American Journal of Botany 97(5): 770–781. 2010. POPULATION DYNAMICS OF EMPETRUM HERMAPHRODITUM (ERICACEAE) ON A SUBARCTIC SAND DUNE: EVIDENCE OF RAPID COLONIZATION THROUGH EFFICIENT SEXUAL REPRODUCTION1 Stéphane Boudreau2,4, Pascale Ropars2, and Karen Amanda Harper3 2Département de Biologie and Centre d'études nordiques, Université Laval, Chaire de recherche nordique en écologie des perturbations, Local 3047 A/B, Pavillon Alexandre-Vachon, 1045, avenue de la Médecine Québec, Canada G1V 0A6; and 3School for Resource and Environmental Studies, Kenneth C. Rowe Management Building, Dalhousie University, 6100 University Avenue, Suite 5010, Halifax, Nova Scotia, B3H 3J5, Canada The importance of sexual reproduction for clonal plant species has long been underestimated, perhaps as a consequence of the difficulty in identifying individuals, preventing the study of their population dynamics. Such is the case for Empetrum hermaphroditum, an ericaceous species, which dominates the ground vegetation of subarctic ecosystems. Despite abundant seed production, seedlings are rarely observed. Therefore, prevalent seedling recruitment on a subarctic dune system provided an opportunity to study the population dynamics and spatial pattern of the colonization phase of this species. We established a 6-ha grid on the dune systems that extended from the shoreline to the fixed dunes and mapped and measured all E. hermaphroditum individuals in the grid. Moreover, we sampled 112 individuals just outside the grid to identify any allometric relationship between the size and age of the individuals, which allowed us to reconstruct population expansion. The overall size structure suggests that the population is still expanding. In the last 50 yr, E. hermaphroditum advanced more than 200 m in the dune system. Expansion started in the 1960s simultaneously at different distances from the shoreline. Colonization did not proceed gradually from the fixed dune toward the shoreline but instead individuals established earlier in the troughs between the dunes, with an increasingly clumped spatial pattern as the population filled in with time. Key words: allometric relationship; clonal species; dendrochronology; dune colonization; Empetrum hermaphroditum; Ericaceae; population dynamics; seedling establishment; spatial pattern; subarctic environments. Few studies have looked at the population dynamics of clonal species. The difficulty in identifying individuals in the field precludes the use of basic ecological tools such as age/size structures. As a result, the potential of many clonal species to reproduce sexually has long been underestimated. However, while clonal growth allows a species to maintain itself on a site, population expansion relies mainly on the colonization of new areas via the production and dispersal of viable seeds and successful seedling establishment (Eriksson, 1989). The prevalence of both reproductive strategies in clonal species is often linked to the influence of the abiotic environment. In harsh environments such as in subarctic and arctic areas, primary productivity is often limited by suboptimal temperatures and by the short duration of the growing season. Moreover, pedogenic processes associated with soil development are slow, resulting in substrates with low nutrient availability (Kershaw, 1983; Harper and Kershaw, 1997). It is therefore not surprising to observe that in such environments, where costs of sexual reproduc1 tion are high and probability of successful seedling establishment is low, more than 90% of the plant species are able to spread vegetatively (Bliss, 1971). Crowberry, Empetrum nigrum L. subsp. hermaphroditum (Hagerup) Böcher (referred to as E. hermaphroditum from now on), found in temperate, subarctic, and arctic biomes in North America, northern Europe, and Asia (Szmidt et al., 2002), is one of these species. Despite abundant berry and seed production (588 viable seeds·m−2 reported by Vieno et al., 1993), this 15-cm tall, creeping, evergreen shrub is believed to rely mainly on vegetative propagation to maintain its population. In forested ecosystems, it can dominate the shrub layer and interfere with germination of tree seeds and seedling growth (Zackrisson et al., 1995, 1997; Nilsson et al., 2000) and with forest productivity (Zackrisson et al., 1996; DeLuca et al., 2002; Wardle et al., 2003). In such forested ecosystems, vegetative propagation makes it difficult to delimit individuals, preventing the study of population dynamics. Field observations of seedlings are rare in old forests despite massive fruit production (Eriksson, 1989; same results for E. nigrum reported in Bell and Tallis, 1973), although seedling establishment can be observed on forest sites where the mineral soil has been exposed and on decomposing logs or stumps (M.-C. Nilsson, personal observation, in Nilsson and Wardle, 2005). These findings contrast drastically with recent observation of abundant recruitment on a nonforested subarctic dune system near Kuujjuarapik-Whapmagoostui in northern Québec (S. Boudreau, personal observation). On this dune system, which supports early successional communities, E. hermaphroditum seedling recruitment appears to be more frequent, resulting in an apparent rapid population expansion at this particular location. Therefore, this unique situation pres- Manuscript received 30 September 2009; revision accepted 15 March 2010. The authors thank Ian Boucher, Julie Faure-Lacroix, Pierre-Samuel Proulx, and Marie-Pascale Villeneuve-Simard for help in the field and laboratory. They also thank the Centre d’études nordiques for logistical support. This project was financially supported by the Ministry of Indian and Northern Affairs of Canada (NTSP Program) and by the Natural Sciences and Engineering Research Council of Canada (NSERC) via the Northern Research Chair on Disturbance Ecology. 