Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Oecologia (2003) 136:551–557 DOI 10.1007/s00442-003-1295-3 POPULATION ECOLOGY Stacey A. Forsyth Density-dependent seed set in the Haleakala silversword: evidence for an Allee effect Received: 23 December 2002 / Accepted: 15 April 2003 / Published online: 29 May 2003 Springer-Verlag 2003 Abstract Plant species may be subject to Allee effects if individuals experience a reduction in pollination services when populations are small or sparse. I examined temporal variation in reproductive success of the monocarpic Haleakala silversword (Argyroxiphium sandwicense subsp. macrocephalum) over five years, to determine if plants flowering out of synchrony with most of the population (i.e., in low flowering years) exhibited lower percent seed set than synchronously-flowering plants (i.e., those flowering in high flowering years). Through two pollination experiments conducted over multiple years, I also measured pollen limitation and selfincompatibility in this species. The number of flowering plants varied greatly among years, as did reproductive success. Percent seed set was significantly correlated with the number of plants flowering annually, such that plants flowering in high flowering years (1997 and 2001) exhibited significantly higher percent seed set than did plants flowering in low flowering years (1998–2000). In the 3-year pollen limitation study, plants flowering asynchronously were pollen-limited, whereas plants flowering synchronously were not. This species is strongly self-incompatible. Results of this study demonstrate that the Haleakala silversword experiences reduced reproductive success in low flowering years, and suggest that this Allee effect is pollinator-mediated. Allee effects in plants are an understudied yet potentially important force with implications for the population dynamics and conservation of rare species. Keywords Argyroxiphium · Monocarpy · Pollen limitation · Self-incompatibility · Synchronous flowering S. A. Forsyth ()) Department of Ecology and Evolutionary Biology, Biological Sciences West 310, University of Arizona, Tucson, AZ 85721, USA e-mail: [email protected] Tel.: +1-520-6213458 Fax: +1-520-6219190 Introduction Theoretical and empirical work in conservation biology suggests that small populations are more likely to go extinct than are larger populations, in part due to chance fluctuations in population size (Shaffer 1981; Gilpin and Soule 1986). Small populations may also suffer additional disadvantages due to Allee effects, in which individuals experience reduced survival and/or fertility at low population sizes (Allee 1931, 1938). In species exhibiting Allee effects, individual fitness is reduced as population size decreases, with extinction becoming more likely as a population declines. Allee effects may be due to a variety of genetic, demographic, and/or ecological factors, including increased levels of inbreeding depression or genetic drift, skewed sex ratios, reduced availability of mates, and decreased hunting success with smaller pack size (Allee 1938; Courchamp et al. 1999; Stephens et al. 1999; Courchamp and MacDonald 2001). Although the role of Allee effects in the ecology and conservation of animals has received much consideration (Kuussaari et al. 1998; Courchamp et al. 1999; Courchamp and MacDonald 2001), their importance in plants is poorly documented and requires further study (Lamont et al. 1993; Hackney and McGraw 2001). Such studies are particularly important for rare plants, for which the evaluation of Allee effects should be an essential component of conservation efforts (Groom 1998). In animal species, reproductive success may decline in small populations because individuals in these populations are mate-limited (Allee 1931, 1938; Kuussaari et al. 1998; Courchamp et al. 1999). Similar effects can occur in small or isolated plant populations, particularly in species that depend on animal vectors for pollination (Groom 1998). Plant abundance and/or density may impact plant-pollinator interactions, and therefore plant reproductive success, by affecting both the quantity and quality of pollination services received (Kunin 1993; Lamont et al. 1993; Ingvarsson and Lundberg 1995; Groom 1998; Hackney and McGraw 2001). Small and/or isolated plant populations may be less apparent or 552 attractive to pollinators, resulting in a lower number of pollinator visits to plants in these populations (Schemske 1980; Jennersten 1988). Lower visitation rates may lead to insufficient pollen transfer, and therefore to decreased seed set, for plants located in these areas. In cases where pollinator visitation is not affected by population size or density, plants in small populations may still suffer reduced seed set if the effectiveness of pollination, due to pollen quality, declines (Kunin 1993; Lamont et al. 1993; Wolf and Harrison 2001). This may occur through the deposition of foreign rather than conspecific pollen, or through the transfer of incompatible