Am. J. Bot. Join the BSA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (11)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Houle, G.
Right arrow Articles by Lapointe, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Houle, G.
Right arrow Articles by Lapointe, L.
Agricola
Right arrow Articles by Houle, G.
Right arrow Articles by Lapointe, L.
(American Journal of Botany. 2001;88:594-607.)
© 2001 Botanical Society of America, Inc.

Spatiotemporal dynamics of Floerkea proserpinacoides (Limnanthaceae), an annual plant of the deciduous forest of eastern North America1

Gilles Houle 2,3, 4, Margaret F. McKenna 2,3 and Line Lapointe 2

2Département de biologie and 3Centre d'études nordiques, Université Laval, Sainte-Foy, Québec, Canada G1K 7P4

Received for publication January 11, 2000. Accepted for publication May 30, 2000.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 STUDY SPECIES
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Because environmental filters are temporally and spatially heterogeneous, there often is a lack of significant relationship between the spatial patterns of successive life stages in plant populations. In this study, we determined the spatiotemporal relationships between different life stages in two populations of an annual plant of the deciduous forests of eastern North America, Floerkea proserpinacoides. Demographic surveys were done over a 4-yr period, and experiments were performed in the field and under controlled conditions to test for the effects of various environmental factors on population dynamics. There was a general lack of relationship between the spatial patterns of seed bank and seedling density, and a lack of similarity between their spatial correlograms. This was related mostly to the effects of spatially variable environmental filters operating on germination and emergence. However, environmental filters acting on plant survival were stable through time and contributed to stabilize the density and spatial patterns of the populations. Despite density-dependent presenescence mortality, spatial patterns of seedlings and mature individuals were similar and their correlograms were alike, suggesting that mortality did not fully compensate for density. Estimated fecundity was negatively correlated with population density over the study period. Although flower production started only 2–3 wk after emergence, seed maturation mostly occurred at the end of the life cycle, just before the onset of plant senescence. Yet, individual fecundity was low for an annual plant, i.e., 3.0 ± 0.5 mature seeds/plant (mean ± 1 SE). Seed predation by vertebrates was not significant. Low soil moisture had little effect on the total number of seeds germinating, although it slowed down the germination process. In quadrats where leaf litter was experimentally doubled, seedling emergence was lower than in control quadrats; in quadrats where leaf litter was completely removed, emergence did not differ from that in control quadrats. Susceptibility to drought stress was higher for seedlings than for mature plants. Although the species does not maintain a long-term persistent soil seed bank, other factors, such as density-dependent fecundity and autogamy, may temper population fluctuations through time and reduce the probability of local extinction.

Key Words: annual plant • demography • environmental filters • Floerkea proserpinacoides • Limnanthaceae • Moran's I • partial Mantel test • seed bank • seed predation


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
STUDY SPECIES
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Traditionally, plant demography has been concerned with the temporal sequence of events from seed dispersal to germination, seedling emergence, and subsequent survival, maturation, and reproduction, all of which are affected by various environmental factors (Harper, 1977 ). Yet, population processes occur not only in time, but also in space, with significant implications for population dynamics and persistence (Sterner, Ribic, and Schatz, 1986 ; Horvitz and Schemske, 1995 ). For example, not only the number, but also the dispersion pattern of the seeds from the seed rain influence the pattern of seedling recruitment and subsequent demographic processes (Janzen, 1970 ; Connell, 1971 ; Hubbell, 1980 ; McCanny and Cavers, 1987 ; Houle, 1995 ). Because the environment is spatially heterogeneous (Kotliar and Wiens, 1990 ; Bell and Lechowicz, 1991 ; Bell, Lechowicz, and Schoen, 1991 ; Lechowicz and Bell, 1991 ), patterns of recruitment may bear little resemblance to seed rain dispersion patterns (after primary dispersal). Environmental filters, i.e., the combination of factors locally affecting the specific number of individuals of a given cohort as time progresses, are most often spatially heterogeneous. Under such conditions, there often is a lack of significant relationship between the spatial patterns of successive life stages (Herrera et al., 1994 ; Houle, 1992, 1995, 1998 ).

However, finding a positive (i.e., concordance), a negative (i.e., discordance), or no relationship (i.e., independence) between life stages does not reveal the identity and the relative importance of the factors that control spatiotemporal relationships within populations (Bascompte and Solé, 1995 ; Schupp and Fuentes, 1995 ). Indeed, similar outcomes may result from different processes. For example, concordance between the spatial patterns of seed rain and seedling recruitment may be related to undercompensating density-dependent (e.g., Hubbell, 1980 ) or to spatially uniform (and thus density-independent) mortality (McCanny and Cavers, 1987 ; Houle, 1995 ). Conversely, similar processes may lead to different outcomes. For example, density-dependent mortality may lead to discordance between the spatial patterns of seed rain and of seedling recruitment if it is overcompensating (Janzen, 1970 ; Connell, 1971 ) or to concordance, if it is undercompensating (Hubbell, 1980 ; see also: Houle 1992, 1995, 1998 ). It is thus important, if one wants to understand how patterns develop through time and space, to combine descriptive and experimental approaches, first to describe the relationships and then to identify the causal factors.

Here, we report the results of a 4-yr study on the spatio-temporal dynamics of natural populations of the rare, annual Floerkea proserpinacoides Willd. (Limnanthaceae), in Québec (Canada). Smith (1983a, b, c) studied density-dependent growth, mortality, and fecundity in a Floerkea population in Wisconsin: he showed that increased density negatively affected overall plant performance. Consequently, we may expect density-dependent processes to be important for the regulation of the species' populations in space and time. Baskin, Baskin, and McCann (1988) and Houle, McKenna, and Lapointe (1998) showed that Floerkea seed germination occurs during the early part of the winter, after a period of warm and, then, of cold stratification. They showed that very few seeds remain dormant after such a treatment. Consequently, the maintenance of a long-lived seed bank seems unlikely for the species, a somewhat uncommon characteristic for an annual plant and one that may place the populations at a high risk of local extinction. Other studies (e.g., Russell, 1919 ; Curtis, 1959 ; Struik, 1965 ; Ornduff and Crovello, 1968 ; Gauthier and Rousseau, 1973 ; Parker and Bohm, 1979 ; Rogers, 1982 ; McKenna, 1999 ; McKenna and Houle, 1999 ) have also considered some aspects of the taxonomy, physiology, and ecology of the species.

Our main objectives were to identify the spatiotemporal relationships between successive demographic stages in the annual plant F. proserpinacoides and to experimentally evaluate the relative importance of some of the factors controlling these relationships. To do this, we studied seed bank dynamics, seedling emergence schedule, and plant survival and reproduction in permanent quadrats over a 4-yr period. Spatial statistics were used to characterize the spatial structure of successive demographic stages, and experiments were performed to test for the effects of soil moisture, litter accumulation, and predation on specific demographic transitions.