4 Author for correspondence (e-mail: stephane.boudreau@bio.ulaval. ca); phone: + 418-656-2131 #3857; fax: + 418 656-2043 doi:10.3732/ajb.0900304 American Journal of Botany 97(5): 770–781, 2010; http://www.amjbot.org/ © 2010 Botanical Society of America 770 Boudreau et al.—EMPETRUM population expansion on a dune system May 2010] ents an opportunity to study the population dynamics of the early establishment phase of a clonal species. In this study, we investigated the population dynamics and spatial pattern of an E. hermaphroditum population on a subarctic dune system. Specifically, our objectives were (1) to identify allometric relationships between the age and the size of the E. hermaphroditum individuals, (2) to reconstruct the expansion of E. hermaphroditum population on the dune system, (3) to determine how colonization has proceeded along the gradient from the fixed dunes to the shoreline, and (4) to assess changes in spatial pattern through time across the dune system. Trends in spatial pattern can reveal insight into the processes affecting colonization. Aggregation of plants is common in early stages of establishment due to facilitation (Cutler et al., 2008), as has been found on dune sites (Franks, 2003). Subsequently, clumps enlarge and coalesce creating a more homogeneous spatial pattern (Yarranton and Morrison, 1974; Cutler et al., 2008). MATERIALS AND METHODS Study area—The study area is located on the Hudson Bay coast at the mouth of the Great Whale River in subarctic Québec, Canada (Fig. 1), near the Inuit village of Kuujjuarapik. The large quantities of sediments brought in by the river combined with the rapid isostatic rebound (Allard and Tremblay, 1983; Filion and Morisset, 1983) result in an important beach progradation (Ruz and Allard, 1994) and thus a relatively rapid succession on coastal dunes (Houle, 1996). The study site extends from the upper shore to the fixed dunes. It corresponds to a section of the dune for which the recruitment of E. hermaphroditum is believed to be representative of the 3-km-long dune system. Moreover, it is relatively unaffected by anthropogenic activities, compared to other areas. Elevation measures (with an optical theodolite) revealed that the study site fully encompasses two dune ridges perpendicular to the shore (located respectively at 50 m and 175 m from the vegetation line on the front of the foredune) and the beginning of a third one (Fig. 2). The first trough on the back of the foredune is ca. 60 m wide, while the second one varies between 60 m and 110 m. The low- 771 est elevation (used as the “0” reference) is found on the upper shore, while the highest point (434 cm) was measured on the second dune ridge. Climatic data provided by the Kuujjuarapik–Whapmagoostui weather station (55°16′N, 77°45′W; 10.4 m a.s.l.) reveal an annual mean temperature of –4.4°C for 1971–2000 with the highest and lowest mean monthly temperatures recorded in August (11.4°C) and January (–23.6°C), respectively. Annual precipitation averages 650 mm, of which 37% falls as snow (Environment Canada, 2007). Data collection—In May 2007, we established a 300 m × 200 m grid on the sand dune (length perpendicular to the shore) extending from the first plant found on the upper shore to the fixed dunes. The grid was bordered by large E. hermaphroditum clones at its farthest end, i.e., 300 m from the vegetation line on the foredune. However, the colonization of E. hermaphroditum in the section covered by the grid appeared to be recent. Every E. hermaphroditum individual was mapped by subdividing the grid into 10 m × 10 m grid cells and walking along 2-m-wide transects within each cell. Mapping was done after snowmelt and before the growth season to maximize the detection of small E. hermaphroditum individuals, which are easy to detect at that time of year because their reddish leaves contrast with the yellow dead herbaceous cover. For each individual, we recorded its coordinates in the grid, the shape of the crown (elliptical, semielliptical, triangular) and its diameter along two perpendicular axes. On sandy substrate with little microtopographic variation, E. hermaphroditum generally has an almost circular crown. To infer germination bed characteristics for E. hermaphroditum, we categorized the vegetation around smaller individuals (crown < 314 cm2 or 10 cm radius). On the basis of visual inspection of the topographic map and the spatial pattern of E. hermaphroditum (Fig. 2), we divided the study area into five zones of relatively homogeneous patterns of E. hermaphroditum (Table 1). Zone A extends from the shore line to the foredune ridge, while zone C encompasses the second dune ridge. Zones B and D correspond roughly to the first and second trough, respectively. Zone E extends across a third ridge to the end of the grid. Zone D was further subdivided into two subsections based on apparent differences in spatial pattern. Vegetation cover was assessed in August 2007 by taking digital images of the vegetation inside 50 cm × 50 cm quadrats located every 10 m along ten 300 m-long transects spaced 20 m apart and running from the upper shore to the fixed dunes. Vegetation cover was evaluated on the photographs with a pinpoint method, using a 64-dot grid that was superimposed on the image. The dominant plant functional group (lichen, moss, herbaceous plants, Lathyrus japonicus, E. hermaphroditum, Vaccinium vitis-idaea, or Sibbaldiopsis tridentata) or mineral soil under each dot was identified. Reconstruction of population expansion—To infer the age of