pollen, if a selfincompatibility mechanism is present (Feinsinger et al. 1991; Kunin 1993; Aizen and Feinsinger 1994; Groom 1998; Ramsey and Vaughton 2000). Essentially, the efficiency of pollination, in terms of both the quantity and quality of pollen received, declines as patches become too small or the distance between plants becomes too great (Courchamp et al. 1999). Obligate outcrossers that rely on animal vectors for pollination are most likely to be affected by Allee effects (Huenneke 1991). Previous studies that have examined Allee effects in plants have focused on reproductive success in differentsized populations or in patches of varying density (Lamont et al. 1993; Groom 1998; Hackney and McGraw 2001). Some of these studies have manipulated plant population size or density and examined subsequent effects on pollinator behavior and/or efficiency (Feinsinger et al. 1991; Kunin 1993, 1997; Bosch and Waser 2001; Hackney and McGraw 2001; Mustajrvi et al. 2001), whereas others have taken advantage of existing variation in plant population size or density (Lamont et al. 1993; gren 1996; Roll et al. 1997; Groom 1998; Bosch and Waser 1999; Mavraganis and Eckert 2001). However, in many species, the abundance of flowering plants varies greatly among years, even if overall population size remains fairly constant. This is particularly true for longlived species exhibiting synchronized flowering, in which years of abundant flowering are separated by one to a few years of little or no flowering (Janzen 1976; Taylor and Inouye 1985; Kelly 1994). In these species, temporal variation in flowering plant abundance and density may affect reproductive success via the mechanisms described above. The Haleakala silversword (Argyroxiphium sandwicense subsp. macrocephalum; Asteraceae: Madiinae) is a long-lived, monocarpic rosette plant endemic to Haleakala volcano on the Island of Maui, Hawaii. As is the case with many other monocarpic plant species, A. sandwicense subsp. macrocephalum (subsequently, silversword) flowering exhibits a boom and bust pattern, with years of abundant flowering separated by one to a few years of very little flowering (Loope and Crivellone 1986). In this study, I use the natural annual variation in the abundance of flowering plants, combined with observations and pollination experiments, to test for an Allee effect in the Haleakala silversword. I quantify silversword seed set over five years and examine the relationship between annual flowering plant abundance and reproductive success. Specifically, I address the following questions: 1. To what degree does seed set vary among years? 2. Does among-year variation in seed set correlate with annual variation in flowering plant abundance? 3. Is the silversword pollen-limited? 4. To what degree is the silversword self-incompatible? Questions 1 and 2 test for the presence of an Allee effect; questions 3 and 4 examine a potential mechanism underlying this effect. Following these analyses, I discuss the conservation implications of an Allee effect in this species. Materials and methods Study site I conducted this research at Haleakala National Park (HALE), Maui, Hawaii (20440 N, 156130 W). This area is comprised of alpine desert (>3,000 m) and subalpine shrubland (2,000–3,000 m) habitats, which are characterized by low annual precipitation, large diurnal variation in temperature, and relatively bare cinder and ash soils (Juvik and Juvik 1998). Vegetation in this area is sparse and primarily comprises a few dominant species, in addition to the silversword. Annual precipitation at Haleakala summit (3,341 m) averages 500–800 mm, with the majority of rainfall occurring between November and April (Kobayashi 1973; Juvik and Juvik 1998). Precipitation is slightly greater in the silversword habitats (2,000–3,100 m), due to the cloud layer associated with a trade wind inversion (Loope and Medeiros 1994). Study species The Haleakala silversword, a member of the Hawaiian silversword alliance, is one of five species in the genus, and one of two subspecies. The other, the Mauna Kea silversword (Argyroxiphium sandwicense subsp. sandwicense), is endemic to the Island of Hawaii. The Haleakala silversword occurs on the cinder cones and lava flows within Haleakala crater (2,000–3,000 m), as well as on the outer slope near the volcano’s summit (3,000–3,341 m). After growing for many years as a basal rosette, individuals produce a tall flowering stalk (capitulescence) bearing hundreds of capitula, and die after setting seed. Capitulescence length and number of capitula are both strongly correlated with rosette diameter (Forsyth 2002). Flowering occurs throughout the summer months, typically peaking in July; during this time, flowering individuals are frequently visited by native and alien bees, wasps, flies, and moths (Forsyth, unpublished data). Nearly all pollination is effected by diurnal visitors, primarily native yellow-faced bees (Hylaeus spp.; Forsyth, unpublished data). Due to its extensive interactions