   STUDY SPECIES
TOP
 ABSTRACT
 INTRODUCTION
 STUDY SPECIES
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Floerkea proserpinacoides is a North American taxon with a disjunct distribution. In the eastern part of North America, it is present from West Virginia, North Carolina, Tennessee, and Missouri to Ontario, Québec, and Nova Scotia (Gauvin, 1983 ). The species reaches the northern limit of its distribution in North America in Québec, at Île aux Grues (Gauthier and Rousseau, 1973 ).

Floerkea is a small annual plant (<30 cm high). The stem becomes decumbent as it elongates and adventitious roots are often produced at its base (Russell, 1919 ). The plant produces small autogamous flowers, one per node, except at the 2–3 lowest nodes where flower-bearing branches may develop (Ornduff and Crovello, 1968 ; Smith, 1983b ). There are one to three nearly distinct carpels per flower. The fruit is a small, egg-shaped nutlet (~1.0–2.0 mm in diameter, at maturity) bearing small tubercules on its distal part, and containing a single embryo. The floral peduncle elongates as the fruits mature, and it may reach ~5 cm in length. The fruits are apparently barochorous, having no structures to assist in dispersal (no wings, hooks, nor elaiosomes), although they can float in water (Houle, McKenna, and Lapointe, 1998 ).

Seed germination occurs in winter, from December through February (Baskin, Baskin, and McCann, 1988 ; Houle, McKenna, and Lapointe, 1998 ), contrary to what was reported by Russell (1919) and Struik (1965) . Yet, seedlings do not emerge until the following spring (April). Plants start producing flowers when only 2–3 wk old (May). Senescence, which occurs within 9–10 wk after emergence, coincides with canopy closure (mid-June), at a time when irradiance decreases and temperatures increase (Houle, McKenna, and Lapointe, 1998 ). Most nutlets mature just before the onset of plant senescence (early June); dispersal occurs in mid- to late June. Seeds remain dormant until the following fall.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 STUDY SPECIES
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 LITERATURE CITED
 
Study sites
Two sites, ~350 km apart, were chosen within the species range in Québec. The first site, Île de Beaujeu, is located in the southwestern part of Floerkea distribution in Québec, and the other, Île aux Grues, is at the northeastern limit of the species range.

Île de Beaujeu (45°16' N, 74°10' W) is part of the Salaberry-de-Valleyfied archipelago, a group of islands in the St. Lawrence River (Lake St. François), southwest of the island of Montréal. A relatively open, mesic forest dominated by Tilia americana L., Ulmus spp., Celtis occidentalis L., and Rhus typhina L. covers most of the island. The demographic study was carried out in a stand with a ~60% canopy cover (mostly of Rhus typhina), occupying lower and wetter grounds on the island. Other herbaceous species found with Floerkea include Impatiens capensis Meerb. and Matteuccia struthiopteris (L.) Todaro. Mean annual temperature at the Les Cèdres weather station (45°18' N, 74°03' W; 47 m above sea level), in the general area of Île de Beaujeu, is 6.4°C, and annual precipitation totals 928 mm, of which 21% falls as snow (Atmospheric Environment Service, 1993 ). There are 2130 degree-days above 5°C for the area.

Île aux Grues (47°02' N, 70°33' W) is part of the Montmagny archipelago, in the St. Lawrence River, east of Quebec City. Most of Île aux Grues has been cleared for agriculture, but its southwestern tip is still covered by a mesic deciduous forest. The most abundant tree species are Acer saccharum Marsh., Fagus grandifolia Ehrh., Ostrya virginiana (Mill.) K. Koch, and Tilia americana L. Other than Floerkea, herbaceous species present include Dicentra cucullaria (L.) Bernh. and Trillium erectum L. Mean annual temperature at the Montmagny weather station (46°58' N, 70°35' W, 15 m above sea level) is 4.4°C, and total precipitation amounts to 1087 mm, of which 24% falls as snow (Atmospheric Environment Service, 1993 ). There are 1685 degree-days above 5°C for the area.

Seed bank and plant demography
At Île de Beaujeu, seven parallel transects, 5 m long and 1.5 m apart, were positioned in the stand. Ten contiguous 50 x 50 cm quadrats were delimited along each one of these transects, for a total of 70 quadrats. At Île aux Grues, three parallel 9-m-long transects were established within the forest. Along each transect, 18 contiguous 50 x 50 cm quadrats were positioned, for a total of 54 quadrats. Layout of the quadrats differed between sites because of differences in topography. At both study sites, a 10 x 10 cm subquadrat was marked in one corner of each quadrat. Seedling emergence was followed weekly in these subquadrats from 1995 to 1997, and every 2 wk in 1998 (only in 1995 and 1996 for the Île de Beaujeu site). Seedlings were marked with color-coded drinking straw sections to facilitate the census of weekly cohorts. In 1995, seedling emergence was almost completed, at both sites, when the demographic survey began.

In mid-May (after seedling emergence), in early July (after seed dispersal), and in late September (after leaf litter fall and before seed germination) from 1995 to 1998, a 4.4-cm-diameter soil core was taken to a depth of 10 cm within each 50 x 50 cm quadrat, close to, but outside the 10 x 10 cm subquadrat (no samples were taken in 1997 and 1998 at the Île de Beaujeu site). After the first sampling of mid-May 1995, great care was taken not to core in areas that had been previously sampled. Each core was separated into five 2-cm-thick samples. These samples were brought to the laboratory where they were washed through a 710-µm mesh sieve under a continuous flow of water. The material retained in the sieve was allowed to dry at room temperature and Floerkea nutlets were retrieved. Seeds were tested for viability with a 1% tetrazolium chloride solution. Only completely colored embryos were counted as viable.

To test for significant fluctuations through time in seed and seedling abundance, repeated-measures analyses of variance (ANOVAs) were used. Profile contrasts (i.e., comparisons between adjacent years) allowed us to determine when significant interannual fluctuations occurred, if the repeated-measures ANOVAs indicated overall significant differences through time (SAS Institute Inc., Cary, North Carolina, 1989). Huynh-Feldt-adjusted P values are reported for the time factor in all repeated-measures ANOVAs.