the mapped individuals, we sampled 112 individuals of E. hermaphroditum whose diameter ranged from 2.5 cm to 130 cm (area: ca. 20 cm2 to ca. 53 000 cm2) collected outside but in the vicinity of the plot between mid-May and mid-June 2007. The main stems were uprooted from the periphery to the center of the individual, and the root collar was harvested by cutting the main roots and stems. Samples were cleaned and left to dry at room temperature for at least 8 wk before processing. For minimal age determination, thin sections (ca. 20 µm) of E. hermaphroditum root collar were cut with a sledge microtome after samples had been boiled for 2 h. The thin sections were then stained with a 1% solution of safranin (w/v in distilled water) and subsequently mounted on slides using a lowviscosity mounting medium. Growth rings were then counted using a dissecting microscope. We used the allometric relationship between crown size and age to reconstruct expansion of the population in the grid through time. We assumed that mortality of E. hermaphroditum was rare in this ecosystem because of the complete absence of dead individuals during our field observations. Our assumption is also supported by the consideration that decomposition rate on a subarctic dune system, i.e., a cold and dry environment, is low. The allometric relationship allowed us to infer and map the E. hermaphroditum population structure for every decade (1967, 1977, 1987, 1997, and 2007). We acknowledge, however, that it is probable that some seedlings became established and died at a young age (1–2 yr) although Hansen (1998, in Tybirk et al., 2000) reported a high survival rate for this particular age class. Fig. 1. Study area in subarctic Québec, Canada. Spatial pattern analysis—We used point pattern analysis to assess the degree of aggregation or amount of clumping of individuals of E. hermaphroditum at different spatial scales. Point pattern analysis examines mapped positions American Journal of Botany 772 to compare the observed number of other individuals to the expected number within circles or rings of different sizes or scales around each individual (Fortin and Dale, 2005). Because the spatial pattern of E. hermaphroditum in the grid was likely to vary across the dunes and valleys and therefore be nonstationary (i.e., the spatial pattern was not expected to be constant throughout the grid), few methods were available for analyzing the spatial pattern of the entire study area. One of the only available methods appropriate for nonstationary data are Getis and Frankin’s L(d), a version of the better-known Ripley’s K function in which L(d) is calculated separately for each point (Getis and Franklin, 1987; Perry, 2004): Lˆ i d A nj 1 k ij / n 1 where Σkij is the summation over all points j within distance d of point i, n is the number of points, and A is the area (Perry, 2004). Results of this local analysis are often displayed using a contour plot to show areas of aggregation and areas of regularity for each point rather than simply calculating a global average (Perry, 2004). Although various distances d were used with edge correction, only the results using d = 10 are presented because this scale appears to show the optimum amount of variation within our study area. All our spatial analyses were conducted using SpPack, a Microsoft (Redmond, Washington, USA) Excel Add-In (Perry, 2004). We also used another approach to overcome the problem of analyzing nonstationary data; we performed spatial analyses within the five previously identified zones (zones B, C, D1, D2, E) where the spatial pattern of E. hermaphroditum appeared to be relatively stationary. Within each zone, neighborhood density function (NDF) analyses were conducted. The NDF is similar to the more common Ripley’s K function, but is able to isolate patterns at different distances because it is not cumulative. Also known as the O-ring statistic (Wiegand and Moloney, 2004), it is a function of the number of individuals found within annuli of different distance classes from each individual (Perry et al., 2006). Univariate NDF analyses with edge-weighted area correction were performed on E. hermaphroditum individuals within each zone for each decade. Analyses were only performed if there were at least 10 individuals in every zone decade combination to ensure a sufficient sample size for determining the spatial pattern. Distance classes included 0.1-m intervals up to 2 m and 1-m intervals up to half the width of the zone. Significant differences from a random distribution [Vol. 97 (NDF = 1) were determined using 95% confidence intervals calculated using Monte Carlo tests with 499 replicates. RESULTS Vegetation of the study site— The abundance of the main plant functional groups varied with distance from the upper beach (Table 1). From the upper shore to the foredune crest (zone A), the vegetation cover is dominated mainly by Leymus mollis and to a lesser extent by Lathyrus japonicus. On the back of the foredune and in the first trough (zone B), herbaceous cover declined, while the cover of mosses, lichens, and Sibbaldiopsis tridentata increased. The second dune ridge (zone C) was dominated by herbaceous species and mosses, while lichen abundance decreased. However, lichens dominated the second trough (zones D1 and D2), reaching a cover of over 60%. Finally, zone E was dominated by mosses and lichens. Empetrum hermaphroditum cover was concentrated in the first trough and at the end of the second one, in front of a third dune ridge located just outside the study area. Population structure of E. hermaphroditum— A total of 1197 individuals were mapped and measured in the grid (Fig. 2). Individual crown area ranged from 0.4 cm2 