starts flowering only 2–3 wks after emergence and continues to do so until senescence, which occurs 8–9 wks after emergence (Struik, 1965). An interpretation of this unusual behavior is that an annual plant may not have the time to reproduce before resource availability deteriorates with canopy closure, unless reproduction begins early in the life-cycle (McKenna & Houle, 2000a). Indeed, Smith (1984) found a disadvantage to late flowering in F. proserpinacoides– associated with the production of a basal branch instead of an axillary flower – using a modeling approach.

To determine the presence of potential conflicts between growth and reproduction and to estimate the advantage of early reproduction in F. proserpinacoides, I manipulated the reproductive schedule of a series of plants and compared their proximate fitness (seed production) with that of controls in a natural setting. I expected late-flowering plants to maintain a higher growth rate, reach a higher biomass, attain a greater longevity, and have a higher probability of survival and an overall higher reproductive success than control (i.e. early flowering) plants.

Study species

Floerkea proserpinacoides (Limnanthaceae) is a spring ephemeral annual of the eastern deciduous forest, ranging from West Virginia, North Carolina, Tennessee and Missouri to Ontario, Québec and Nova Scotia (Fernald, 1950; Gleason & Cronquist, 1963; Steyermark, 1963). It is a small plant (< 30 cm high) with alternate leaves typically comprising three to five leaflets. As the stem grows, the plant becomes decumbent and develops adventitious roots at the base (Russell, 1919). The plant produces small (approx. 3 mm diameter), autogamous flowers, one per node, except at the lowest nodes (Ornduff & Crovello, 1968; Smith, 1983). The petals are minute, white and caducous, while the sepals are somewhat larger, green and persistent. Each flower bears one to three practically distinct carpels. The fruit, which matures as the plant starts to senescence, is a small, egg-shape nutlet (approx. 2.0–3.5 mm long) bearing tubercles on its distal end. Both the green sepals and the flower peduncle progressively enlarge after anthesis.

The seeds of F. proserpinacoides germinate from early December to February (Baskin et al., 1988; Houle et al., 1998). Root growth occurs during the winter, but seedlings do not emerge until the following spring (April in Québec). Flowers are produced continuously during the growth period and may appear as early as when the plant is 2–3 wks old. Senescence typically occurs within 8–9 wks after emergence, as the canopy closes (Struik, 1965; Houle et al., 1998).

Although Smith (1983, 1984) reports that F. proserpinacoides plants often branch at the lower nodes, we have only infrequently observed branched plants in the 20 natural populations in which we have worked over the last 7 yrs in southern Québec (McKenna & Houle, 2000b; Houle et al., 2001). However, plants often branch when exposed to high temperatures, high light and high nutrient availability in the greenhouse and the growth chamber (G. Houle, personal observation).

Materials and Methods

A total of approximately 1000 seedlings of F. proserpinacoides were collected from several locations in a deciduous forest, at Île aux Grues, Québec, Canada (47°02′ N, 70°35′ W), on April 26, 2001 (approximately 2-wk-old-seedlings). From these, 110 seedlings of equal size (first leaf approx. 1.0 cm long) were selected. These were carefully washed to disentangle their root system from that of neighbors. Twenty of these seedlings were used to provide data on initial biomass: seedlings were separated into leaves, stems and roots, dried at 75°C for 24 h and weighed. The other 90 seedlings were transplanted individually (May 1, 2001) into 170 cm3 peat pots filled with a commercial soil mixture (Tri-Sol, Jardins Hamel, Ste-Foy, QC, Canada). A clear acetate tube with perforations to allow for air circulation was placed around each plant to ensure that seeds would not fall outside the pot. The pots were then placed into the ground in 45 bocks of two, in a deciduous forest on the campus of Université Laval, Québec. Blocks were positioned every 25 cm, along an 11-m long transect. Within a block, pots were approx. 10 cm apart.

Forest composition at the transplantation site was comparable to the Île aux Grues site, with sugar maple (Acer saccharum Marsh.) and white ash (Fraxinus americana L.) as the dominant species, and white birch (Betula papyrifera Marsh.), yellow birch (Betula alleghaniensis Britton), red maple (Acer rubrum L.) and red oak (Quercus rubra L.) as associate species.