with pollinators, herbivores and seed predators, the silversword is likely a key component of HALE’s high-elevation ecosystem (Loope and Medeiros 1994). Once common throughout Haleakala crater, the silversword suffered severe population declines in the late 1800s and early 1900s, primarily due to browsing by introduced goats and cattle, as well as human vandalism (Loope and Crivellone 1986). Following federal protection from these threats, the population rebounded to an estimated size of 48,000 individuals by 1982 (Loope and Crivellone 1986). A silversword monitoring study conducted from 1982 to 2002, however, suggests that the population may once again be in decline (Forsyth 2002). A current threat to the silversword is the invasive Argentine ant (Linepithema humile), a predator of native insect species, including Hylaeus spp. (Cole et al. 553 1992; Loope and Medeiros 1994). The Haleakala silversword was federally listed as threatened in 1992 (USFWS 1992). Silversword flowering I obtained long-term silversword flowering records from the U.S. Geological Survey, Biological Resources Division, Haleakala National Park. With the exceptions of 1977 and 1981, flowering surveys have been conducted every year since 1969. Surveys are conducted in October, following the flowering season, in order to quantify the number of silverswords flowering in each area of HALE, as initially delineated by Kobayashi (1973). I used the total number of plants flowering throughout HALE as an estimate of flowering for each year, because annual flowering densities in each of the four areas used in this study reflected flowering intensity crater-wide (Loope and Forsyth, unpublished data). Temporal variation in seed set In order to assess temporal variation in seed set, I collected achenes (fruits) from flowering individuals every year from 1997 to 2001. Achenes were collected from plants located in four areas of HALE crater: Sliding Sands Trail (2,700 m), Silversword Loop (2,400 m), Puu Kauaua (2,400 m), and Kalahaku (3,050 m). These sites were selected based on relatively high densities of silverswords in these areas. For each plant, I measured rosette diameter and collected three naturally-pollinated capitula, in order to quantify seed set. Capitula were collected in late September, approximately 8– 10 weeks after flowering, when achenes were mature and close to dispersal. I examined achenes under a dissecting microscope and used the proportion of all dissected achenes that contained a filled embryo as a measure of percent seed set for that plant. Within each year, seed set estimates from the four areas were pooled together, due to low numbers of flowering plants in some years, and average annual percent seed set was compared to the number of plants flowering that year. Pollen limitation and self-incompatibility experiments I conducted pollen limitation and self-incompatibility experiments on individuals located throughout HALE crater from 1997 to 1999. In order to limit damage to cinder cones, the majority of study plants were located at Silversword Loop, Puu Kauaua, and Kalahaku. In all years, access to flowering plants limited sample sizes in these experiments. In 1997, 1998, and 1999, I conducted pollen limitation experiments on 6, 8, and 16 individuals, respectively. Self-incompatibility experiments were conducted in 1998 and 1999 on 8 and 13 individuals, respectively. Heavy seed predation by Rhynchephestia rhabdotis (Lepidoptera: Pyralidae) reduced sample sizes in these experiments to 5, 4, and 13 plants (pollen limitation) and 5 and 10 plants (self-incompatibility), respectively. In each experiment, I randomly selected experimental capitula, although capitula exhibiting signs of R. rhabdotis infestation were replaced by nearby uninfested capitula. To determine if seed set was pollen-limited, I conducted a pollen augmentation experiment. If seed production in pollen-augmented flowers is increased relative to naturally pollinated flowers, the plant is considered to be pollenlimited at that place and time (Hainsworth et al. 1985; Zimmerman and Pyke 1988). I selected six experimental capitula on each plant and randomly assigned each to one of two treatments, for a total of three capitula per treatment per plant: open-pollinated with no supplemental pollen added (OPEN), or open-pollinated with supplemental pollen added (AUGMENT). In the AUGMENT treatment, the supplemental pollen was a mix of pollen collected from at least three other individuals. I collected donor pollen in a vial by gently brushing the vial against several pollen-bearing capitula, and applied pollen to recipient capitula using a paintbrush to cover all receptive stigmas. This procedure was repeated twice for each capitulum, in order to pollinate the maximum number of stigmas. Each experimental capitulum was marked with a colorcoded thread to distinguish treatment and permitted to mature. Achenes were collected and analyzed as described above. If seed production is limited by resources as well