to >230 000 cm2, and most individuals had an elliptical crown (90.1%), with others having circular (5.1%), triangular (3.7%), semielliptical (0.8%), or rectangular (0.2%) crowns. In most cases, individuals were identified by following the main stems up to the root collar. However, we probably underestimated the number of Table 1. Cover of different functional groups of plant species in the study area as inferred by the percentage of hits of each functional group on the picture (64 dots per picture). Zone A A A A A A A A B B B B B B C C C C D1 D1 D2 D2 D2 D2 D2 E E E E E a Distance (m) Herbs* Lathyrus Mosses Sibbaldiopsis Lichens Empetrum Vaccinium 0–10 10–20 20–30 30–40 40–50 50–60 60–70 70–80 80–90 90–100 100–110 110–120 120–130 130–140 140–150 150–160 160–170 170–180 180–190 190–200 200–210 210–220 220–230 230–240 240–250 250–260 260–270 270–280 280–290 290–300 11.1 22.5 92.3 97.7 100.0 97.5 98.1 96.9 61.9 49.1 46.7 41.3 30.6 42.5 38.1 31.3 45.8 38.6 15.2 27.7 9.8 9.1 11.2 17.4 19.1 12.7 19.1 7.2 23.4 19.3 — 16.6 2.3 2.3 — 0.6 1.3 0.6 0.5 0.3 0.6 — 0.6 — — — — — — — — — 0.7 0.2 — 0.2 0.2 — 0.2 0.4 — — — — — — 0.2 2.0 29.8 28.3 17.0 14.1 20.0 16.3 42.2 26.3 36.6 45.3 24.8 24.2 16.2 10.0 14.1 9.8 29.3 23.4 15.4 16.0 21.7 23.2 — — — — — — — — 6.4 4.1 9.4 9.7 5.0 7.8 7.2 11.2 10.5 13.2 10.0 10.4 12.1 13.1 10.9 7.6 10.5 9.8 6.8 8.2 10.4 8.0 — — — — — — — — 0.5 18.0 14.5 32.8 40.3 10.0 12.5 31.3 6.9 2.7 50.0 36.1 59.2 65.6 59.8 64.8 38.3 44.3 42.2 45.1 23.6 44.1 — — — — — — — — — 0.3 11.4 — — 20.0 — — — — — — — — — — 2.7 — 16.2 22.5 18.4 2.0 — — — — — — — — — — — — — — — — — — — 1.2 2.7 2.2 1.6 — — 0.2 — 0.8 2.3 1.6 Including Leymus mollis May 2010] Boudreau et al.—EMPETRUM population expansion on a dune system 773 Fig. 2. Topography, location, and relative size of the individuals identified inside the 6-ha grid along the dune system. The 43 aggregates located farthest away from the shore are not shown. See Fig. 5 to visualize their approximate positions (1967 map). Colors represent the topographical gradient. individuals in the grid because 7% of the individuals sampled for the development of allometric relationships were in fact two individuals with intermingling root collars that were impossible to differentiate in the field. We also recorded 43 aggregates, defined here as E. hermaphroditum patches too large to be identified with certainty as one individual. All the aggregates were recorded in the last 50 m of the grid on the fixed dunes and are believed to be relatively old. No dead or senescing individuals were recorded inside the grid, suggesting that once well-established, E. hermaphroditum survival is high. The size structure of E. hermaphroditum followed an inverse J-shaped curve in all zones, typical of a population with significant recruitment (Fig. 3). However, zones C and D2 presented an almost equal number of individuals in the first few diameter classes, but the other zones did not. In Zone B, there are many intermediate size classes with numerous individuals resulting in American Journal of Botany 774 Fig. 3. [Vol. 97 Size structure of Empetrum hermaphroditum individuals in the different zones. a more even distribution, while zone E contains all the large aggregates. Zone D1 has the steepest curve with few individuals over 70 cm diameter. Overall, almost half of the individuals (45%) had a mean diameter <20 cm (or were younger than 9.8 yr, according to the allometric relationship), suggesting that the population is rapidly expanding. Most individuals originated from seeds, as revealed by their isolated position on the dune and by the morphology of their root collar. Young E. hermaphroditum in- dividuals were mainly associated with mosses and lichens, the two dominant or subdominant functional groups in the vicinity of the seedlings (73.9% and 63.3% of the cases, respectively). Seedling establishment on other substrates was much less frequently observed. Allometric relationships— Of the 112 individuals sampled, 77 were used to build the allometric relationship between the crown size (cm2) and the age (yr). Samples of various sizes May 2010] Boudreau et al.—EMPETRUM population expansion on a dune system were excluded from the analysis if they were composed of more than one individual (intermingled collars; 8), if it was impossible to get proper thin sections (sample rotten inside; 15), or if accurate dating could not be obtained from thin sections (12). Minimum age of the sampled individuals varied between 6 and 35 yr old. There was a significant relationship between the crown area and the minimum age of the individuals (Fig. 4). The crown area–minimum age relationship follows a power function (y = 0.032x4.0303, F1, 74 = 736.4, P < 0.001) with a high determination coefficient (R2) of 0.909. Consequently, crown area is a good predictor of the age of an individual and was used to reconstruct the E. hermaphroditum population expansion in the study area. To do so, we calculated the establishment year of the 1197 mapped individuals. As a result, our reconstruction does not take into account individuals that died before 2007. However, because field observations suggest a low mortality, we believe that our reconstruction is a reliable representation of the population expansion. Population expansion— Trends in the abundance of E. hermaphroditum with time varied considerably among the five zones (Figs. 5, 6). Prior to 1967, only 49 individuals, including the 43 aggregates, were present on the farthest portion of the second trough zone E. However, numerous other aggregates located just outside the grid were also believed to be well established at that time. By 1967, E. hermaphroditum were found in all zones but zones C and D1. The