Spatial patterns of seeds, seedlings, and mature individuals
Moran's I
To determine the spatial dispersion patterns of seed bank and of seedling density, we performed autocorrelation analyses. There is spatial autocorrelation when the value that a variable takes, at a given location, can be predicted from values that the same variable takes at other points of known position (Legendre and Fortin, 1989 ). One commonly used, easily interpretable index of spatial autocorrelation is Moran's I. This index may vary between -1 (repulsion) and +1 (contagion); the expected value of the index in the absence of spatial autocorrelation (randomness) approaches 0 (Cliff and Ord, 1981 ). I can be calculated for different distance classes (d) and each Id can be tested for significance. A graphical representation of I as a function of d is called a spatial correlogram. Before testing for the significance of individual values of Id, a correlogram must be globally significant, i.e., at least one of the Id's must be significant at 0.05/k, with k representing the number of distance classes considered (Bonferroni criterion). The general shape of the correlogram is representative of the spatial pattern from which it is calculated, although similar-shape correlograms may sometimes underlie different spatial patterns (Legendre and Fortin, 1989 ; Houle, 1998 ). To optimize the spatial information in our data sets, we chose a 0.5-m step for both the Île de Beaujeu (20 distance classes, 4830 point pairs) and the Île aux Grues data (17 distance classes, 2862 point pairs). The coordinates of each quadrat center were used for the spatial analyses on seed, seedling, and mature individual abundance.

Partial Mantel tests
To estimate the relationship between seed bank and seedling density, while removing the possible spurious effect of spatial autocorrelation, we used partial Mantel tests (Smouse, Long, and Sokal, 1986 ; Legendre and Fortin, 1989 ; Fortin and Gurevitch, 1993 ). A partial Mantel statistic is similar to a partial correlation coefficient. It accounts for spatial autocorrelation by computing matrices of residuals of the linear regression of two variables (two distance matrices, one for each variable) over the values of a third variable (the distance matrix for a third variable, e.g., location in space; Oden, 1992 ). We calculated a partial Mantel statistic for several pairwise comparisons of seed bank and seedling densities, e.g., (1) the September 1995 seed bank and the 1996 seedling densities (emergence depends on the previous-fall seed bank), (2) the 1996 seedling and the July 1996 seed bank densities (seed bank abundance should reflect current-year fecundity), and (3) 1995 and 1996 seedling densities or July 1995 and July 1996 seed bank densities (to test for concordance in patterns of abundance, through time). Because a given data set was used several times for the partial Mantel tests (from two to five times), the alpha level of a given test was adjusted according to the sequential Bonferroni correction (Rice, 1989 ). Both autocorrelation analyses (Moran's I) and partial Mantel tests were performed with the R software package of Legendre and Vaudor (1991) .

Density-dependent mortality
To determine whether plant mortality was density dependent, Spearman's rank correlation tests were performed between the mortality rate during a given 1-wk period and the number of individuals present at the beginning of the period for each 10 x 10 cm subquadrat. There was a maximum of 70 (Île de Beaujeu) or 54 (Île aux Grues) observations considered in these analyses for each weekly period in each year. Although mortality rate might have been affected by seedling age at each period, considering this additional factor in the analyses would have implied (1) a much larger number of correlations, (2) a low N for each test, and (3) a loss of statistical power.

Concordance tests
To estimate the relationship among three or more variables, we used the Kendall concordance test (Sokal and Rohlf, 1995 ). This is a nonparametric test and is similar to nonparametric correlation tests commonly used to estimate the relationship between two variables.

Seed production and maturation
To determine the schedule of seed maturation, five plants were haphazardly sampled weekly from early May to late June in 1997 from a marked area at Île aux Grues. The total number of nutlets per plant and the number of mature nutlets were determined. A mature nutlet was a plump, green fruit at least 1.5 mm long and with small tubercules on its distal part. From our experience, seeds from such nutlets have a germination percentage close to 100%.

Seed predation by vertebrates
Smith (1983c) performed an experiment on Floerkea seed dispersal by ants. He showed that ants remove the seeds of another herbaceous species, namely Sanguinaria canadensis L., but do not disperse nor eat Floerkea seeds. However, no test has been done on seed predation by vertebrates from the time of seed dispersal to the time of seedling emergence. We thus performed such an experiment using small exclosures.

In early July 1996, immediately after seed dispersal, we haphazardly positioned five blocks of three quadrats at Île aux Grues. Quadrats within a block were either covered with a complete cage (50 x 50 cm wire netting top, with 25-cm high sides), with a half cage (50 x 50-cm wire netting top attached to 25-cm high stakes, but without sides), or left uncovered (control). Mesh size of the wire netting was 1 cm. Seedling emergence was monitored in a 10 x 10 cm subquadrat, in the center of these 50 x 50 cm quadrats in 1997. In mid-May 1997, a 4.4 cm diameter x 6 cm depth soil core was taken in each quadrat to estimate seed bank density. The soil cores were treated in the laboratory as described in the section Seed bank and plant demography (above). The data (seedling and seed densities) were analyzed in one-factor factorial analyses of variance in a randomized complete block design (Montgomery, 1984 ).

Seed germination and soil moisture
Microtopography often has a significant influence on soil moisture and thus on seed germination. To determine the effect of soil moisture on Floerkea seed germination, we performed an experiment under controlled conditions, in a germination cabinet. One litre of oven-dried, steam-sterilized sand (2 h at 121°C) was placed in each of six plastic bags with enough distilled water to bring the substrate moisture to 20, 15, 10, 7.5, 5, and 2.5% (Etherington, 1993 ). The moist substrate was mixed several times during a 24-h period to ensure homogeneity. Then, ~70 cm3 of substrate were placed in each one of five 10-cm diameter petri dishes for each one of the six moisture levels. Twenty Floerkea nutlets from the Île de Beaujeu population (1995 cohort) were evenly positioned on the substrate in each dish. The dishes were sealed with Parafilm® and placed in a germination cabinet, in the dark, at 5°C. After 8 wk of stratification, the conditions in the germination cabinet were adjusted to provide alternating temperatures of 7°C and 15°C (night, day) with a 14-h photoperiod. Seed germination was verified weekly, for an 8-wk period. Radicle emergence (>1 mm) was used as germination criterion. Germination velocity was calculated as in Mugnisjah and Nakamura (1986) :

where Gw represents the number of seeds germinating during week w; w is the number of weeks since the beginning of the germination test; and Gt is the total number of seeds having germinated during the whole period. The higher the value of the index, the faster germination is.

Total germination percentage and germination velocity were analyzed with one-way analyses of variance. Both variables were arcsine transformed prior to analysis. Because the ANOVA on seed germination velocity indicated that there were significant differences among soil moisture levels (see RESULTS below), differences between treatments were identified with an LSD test.