as pollen, seed set in pollen-augmented flowers may occur at the expense of nonaugmented flowers on the same plant, as resources are reallocated to flowers that receive more pollen (Lee 1988; Zimmerman and Pyke 1988). To evaluate this effect in the silversword, in 1999 I compared seed set in open-pollinated capitula on plants that received the augmentation treatment (n=9 plants) to seed set in open-pollinated capitula on control plants that did not receive the augmentation treatment (n=7 plants). For each plant, openpollinated seed set was the proportion of all dissected achenes (pooled from three capitula per plant) that were filled. Achenes from open-pollinated capitula on treatment and control plants were collected and analyzed as described above. Previous evidence suggests that some species in the Hawaiian silversword alliance (Argyroxiphium, Dubautia, and Wilkesia) are self-incompatible (Carr et al. 1986). To measure self-incompatibility in the silversword, I conducted a pollinator exclusion experiment using nylon mesh bags to prevent pollinators from visiting experimental capitula. This method effectively excludes insect visitors but does not alter the microclimate surrounding the developing capitulum (Kearns and Inouye 1993). I selected nine capitula per plant and randomly assigned each to one of the following three treatments, for a total of three capitula per treatment per plant: unbagged and pollinated naturally (OPEN); bagged, with no pollen added (BAG); or bagged, with self-pollen added (SELF). BAG and SELF capitula were enclosed in mesh bags as soon as they were distinct from the flowering stalk. Selfpollen was collected from at least three other capitula on the same plant and applied to receptive stigmas as described above. Experimental capitula were collected after achene maturation and analyzed as previously described. Statistical analyses Percent seed set data were logit transformed for statistical analyses. Among-year variation in seed set was examined with a one-way analysis of variance; Tukey-Kramer tests were used to evaluate significant differences between all pairs of years. To examine the relationship between flowering plant abundance and percent seed set, I plotted the average percent seed set (€1 SE) for each of the five years against the number of plants that flowered in that year. I used the transformation Y ¼ lnð1 f =fmax Þ ð1Þ where f equals the average percent seed set for each year and fmax equals the maximum fertility observed in the 5-year period, or 0.45, in order to linearize the relationship. I then used linear regression to analyze the relationship between annual flowering plant abundance and percent seed set. In order to examine other factors potentially related to plant seed set, I used linear regression. Specifically, I tested whether individual seed set was dependent on plant size (rosette diameter), and whether average annual percent seed set was correlated with annual precipitation. Annual precipitation records for HALE were obtained from the Western Regional Climate Center. To analyze the results of the pollen limitation and selfincompatibility experiments, I used two-way ANOVAs, with treatment and year as the main effects. In the pollen limitation experiment, I compared two treatments (OPEN and AUGMENT) and three years (1997–1999), and in the self-incompatibility experiment, I compared three treatments (OPEN, BAG, and SELF) and two years (1998 and 1999). Open-pollinated seed set data from augmented and non-augmented plants were logit-transformed and compared with a student’s t-test. All statistical analyses were performed with JMPin software (SAS 2000). 554 Results From 1997 to 2001, the number of flowering silverswords ranged from a low of 167 (~0.33% of the total population) in 1998 to a high of 2,687 (~5.37% of the total population) in 1997. Consistent with the annual variation in flowering, percent seed set differed significantly among years (F4, 48=6.788, P=0.0002). Seed set was significantly greater in 1997 and 2001, both high flowering years, than in 1998, 1999, and 2000, all low flowering years (Tukey-Kramer test, a=0.05). Average annual percent seed set was positively correlated with flowering plant abundance, reaching a maximum of about 45% seed set in 1997 (Fig. 1A). After transformation (Eq. 1), the number of plants flowering each year explained 99.5% of the variation in seed set among years (Fig. 1B). There was no relationship between plant size (rosette diameter) and percent seed set, for all years combined, as well as for each year, 1997–2001 (Fig. 2). Average annual percent seed set was not correlated with annual precipitation (R2=0.554, df=4, P=0.149). In the pollen limitation experiment, treatment (F1,38= 20.677, P<0.0001), year (F2,38=9.426, P=0.0005), and the treatment by year interaction (F2,38=3.963, P=0.0273) were significant effects. Pollen augmentation significant- Fig. 2 Percent seed set of open-pollinated silversword capitula (1997–2001) and plant size. Symbols distinguish the five years, with clear symbols indicating individuals that flowered in highflowering years (1997 and 2001), and filled symbols indicating individuals that flowered in low-flowering years (1998, 1999, and 2000) Fig. 3 Average percent seed set (€1 SE) for open-pollinated and pollen-augmented silversword capitula, 1997–1999. 