first individuals in those two zones are believed to have established in 1971 and 1978, respectively. Both the number of individuals as well as the area occupied by E. hermaphroditum increased substantially over the past few decades in zones B and E but with more moderate abundance in zone B. In zone D1, E. hermaphroditum individuals had become numerous, reaching similar densities as zone E by 2007, but with very little cover. Zones C and D2 also had low cover of E. hermaphroditum with low and moderate densities, respectively. Population growth followed an exponential growth curve in all zones (Table 2). Spatial pattern— In 2007, aggregation of individual E. hermaphroditum within 10 m was strongest within the first trough (100 m), along the edge of the second trough nearest the Bay (200 m) and along the slope facing the Bay of the same trough (250–300 m; Fig. 7). The highest amount of aggregation, with Fig. 4. Allometric relationship between the crown size of the Empetrum hermaphroditum individuals and their minimum age (N = 77). 775 values of L(d) greater than 20, was found farthest from the shore. These results for local spatial pattern analysis illustrate the nonstationary nature of the data and provide evidence for the zones we have delineated. The spatial pattern of E. hermaphroditum varied not only with distance from shore but also with time. In all zones, E. hermaphroditum was clumped at more distances and at finer scales in 2007 compared to previous decades (Figs. 8, 9). Thus, there appears to be a shift toward a higher degree of clumping at close proximity or denser patches of E. hermaphroditum individuals as density increased. This trend was apparent even at very fine scales of 0.1-m intervals. In 1967 and 1977, E. hermaphroditum individuals were regularly spaced at intermittent distances up to 25 m in zone E, but this even pattern consisted mostly of the aggregates, which likely consisted of more than one individual each. Despite the presence of these aggregates, zone E provides a good example of the progressive development of the spatial pattern from clumping at only a few distances in earlier decades to significant clumping at most distances up to 25 m in 2007, a trend that is also pronounced in zone B but present in all zones. Particularly in 2007, NDF values at very fine scales (up to 0.5 m) were much greater, indicating greater aggregation, in zones B, C, and E compared to zones D1 and D2; however, this was significant only at 0.1 m for zone C. At broader scales, zones B and E had significant clumping up to 25 m in 2007, whereas zone D1 had significant clumping only up to 6 m and E. hermaphroditum individuals in zones C and D2 were only clumped at intermittent distances beyond 5 m. DISCUSSION Seedling establishment was prominent throughout our study site beyond the first dune ridge. This result is in opposition with the generally accepted idea that sexual reproduction is not important for the dynamics of clonal species. In fact, our study does not corroborate previous studies on Empetrum sp., which reported that seedling establishment was rare (Bell and Tallis, 1973; Eriksson, 1989) and that layering was the most important process for population maintenance. These studies, however, were mainly conducted in late-successional sites (forested sites) in which the dense vegetation layer at the soil level and the asymmetric competition with the shrub and tree component might limit seedling establishment. Moreover, a recent study in northern Sweden suggests that seedling establishment in forested ecosystems is more common than previously thought, resulting in a rather high genetic diversity even in old-successional sites (Szmidt et al., 2002). The results presented here showed that Empetrum hermaphroditum has the ability to colonize a subarctic dune system on a decadal time scale via efficient seed production and seedling establishment. Most of the 1197 individuals identified inside the 6-ha plot originated from seed as revealed by their proper root collar and their isolated position on the dune. The numerous small individuals recorded in the grid suggest that the population is still rapidly expanding. At this stage, layering is not believed to be an important process for population dynamics in the grid. Layering occurs mainly when the crown of old individuals starts to die out near the root collar. At that point, the adventitious roots ensure the survival of the distal extremity of the main stems. In our study, however, most individuals had a healthy crown showing no sign of mortality. 776 American Journal of Botany [Vol. 97 Fig. 5. Maps of the locations of individual Empetrum hermaphroditum in the five decades. This reconstruction of the population expansion from 1967–2007 is based on the allometric relationship. The E. hermaphroditum population expansion in our study area occurred in the last 55 yr. Based on our allometric relationship, ca. 98% of the individuals are younger than 55 yr old. With the exception of the large aggregates previously established in the grid, which were probably responsible for most of the recruitment at the beginning of the population expansion, E. hermaphroditum advanced >220 m in the last 50 yr. It colonized almost simultaneously the second trough (zones D1 and D2), where lichens are abundant, and the first trough (zones B) where graminoid species dominate and where the substrate is more mobile. Only the second dune ridge (zones C) was not colonized by E. hermaphroditum, probably a result of high wind exposure that led to the near absence of snow during the winter. These results do not support the hypothesis of a colonization front slowly advancing from the fixed dune to the foredune. Colonization of the first trough occurred via the establishment of a few individuals that have probably acted as seed providers for subsequent seedling establishment. Such rapid colonization of the dune system suggests that it was triggered by some environmental change because the rate of colonization is much faster than the rate of dune progradation. Different hypotheses may explain this rapid expansion of E. hermaphroditum. First, more favorable climatic conditions could have increased the reproductive output of individuals, May 2010] Boudreau et al.