Seedling emergence and litter
Smith (1983c) reported that low seedling emergence was correlated with high leaf litter mass in a Wisconsin Floerkea population. However, his results did not come from a controlled experiment and other factors related to leaf litter mass (e.g., topography, which also influences soil moisture) could have been responsible for the observed correlation. We thus performed a field experiment to determine the specific effect of leaf litter thickness on Floerkea seedling emergence. We set up an experiment at Île aux Grues, in a sector where Floerkea was not present. Five blocks of three treatments each were haphazardly positioned within the forest. The treatments were applied to 50 x 50 cm quadrats in early July 1996: 50 nutlets from the Île aux Grues population (1996 cohort) were placed in the central section of each quadrat. A 1-mm mesh fiberglass net was fixed in place with stakes at each corner, on each quadrat. In late September, after leaf fall, the litter present on the nets was either replaced on the quadrat (control), removed (litter removal), or doubled with litter from the litter-removal quadrat (litter addition). Nets were then replaced on the quadrats. In spring 1997, the fiberglass nets were removed and seedling emergence was monitored. Percentage emergence (after arcsine transformation) was analyzed in a one-factor factorial analysis of variance in a randomized complete block design (Montgomery, 1984 ). Because the ANOVA on seedling emergence indicated significant differences among litter treatments (see RESULTS below), differences between treatments were identified with an LSD test.

To determine the absolute value of our litter treatments (control, litter removal, and litter addition), we measured leaf litter thickness and mass at the experimental site, at Île aux Grues. After leaf fall, in early November 1997, we inserted a pin through the litter five times in each of ten 50 x 50 cm quadrats and counted the number of leaves transpierced. The leaf litter was also collected from each quadrat, dried in an oven at 75°C for 48 h, and weighed.

Seedling survival and drought stress
In spring 1996, we set up a glasshouse experiment at the Université Laval campus to determine the effect of a drought stress at different developmental stages on Floerkea plant survival. Plants at five stages of development were deprived of water for variable lengths of time. Seedlings (individuals with a single leaf) were collected at Île aux Grues in April 1996. They were placed individually in 170-cm3 plastic pots containing sand and allowed to acclimate to their new environment for a week before the beginning of the experiment. Twenty-five pots representing all treatment combinations (five developmental stages x five drought stress levels, see below) were placed in each of six trays (a blocking factor). Plants were watered with a complete Hoagland solution once a week. On the other days, tap water was used to water the plants. For watering, 1 L of tap water or Hoagland solution was poured into each tray. The pots were left to saturate for ~15 min and then the excess liquid was removed from the tray bottom with a sponge. When a plant was subjected to a drought stress, its pot was removed from the tray before the daily watering and placed on the greenhouse bench next to the tray. It was then replaced in the tray after 0, 24, 48, 72, or 96 h of drought stress. Developmental stages corresponded to 1, 2, 3, 4, or 5 wk after transplantation.

Each time a drought stress treatment was applied (five times, once for each developmental stage), the daily decrease in soil moisture was monitored using six pots filled with sand, but without any plant. These sand-filled pots were saturated with water at the beginning of the drought stress treatment and then weighed daily.

Plant longevity (number of days between the time of transplantation and death or the end of the experiment) was analyzed in a two-factor factorial analysis of variance in a randomized complete block design (Montgomery, 1984 ). A G test was used to compare the number of surviving individuals among the different drought stress–developmental stage combinations.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 STUDY SPECIES
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 LITERATURE CITED
 
Seed bank and plant demography
The intra-annual seed bank dynamics of Floerkea were quite similar at the two study sites. After seedling emergence was completed, in mid-May, there were few viable seeds still present on or in the soil (Fig. 1). After seed production and dispersal, seed bank density increased significantly (except in 1998, at Île aux Grues) and appeared to remain relatively constant until the following spring (repeated-measures analyses of variance on seed bank density: Île de Beaujeu—significant temporal variations in seed density at P = 0.0001, with significant profile contrasts between May 1995 and July 1995, September 1995 and May 1996, and May 1996 and July 1996; Île aux Grues—significant temporal variations in seed density at P = 0.0001, with significant profile contrasts between May 1995 and July 1995, September 1995 and May 1996, May 1996 and July 1996, September 1996 and May 1997, May 1997 and July 1997, and September 1997 and May 1998). At both study sites, most of the seeds (from 66 to 98% of the July and September seed bank) were found in the 0–2 cm soil layer.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 1. Population dynamics of Floerkea proserpinacoides. Density (no./m2) of seedlings, mature individuals, and viable seeds (in boxes). Mean ± 1 SE. Seed rain density (no./m2) is estimated (minimal value) as the difference between seed bank density in July and May of the same year. Left panel: Île de Beaujeu, 1995 and 1996; right panel: Île aux Grues, 1995–1998

 
Seed bank density not only varied significantly among seasons within years, but also among years, within seasons. However, these interannual fluctuations were not synchronous between the two sites: at Île de Beaujeu, seed bank density in July 1996 was 1.8 times that of July 1995; in September 1996, it was 1.5 times higher than that of September 1995. Yet, as revealed by repeated-measures analyses of variance, these differences between years were only marginally significant (P = 0.0690 and P = 0.0584 for July and September, respectively). At Île aux Grues, seed bank density did not differ between July 1995 and July 1996, but it was higher in July 1997 than in July 1996 (1.8 times) and lower in July 1998 than in July 1997 (repeated-measures analysis of variance: significant temporal variations in seed density at P = 0.0071, with no significant contrast between 1995 and 1996 at P = 0.6162, marginally significant contrast between 1996 and 1997 at P = 0.0895, and significant contrast between 1997 and 1998 at P = 0.0008). At both sites, interannual fluctuations in seed bank density were mostly due to fluctuations in seed density in the 0–2 cm soil layer.

Viability of the seeds in the seed bank did not vary greatly from July to September within a given year (average viability of 95 ± 2% and 84 ± 8% for July and September, respectively), but it was lower in May, after seedling emergence (average viability of 61 ± 11%).

Although the specific time course of seedling emergence varied among years (e.g., later in 1997 than in 1996 at Île aux Grues) and between sites (e.g., earlier at Île de Beaujeu than at Île aux Grues in 1996), it usually began early in the spring (April) and was over by mid-May (Fig. 2). In 1996, a year for which we can compare the complete demography schedule of both study sites, Floerkea growing season was 2 wk shorter at Île aux Grues than at Île de Beaujeu, not because of earlier senescence but because of later emergence (Fig. 2).



