Senescence

MARTIN H. BENDER1, JERRY M. BASKIN1, AND CAROL C. BASKIN1,2 (1School of Biological Sciences, University of Kentucky, Lexington KY 40506-0225, 2Department of Agronomy, University of Kentucky, Lexington, KY 40546-0091). Role of intraspecific competition in mass seeding and senescence in Polymnia canadensis, a primarily monocarpic species. J. Torrey Bot. Soc. 129:107-114. 2002.-Mass seeding and senescence in 2-year population cycles previously were documented in the primarily monocarpic species Polymnia canadensis. In the present study, small individuals of this species died in competition with large ones in field trials. These results suggested that intraspecific competition followed by synchronous semelparity in a dominant age-class could result in mass flowering and senescence and thus could play a role in population cycling in P. canadensis. It was also found in some trials that other environmental factors such as drought could eliminate small individuals, perhaps sometimes in interaction with competition.

Key words: intraspecific competition, mass seeding and senescence, monocarpic species, plant population cycles, Polymnia.

Biologists have long attempted to explain the regularity of population fluctuations in animals (Elton 1942; Pianka 1988) and the oscillatory dynamics in laboratory and natural populations of plants (Symonides et al. 1986; Thrall et al. 1989) and of animals (Nicholson 1954; Nisbet and Gurney 1982). In particular, causes of cyclic population change have been explored in animals (Ricker 1962; Botsford and Wickham 1979; Hoppensteadt 1982) and in plants (Watt 1947, 1955; Kershaw 1964; Rabotnov 1985; Marrs and Hicks 1986; Solbrig et al. 1988). The plant population cycle examined in the present investigation was mass seeding and senescence, in which synchronous dieback of a population as a result of monocarpic senescence is followed by massive germination that reestablishes the population mostly as a single age-class. This phenomenon has been reported for semelparous bamboo species (Janzen 1976; Gadgil and Prasad 1984) and for a few biennials in patches (Kelly 1985, 1989; Baskin and Baskin 1992).

Preliminary observations revealed wide variation in the life history of Polymnia canadensis L. (Asteraceae), a species of the North American Temperate Deciduous Forest. Hence, we conducted four ecological studies, in addition to the present investigation, to document its life history variation and determine causes of that variation. One study monitored the demography of the species and revealed that it is primarily monocarpic with life cycles ranging from winter annual to polycarpic perennial (Bender et al. 2000). It proved to be mainly a facultative biennial, a life cycle type that is rare in deciduous forests.

In a second study, Bender et al. (in press) documented the physiological requirements for flowering in P. canadensis to ascertain their role in its life history variation. Vernalization, or a set of physiological changes that occur during the cold season at approximately 0-10°C, was required prior to summer flowering in individuals from a wide part of the geographic range of this species, thus allowing most field individuals to be biennial, few to be winter annual, and none to be summer annual. Vernalized plants flowered under short- and long-day conditions, which in conjunction with its indeterminate growth habit, would allow field individuals to flower throughout the growing season in good conditions.

A third study examined seed dormancy and germination characteristics of P. canadensis to determine if they contributed to the rarity of its facultative biennial life cycle in deciduous North American forests (Bender 1991). Its seeds had physiological dormancy and germinated better in light than in darkness and to higher percentages after cold stratification, but they also formed a persistent soil seed bank. Since its seed dormancy and germination characteristics were not uncommon in plant species and were not particularly associated with any one life cycle type in woodlands, they did not contribute to the rarity of its life cycle type in deciduous forests. A fourth investigation found that this species exhibited extensive phenotypic plasticity but little genetic difference between populations with regards to date of anthesis, plant height, and life cycle (Bender 1991).

Bender et al. (2000) used population structure diagrams (i.e., proportion of population in various age-classes) to document cycles of mass seeding and senescence in monospecific stands containing mostly one age-class of P. canadensis at two field sites, of which one is included in the present investigation. This phenomenon was also observed for two or three years at periodically revisited sites in Clark and Madison Counties, Kentucky, Lee County, Virginia, and Lincoln County, Missouri (Bender 1991). Since P. canadensis was mostly biennial, the expected cycle would be mass reproduction and senescence once every two years when there was mostly one age-class in a patch.