1997: n=5 plants; 1998: n=4 plants; 1999: n=13 plants. Letters indicate statistical differences: the pollen augmentation treatment significantly increased seed set in 1998 and 1999, but not in 1997 Fig. 1. A Raw and B transformed (relative to fmax) percent seed set of open-pollinated silversword capitula for each year, 1997–2001, as a function of the number of plants that flowered that year (n). Markers represent average percent seed set (€1 SE) for each year. n=6 (1997); 5 (1998); 15 (1999); 7 (2000); 20 (2001). Transformed percent seed set =ln (1f/fmax), where f = average percent seed set for each year, 1997–2001, and fmax=0.45 (maximum percent seed set observed in the 5-year period). R2=0.995; F=602.380, P<0.0001 ly increased seed set in 1998 and 1999, but not in 1997 (Fig. 3). This suggests that the silversword was strongly pollen-limited in 1998 and 1999, both low flowering years, but not in 1997, a high flowering year. The pollen augmentation treatment produced consistent results across years, yielding 35–40% seed set in all years (1997: 38.92€3.63; 1998: 34.80€5.99; 1999: 36.72€4.87). Openpollinated seed set did not differ between augmented and non-augmented plants (t=0.278, df=14, P=0.7854), indicating that the pollen augmentation treatment did not affect seed set in open-pollinated capitula on the same plant. The self-incompatibility experiment demonstrated that the silversword is strongly self-incompatible (Fig. 4). 555 Fig. 4 Average percent seed set (€1 SE) for open-pollinated, bagged, and self-pollinated silversword capitula, 1998–1999. 1998: n=5 plants; 1999: n=10 plants. Letters indicate statistical differences: in both years, seed set was significantly greater in openpollinated capitula than in bagged and selfed capitula There was a highly significant treatment effect (F2, 39= 17.329, P<0.0001), but no effect of year (F1, 39=1.594, P=0.2140) or the treatment by year interaction (F2, 39= 0.262, P=0.771). In both 1998 and 1999, percent seed set in bagged and selfed capitula was significantly lower than percent seed set in open-pollinated capitula (Fig. 4). Seed set in the BAG and SELF treatments were statistically indistinguishable, averaging 0.31€0.12% and 0.40€0.13% across years, respectively, whereas seed set in openpollinated capitula in these two years averaged 5.99€ 1.13%. Discussion In the 5-year study period, the number of silverswords flowering annually varied greatly among years, with three low flowering years (1998–2000) bracketed by two high flowering years (1997 and 2001). The 30-year flowering record suggests that this among-year variation is typical for the silversword; relatively high flowering years are commonly separated by one to three low flowering years, with the number of plants flowering annually ranging from zero in 1970 to nearly 7,000 in 1991 (L.L. Loope, personal communication). It is currently unknown what triggers silverswords to flower; attempts to correlate annual flowering abundance with precipitation, El Nio, and other environmental factors have been unsuccessful (Loope and Crivellone 1986; Rundel and Witter 1994). Synchronous flowering at multi-year intervals has been observed in other long-lived monocarpic species (Janzen 1976; Foster 1977; Taylor and Inouye 1985; Kelly 1994), and in general, the causes for such flowering synchrony are rarely understood (Taylor and Inouye 1985). Annual variation in flowering was highly correlated with significant variation in percent seed set among years. Although the observed among-year variation in seed set could be due to other, potentially confounding factors, attempts to correlate seed set with other variables were not successful. For example, percent seed set may vary among individuals as a function of plant size, due to differences in resource availability or pollinator attraction. In this study, however, there was no correlation between plant size and percent seed set, both when data from the five years were pooled together (n=49 plants), as well as when data from each year were examined separately. In addition, annual precipitation was not correlated with average annual percent seed set and does not appear to be a confounding factor in this system. This is also implied by the consistent results of the pollen augmentation experiment: from 1997 to 1999, pollen augmentation consistently yielded seed set levels of about 35–40%, despite differences in annual precipitation among these three years. This suggests that reproductive success is determined by factors related to pollination, such as flowering plant abundance, rather than by precipitation. Although it would be ideal to manipulate flowering plant abundance or density, to