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 2. (A) Cumulative emergence (thin, continuous line), cumulative death (thin, broken line), and live individuals (thick, continuous line) of Floerkea proserpinacoides. The arrow indicates the onset of plant senescence. (B) Weekly plant mortality. An asterisk indicates significant density-dependent mortality for the period. Left panels: Île de Beaujeu, 1995 and 1996; right panels: Île aux Grues, 1995–1997

 
Seedling density was variable among quadrats within years and within sites (coefficient of variation: from 90 to 122%): it reached a maximum of 6900 seedlings/m2 at Île de Beaujeu (1996) and 7400 seedlings/m2 at Île aux Grues (1998). At the Île de Beaujeu site, mean density varied from 839 seedlings/m2 in 1995 to 1489 seedlings/m2 in 1996 (significant temporal fluctuations at P = 0.0001, repeated-measures analysis of variance; Fig. 1). In contrast, mean density of mature individuals (i.e., those still alive at the onset of senescence) was almost identical in 1995 and 1996 (P = 0.2667, repeated-measures analysis of variance). This is perhaps a consequence of presenescence density-dependent mortality (Fig. 2). At Île aux Grues, mean annual seedling density differed significantly between 1997 and 1998 (repeated-measures analysis of variance, P = 0.0030; significant profile contrast between 1997 and 1998 at P = 0.0002), but not between 1995 and 1996, or between 1996 and 1997 (profile contrasts at P = 0.2822 and P = 0.3859, respectively). At Île aux Grues, density of mature individuals differed significantly only between 1997 and 1998 (P = 0.0176, repeated-measures analysis of variance; P = 0.4910, P = 0.7713, and P = 0.0023 for profile contrasts between 1995–1996, 1996–1997, and 1997–1998, respectively).

Within sites, Floerkea seedling emergence was not as synchronous as we expected it to be for a species with such a short life cycle (Figs. 2 and 3). Although most seedlings emerged before the end of April, a substantial proportion of the seedlings emerged later (e.g., 45% at Île de Beaujeu and 35% at Île aux Grues, in 1996). Yet, the onset of senescence was relatively synchronous among individuals on a given site, and it was independent of the age of the plants (Figs. 2 and 3); it occurred within a 2-wk period starting in early to mid-June. Consequently, longevity of the late-emerging seedlings was short (Fig. 3), and late-emerging plants did not have the time to mature seeds. At the onset of senescence, a substantial proportion of the seedlings had died: 36 ± 4% (1995) and 49 ± 3% (1996) at Île de Beaujeu, and 15 ± 3% (1995), 33 ± 4% (1996), 24 ± 3% (1997), and 31 ± 3% (1998) at Île aux Grues.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 3. Survival of weekly cohorts of Floerkea proserpinacoides at Île de Beaujeu and Île aux Grues, in 1996.

 
Spatial patterns of seeds, seedlings, and mature individuals
Autocorrelation: seed bank density
Within a given year and a given site, the correlogram of the July and the September seed banks have little in common (Figs. 4 and 5). For example, in 1995, the July seed bank correlogram is globally significant for the Île de Beaujeu population, while that of September is not significant (Fig. 4). Between years, for a given season and a given site, the correlograms do not bear much resemblance either. At Île aux Grues, July 1995 and July 1998 correlograms are not globally significant. However, the July 1996 correlogram is significant and suggests a gradient in seed density, with positive values of Moran's I at small distance classes and negative values at greater distances (Fig. 5). In July 1997, still at Île aux Grues, seeds in the seed bank are dispersed in patches of <=1 m, with patches of high density separated by ~7 m (significant negative values of Moran's I at distance classes of 2.5 and 3.5 m, and positive, although not quite significant values, at 7 m).



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 4. Spatial correlogram for seed bank density (July and September) and for the density of seedlings and mature individuals, at Île de Beaujeu in 1995 and 1996. Filled squares indicate significant values of Moran's I (for globally significant correlograms). For the correlogram of the density of the September seed bank and that of the density of mature individuals in 1996, the significance of the extreme value (distance class 9 m and distance class 7 m, respectively) is indicated by an asterisk.

 


View larger version (19K):
[in this window]
[in a new window]
  		Fig. 5. Spatial correlogram for seed bank density (July and September) and for the density of seedlings and mature individuals, at Île aux Grues, from 1995 to 1998. Filled squares indicate significant values of Moran's I (for globally significant correlograms). For the correlogram of the density of mature individuals in 1997, the significance of the extreme value (distance class 0.5 m) is indicated by an asterisk.

 
Autocorrelation: seedling density
There is significant spatial autocorrelation (positive) for seedling density at a scale of <=1 m in 1995, 1997, and 1998 for the Île aux Grues population (also positive, but not significant in 1996), and in 1996 for the Île de Beaujeu population (also positive, but not significant, in 1995; Figs. 4 and 5). Autocorrelation is also present at 6–7 m (positive) and at 8–8.5 m (negative) for the Île de Beaujeu population in 1996, and at 2.5–3.5 m (negative) for the Île aux Grues population in 1995, 1997, and 1998 (also negative, but not significant, in 1996). In 1995 for Île de Beaujeu and in 1996 for Île aux Grues, the seedling correlograms are not globally significant, i.e., the spatial pattern of seedling density is not different from random. Despite differences in the significance of the specific values of Moran's I and significant differences in seedling density, the correlograms for Île aux Grues are relatively similar between years (Kendall concordance test among correlograms: W = 0.930, P = 0.0001). This suggests an overall patchy dispersion pattern with high-density patches of ~1 m separated by ~5 m (Fig. 5). The correlograms for Île de Beaujeu differ more, from one year to the next, than those for Île aux Grues (Fig. 4; Kendall concordance test between correlograms: W = 0.689, P = 0.1247).

Autocorrelation: density of individuals at senescence
Within a given year and a given site, the correlograms of the seed-producing individuals (i.e., those having survived to the onset of senescence) are almost identical to those of the seedlings (Figs. 4 and 5). This is despite significant density-dependent mortality between the seedling and the mature stages (Fig. 2). At Île aux Grues, correlograms are very similar among years (Kendall concordance test among correlograms: W = 0.822, P = 0.004). At Île de Beaujeu, correlograms are more variable between years (Kendall concordance test: W = 0.538, P = 0.3676); this is also the case for the seedlings (see above, Autocorrelation: seedling density). Moreover, the correlogram for the 1995 density of mature individuals at Île de Beaujeu is globally significant and shows contagion at scales of 0.5 and 6 m and repulsion at a scale of 3.5 m, while that for the seedlings, although similar in shape, is not globally significant (Fig. 4).

Partial Mantel tests
The results of the partial Mantel tests suggest that there is significant variability through time in the spatial patterns of seed and seedling density. Of eight comparisons between seed and seedling density patterns, only one is significant for the Île de Beaujeu site (Table 1): that between the 1995 and the 1996 seedling cohorts (although their correlograms are different: see above, Autocorrelation: seedling density, and Fig. 4). For the Île aux Grues site, five out of 28 comparisons between seed and seedling density patterns are significant (Table 2): those between seedling density and the July seed bank, in 1995 (different correlograms) and in 1997 (similar correlograms), and between 1996 and 1997, between 1996 and 1998, and between 1997 and 1998 seedling cohorts (similar correlograms; Fig. 5). For Île aux Grues, 1995 and 1996 seedling density patterns are not significantly correlated, although their correlograms are relatively similar (Fig. 5).