—EMPETRUM population expansion on a dune system 777 Fig. 6. Trends in (A) density and (B) proportion of area of Empetrum hermaphroditum over time for the five zones of the study area. and, therefore, increased seedling establishment and survival. Subarctic Québec is now experiencing rapid warming since the 1990s after long being considered a zone of climatic inertia (Allard et al., 1995). Laliberté and Payette (2008) showed a simultaneous expansion of white spruce along the Hudson Bay coast, supporting our results. The rapid colonization of E. hermaphroditum might also be linked to increased disturbance caused by human settlement in the area following the construction of the military base in the 1950s. Such disturbances, like increased trampling or ATV traffic might have created suitable sites for seed germination. This hypothesis is, however, not supported by our results on the germination seedbed. In fact, it appears that E. hermaphroditum seeds can germinate in either mosses or lichens, i.e., undisturbed vegetation. Moreover, the analyses of aerial photographs showed that the particular area where the grid was located had undergone only minor disturbance since the late 1940s. Finally, establishment following disturbance would more likely have occurred all at once rather than with the gradual increase in density that we observed. Another hypothesis that might explain the rapid advance of E. hermaphroditum is the abundance of foxes on the dune system. Although no data are available to characterize its abundance, several dens were observed during the mapping exercise. Because red foxes have previously been shown to be abundant near residential areas (Harris and Rayner, 1986), we hypothesize that the red fox population increased on the dune system after community settlement. Foxes, as other carnivorous animals, often complete their diet with fruits (Serafini and Lovari, 1993; Lovari et al., 1994), seeds (Sargeant et al., 1986), or coniferous cones (Sklepkovych, 1994) depending on their prey availability. Such a hypothesis is supported by personal field observations of red fox scats containing numerous E. hermaphroditum seeds. There is at least one report in the literature of increased seed germination and seedling survival following Table 2. Zone B C D1 D2 E All zones transit via the digestive tract of red fox (Celtis australis; Traba et al., 2006). Trends in the spatial pattern of this E. hermaphroditum population reveal some insight into the processes affecting establishment. Overall, there was significant clumping over many scales. Cutler et al. (2008) also found strong autocorrelation or clumping in E. nigrum colonizing lava flows, but only at shorter distances of less than 7 m, after which there was significant negative autocorrelation. However, they investigated the spatial pattern of frequency rather than mapped individuals. Aggregation of plants on primary succession sites such as initial colonization on dunes likely reflects the effects of facilitation on seedlings. Franks (2003) found more seeds and greater seedling emergence under plants as compared to open dune sites, supporting nucleation of the expansion of patches (nuclei) as sites of colonization (Blundon et al., 1993). In our study, E. hermaphroditum individuals became more densely aggregated over time. Early establishment began with an apparently random pattern with only slight clumping at intermediate scales. Individuals arriving later established very close to other plants, even as little as 10 cm apart, resulting in clumps of various sizes. Clumping at fine scales of small plants in earlier decades that later died is unlikely because we observed no dead plants. Although spatial patterns becoming more intense in early succession have been found in other studies (e.g., Dale and Blundon, 1990), clumps usually enlarge and coalesce (Yarranton and Morrison, 1974; Cutler et al., 2008). Our apparent contradictory findings may be because we studied the spatial locations of individuals instead of looking at cover as a measure of abundance. Although clones may amalgamate into large, continuous patches, individual E. hermaphroditum plants would still be aggregated within the patches. By considering individuals rather than cover of E. hermaphroditum, we were able to provide evidence of nucleation with individuals establishing close to other plants as colonization proceeds. Exponential models of population growth from 1960 to 2007 in the different zones. Period Equation R2 F df P 1965–2007 1971–2007 1979–2007 1963–2007 1960–2007 1960–2007 y = 2.0343e0.1148x y = 0.3625e0.1059x y = 0.0069e0.2268x y = 0.2211e0.1552x y = 29.947e0.0602x y = 28.615e0.0803x 0.912 0.949 0.969 0.935 0.968 0.985 423.6 646.0 840.9 634.0 1380.3 2963.0 1, 41 1, 35 1, 27 1, 43 1, 46 1, 46 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 Notes: The inversed J-shape of the overall size structure (all zones) combined with the absence of dead individuals suggests that the population is still in an exponential growth phase. For this reason, the exponential model was tested first. Because all regressions were significant, other models were not tested. 