confirm that this factor is responsible for the observed variation in percent seed set, such experimental manipulation is not feasible due to the conservation status of this species. Kunin (1993) suggested that most self-incompatible plants will have a critical minimum population density threshold, below which pollen limitation is likely to become an important factor influencing reproductive success. Pollen limitation in small or sparse populations can occur via effects on the quantity and/or quality of pollen received (Kunin 1993; Lamont et al. 1993; Bosch and Waser 2001). In poor flowering years, low densities of flowering plants may attract fewer pollinators than larger flowering displays, which can result in reduced visitation, and consequently pollen limitation (Sih and Baltus 1987; Groom 1998; Ramsey and Vaughton 2000). In addition, individuals flowering in low flowering years are less likely to overlap in time with nearby flowering plants. This is particularly true for the silversword, for which the flowering season extends across several months but the flowering period of an individual is only one to three weeks (Forsyth, unpublished data). Flowering individuals, especially those flowering at the beginning or end of the flowering season, may be isolated in both time and space from any other flowering plant. In such cases, there is little outcross pollen available and the probability that receptive stigmas will receive compatible outcross pollen declines as the number of flowering plants decreases. Similarly, if flowering plants are sparsely distributed, pollinators are more likely to forage on multiple flowers within one plant, instead of visiting several different plants (de Jong et al. 1993; Bosch and Waser 2001). This is particularly true for species exhibiting a “big bang” flowering habit, in which large numbers of flowers within a plant are on display simultaneously (Augspurger 1980). Several studies have noted that pollinators tend to visit more flowers in larger inflorescences (Klinkhamer et al. 1989; Mustajarvi et al. 2001). Such an increase of withinplant foraging would reduce the amount of outcross pollen deposited on a plant, while increasing the amount 556 of geitonogamous (within-plant) self-pollination (de Jong et al. 1993; Harder and Barrett 1995; Bosch and Waser 2001). Given the silversword’s strong self-incompatibility mechanism, geitonogamous pollination would lead to low seed set (reduced female fitness) as well as pollen discounting (reduced male fitness). Pollen, however, does not appear to be the only factor limiting silversword reproduction. Even in high flowering years, seed set did not exceed 45%, suggesting that silversword seed set is limited by factors other than compatible pollen in all years. Resource limitation is also suggested by the pollen augmentation experiment, which consistently produced only 35–40% seed set in each of the three years in which the experiment was conducted. These levels of seed set (<40%) suggest that silversword seed set is limited by resources as well as by pollen. The observed effect of flowering plant abundance on reproductive success may help to explain the synchronized flowering pattern observed in this species. The evolution of flowering synchrony to maximize reproductive success has been hypothesized for other long-lived monocarpic species, including Swertia radiata, a longlived gentian (Taylor and Inouye 1985). Synchronized flowering may also affect levels of seed predation, as predicted by the predator satiation hypothesis (Janzen 1976). This hypothesis, that synchronized flowering serves to minimize seed predation by swamping seed predators with an abundance of seed in good flowering years, and starving them in poor flowering years, has been proposed for other long-lived plant species (Janzen 1976; Augspurger 1981; Kelly 1994). Although not addressed in this study, previous observations suggest that this interaction may occur with the silversword. In some years, a large proportion of silversword achenes are lost to two native predispersal seed predators, but in good flowering years, seed predation may be less than 50% (Kobayashi 1973; Loope and Medeiros 1994). Alternating years of synchronized flowering with low flowering years may help to minimize seed predation over the long term. Conservation implications The presence of an Allee effect in the Haleakala silversword has important implications for the persistence of this species, because at small population sizes, low numbers of flowering plants will lead to reduced reproductive success for most individuals. This will in turn likely impact future population size. Although the link between seed production and recruitment is not precisely known, due to seed dormancy, population monitoring studies indicate that recruitment is greater following years of high flowering (Forsyth and Loope, unpublished data). The mechanism underlying this Allee effect suggests that population persistence is not only dependent on population