Invading hybrids

Wikimedia - João de Deus Medeiros - Schinus terebinthifolius

Editor's Introduction

Hybrid Vigor for the Invasive Exotic Brazilian Peppertree (Schinus terebinthifolius Raddi., Anacardiaceae) in Florida

Scientists have been interested in understanding why certain species become so dominant when introduced to new regions while others do not, this is especially interesting in the case of an invasive species. A successful “invasion” begins with small numbers of founding individuals and studies have shown greater fitness for future generations when they are hybridized over the years. The “invasion” of Brazilian pepper trees in South Florida has inspired this study. The authors of this paper work to understand more about species’ fitness due to hybridization.

NOTE: This article is part of a Collection of student-annotated papers that are the product of the SitC team’s research into best practices for using primary literature to support STEM education. For this reason, these papers have undergone an alternate review process and may lack educator guides. To learn more, visit the main Collection page: SitC Lab.

Paper Details

Original title
Hybrid Vigor for the Invasive Exotic Brazilian Peppertree (Schinus terebinthifolius Raddi., Anacardiaceae) in Florida
Original publication date
Reference
Vol.172 No.5 pp.655-663 ref.42
Issue name
International Journal of Plant Sciences
DOI
10.1086/659457

Abstract

How can successful invaders overcome reduced genetic variation via small founder population sizes to persist, thrive, and successfully adapt to a new set of environmental conditions? An expanding body of literature posits hybridization, both inter- and intraspecific, as a driver of the evolution of invasiveness via genetic processes. We studied Brazilian peppertree (Schinus terebinthifolius), a tree species native to South America that is a successful invader throughout Florida. The tree was introduced separately to the east and west coasts of Florida more than 100 years ago from genetically distinct source populations. We conducted a common garden experiment to compare the early life-stage performance of hybrids versus their progenitors. We hypothesized that hybrids would outperform their progenitors due to the positive genetic effects of intraspecific hybridization (i.e., hybrid vigor). Hybrid seeds germinated at higher rates than eastern seeds. Over the 8-mo experiment, a greater proportion of hybrid seedlings survived than did western seedlings, and hybrids attained greater biomass than the western types. The cumulative hybrid advantage of both seed germination and seedling survival led to the establishment of nearly 45% more hybrid seedlings versus either progenitor. Documenting fitness advantages for hybrids over their progenitors is a requisite finding to consider hybridization as a factor in the success of invasive species.
Keywords: invasive species, Schinus terebinthifolius, multiple introductions, hybridization, common garden experiment, hybrid vigor, intraspecific.

Report

INTRODUCTION


While the search for shared characteristics among successful invaders continues (Kolar and Lodge 2001), several hypotheses have been proposed to elucidate why certain species become so dominant when introduced to new regions. Darwin (1859) suggested that some introduced species are released from natural enemies found in their native ranges, resulting in their uncontrolled growth in new areas. Another idea posits that exotic species may become dominant in the new region through rapid genetic changes wrought under new selective forces (Ellstrand and Schierenbeck 2000; Lee 2002; Allendorf and Lundquist 2003; Stockwell et al. 2003; Hierro et al. 2005). Successful exotics also may evolve ‘‘invasiveness’’ from release of their specialist herbivores allowing them to reroute those resources to competitive traits (sensu Blossey and Notzold 1995).

The successful establishment of invasive species generally begins with small numbers of founding individuals. The progression from low-density founding populations to often high-density expanding invasion fronts exposes a genetic par
adox for the few introduced species that become economically and ecologically damaging. How can successful invasive species overcome reduced genetic variation via small founder population sizes to thrive, persist, and adapt to a new set of environmental conditions? The results of many studies conducted to gauge the genetic variation of species in their native versus introduced ranges are equivocal; some have greater variation in their introduced range, and others do not (see reviews: Barrett and Husband 1990; Lee 2002; Lambrinos 2004; Novak and Mack 2005; Dlugosch and Parker 2008). While the search for commonalities in the invasion process through genetic changes continues, recent studies have focused on the role that multiple introductions may play in bolstering genetic diversity of invasive populations.

For several exotics, intraspecific hybridization is achieved through multiple introductions over time that results in admixing of individuals from geographically isolated source populations in their new adventive range; successful invasion may be facilitated by the influx of genetic variation from these multiple introductions, which produces novel genetic combinations. Current research has focused on hybridization, both inter- and intraspecific, as a means of promoting the evolution of invasiveness (Kolbe et al. 2004; Facon et al. 2005; Rieseberg et al. 2007). Recent research testing whether advanced generation hybrids are more fit than their progenitors is still equivocal, with results showing there is often no proof of an inherent fitness advantage to these hybrids (Hardiman and Culley 2010). Evidence of greater fitness (i.e., survival, growth, and future reproduction) for hybrids versus their progenitors is the first necessary finding to propose the case of hybridization promoting invasiveness.