This two-year cycle was documented by photographs of a stand of P. canadensis at another field site that is also in the present investigation (Bender 1991). The population consisted mainly of an adult age-class in 1985 and 1987. There was a dense stand of juveniles in 1986, but droughts in fall 1987 and early summer 1988 allowed only a thin stand of juveniles in 1988 followed by more establishment of juveniles in 1989. This stopped the cycle with a mixed stand of juveniles and adults in 1989 and 1990.

As illustrated by the inability of subsequent adults from the thin stand of juveniles in 1988 to prevent the establishment of more juveniles in 1989, we hypothesized in the current investigation that the cycle of mass seeding and senescence could result from intraspecific competition between cohorts that eliminates almost all but the dominant age-class. Other studies have found that seedlings of various species were suppressed by conspecifics, but none of the species were monocarpic, which would permit mass senescence of a dominant age-class followed by mass seeding (Putwain and Harper 1970; Fowler and Antonovics 1981; Newell et al. 1981; Shaw and Antonovics 1986). Moreover, intraspecific competition experiments have not been conducted on the few plant species in which mass seeding and senescence has been observed (Janzen 1976; Gadgil and Prasad 1984; Kelly 1985, 1989; Baskin and Baskin 1992). Hence, some field trials of intraspecific competition were conducted during 1986-1990 with P. canadensis to determine if large conspecifics caused fewer small individuals to achieve maturity and if they eliminated most small individuals. The former result would show that competition from large conspecifics was reducing the resources available to small individuals, and the latter result would permit the cycle of mass seeding and senescence.

Methods. FIELD SITES. Polymnia canadensis L. is common in woodlands along the Kentucky River in central Kentucky, often in dense monospecific stands in the herbaceous layer. Hence, there were three field sites, 10-20ha in size, along or near the Kentucky River in Jessamine County: Camp Nelson, Scotts Grove, and High Bridge (around 37°47'N, 84°37'W). The sites were located over limestone bedrock in the Bluegrass section of the Western Mesophytic Forest Region (Braun 1950). Camp Nelson and High Bridge were dry woodlands in silty clay on steep, south-facing, gorge slopes, but Scotts Grove was a mesie woodland on level silt loam. Soil depth (cm, mean ± SE, n = 20) was significantly greater (Tukey's test, P

Dominant trees at the three sites included Quercus muehlenbergii Englem. and Q. shumardii Buckl., but Acer saccharum Marsh, was prevalent only in the mesic woods at Scotts Grove. Because of the tree canopies, the photosynthetic (400-700nm) photon irradiance on cloudless days between 12 noon and 2 PM daylight savings time at a height of 1.25m was not significantly different (Tukey's test, P

The average annual precipitation during 1941-1970 at the Lexington Bluegrass Airport was 113 cm (Wallis 1977), while the respective annual values (cm) at Dix Dam near High Bridge during 1986-1989 were 115, 91, 114 and 148 (NOAA 1986-1989). The precipitation was close to normal during the first half of 1990, at which time the investigation was terminated (NOAA 1990). There were droughts during Sept.-Nov. 1987 and June-July 1988 in which soil moisture content was near 10% on a dry basis, far below the wilting coefficient of 26% at the three sites (Bender et al. 2000).

TRIALS. All trials were conducted in monospecific stands of F. canadensis to avoid any confounding influence of interspecific competition. Seedlings 5-8 cm in height were dug up in the field on reported planting dates (Table 1) and immediately transplanted with some watering into nearby 1 m × 1m quadrats from which natural seedlings had been removed. Mortality within the first few weeks was replaced with transplanted seedlings. Each treatment consisted of three, randomly assigned quadrats with nine seedlings per quadrat. However, with this few seedlings per quadrat, we considered only the total of 27 seedlings per treatment in the analysis of the data and did not regard the quadrats to be replicates.

Since the hypothesis of the investigation was that large conspecifics could eliminate smaller individuals, the transplanted seedlings were widely spaced so they experienced competition from the larger conspecifics but not between themselves. That is, the nine seedlings in each quadrat were arranged in a 3 × 3 square array with a spacing of 30 cm between seedlings. This could be regarded as a conservative test of the hypothesis since more seedlings would likely be eliminated by additional competition among closely spaced seedlings in the wild.