View this table:
[in this window]
[in a new window]
  		Table 1. r values from the partial Mantel tests on the demographic variables for Île de Beaujeu, Québec (P values are given in parentheses). Significant tests (P from <= 0.017 to <= 0.05, sequential Bonferroni correction) are in boldface

 

View this table:
[in this window]
[in a new window]
  		Table 2. r values from the partial Mantel tests on the demographic variables for Île aux Grues, Québec (P values are given in parentheses). Significant tests (P from <= 0.01 to <= 0.05, sequential Bonferroni correction) are in boldface

 
Between-year correlations in the spatial patterns of mature plants (i.e., those still alive at the onset of senescence) were always significant for both study sites and varied between 0.0819 and 0.5885. Within-year correlations between the spatial patterns of seedlings and of mature plants were always significant for both study sites and varied between 0.2631 and 0.9342. Using the data on mature individuals for the tests relating the spatial pattern of the July seed bank to that of Floerkea individuals, within a given year, did not increase the number of significant tests for either Île de Beaujeu or Île aux Grues (results not shown).

Stability through time of environmental filters: Île aux Grues
As the demographic data set is larger for the Île aux Grues site, further analyses can be done. If we estimate emergence probability as the ratio of the number of emerged seedlings in yeart+1 to the number of seeds in the September seed bank of yeart, for each quadrat, we get values of 0.62 ± 0.06, 0.62 ± 0.06, and 0.55 ± 0.07 for the 1995–1996, the 1996–1997, and the 1997–1998 periods, respectively. If we compare, with a Kendall concordance test, the relationship through time of the spatial patterns of emergence probability for these three periods, we have a value of W = 0.252 (P = 0.1528), indicating significant change through time (i.e., no concordance) in emergence probability patterns.

Relative stability through time in the patterns of seedling density is demonstrated by the significant correlations already reported (for partial Mantel tests on seedling density patterns from 1995 to 1998, see above, Spatial patterns of seeds, seedlings, and mature individuals).

If we compare, with a Kendall concordance test, the relationship among years (1995–1998) in the spatial patterns of plant survival to the onset of senescence (see above, Spatial patterns of seeds, seedlings and mature individuals), we have a value of W = 0.364 (P = 0.031), indicating no significant change through time (i.e., concordance) in the patterns of survival to senescence.

Furthermore, if we estimate mean plant fecundity as the ratio of the number of seeds present in the July seed bank minus the number of seeds present in the May seed bank to the number of plants alive at the onset of senescence, for each quadrat, we obtain values of 2.1 ± 0.8, 2.5 ± 0.9, 3.5 ± 1.1, and 0.5 ± 0.1 nutlets/plant, for 1995, 1996, 1997, and 1998, respectively. A Kendall concordance test on the relationship through time of the spatial patterns of estimated fecundity for these four years gives a value of W = 0.148 (P = 0.9721) and indicates significant change through time (i.e., no concordance) in estimated fecundity patterns. Over the 4-yr period, estimated fecundity was negatively related to the density of mature individuals (Fig. 1).

Seed production and maturation
Although each plant had an average of 3.0 ± 0.9 nutlets by early June (Île aux Grues, 1997), no nutlets matured before mid-June (mean ± 1 SE: 5.0 ± 0.6 nutlets/plant, but only 3.0 ± 0.5 mature nutlets/plant). At senescence (late June), each surviving plant had produced an average of 6.0 ± 0.6 nutlets, but of these, only 3.8 ± 0.2 nutlets/plant were mature. We have previously estimated mean fecundity (see above, Stability through time of environmental filters: Île aux Grues) as 3.5 ± 1.1 nutlets/plant for 1997, using data on the seed bank density in July and in May, and the density of the individuals present at the onset of senescence. The two fecundity values presented here are thus very similar and provide support for our fecundity estimates for the other years.

Seed predation by vertebrates
Caging had no significant effect on total seedling density (P = 0.6218; mean density ± 1 SE: 2120 ± 320 seedlings/m2). Seed bank density in May 1997 after seedling emergence was completed did not differ between treatments (P = 0.4096), and it amounted to 44 ± 44 viable seeds/m2 (mean ± 1 SE).

Seed germination and soil moisture
Approximately 70% of the seeds germinated during the 6-wk germination period independently of the soil water potential (Fig. 6). Although total germination percentage did not differ significantly among treatments (P = 0.9302), germination velocity was reduced at low (i.e., <=7.5%) substrate moisture levels (P = 0.0001); the radicle was also shorter under these conditions.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 6. Cumulative germination (left; mean of five replicates) and germination velocity (right; mean + 1 SE) of Floerkea proserpinacoides seeds according to substrate moisture. Different letters indicate significant differences among substrate moisture treatments for germination velocity (protected LSD tests). Seeds of the 1995 cohort from Île de Beaujeu were used for the experiment

 
Seedling emergence and litter
There were 3.5 ± 0.3 layers of leaves on the forest floor after leaf fall in November and leaf litter mass was 78.7 ± 2.6 g/2500 cm2 (mean ± 1 SE). Doubling the leaf litter thickness to seven leaf layers significantly reduced seedling emergence the following spring to ~33% of the value recorded for the control (Fig. 7; P = 0.0002). Unexpectedly, complete litter removal did not enhance seedling emergence.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 7. Seedling emergence (%) in 50 x 50 cm quadrats in the litter experiment (see MATERIALS AND METHODS for details on the experiment). Different letters indicate significant differences among treatment means (protected LSD tests). Mean + 1 SE.

 
Seedling survival and drought stress
In the drought stress experiment, soil moisture declined rapidly with time as water evaporated from the soil (Fig. 8). However, because the pots used to estimate soil water loss contained no plants, values represented in Fig. 8 may slightly underestimate the values of soil water loss for those pots with a plant.