Treatment application began after 10 d of acclimation (May 10, 2001). Plants were then approximately 4-wks-old-and had their first flower. There were two treatments: a control (C) and a flower-removal (–F) treatment. On –F plants, all flowers were cut with their peduncle every 3.5 d; cut flowers were kept, dried at 75°C for 24 h and then weighed. When flowers were cut, the acetate tube around each plant was removed and then replaced after the treatment. Control plants were left intact, but their acetate tube was also removed and replaced every 3.5 d. Flower removal continued until May 28, 2001. The plants were then allowed another 10 d of growth, after which the pots from 30 blocks were removed from the forest and brought to the laboratory (June 7). At this time, control plants had started to senesce and tree canopy development was completed (plants were approximately 8 wks old). Plants were separated into roots, stem (including leaf petioles), leaves, flowers, and seeds. Leaf area was determined with a leaf-area meter (CI-202, CID Inc., Vancouver, WA, USA). Flowers present at harvesting were counted. The soil surface in the pots was searched for seeds that would have fallen from the plants and the number of seeds was noted for each plant. All plant material was dried in an oven at 75°C for 24 h and then weighed.

Plants from the remaining 15 blocks were left in the field and checked daily for death (date of death noted) until all plants had died. Then, pots were removed from the forest and seeds were retrieved from the dead plants and the soil surface in the pots.

Statistical analyses

Differences between treatments in total biomass, leaf surface, and number of flowers and seeds produced by those plants collected on June 7, 2001 were tested first in a multiple analysis of variance (manova, Wilks’λ). Because there was an overall difference between treatments (see the Results section), a paired t-test was performed on each variable to detect specific differences between treatments. A manova was similarly performed on the four biomass ratios (stem ratio, stem mass : total biomass; leaf ratio, leaf mass : total biomass; root ratio, root mass : total biomass; reproductive ratio, reproductive biomass : total biomass). Again, because there was an overall difference between treatments (see the Results section), specific differences between treatments for each variable were tested with paired t-tests. The number of mature seeds (≥ 1.5 mm long) and longevity of those plants left in the field until senescence were analysed with paired t-tests.

Relative growth rate (RGR) over the 38 d of growth for the first series of plants was calculated as RGR = ln(W2 − W1)/38, with W2 as the biomass at harvest and W1 as the mean initial seedling biomass (i.e. 2.3 mg; see the Results section). Leaf area ratio (LAR) was calculated as leaf area divided by total biomass; unit leaf rate (ULR) was calculated as RGR/LAR. Two series of RGR-related variables were calculated, one with vegetative biomass only (RGRv, LARv and ULRv) and one with total biomass (i.e. including reproductive tissues: RGRt, LARt and ULRt). Specific leaf area (SLA) was calculated as leaf area divided by leaf mass. The RGR, LAR, ULR and SLA were compared between control and –F plants with manovas (one for the variables calculated with vegetative biomass only, and one for those calculated with total biomass). Because the manovas indicated significant overall differences, paired t-tests were used to detect specific differences between treatments for each variable.


Seedling biomass at transplantation was 2.3 ± 0.1 mg and represented only 2% of the total plant mass at harvest. All the plants from the first group (30 blocks) were alive at harvest, on June 7: there was thus no difference in plant survival between treatments. The result of the manova on total biomass, leaf area and flower and seed number at harvest indicated that there was a significant global difference between treatments (P = 0.0001). Consequently, paired t-tests were used to test for treatment difference for each variable. These tests indicated that total biomass, leaf area and number of flowers at harvest did not differ between control and –F plants (Fig. 1; P = 0.9276, P = 0.6004, and P = 0.2138, respectively). However, 9.1 ± 0.4 flowers representing a biomass of 2.8 ± 0.1 mg had been removed from –F plants during treatment application and before harvest. Thus, –F plants produced a total of 19.2 ± 0.9 flowers and 108.6 ± 7.4 mg of biomass per plant (P = 0.0001 and P = 0.8978, respectively, for comparisons with control plants). At harvest, –F plants had produced approximately half the number of seeds of control plants (Fig 1; P = 0.0001); however, none of the seeds of the –F plants were mature. In sharp contrast, 77% of the seeds of control plants were mature at harvest. It is noteworthy that none of the plants had branched at the base.

Figure 1.

Total biomass and leaf area, and flower and seed number of Floerkea proserpinacoides plants according to treatment (open bar, control; shaded bar, flower-eradication treatment). Values are mean + SE. The different letters indicate a significant difference between treatments for the variable ‘seed number’ (P ≤ 0.05). There is no significant difference between treatments for the other three variables (P > 0.05).

The result of the manova on the four biomass ratios was significant, indicating an overall difference between treatments (P = 0.0001). Consequently, paired t-tests were used to test each variable for difference between treatments. These tests indicated that root ratio only differed marginally between control and –F plants (Table 1). Stem and leaf ratios were higher for the –F plants than for the control plants. However, reproductive ratio was four times higher in control plants than in –F plants (Table 1), essentially because control plants produced more seeds.