Each trial contained three treatments, one in which all large conspecifics were removed from the quadrats so that transplanted seedlings did not experience any intraspecific competition. In the second and third treatments, the large conspecifics were left intact, but in the latter, root competition was prevented for each seedling by surrounding it with a tin can 7.5 cm in diameter and 11 cm in depth with both ends removed and the top flush with ground level. This was done to determine if above-ground competition alone could eliminate the seedlings. A fourth treatment permitting only root competition was not attempted because the investigation was not an experiment to disentangle the confounding effects of root and light competition. For example, it was doubtful that the roots of large conspecifics denuded of their above-ground biomass would be functioning near the level of roots in unaltered plants, thus hindering any interpretation of below-ground competition. Also, an alternative method would be to enclose the above-ground portion of the conspecifics in narrow, open containers, but this was not feasible because this portion was much larger vertically and horizontally than the roots, which were easily surrounded within containers in the third treatment.

Six trials were conducted during 1988-1990 at the field sites. Five other trials were attempted during 1986-1987, but are not reported in the results section because all transplanted seedlings were eliminated by the above droughts. Nonetheless, the results are significant and are very briefly noted in the discussion section. These five trials contained only the first two treatments listed above, namely no competition and above-and below-ground competition.

The six trials during 1988-1990 consisted of three pairs of trials, each pair planted on a different date (Table 1). In each pair, one trial was located at a mesic site, and the other at a dry site, not for statistical testing but for qualitative comparison to avoid site-specific results associated with only one site.

To determine if large conspecifics allowed significantly fewer seedlings to reach maturity, a simple indicator was flowering designation. Each trial was conducted long enough until all seedlings had either eventually flowered (i.e., reached maturity) or experienced juvenile mortality. Thus, flowering designation for the seedlings in each treatment was two final frequencies at the end of a trial, namely the number flowering and the number non-flowering (i.e., the latter due to juvenile mortality). The flowering in each trial occurred within one growing season, The data was graphed as the proportion of seedlings that eventually flowered,

In demonstrating that most seedlings were eliminated in the presence of large conspecifics, the mean frequencies for flowering designation were simply reported and interpreted. It was not possible to define what proportion of seedlings must be eliminated in order to obtain a recognizable cycle of subsequent mass seeding and senescence. In other words, we could not assign a critical value that was biologically meaningful in this context in a null hypothesis for comparison with mean proportions or frequencies.

DATA ANALYSIS. To compare the effects of the three competition treatments, the data for flowering designation can be analyzed as proportions or as frequencies. In the former case, a nonparametric test would have been considered in this study because there were many values of zero in the data. However, neither the Kruskal-Wallis test or Friedman's test, which are nonparametric analogs of analysis of variance, were appropriate for comparison of the three treatments. This was because the distributions of flowering designation for the two trials in spring 1988 were very different from the four trials in fall 1988 and spring 1989; i.e., the two nonparametric tests assume distributions identical in shape (Sokal and Rohlf 1981).

To compare the three competition treatments by analyzing the data for flowering designation as frequencies, tests of independence between competition treatment and flowering designation were conducted by G-tests, also known as log likelihood-ratio tests (Sokal and Rohlf 1981). Six G-tests, one for each trial, were conducted on 3 × 2 contingency tables for the three levels of competition and the two frequencies of flowering designation (SYSTAT 1998). No tests were conducted between trials because the two types of sites, dry and mesic, were included just for qualitative comparison to avoid any site-specific results associated with only one site. The G-statistic was adjusted with Williams' correction since some frequencies of flowering had values of zero (Sokal and Rohlf 1981). Expected frequencies for the cells in the contingency tables were calculated and examined since G-tests reject the null hypothesis too often if an expected frequency is too small (Sokal and Rohlf 1981).

The critical value for testing the above adjusted G-statistic was obtained from the chisquare distribution based on Sidak's multiplicative inequality, or [chi]^sup 2^^sub [alpha][k,v]^, where [alpha] = 0.05 is the experimentwise Type 1 error rate (the experiment being the six trials), k = 6 is the number of intended tests, and v = 2 is the degrees of freedom of the contingency tables (Sokal and Rohlf 1981). The Sidak inequality set the Type I error rate at 1 -(1 - 0.05)^sup 1/6^, or 0.0085, for each G-test such that the probability of making at least one Type I error in the six G-tests was kept at [alpha] = 0.05. The null hypothesis in each test was that the respective frequencies for flowering designation were equal among the three competition treatments. Since the alternative hypothesis was simply departure from the expected frequencies resulting from the null hypothesis (i.e., not specifying whether the observed frequencies were too deviant or too close to the expected frequencies), these tests were one-sided or one-tailed, as is typically the case in tests of independence (Sokal and Rohlf 1981). Hence, critical values for chi-square based on the Sidak inequality were obtained for [alpha] = 0.05 from the statistical table published by Rohlf and Sokal (1981).