View larger version (35K):
[in this window]
[in a new window]
  		Fig. 8. Soil moisture depletion (above; mean ± 1 SE) and longevity of individuals according to the stage at which a drought stress was imposed (below; mean + 1 SE) as a function of drought stress duration. From the darkest to the lightest shaded bars: stress imposed at 1, 2, 3, 4, or 5 wk after transplantation (see MATERIALS AND METHODS for details on the experiment)

 
Mortality of the control plants was very low during the 35-d period of the experiment (e.g., none of the control plants of stage 1 died). However, a prolonged drought stress (>=3 d) significantly reduced Floerkea plant longevity (Fig. 8). This effect was particularly marked when the stress was applied at early stages of plant development (significant interaction between developmental stage and drought stress duration in the analysis of variance, P = 0.0001). A G test for independence between the factors plant stage and drought stress duration revealed a marginally significant effect on plant survival (G = 25.1 with 16 df and 0.05 < P < 0.10): survival was lower when a prolonged drought stress (3–4 d duration) was applied in the earlier stages of development (stages 1–3).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 STUDY SPECIES
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
LITERATURE CITED
 
Seed bank dynamics
Unlike several annual plant species (Fenner, 1985 ; Thompson, 1992 ), Floerkea does not maintain a long-term, persistent soil seed bank. After seedling emergence in April and May, few seeds are still present on/in the soil. Those that remain have a lower viability than those present in July, immediately after seed dispersal, or in September, immediately before germination, and their potential contribution to the following-year seedling cohort is extremely low (Houle, McKenna, and Lapointe, 1998 ). The lack of a long-term persistent seed bank has important implications for the population dynamics of an annual species; indeed, it makes populations much more vulnerable to local extinction (Baskin and Baskin, 1978 ; Pacala, 1986 ; Menges, 1991 ; Kalisz and McPeek, 1993 ). However, in habitats where there is a high probability for an annual plant to successfully complete its life cycle (e.g., in mature forests), the optimal germinable seed fraction is often high and seed dormancy is much less important (Cohen, 1966 ; Watkinson, 1978 ; Carey and Watkinson, 1993 ). Successful Floerkea plants produce an average of only four mature nutlets each (Struik and Curtis, 1962 ; this study). In other annual species often associated with Floerkea, such as Collinsia verna and Galium aparine (~20 seeds/plant), and Impatiens capensis and I. pallida (~40 seeds/plant), the lack of a long-term persistent seed bank appears to be compensated for by a somewhat larger reproductive output (Struik and Curtis, 1962 ; Leck, 1979 ; Kalisz, 1991 ). The absence of a persistent seed bank and the low reproductive output of Floerkea appear to be compensated for by a high survival of the seeds from dispersal to emergence. However, significant interannual variations in seed output, such as those reported at Île aux Grues over our 4-yr study, could place the populations at a high risk of extinction, as a result of random environmental fluctuations. Negative density-dependent fecundity (Smith, 1983b ) and autogamy may assure that even under low plant density, local populations have a high probability of persistence (i.e., no Allee effect: e.g., Groom, 1998 ).

The spatio-temporal dynamics of the seed bank are essentially the result of important events in the life cycle of the plants, from seed germination through seed production, dispersal, and subsequent survival (Parker, Simpson, and Lack, 1989 ; Kalisz, 1991 ; Chambers and MacMahon, 1994 ; Chambers, 1995 ). Seed predation, pre- or postdispersal, does not have a significant influence on Floerkea seed bank dynamics. This is shown in our seed predation experiment and also by the repeated observations that seed bank density does not vary significantly between seed dispersal in early summer and seed germination in the fall. This is despite the fact that most of the seeds remain directly on the soil surface or at very shallow depths. Although local resource availability does affect the reproductive success of Floerkea (Houle, McKenna, and Lapointe, 1998 ), and thus seed bank dynamics, potential enemies seem to have little effect on these processes (see also Smith, 1983c ; Houle, McKenna, and Lapointe, 1998 ). Null or low seed predation may be related to a variety of secondary metabolites (e.g., flavonoids) found in the Limnanthaceae (Parker and Bohm, 1979 ; Bartelt and Mikolajczak, 1989 ).

Interannual fluctuations in seed bank density were not synchronous between the two study sites. Local environmental, i.e., climatic, factors are most probably involved (see: MATERIALS AND METHODS: Study sites). Smith (1983b) has shown the significance of the length of the growth period for the reproductive success of Floerkea plants. On both of our study sites, the year of higher seed bank density in July, after seed dispersal (1996 for Île de Beaujeu and 1997 for Île aux Grues), was not related to a significantly higher density of mature plants (i.e., those present at the onset of senescence). A higher individual fecundity (see RESULTS: Stability through time of environmental filters: Île aux Grues) and a greater proportion of plants bearing mature nutlets are the most probable causes of these increases in seed bank density. Both potential causes may be related to an extended growing season by means of an earlier seedling emergence.

The spatial pattern of seed bank density varied considerably through time. Indeed, there was no significant correlation in the spatial patterns of seed bank density between seasons, and correlograms differed between seasons within a given year and between years for a given season. This spatiotemporal variability in the seed bank may be influenced by several factors among which are seed dispersal, the spatial patterns of seed germination and of seedling emergence, and those of plant survival and fecundity (see also Kalisz, 1991 ; Chambers, 1995 ). Because seed bank samples were taken at adjacent, but not exactly the same geographical positions (within ~5–10 cm) within the 50 x 50 cm quadrats, density variation between sampling periods may reflect a high spatial variability at a scale finer than that of the quadrat size used. Similarly, the absence of significant concordance between seed bank abundance and emergence and fecundity may be related to fine-scale spatial variability. These factors will be analyzed for their effects on population dynamics in the following sections.

Seed dispersal
Nutlets of Floerkea are barochorous: at maturity, they fall to the ground near the parent plant. Houle, McKenna, and Lapointe (1998) experimentally demonstrated that once the seeds are on the ground their mobility is restricted in the forest and most of them remain within 10 cm of the initial site of dispersal up to their germination. Rain splash may be a significant factor for the fine-scale movement of the seeds after primary dispersal. Such fine-scale movement may explain the differences we observed, from July to September, in the spatial pattern of seed bank density and, consequently, local (i.e., microsite) seed bank dynamics. Seed movement should be very limited once the leaf litter has fallen (towards the end of September).

Seed germination and seedling emergence
Despite the low temperatures (~0°C) and heavy snow cover in December, seeds of Floerkea germinate during the early winter in Québec (Houle, McKenna, and Lapointe, 1998 ), as they do in the more southern parts of the species range (Smith, 1983a ; Baskin, Baskin, and McCann, 1988 ). The timing of germination may be particularly important for a species whose photosynthetic part of the life cycle is very short and restricted to the period between snowmelt and tree canopy closure. Yet, the onset of emergence was variable between years on a given site and between sites for a given year. Although the 2-wk difference that we observed in the initiation of seedling emergence does not seem large, it might be significant for such a short-lived plant. Indeed, seedlings that emerge later during the season have a much reduced longevity and most often do not have the time to mature their nutlets before senescence (Smith, 1983a ; see also: Kalisz, 1986 ; Puntieri and Hall, 1996 ). Differences in emergence do not seem to be related to the persistence of the snow cover as such, because the snow has already melted by the time seedlings start to emerge. Differences in soil and air temperatures might be responsible for the between-year and between-site variations in the initiation of seedling emergence.