size, but also on the spatial distribution of flowering plants, annual flowering patterns, and pollinator activity and efficiency as well. Also, a pollinator-mediated Allee effect could be exacerbated by reduced pollinator abun- dance or efficiency, which in this system is a likely consequence of the current Argentine ant invasion (Cole et al. 1992; Forsyth 2002). Density-dependent seed set in the Haleakala silversword also has important implications for restoration efforts on the Hawaiian volcano, Mauna Kea. The critically endangered Mauna Kea silversword (A. sandwicense subsp. sandwicense) is currently the focus of a large-scale reintroduction effort on the Island of Hawaii (Robichaux et al. 2000). Because the Mauna Kea silversword is also strongly self-incompatible (Carr et al. 1986), low densities of individuals will likely exhibit reduced reproductive success. Outplanting individuals at higher densities may serve to alleviate the negative effects of density-dependent seed set. In combination, results of this study provide strong evidence for a pollinator-mediated Allee effect in the Haleakala silversword. Allee effects may be prevalent in plants, particularly in self-incompatible species that rely on animal vectors for pollination. As plant population sizes begin or continue to decline, due to factors such as habitat fragmentation, harvesting, or invasive species, extinction risks may be heightened due, in part, to Allee effects. Such possibilities point to the need to better understand factors influencing plant reproduction, particularly for rare and endangered species, including how reproductive success is affected by changes in population size. Evaluating Allee effects and determining threshold population sizes, below which individual reproduction is reduced, should be a critical step in the management and conservation of rare plant species. Acknowledgements I thank Clare Aslan for assistance in the field and laboratory, and Regis Ferriere for inspiring discussion. Much thanks to Lloyd Loope and Ron Nagata for logistical support at Haleakala National Park. Rob Robichaux, Judie Bronstein, Lucinda McDade, Molly Hunter, Nick Waser and an anonymous reviewer provided valuable feedback on earlier drafts of this manuscript. This project was funded by the National Science Foundation (Graduate Research Fellowship), the National Parks Conservation Association, and the Department of EEB at the University of Arizona. This manuscript was submitted to the University of Arizona in partial fulfillment of requirements for the degree of Doctor of Philosophy. References gren J (1996) Population size, pollinator limitation, and seed set in the self-incompatible herb Lythrum salicaria. Ecology 77:1779–1790 Aizen MA, Feinsinger P (1994) Forest fragmentation, pollination, and plant reproduction in a chaco dry forest, Argentina. Ecology 75:330–351 Allee WC (1931) Animal aggregations, a study in general sociology. University of Chicago Press, Chicago Allee WC (1938) The social life of animals. W.W. Norton, New York Augspurger CK (1980) Mass-flowering of a tropical shrub (Hybanthus prunifolius): influence on pollinator attraction and movement. Evolution 34:475–488 Augspurger CK (1981) Reproductive synchrony of a tropical shrub: experimental studies on effects of pollinators and seed preda- 557 tors on Hybanthus prunifolius (Violaceae). Ecology 62:775– 788 Bosch M, Waser NM (1999) Effects of local density on pollination and reproduction in Delphinium nuttallianum and Aconitum columbianum (Ranunculaceae). Am J Bot 86:871–879 Bosch M, Waser NM (2001) Experimental manipulation of plant density and its effect on pollination and reproduction of two confamilial montane herbs. Oecologia 126:76–83 Carr GD, Powell EA, Kyhos DW (1986) Self-incompatibility in the Hawaiian Madiinae (Compositae): an exception to Baker’s Rule. Evolution 40:430–434 Cole FR, Medeiros AC, Loope LL, Zuehlke WW (1992) Effects of the Argentine ant on arthropod fauna of Hawaiian highelevation shrubland. Ecology 73:1313–1322 Courchamp F, Macdonald DW (2001) Crucial importance of pack size in the African wild dog Lycaon pictus. Anim Conserv 4:169–174 Courchamp F, Clutton-Brock T, Grenfell B (1999) Inverse density dependence and the Allee effect. Trends Ecol Evol 14:405–410 de Jong TJ, Waser NM, Klinkhamer PGL (1993) Geitonogamy: the neglected side of selfing. Trends Ecol Evol 8:321–325 Feinsinger P, Tiebout HM III, Young BE (1991) Do tropical birdpollinated plants exhibit density-dependent interactions? Field experiments. Ecology 72:1953–1963 Forsyth SA (2002) Demographic modeling of Hawaiian silverswords, and its implications for conservation. Ph.D. dissertation, Department of Ecology and Evolutionary Biology, University of Arizona, Tucson Foster RB (1977) Tachigalia versicolor is a suicidal neotropical tree. Nature 268:624–626 Gilpin ME, Soule ME (1986) Minimum viable populations: processes of species extinction. In: Soule ME (ed) Conservation biology: the science of scarcity and