If the null hypothesis was rejected in any trial, then three post hoc pairwise comparisons were done for the three treatments by three adjusted G-tests of independence conducted on 2 × 2 contingency tables of two treatments and the two frequencies of flowering designation (Sokal and Rohlf 1981). Again, expected frequencies for the cells in the contingency tables were examined. Critical values were obtained from the chi-square distribution based on Sidak's multiplicative inequality, but in this case, [alpha] = 0.05, k = 3, and v = 1. Fisher's exact test could not be applied instead of G-tests because it requires the marginal totals in a contingency table to have fixed values, which did not hold in the separate marginal totals for flowering designation (Sokal and Rohlf 1981, p. 735).

Results. In the six trials, seven of the twelve treatments containing competition had significantly lower flowering frequency than the treatments without competition. In other words, this was true for four of the six treatments containing above- and below-ground competition and for three of the six treatments containing only above-ground competition. These two cases occurred in both fall 1988 trials and in the spring 1989 trial at the dry site, including the treatment with above- and below-ground competition in the spring 1988 trial at the mesic site (Fig. 1).

The above results were obtained from the six G-tests of independence (Table 2), and calculation of post hoc pairwise comparisons was required only for the spring 1988 trial at the mesic site. The null hypothesis was not rejected in the spring 1988 trial at the dry site and in the spring 1989 trial at the mesic site. The null hypothesis was rejected in the two fall 1988 trials and the spring 1989 trial at the dry site, but calculation of pairwise comparisons was not necessary because both treatments containing competition had the same flowering frequency, namely zero, leading to obvious conclusions (Fig. 1). The spring 1989 trial at the mesic site was the only trial in which a cell in the contingency table had a value less than three for the expected frequency. This would have been a problem if the null hypothesis had been rejected (methods section), but it was not, as noted above.

In the pairwise comparisons for the spring 1988 trial at the mesic site, the treatment with above- and below-ground competition had a significantly lower flowering frequency than either the treatment without competition, adjusted G = 5.863, or the treatment with only above-ground competition, adjusted G = 9.985 ([chi]^sup 2^^sub 0.05[3,1]^ = 5.701). Unexpectedly, there was a greater frequency of flowering plants in the treatment containing only above-ground competition than in the treatment without competition, but this difference had an adjusted G = 0.650 that was not significant.

In the six trials, most or all seedlings were eliminated by the presence of large conspecifics in ten of the twelve treatments containing competition. All seedlings encountered juvenile mortality in both treatments containing competition in the two fall 1988 trials and the two spring 1989 trials (Fig. 1). Most seedlings were eliminated in the treatment containing above- and below-ground competition in the spring 1988 trials; i.e., only 22 and 15% of the seedlings flowered at the dry and mesic sites, respectively (Fig. 1). However, most seedlings were also eliminated in the treatment with no conspecifics in the spring 1989 trial at the mesic site, in which only 11% of the seedlings flowered, much lower than the range of 33-52% for this treatment in the other five trials (Fig. 1).

Each pair of dry and mesic trials within trial date had mostly similar results with one exception. Both spring 1988 trials were the only ones in which some seedlings survived to flower in either treatment containing competition (Fig. 1). There was complete juvenile mortality in the treatments containing competition in both trials in fall 1988 and again in spring 1989. The proportion of seedlings that eventually flowered in the treatment with no competition was fairly similar in each pair of trials within trial date, except in spring 1989 only 11% of the seedlings survived to flower at the mesic site, but 52% at the dry site.

Discussion. Most or all transplanted seedlings were eliminated in the presence of large conspecifics in ten of the twelve treatments containing competition in the six trials. In the eight treatments containing complete juvenile mortality, mass seeding and senescence would clearly be evident after the trial because mostly the dominant age-class would remain with synchronous maturity and monocarpic dieback, followed by massive germination. In the other two treatments in which only 15 and 22% of the transplanted seedlings survived to flower, these proportions may be low enough to allow mass seeding and senescence to be discernible in the above cycle of the dominant age-class. Although juvenile mortality was complete or nearly so in ten of the twelve treatments containing large conspecifics, competition was a significant factor in only seven of these treatments, namely four treatments containing above- and below-ground competition and three treatments containing only above-ground competition.