778 American Journal of Botany [Vol. 97 Fig. 7. Pattern of aggregation and regularity for Empetrum hermaphroditum individuals (circles) at a scale of d = 10 in the study area. Distances along the x- and y-axes in m are the same as in Fig. 2. Orange and yellow areas have the most aggregated clumps and the blue shading represents the areas with the most regularly spaced individuals. The establishment of new E. hermaphroditum seedlings close to other E. hermaphroditum plants suggests that competition was becoming less important with continued establishment of the population, perhaps due to increased availability of resources. However, Cutler et al. (2008) proposed that there is less competition in the earlier stages when stressful conditions led to positive facilitative interactions. Seed dispersal does not appear to be a limiting factor in early stages because there was little clumping in earlier decades, especially at short distances. Clumping might have become more prevalent in later stages as seedlings established in the remaining available microsites as dictated by microtopography. Cutler et al. (2008) also found evidence that pattern can be structured by edaphic gradients or physical conditions; an unexpected increase in pattern intensity (more dense aggregation) in later stages was observed on a lava flow including for E. nigrum. Differences in the spatial pattern among the different zones provide further insight into the establishment of the E. hermaphroditum population. The population was already established in zone E, the part of the study area farthest from the shore, and early establishment was also observed in zone B, the first trough. In both zones, there was significant clumping at greater distances, indicating densely aggregated E. hermaphroditum at scales up to 25 m, compared to the other zones where establishment started later. Again the clumping may be due to the populations filling in as seedlings established in the remaining available microsites in these zones. In zone D2, the part of the trough next to the well-established population in zone E, establishment was only somewhat later. Here there was also considerable clumping by 2007, although not at fine scales. With only moderate density and low cover, competition may still be preventing clumping at fine scales. Zone D1 presents an interesting pattern with sudden dense late establishment in a narrow row along the shore side of the trough that was clumped by 2007. Such a pattern could be related to the presence of a small trail from animal or human activity. Finally, zone C, the second ridge which is more exposed to wind and salt spray, is one of the latest zones for establishment and is still very sparse. The extremely high peak of clumping at 0.1 m in this zone is an exception to the trend of E. hermaphroditum becoming more densely aggregated with time and is probably a result of two individuals in close proximity, which may have skewed the results due to the low density within the zone. Otherwise, clumping was minimal at fine scales. Conclusion— To our knowledge, this is the first study to emphasize the population dynamics of E. hermaphroditum. The open vegetation structure of the dune system combined with the May 2010] Boudreau et al.—EMPETRUM population expansion on a dune system 779 Fig. 8. Neighborhood density function (NDF) results for the different zones at scales of 0–2 m and 0–25 m in different decades. Higher values of NDF indicate aggregation of Empetrum individuals at that distance. Note the different scale for NDF for zone C at scales 0-2 m. Note that for zone C in 1997, there is a peak at 1.4 m that is almost identical (and therefore hidden) to the one in 2007. Otherwise, it is always zero for the graph of 0 to 2 m. 780 American Journal of Botany [Vol. 97 Fig. 9. Significance of univariate neighborhood density function results for the different zones at scales of 0–2 m and 0–25 m in different decades. Solid lines and filled circles indicate significant aggregation of Empetrum individuals at those distances; dotted lines and open circles indicate significant regularity of Empetrum individuals at those distances. May 2010] Boudreau et al.—EMPETRUM population expansion on a dune system recent population expansion allowed us to do so. We demonstrated that the age of an individual could be inferred from its size via the development of a strong allometric relationship between crown size and minimal age. According to our reconstruction, E. hermaphroditum population advanced >220 m toward the shoreline in the last 55 yr. However, colonization did not proceed as a gradual advance of the population. Instead, the first and second troughs were colonized at the same time. Finally, the spatial pattern of the population reveals greater clumping with time in areas where establishment started earlier, suggesting that individuals fill any remaining microsites without intraspecific competition in later stages of population expansion. LITERATURE CITED Allard, M., and G. Tremblay. 1983. La dynamique littorale des îles Manitounuk durant l’Holocène. Zeitschrift fur Geomorphologie 47 (Supplement): 61–95. Allard, M., B. L. Wang, and J. A. Pilon. 1995. Recent cooling along the southern shore of Hudson Strait, Quebec, Canada, documented from permafrost temperature-measurements. Arctic and Alpine Research 27: 157–166. Bell, J. N. B., and J. H. Tallis. 1973. Biological flora of British-Isles— Empetrum nigrum L. Journal of Ecology 61: 289–305. Bliss, L. C. 1971. Arctic and alpine plant life cycles. Annual Review of Ecology and Systematics 2: 405–438. Blundon, D. J., D. A. Macissac, and M. R. T. Dale. 1993. Nucleation during primary succession in the Canadian Rockies. Canadian Journal of Botany 71: 1093–1096. Cutler, N. A., L. R. Belyea, and A. J. Dugmore. 