Table 1.  Biomass ratios1 of Floerkea proserpinacoides plants according to treatment and P-values for difference between flower-removal treatments (mean ± SE)
VariableTreatment control−F2P3
  1. 1 Root ratio, root mass/total biomass; stem ratio, stem and petiole mass/total biomass; leaf ratio, leaf mass/total biomass; reproductive ratio, reproductive mass/total biomass. 2 −F, flower removal treatment (see text for detail). 3 Values in bold type are significant at P ≤ 0.05 (paired t-test).

Root ratio0.027 ± 0.0020.033 ± 0.0030.0547
Stem ratio0.405 ± 0.0090.647 ± 0.0070.0001
Leaf ratio0.149 ± 0.0050.217 ± 0.0080.0001
Reproductive ratio0.419 ± 0.0100.102 ± 0.0070.0001

The results of the manovas for the four variables associated with growth (RGR, ULR, LAR and SLA) indicated a significant treatment effect at P = 0.0001 for both the series of variables calculated with vegetative biomass only (RGRv, ULRv and LARv) and that calculated with vegetative and reproductive biomass combined (RGRt, ULRt, and LARt). Consequently, paired t-tests were used to test each variable for difference between treatments. It was found that RGRv was approximately 12% higher for –F than for control plants (Table 2); LARv was 37% higher, but ULRv was 31% lower for control than for –F plants (Table 2). Specific leaf area was 21% lower for –F plants than for controls (Table 2). It was found that RGRt, LARt and ULRt did not differ significantly between treatments (Table 2).

Table 2.  Relative growth rate and associated variables for Floerkea proserpinacoides plants according to treatment and P-values for difference between flower-removal treatments (mean ± SE)
Variable1Treatment control−F2P3
  • 1

    RGR, relative growth rate; ULR, unit leaf rate; LAR, leaf area ratio; RGRv, ULRv, and LARv are variables calculated with vegetative biomass only; RGRt, ULRt, and LARt are variables calculated with both vegetative and reproductive biomass.

  • 2

    −F, flower-removal treatment.

  • 3

    Values in bold type are significant at P ≤ 0.05 (paired t-test).

RGRv (d−1)0.1047 ± 0.00260.1171 ± 0.00200.0016
ULRv (g cm−2 d−1)0.0009 ± 0.00010.0013 ± 0.00010.0023
LARv (cm2 g−1)137.16 ± 10.46 99.89 ± 5.930.0033
RGRt (d−1)0.1197 ± 0.00230.1201 ± 0.00200.9055
ULRt (g cm−2 d−1)0.0019 ± 0.00020.0015 ± 0.00010.0779
LARt (cm2 g−1) 80.10 ± 6.30 90.17 ± 5.620.2233
Specific leaf area (cm2 g−1)520.89 ± 30.64409.68 ± 17.450.0062

For those plants left to senesce naturally (15 blocks), longevity was significantly higher for the –F plants than for the control plants (approx. 20% higher (mean ± SE): 41.0 ± 0.7 d and 49.1 ± 1.5 d, for control and –F plants, respectively; P = 0.0007). Among those, –F plants produced somewhat fewer mature seeds than control plants (approx. 20% fewer), but the difference was not significant (mean ± SE: 9.0 ± 2.7 and 7.1 ± 2.7 seeds per plant, for control and −F plants, respectively; P = 0.6432).


When the reproductive schedule of F. proserpinacoides was experimentally modified, late-reproducing individuals maintained a higher relative growth rate (RGRv), but they did not have a higher survival probability and did not bear more flowers (data of June 7). The –F plants had a greater longevity but did not produce more mature seeds than control plants when left to senesce naturally. Thus, although there is a ‘somatic’ cost to early reproduction in F. proserpinacoides, it does not appear to have any significant effect on proximate fitness (Fox, 1992a,b).

In F. proserpinacoides, seed maturation occurs mostly during the later part of the life cycle, at a time when growth has practically stopped. Indeed, a study in which plants were sampled weekly from early May to late June in 1997, in the natural population from which the seedlings used in this experiment were taken, showed that no mature nutlets were present on the plants before the middle of June (Houle et al., 2001). Nutlet maturation thus essentially occurs during the last 1–2 wks of the plant life-cycle. Because flower production reduces vegetative growth by only a small percentage, flowering can start relatively early in the plants’ life. However, seed maturation, a process that is much more costly (Stephenson, 1981), occurs mostly towards the end of the life cycle.