The five treatments containing large conspecifics in which competition was not a significant factor were in the two spring 1988 trials and the spring 1989 trial at the mesic site. The former trials were the only ones in which there was not complete juvenile mortality in the treatments containing competition. This lower mortality might have been due to reduced competition from large conspecifics as a result of the summer 1988 drought (methods section). That is, the drought could have hindered the large conspecifics more than the seedlings because the conspecifics experienced more sunlight and possibly subsequent evapotranspiration than the seedlings located underneath the canopy of the conspecifics.

In the other exception, namely the spring 1989 trial at the mesic site, the complete juvenile mortality in the two treatments containing large conspecifics was not the result of competition alone because nearly complete juvenile mortality occurred in the treatment without conspecifics. The latter result was inexplicable because precipitation was above average during the trial (methods section). Hence, competition may have been present to some extent, but some uncontrolled environmental variable(s) appeared to be the dominant factor in this trial, perhaps in interaction with competition.

The dry and mesic trials within each trial date had mostly similar results for the three treatments with the exception of one treatment in spring 1989. Mostly similar results might be expected since the environmental differences between the sites were not large. Soil depth was greater at the mesic site than at the dry sites (methods section), but weekly soil moisture content was not very different between the sites (Bender et al. 2000). Because of the tree canopies, the photosynthetic photon irradiance at the height of the large conspecifics was not different between one dry site, High Bridge, and the mesic site, but was greater at the other dry site, Camp Nelson (methods section). The one exception in which the treatment with no competition in spring 1989 had much lower flowering frequency at the mesic site than the dry site was inexplicable since precipitation was above normal during the trial (methods section).

As noted above, above-ground competition alone from the large conspecifics was strong enough to significantly reduce the number of seedlings eventually reaching maturity. A similar claim cannot be made about below-ground competition since we did not attempt a treatment permitting only root competition. This study was not an experiment to disentangle the confounding effects of above- and below-ground competition, but was an exploration of the role of intraspecific competition in the cycle of mass seeding and senescence. It was simple to include a treatment containing only above-ground competition, but not one containing only below-ground competition (see methods section).

The investigation confirmed our hypothesis that intraspecific competition could eliminate smaller individuals of P. canadensis, thus permitting a mostly dominant age-class to initiate a cycle of mass seeding and senescence. However, we found that other environmental factors can also set the stage for this cycle by eliminating smaller individuals, as noted above for the spring 1989 trial at the mesic site. This was also true for the five trials during 1986-1987 that were not reported in the results (see methods section), in which drought eliminated all transplanted seedlings. Likewise, mass seeding by biennials in chalk grassland in England depends mainly on environmental conditions that allow only one of the few age-classes to establish in a patch (Kelly 1985, 1989). In contrast, mass seeding in bamboo species occurs at regular intervals that mainly are determined genetically (Janzen 1976).

We found that environmental factors can also disrupt the cycle of mass seeding and senescence. For example, we noted above in the two spring 1988 trials that drought may have resulted in reduced competition from large conspecifics that allowed a large proportion of seedlings to survive. This would yield a mixed population of juveniles (the former seedlings) and adults (the large conspecifics), in which the dispersed stand of juveniles would mask any synchronous monocarpic dieback of the adults or subsequent massive germination, thus ending the perceptible cycle of mass seeding and senescence.

A brief survey of the mechanisms that enable some plant species to persist in dense monospecific stands suggests possible mechanisms for maintenance of stands of P. canadensis, whether by cyclic patches of mass seeding and senescence or by non-cyclic patches. A near-monoculture of Epilobium angustifolium L. was able to persist for 30 years due to the mineral cycle brought about by the population itself (van Andel 1975). Early and synchronized seedling emergence, high seedling density, rapid seedling growth, and low seedling mortality enabled Impatiens pallida Nutt. to quickly preempt large gaps in disturbed woodland (Cid-Benevento and Schaal 1986). Early and nearly synchronous, complete germination and limited seed dispersal produced dense, even-aged stands of I. capensis Meerb. with light-blocking capacity that quickly gained dominance of habitats (Winsor 1983). Brassica nigra (L.) Koch dominates California annual grasslands by means of allelopathic suppression of other species (Bell and Muller 1973). Allium ursinuin L. persists in monocultures by several mechanisms: dense stands by the production of many seeds with limited dispersal; rapid growth; allelochemical suppression; and senescence of leaves causing mechanical suppression of young seedlings of other species (Ernst 1979).