Brazilian peppertree is a fast-growing, dioecious tree native to Brazil, Argentina, Paraguay, and Uruguay (Ewel 1986). It has successfully colonized most of Florida, covering more than 280,000 ha in southern Florida alone (Ferriter 1997; Schmitz et al. 1997). Historical records (Morton 1978) reveal two separate introductions as an ornamental plant in Florida ~100 years ago near Punta Gorda on the west and Miami on the east coast of the state. Recent genetic analyses indicate that these introductions originated from different source areas in Brazil and have since hybridized extensively throughout Florida (Williams et al. 2005). The west coast introduction likely comes from coastal Brazil near 27° south, while the provenance of the east coast introduction is near 13° south in Salvador, Bahia (G. Wheeler, unpublished data). Nearly 50 years passed after its arrival before Brazilian peppertree was reported to invade natural areas (Ewel et al. 1982), and now there is evidence of extensive intraspecific hybridization between these two types, with most presentday populations being a mix of hybrids (Williams et. al 2007). This invasive plant spreads predominantly via seedlings that become established following the copious production of small, animal-dispersed seeds. Recently, it has been observed that the fruit is attacked by an introduced torymid wasp, Megastigmus transvaalensis, originally from South Africa (Habeck et al. 1989), that can destroy a significant proportion of the fruit (Wheeler et. al 2001). The extent of the possible negative effect this seed predator has on the population dynamics of this invasive plant is still unknown.

The broad goal of our research was to further understand the basic biology of Brazilian peppertree regarding seed germination and seedling establishment as the spread of this invasive relies on sexual reproduction. Specifically we chose to investigate whether intraspecific hybrids outperformed their progenitors in the early life stages and whether any potential advantages at each of these life stages translated into a greater measure of total fitness for hybrids. To accomplish this, fruit were collected from across the species’ range in Florida to provide a more complete survey of population variation. A subsample of the fruit was examined for damage from the introduced South African seed predator. Another subsample of fruit was weighed to measure differences among sites and the three Florida types. We then conducted a common garden experiment to gauge differences in the germination of seeds among the three Florida types (eastern introduction, western introduction, and their hybrids), as well as subsequent seedling survival and biomass accumulation. Once results were obtained from these analyses, a separate tally on the relative fitness of each of the types across the life stages (i.e., seed weight, seed germination, seedling survival, and seedling biomass) was made to then compare the total fitness scores. To validate that our sampling protocol, which followed that of Williams et al. (2005, 2007), correctly designated the types we wished to test, we genotyped a sample of seedlings grown from the same cohort used in the common garden experiment. Our research is the first to consider differences among the three Florida types of Brazilian peppertree, and evidence of hybrid advantage will guide future research aimed at managing this costly invasive species.


MATERIAL AND METHODS

Study Species

Brazilian peppertree is a woody perennial tree up to 13 m in height (Ewel et al. 1982) generally having a domed shape with the lowermost lateral branches extending outward from the usually short, multistemmed trunk. Dead branches are retained, creating a dense thicket under the canopy. Conditions are usually highly shaded beneath the tree’s canopy, with very little vegetation. It has been suggested that the phenolic compounds in the exocarp of the fruit may act as germination inhibitors for competing species, as well as for Brazilian peppertree seeds themselves (Nilsen and Muller 1980). Brazilian peppertree flowers predominantly in the fall in Florida, with a smaller flowering pulse in April–May (Ewel et al. 1982). An extended peak of fruiting soon follows with most mature fruit produced from December to March, although fruit may be found on trees throughout the year (Ewel et al. 1982). Large females ripen thousands of bright red drupes that each contains a kidney-shaped seed covered with a sticky mesocarp. In Florida, most seeds germinate in January–February while mortality of seedlings occurs throughout the year (Ewel et al. 1982).


Design and Experiment

From January 2 to 18, 2008, we collected mature fruit from 10 plants at 12 sites each across the range of Brazilian peppertree in Florida (table 1). At each site, a subsample of female plants separated by at least 20 m was haphazardly chosen from which to collect fruit; in general, these 10 sampled plants represented ~10% of the female plants at most sites. The fruit collection protocol was designed to sample the three types (i.e., eastern, western, and their hybrids) found in Florida, as delineated by recent genetic analyses (Williams et al. 2005, 2007). To assess the impact of the Brazilian peppertree seed predator Megastigmus transvaalensis (Hussey) (Hymenoptera: Torymidae) on collected fruit, three replicates of 100 fruit lots from each of the 120 plants (i.e., 300 fruit x 120 plants = 36,000 fruit in total) were visually scored for exit holes of emerged flies. Fruit with exit holes were considered nonviable (Wheeler et al. 2001). To gauge differences in seed weight among the collected fruit, a random sample of 20 fruit from each of the 120 plants (i.e., 20 fruit x 120 plants = 2400 fruit in total) was weighed on a Mettler balance after the exocarps of all fruit were removed.