Although ULRv (unit leaf rate calculated on vegetative biomass only) was lower in control than in –F plants when all biomass components (both vegetative and reproductive) were included in the calculation (ULRt), –F and control plants had a similar unit leaf rate. LARv was higher in control plants, but LARt was higher in –F plants although not significantly so. Yet, for both control and –F plants, LAR decreased when reproductive tissues were included in the calculations. Obviously, this decrease was more pronounced in control plants because of a higher investment into seeds. Nevertheless, RGRt for control and –F plants was similar, mostly because ULRt increased dramatically in control plants when reproductive biomass was included. The –F plants produced leaves with a much lower SLA: this was essentially associated with their higher leaf mass (data not shown), because their total leaf area, although somewhat greater, did not differ significantly from that of control plants. Thus, leaves of –F plants may have been used as storage organs, along with the stem (see stem ratio earlier in this section), for subsequent seed maturation.

Are there any advantages for F. proserpinacoides to flower so early in life and to mature seeds so late (McKenna & Houle, 2000a; Houle et al., 2001)? There might be several.

Flowers are very small and have green sepals that continue to grow after flowering, attaining three to five times their initial size. They may contribute to the plant’s carbon budget and, consequently, support at least part of the respiratory cost of the reproductive structures (Bazzaz et al., 1979). Indeed, in the present experiment, when RGR was calculated on overall plant biomass (including reproductive biomass, RGRt), there was no difference between control and –F plants. Consequently, flowering and nutlet initiation may not represent a significant carbon drain for the plant, and early flowering may have little effect on overall growth.

Early production of flowers may also provide the plant with a reserve of ovaries ready to respond to resource pulses or to resource release following the death of neighbors (Stephenson, 1981). Resources accumulated in the leaves and stem may be rapidly channeled towards seed maturation (Bloom et al., 1985) when ambient conditions start to deteriorate with canopy closure. Indeed, decreasing photosynthetic photon flux densities as a result of canopy closure decreases carbon gain in spring ephemerals, at a time when increasing night temperature increases night respiration. Under such conditions, spring ephemerals become unable to maintain a positive carbon gain and they senesce (McKenna & Houle, 2000a). However, the presence of immature nutlets on the plants appears somehow to inhibit further flowering, as demonstrated by the much higher overall number of flowers produced by the –F plants in the present study. Consequently, my results suggest that F. proserpinacoides’ response to resource pulses or resource release following the death of neighbors may occur more through increased seed set within flowers (approx. 60% under natural conditions, Houle et al., 2001) than through increased flowering (e.g. through branching). As suggested by Smith (1984), the rate of canopy closure may be important in determining the appropriate reproductive strategy, i.e. increasing seed set or increasing flowering. Latitudinal differences in the rate of canopy closure may be responsible for latitudinal differences reported in strategy among natural populations of F. proserpinacoides (Smith, 1983, 1984; Houle et al., 2001; this study).

Because in F. proserpinacoides flowers are borne individually in the axis of the leaves (except the lowest leaves), flower production implies leaf production (Smith, 1983). Thus, an investment in flowering cannot occur without an investment in growth (Cohen, 1971; King & Roughgarden, 1982b; Watson & Casper, 1984). Once the plant start to flower, it does so continuously (every subsequent node bears a flower). Consequently, architectural constraints may link growth and reproduction in F. proserpinacoides (Watson, 1984; Watkinson & White, 1985; Preston, 1998). Yet, contrary to what has been observed in other plant species (Geber, 1990), the fate of axillary meristem is fixed (giving a flower), although basal axillary meristems may produce flower-bearing branches (Smith, 1983, 1984). However, in the present experiment, none of the plants (control or –F) branched at the lower nodes.

Nevertheless, experimentally delayed flowering increased longevity but did not increase proximate fitness in F. proserpinacoides, as demonstrated by those plants left to senesce naturally. Flower-removal plants took some extra time (approx. 9 d) to produce new flowers and mature seeds, with no interruption between the two processes, in contrast to control plants for which seed maturation occurred well after flowering. If these traits were genetically determined and linked in a late-flowering mutant (Law, 1979), any conditions that would accelerate canopy development could put that genotype at a high risk of reproductive failure. Early flowering may thus be adaptive in an environment in which the period of high resource availability (mostly light) is variable in duration (Leopold & Jones, 1947; Cohen, 1971; King & Roughgarden, 1982a,b; Fox, 1992b). Environmentally induced senescence may be what triggers seed maturation in F. proserpinacoides and not the converse.


I thank É. Coste, J. Lake and M.-H. Langis for technical help in the field and laboratory, L. Lapointe for her comments on an earlier version of this paper, and the Conseil de recherche en sciences naturelles et en génie du Canada (CRSNGC) for its financial support.