Polymnia canadensis has limited seed dispersal, which allows for production of dense stands (Bender et al. 2000). Seedling mortality is low until mid-summer of the first growing season, when soil moisture is low and the plants are large enough to encounter competition from neighboring conspecifics (Bender et al. 2000). The initially low seedling mortality and the rapid spring growth can produce a dense canopy that shades out other species. It is not known if allelochemical suppression or mineral cycling occurs. Self-replacement by seedlings in rosette openings has been used to explain how Dipsacus sylvestris Hudson (Werner 1977) and Seneciojacobaea L. (McEvoy 1984) persist in the absence of environmental disturbance and may very well apply to P. canadensis.

Conclusions. Intraspecific competition by large conspecifics caused nearly complete elimination of small individuals, thus allowing mass seeding and senescence to occur by synchronous semelparity in the remaining dominant age-class and then massive germination. Thus, our investigation showed that intraspecific competition could play a role in population cycling of Polymnia canadensis. However, in some of the trials, other environmental factors such as drought could also result in nearly complete juvenile mortality with the same result as intraspecific competition leading to mass seeding and senescence. Experiments should be conducted to explore the interaction of environmental factors with intraspecific competition in the population cycling and also to ascertain if allelochemical suppression or mineral cycling contribute to this maintenance of dense monospecific stands of P. canadensis.

4 We thank two anonymous reviewers who improved the analysis and the text.

Literature Cited

ANDEL, J. VAN. 1975. A study on the population dynamics of the perennial plant species Chamaenerion angustifolium (L.) Scop. Oecologia 19: 329-337.

BASKIN, J. M., AND C. C. BASKIN. 1992. Seed germination biology of the weedy biennial Alliaria petiolata. Nat. Areas J. 12: 191-197.

BELL, D. T., AND C. H. MULLER. 1973. Dominance of the California annual grasslands by Brassica nigra. Amer. Midl. Nat. 90: 277-299.

BENDER, M. H. 1991. Contributions to the basic and theoretical population biology of monocarpic perennials ("biennials"). Ph.D. Dissertation. University of Kentucky, Lexington.

BENDER, M. H., J. M. BASKIN, AND C. C. BASKIN. 2000. Ecological life history of Polymnia canadensis, a monocarpic species of the North American Temperate Deciduous Forest: Demography. Plant Ecology 147: 117-136.

BENDER, M. H., J. M. BASKIN, AND C. C. BASKIN. In press. Flowering requirements of Polymnia canadensis (Asteraceae) and their influence on its life history variation. Plant Ecology.

BOTSFORD, L. W., AND D. E. WICKHAM. 1979. Population cycles caused by inter-age, DD mortality in young fish and crustaceans, pages 73-82. In E. Naylor, and R.G. Hartnoll (eds.), Cyclic Phenomena in Marine Plants and Animals. Pergamon Press, Oxford.

BRAUN, E. E. 1950. Deciduous Forests of Eastern North America. Blakiston, Philadelphia.

CID-BENEVENTO, C., AND B. SCHAAL. 1986. Variation in population growth rate in the woodland annual Impatiens pallida (Balsaminaceae). Amer. J. Bot. 73: 1031-1042.

ELTON, C. S. 1942. Voles, Mice, and Eemmings: Problems in Population Dynamics. Oxford Univ. Press, Oxford.

ERNST, W. H. O. 1979. Population biology of Allium iirsinum in northern Germany. J. Ecol. 67: 347-362.

FOWLER, N. L., AND J. ANTONOVICS. 1981. Small-scale variability in the demography of transplants of two herbaceous species. Ecology 62: 1450-1457.

GADGIL, M. D., AND S. N. PRASAD. 1984. Ecological determinants of life history evolution of two Indian bamboo species. Biotropica 16: 161-172.

HOPPENSTEADT, F. C. 1982. Mathematical Models in Population Biology. Cambridge Univ. Press, Cambridge.

JANZEN, D. 1976. Why bamboos take so long to flower. Annu. Rev. Ecol. Syst. 7: 347-391.

KELLY, D. 1985. On strict and facultative biennials. Oecologia 67: 292-294.

KELLY, D. 1989. Demography of short-lived plants in chalk grassland. III. Population stability. J. Ecol. 77: 785-798.

KERSHAW, K. A. 1964. Quantitative and Dynamic Ecology. American Elsevier Publ. Co., New York.