2008. The spatiotemporal dynamics of a primary succession. Journal of Ecology 96: 231–246. Dale, M. R. T., and D. J. Blundon. 1990. Quadrat variance analysis and pattern development during primary succession. Journal of Vegetation Science 1: 153–164. Deluca, T. H., M. C. Nilsson, and O. Zackrisson. 2002. Nitrogen mineralization and phenol accumulation along a fire chronosequence in northern Sweden. Oecologia 133: 206–214. Environment Canada. 2007. National Climate Data and Information Archive [online]. Website http://www.climate.weatheroffice.ec.gc.ca [accessed 4 September 2009]. Eriksson, O. 1989. Seedling dynamics and life histories in clonal plants. Oikos 55: 231–238. Filion, L., and P. Morisset. 1983. Eolian landforms on the east coast of Hudson Bay, northern Québec. In P. Morisset and S. Payette [eds.], Treeline ecology, Proceedings of the Northern Québec Tree-Line Conference, vol. 47, 73–94. Nordicana, Université Laval, Quebec, Canada. Fortin, M.-J., and M. Dale. 2005. Spatial analysis: A guide for ecologists. Cambridge University Press, Cambridge, UK. Franks, S. J. 2003. Facilitation in multiple life-history stages: Evidence for nucleated succession in coastal dunes. Plant Ecology 168: 1–11. Getis, A., and J. Franklin. 1987. Second-order neighborhood analysis of mapped point patterns. Ecology 68: 473–477. Harper, K. A., and G. P. Kershaw. 1997. Soil characteristics of 48-yearold borrow pits and vehicle tracks in shrub tundra along the CANOL no. 1 pipeline corridor, Northwest Territories, Canada. Arctic and Alpine Research 29: 105–111. Harris, S., and J. M. V. Rayner. 1986. Urban fox (Vulpes vulpes) population estimates and habitat requirements in several British cities. Journal of Animal Ecology 55: 575–591. Houle, G. 1996. Environmental filters and seedling recruitment on a coastal dune in subarctic Quebec (Canada). Canadian Journal of Botany 74: 1507–1513. Kershaw, G. P. 1983. Some abiotic consequences of the CANOL Crude Oil Pipeline Project, 35 years after abandonment. Permafrost: Fourth 781 International Conference Proceedings, 595–600. University of Alaska, Fairbanks, Alaska, USA. Laliberté, A. C., and S. Payette. 2008. Primary succession of subarctic vegetation and soil on the fast-rising coast of eastern Hudson Bay, Canada. Journal of Biogeography 35: 1989–1999. Lovari, S., P. Valier, and M. R. Lucchi. 1994. Ranging behavior and activity of red foxes (Vulpes vulpes, Mammalia) in relation to environmental variables, in a Mediterranean mixed pinewood. Journal of Zoology 232: 323–339. Nilsson, M. C., and D. A. Wardle. 2005. Understory vegetation as a forest ecosystem driver: Evidence from the northern Swedish boreal forest. Frontiers in Ecology and the Environment 3: 421–428. Nilsson, M. C., O. Zackrisson, O. Sterner, and A. Wallstedt. 2000. Characterisation of the differential interference effects of two boreal dwarf shrub species. Oecologia 123: 122–128. Perry, G. L. W. 2004. SpPack: Spatial point pattern analysis in Excel using Visual Basic for Applications (VBA). Environmental Modelling & Software 19: 559–569. Perry, G. L. W., B. P. Miller, and N. J. Enright. 2006. A comparison of methods for the statistical analysis of spatial point patterns in plant ecology. Plant Ecology 187: 59–82. Ruz, M. H., and M. Allard. 1994. Foredune development along a sub-arctic emerging coastline, eastern Hudson Bay, Canada. Marine Geology 117: 57–74. Sargeant, A. B., S. H. Allen, and J. P. Fleskes. 1986. Commercial sunflowers: Food for red foxes in North Dakota. Prairie Naturalist 18: 91–94. Serafini, P., and S. Lovari. 1993. Food habits and trophic niche overlap of the red fox and the marten in a Mediterranean rural area. Acta Theriologica 38: 233–244. Sklepkovych, B. 1994. Arboreal foraging by red foxes, Vulpes vulpes, during winter food shortage. Canadian Field Naturalist 108: 479–481. Szmidt, A. E., M. C. Nilsson, E. Briceno, O. Zackrisson, and X. R. Wang. 2002. Establishment and genetic structure of Empetrum hermaphroditum populations in northern Sweden. Journal of Vegetation Science 13: 627–634. Traba, J., A. Sagrario, H. Jesus, and M. C. Clamagirand. 2006. Red fox (Vulpes vulpes L.) favour seed dispersal, germination and seedling survival of Mediterranean hackberry (Celtis australis L.). Acta Oecologica 30: 39–45. Tybirk, K., M. C. Nilsson, A. Michelson, H. L. Kristensen, A. Shevtsova, M. T. Strandberg, M. Johansson, et al. 2000. Nordic Empetrum dominated ecosystems: Function and susceptibility to environmental changes. Ambio 29: 90–97. Vieno, M., M. Komulainen, and S. Neuvonen. 1993. Seed bank composition in a subarctic pine-birch forest in Finnish Lapland: Natural variation and the effects of simulated acid rain. Canadian Journal of Botany 71: 379–384. Wardle, D. A., M. C. Nilsson, O. Zackrisson, and C. Gallet. 2003. Determinants of litter mixing effects in a Swedish boreal forest. Soil Biology & Biochemistry 35: 827–835. Wiegand, T., and K. A. Moloney. 2004. Rings, circles, and null-models for point pattern analysis in ecology. Oikos 104: 209–229. Yarranton, G. A., and R. G. Morrison. 1974. Spatial dynamics of a primary succession: Nucleation. Journal of Ecology 62: 417–428. Zackrisson, O., M. C. Nilsson, A. Dahlberg, and A. Jaderlund. 1997. Interference mechanisms in conifer–Ericaceae–feathermoss communities. Oikos 78: 209–220. Zackrisson, O., M. C. Nilsson, I. Steijlen, and G. Hornberg. 1995. Regeneration pulses and climate vegetation interactions in nonpyrogenic boreal Scots pine stands. Journal of Ecology 83: 469–483. Zackrisson, O., M. C. Nilsson, and D. A. Wardle. 1996. Key ecological function of charcoal from wildfire in the boreal forest. Oikos 77: 10–19.