diversity. Sinauer, Sunderland, Mass., pp 19–34 Groom MJ (1998) Allee effects limit population viability of an annual plant. Am Nat 151:487–496 Hackney EE, McGraw JB (2001) Experimental demonstration of an Allee effect in American ginseng. Conserv Biol 15:129–136 Hainsworth FR, Wolf LL, Mercier T (1985) Pollen limitation in a monocarpic species, Ipomopsis aggregata. J Ecol 73:263–270 Harder LD, Barrett SCH (1995) Mating cost of large floral displays in hermaphrodite plants. Nature 373:512–515 Huenneke LF (1991) Ecological implications of genetic variation in plant populations. In: Falk DA, Holsinger KE (eds) Genetics and conservation of rare plants. Oxford University Press, New York, pp 31–44 Ingvarsson PK, Lundberg S (1995) Pollinator functional response and plant population dynamics: pollinators as a limiting resource. Evol Ecol 9:421–428 Janzen DH (1976) Why bamboos wait so long to flower. Annu Rev Ecol Syst 7:347–391 Jennersten O (1988) Pollination in Dianthus deltoides (Caryophyllaceae): effects of habitat fragmentation on visitation and seed set. Conserv Biol 2:359–366 Juvik SP, Juvik JO (1998) Atlas of Hawaii, 3rd edn. University of Hawaii Press, Honolulu Kearns CA, Inouye DW (1993) Techniques for pollination biologists. University Press of Colorado, Niwot, Colo. Kelly D (1994) The evolutionary ecology of mast seeding. Trends Ecol Evol 9:465–470 Klinkhamer PGL, de Jong TJ, de Bruyn G (1989) Plant size and pollinator visitation in Cynoglossum officinale. Oikos 54:201– 204 Kobayashi, HK (1973) Ecology of the silversword, Argyroxiphium sandwicense DC. (Compositae), Haleakala Crater, Hawaii. Dissertation, Department of Botany, University of Hawaii, Honolulu Kunin WE (1993) Sex and the single mustard: population density and pollinator behavior effects on seed set. Ecology 74:2145– 2160 Kunin WE (1997) Population size and density effects in pollination: pollinator foraging and plant reproductive success in experimental arrays of Brassica kaber. J Ecol 85:225–234 Kuussaari M, Saccheri I, Camara M, Hanski I (1998) Allee effect and population dynamics in the Glanville fritillary butterfly. Oikos 82:384–392 Lamont BB, Klinkhamer PGL, Witkowski ETF (1993) Population fragmentation may reduce fertility to zero in Banksia goodii- a demonstration of the Allee effect. Oecologia 94:446–450 Lee TD (1988) Patterns of fruit and seed production. In: LovettDoust J, Lovett-Doust L (eds) Plant reproductive ecology: patterns and strategies. Oxford University Press, New York, pp 179–202 Loope LL, Crivellone CF (1986) Status of the Haleakala silversword: Past and present. Tech. Rep. 58. National Park Resource Studies Unit, University of Hawaii, Honolulu Loope LL, Medeiros AC (1994) Biotic interactions in Hawaiian high elevation ecosystems. In: Rundel PW, Smith AP, Meinzer FC (eds) Tropical alpine environments: plant form and function. Cambridge University Press, New York, pp 337–354 Mavraganis K, Eckert CG (2001) Effects of population size and isolation on reproductive output in Aquilegia canadensis (Ranunculaceae). Oikos 95:300–310 Mustajrvi K, Siikamki P, Rytknen S, Lammi A (2001) Consequences of plant population size and density for plantpollinator interactions and plant performance. J Ecol 89:80–87 Ramsey M, Vaughton G (2000) Pollen quality limits seed set in Burchardia umbellata (Colchicaceae). Am J Bot 87:845–852 Robichaux R, Bergfeld S, Bruegmann M, Canfield J, Moriyasu P, Rubenstein T, Tunison T, Warshauer F (2000) Reintroducing Hawaii’s silverswords. Endangered Spec Bull 25:22–23 Roll J, Mitchell RJ, Cabin RJ, Marshall DL (1997) Reproductive success increases with local density of conspecifics in a desert mustard (Lesquerella fendleri). Conserv Biol 11:738–746 Rundel PW, Witter MS (1994) Population dynamics and flowering in a Hawaiian alpine rosette plant, Argyroxiphium sandwicense. In: Rundel PW, Smith AP, Meinzer FC (eds) Tropical alpine environments: plant form and function. Cambridge University Press, New York, pp 295–306 SAS (2000) JMP in statistical software, version 4.0.2 (academic). SAS, Cary, N.C. Schemske DW (1980) Floral ecology and hummingbird pollination of Combretum farinosum in Costa Rica. Biotropica 12:169–181 Shaffer ML (1981) Minimum population sizes for species conservation. BioScience 31:131–134 Sih A, Baltus MS (1987) Patch size, pollinator behavior, and pollinator limitation in catnip. Ecology 68:1679–1690 Stephens PA, Sutherland WJ, Freckleton RP (1999) What is the Allee effect? Oikos 87:185–190 Taylor OR Jr, Inouye DW (1985) Synchrony and periodicity of flowering in Frasera speciosa (Gentianaceae). Ecology 66:521–527 U.S. Fish and Wildlife Service (USFWS) (1992) Determination of endangered or threatened status for 15 plants from the Island of Maui, HI. Federal Register 57:20772–20787 Wolf AT, Harrison SP (2001) Effects of habitat size and patch isolation on reproductive success of the serpentine morning glory. Conserv Biol 15:111–121 Zimmerman M, Pyke GH (1988) Reproduction in Polemonium: assessing the factors limiting seed set. Am Nat 131:723–738