MARRS, R. H., AND M. J. HICKS. 1986. Study of vegetation change at Lakenheath Warren: A re-examination of A.S. Watt's theories of bracken dynamics in relation to succession and vegetation management. J. Appl. Ecol. 23: 1029-1046.

McEVOY, P. B. 1984. Seedling dispersion and the persistence of ragwort Senecio jacobaea (Compositae) in a grassland dominated by perennial species. Oikos 42: 138-143.

NATIONAL OCEANIC, AND ATMOSPHERIC ADMINISTRATION (NOAA). 1986-90. Climatological Data, Kentucky, Vois. 81-85. National Climatic Data Center, Asheville, North Carolina.

NEWELL, S. J., O. T. SOLBRIG, AND D. T. KINCAID. 1981. Studies on the population biology of the genus Viola. III. The demography of Viola blanda and Viola pattens. J. Ecol. 69: 997-1016.

NICHOLSON, A. J. 1954. An outline of the dynamics of animal populations. Aust. J. Zool. 2: 9-65.

NISBET, R. M., AND W. S. C. GURNEY. 1982. Modelling Fluctuating Populations. John Wiley and Sons, New York.

PIANKA, E. 1988. Evolutionary Ecology. Harper and Row, New York.

PUTWAIN, P. D., AND J. L. HARPER. 1970. Studies in the dynamics of plant populations: III. The influence of associated species on populations of Rumex acetosa and R. acetosella in grassland. J. Ecol. 58: 251-264.

RABOTNOV, T. A. 1985. Dynamics of plant coenotic populations, pages 121-142. In J. White (ed.), The Population Structure of Vegetation. Dr. W. Junk Publ., Dordrecht.

RICKER, W. E. 1962. Regulation of abundance of pink salmon populations, pages 155-201. In N. J. Wilimovsky [ed.], Symposium on Pink Salmon. Institute of Fisheries, Univ. British Columbia, Vancouver, Canada, 13-15 October 1960.

ROHLF, F. J., AND R. R. SOKAL. 1981. Statistical Tables, Ed. 2. W. H. Freeman and Co., San Francisco.

SHAW, R., AND J. ANTONOVICS. 1986. Density-dependence in Salvia lyrata, a herbaceous perennial: The effects of experimental alteration of seed densities. J. Ecol. 74: 797-813.

SOKAL, R. R., AND F. J. ROHLF. 1981. Biometry: The Principles and Practice of Statistics in Biological Research, Ed. 2. W. H. Freeman and Co., San Francisco.

SOLBRIG, O. T, R. SARANDON, AND W. BOSSERT. 1988. A density-dependent growth model of a perennial herb, Viola fimbriatula. Amer. Nat. 131: 385-400.

SYMONIDES, E., J. SILVERTOWN, AND V. ANDREASEN. 1986. Population cycles caused by overcompensating density-dependence in an annual plant. Oecologia 71: 156-158.

SYSTAT. 1998. SYSTAT 8.0: Statistics. SYSTAT, Inc., Evanston, Illinois.

THRALL, P. H., S. W. PACALA, AND J. A. SILANDER, JR. 1989. Oscillatory dynamics in populations of an annual weed species Ahutilon theophrasti. J. Ecol. 77: 1135-1149.

WALLIS, A. L. JR., 1977. Comparative Climatic Data through 1976. National Climatic Data Center, Asheville, North Carolina.

WATT, A. S. 1947. Pattern and process in the plant community. J. Ecol. 35: 1-22.

WATT, A. S. 1955. Bracken versus heather, a study in plant sociology. J. Ecol. 43: 490-506.

WERNER, P. A. 1977. Colonizing success of a "biennial" plant species: Experimental field studies of species cohabitation and replacement. Ecology 58: 840-849.

WINSOR, J. 1983. Persistence by habitat dominance in the annual Impatiens capensis (Balsaminaceae). J. Ecol. 71: 451-466.

Martin H. Bender1,3, Jerry M. Baskin1, and Carol C. Baskin1,2

1School of Biological Sciences, University of Kentucky, Lexington KY 40506-0225

2Department of Agronomy, University of Kentucky, Lexington, KY 40546-0091

Received for publication May 17, 2001, and in revised form January 2, 2002.

3 Present address: The Land Institute, 2440 East Watar Well Road, Salina, KS 67401

Copyright Torrey Botanical Society Apr-Jun 2002
Provided by ProQuest Information and Learning Company. All rights Reserved