Although the spatial patterns of seed bank density and emergence probability are highly variable through time, patterns of seedling density are stable (most often concordant between years), suggesting that environmental filters operating on seedling establishment are consistent through time. These filters may be related to soil moisture and leaf litter patterns, both influenced by microtopography (Struik and Curtis, 1962 ; Rogers, 1982 ). We have shown that substrate moisture influences seed germination velocity, although it does not affect overall percentage germination. Consequently, in somewhat drier microsites (e.g., on small mounds, a topographic feature common in temperate forests; Beatty, 1984 ), seeds may thus germinate later in the fall; seedlings from such seeds may emerge later the following spring, have a shorter longevity, and not have the time to mature seeds (Smith, 1983c ). The results of our litter experiment showed that few seedlings emerge from under a thick leaf litter. Thus, litter deposition patterns on the forest floor may have a significant influence on within-population patterns of emergence: thus, microsites where litter and seeds tend to accumulate (e.g., small depressions) are not patches of high seedling density.

Plant survival and reproduction
Most likely a result of density-dependent mortality at the seedling stage, interannual fluctuations in density were generally smaller for the seed-producing individuals (i.e., the majority of those present at the onset of senescence) than for the seedlings. For the same reason (i.e., density-dependent mortality), spatial patterns of individuals at the onset of senescence were even more alike between years than they were for the seedlings. Spatial patterns of survival to maturity were concordant among years, which suggests a certain spatial stability through time for the environmental filters operating on survival (in a density-dependent manner). Consequently, for Floerkea, density-dependent processes may be significant for the regulation of density and spatial structure within populations. We propose that mortality as a result of intraspecific competition, e.g., for water, may be a significant density-dependent factor. Indeed, our experiment on water stress showed that seedlings and juveniles are sensitive to water stress (see also Struik and Curtis, 1962 ; Struik, 1965 ). According to our observations and those of Smith (1983a) , Floerkea's root system is limited to the top 3–5 cm of the soil and does not extend horizontally beyond ~2–3 cm from the plant. In high-density patches where a size hierarchy might develop (Smith, 1983c ), extension of the root system may be severely restricted for some individuals and competition for water may become significant, particularly on well-drained microsites (e.g., on small mounds). Juvenile mortality was consistently higher on Île de Beaujeu than on Île aux Grues, possibly because of earlier increases in air temperature in the spring (this site is farther south than Île aux Grues), higher leaf transpiration, higher competition for water, and more severe drought stress.

If density-dependent mortality operating from the seedling through the reproductive stage dampens population fluctuations and contributes to maintain the same spatial pattern of density through the years, it does not modify the autocorrelation patterns from the seedling to the reproductive stage (the correlograms were very similar in shape and partial Mantel tests provided highly significant values of correlation between patterns). Thus, density-dependent mortality does not fully compensate for local density. Consequently, where seedlings are more abundant, more individuals reach the reproductive stage even after density-dependent mortality. Higher density-dependent mortality in low resource microsites could explain these patterns. The scale of autocorrelation is not altered and appears to be related to microtopographic features of the site.

The spatial patterns of fecundity of Floerkea plants varied significantly among years at Île aux Grues. Kalisz (1991) reported such variations in fecundity for the forest annual Collinsia verna. Fine-scale spatiotemporal changes in soil characteristics (e.g., nutrient) and light availability have been reported for several ecosystems (Crozier and Boerner, 1986 ; Endler, 1993 ; Burghouts, van Straalen, and Bruijnzeel, 1998 ; Finzi, Canham, and van Breemen, 1998 ) and might explain spatio-temporal variations in fecundity for the species studied here. Indeed, plant fecundity in Floerkea responds significantly to variations in nutrient and light availability under controlled conditions (Houle, McKenna, and Lapointe, 1998 ; McKenna and Houle, 1999 ). Small interannual fluctuations in resource patterns may influence the timing of emergence and the seedling growth rate, both of which influence the time available for Floerkea plants to mature their seeds (Smith, 1983b ).

We observed that the timing of senescence was similar among years and between sites and that it was correlated to canopy closure, but also to night (much less to day) temperature increases (Houle, McKenna, and Lapointe, 1998 ). Lower photosynthetic rates with tree canopy development, combined with an increase in dark respiration rates at higher night temperatures, may be largely responsible for the onset of senescence in Floerkea. McKenna (1999) showed that plants exposed to higher temperatures (i.e., 21°C/14°C, day/night) senesced much earlier than plants exposed to more typical spring temperatures (i.e., 16°C/7°C, day/night), when sufficient photosynthetically active radiation (PAR) was provided. Even if PAR did not decrease as a result of canopy closure, senescence would thus occur early simply because temperatures increase.

Our study shows that there are significant fine-scale spatio-temporal variations in the demographic processes of Floerkea proserpinacoides, even in habitats that are considered relatively stable through time. However, density-dependent processes appear to be stabilizing and may reduce the risk of local population extinction. Such processes may be particularly significant for population persistence at the northern limit of the species distribution in North America.


   FOOTNOTES
 
1 The authors thank P. Beaudry, S. Bouchard, E. Despland, J.-M. Gagné, I. Gamache, S. Guay, M. Lafortune, P. Lavoie, J. Plamondon, A.-F. Rocheleau, and J. Turcotte for their technical assistance in the field and the laboratory. Financial support was provided by the Fonds de recherche et de développement technologique en environnement of the ministère de l'Environnement et de la Faune du Québec, and by the Conseil de recherche en sciences naturelles et en génie du Canada. Back

4 Author for reprint requests (gilles.houle{at}bio.ulaval.ca ). Back


   LITERATURE CITED
TOP
 ABSTRACT
 INTRODUCTION
 STUDY SPECIES
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Atmospheric Environment Service. 1993 Canadian climate normals, 1961–1990, Québec. Environment Canada, Atmospheric Environment Service, Ottawa, Ontario, Canada

Bartelt, R. J., and K. L. Mikolajczak. 1989 Toxicity of compounds derived from Limnanthes alba seed to fall armyworm (Lepidoptera: Noctuidae) and European corn borer (Lepidoptera: Pyralidae) larvae. Journal of Economic Entomology 82: 1054–1060[ISI]

Bascompte, J., and R. V. Solé. 1995 Rethinking complexity: modelling spatiotemporal dynamics in ecology. Trends in Ecology and Evolution 10: 361–366[CrossRef]

Baskin, J. M., and C. C. Baskin. 1978 The seed bank in a population of an endemic plant species and its ecological significance. Biological Conservation 14: 125–130[CrossRef][ISI]

———, ———, and M. T. McCann. 1988 A contribution to the germination ecology of Floerkea proserpinacoides (Limnanthaceae). Botanical Gazette 149: 427–431[CrossRef]

Beatty, S. W. 1984 Influence of microtopography and canopy species on spatial patterns of forest understory plants. Ecology 65: 1406–1419[CrossRef][ISI]

Bell, G., and M. J. Lechowicz. 1991 The ecology and genetics of fitness in forest plants. I. Environmental heterogeneity measured by explant trials.