The common garden planting area was established in a mowed, open grassy field on the campus of the USDA-ARS Invasive Plant Research Laboratory in Davie, Florida (26°05'6"N, 080°14'31"W). A border of tall trees, ~10 m, running north to south was located on the western side of the common garden planting area such that the plants in the blocks closer to the border of tall trees were progressively shaded as the sun set in the west. We tilled the garden to a depth of 20 cm and then removed the residual plant material from the sandy substrate. Five blocks were delineated, which were separated by ~3 m. In each block, 120 ‘‘pots’’ were added for a total of 600. The pots were made from 10cm-diameter (2-mm-thick) PVC pipe sections that were 12 cm in length. We inserted the pots to a depth of 6 cm in the loose soil and formed an array composed of 12 columns by 10 rows. In each block, three groupings of four columns were separated by 0.5 m to allow access for the surveys.

As each fruit contains only one seed, we used seed as the unit of measure for the common garden experiment. On February 26, 2008, we placed 100 visually inspected sound fruit that had been lightly macerated in our palms into each pot. This procedure completely removed the thin red exocarp from ~90% of the seeds. We performed this additional step as Nilsen and Muller (1980) reported that the soluble extract from the exocarps inhibited seed germination to a minor extent. The density of sown seeds was chosen to mimic what is commonly observed in the field. The seeds were sown in the pots to a depth of ~1 cm. The 100 seeds from each of the 120 mother plants were randomly assigned a pot location in all five blocks for a total of 60,000 planted seeds. We added ~3 cm of supplemental water to each pot per week if there was no rain, for the duration of the dry season (October to late May). Once the wet season began at the end of May, we no longer gave the seedlings supplemental water.

Beginning March 4, 2008, the 600 pots were surveyed weekly for 20 wk and then every 2 wk for an additional eight surveys. At each survey, the number of new seedlings (i.e., germinated since the last survey), the number of dead seedlings, and the total number of living seedlings in each pot were recorded. Dead seedlings were carefully removed after they had been counted.

After 8 mo, the remaining seedlings were harvested to collect growth and biomass measures. We dug up the contents of each pot using small shovels, removed the seedlings, and then gently agitated the roots to shake off loosely attached soil. Seedlings were brought into the lab, rinsed with water to remove any remaining soil, and the number and total fresh weight of seedlings were recorded. All the plant material from each pot was dried at 65°C for 2 wk and then reweighed to quantify total dry weight of seedlings.

To compare total fitness among the three Florida types, we calculated separate values of relative fitness at each of the following life stages: seed weight, seed germination, seedling survival, and seedling dry weight. Data at each life stage were transformed into a relative fitness value by dividing each value by the highest value, where the final relative fit
ness values ranged from 0 to 1.0. Total fitness values were calculated as the product across all life stages.


Genetic Analyses


We genotyped 211 seedlings (table 1) that were grown from the same seed lots as the experimental seedlings at five nuclear microsatellite loci and an intergenic region in the chloroplast DNA (cpDNA; trnS and trnG; Hamilton 1999) as described in Williams et al. (2007). Genotypes were scored using an ABI 3130 genetic analyzer and GENEMAPPER, version 4.0 (Applied Biosystems, www.appliedbiosystems.com). We performed a x2 test on the proportions of individuals with the A or B chloroplast haplotype per site (table 1) to verify that our type designations, which were classified according to the results of the Williams studies (Williams et al. 2005, 2007) and following their site locality sampling, were correct. There were significant differences among the three types (Pearson x2 = 68.57, df = 2, P < 0.001). Additional x2 tests found significant differences between (1) western and hybrid seedlings (Pearson x2 = 12.10, df = 1, P = 0.0010), (2) hybrid and eastern seedlings (Pearson x2 = 26.19, df = 1, P < 0.0010), and (3) western and eastern seedlings (Pearson x2 = 68.23, df = 1, P < 0.001).

We used a Bayesian clustering method implemented in the program STRUCTURE, version 2.3, to estimate the degree of hybridization between the two introductions into Florida (Pritchard et al. 2000; Pritchard and Wen 2003). The membership of each individual in a subpopulation (eastern or western) was estimated as (q), the ancestry coefficient, which varies on a scale from 0 to 1.0 (table 1). We first ran the Monte Carlo Markov Chain for 106 iterations following a burn-in period of 105 iterations for K = 2, using the correlated allele frequencies model and assuming admixture. As with the A and B chloroplast haplotype data, we then verified whether the average western ancestry q values within each category were consistent with their classification (eastern, western, or hybrid) using ANOVA. The ANOVA performed on the western ancestry q values showed significant differences among the three types (F = 68.99, df = 2, P < 0.001). Post hoc tests using Bonferroni correction found significant differences between all the type pairwise comparisons (fig. 1). Therefore, we have congruent evidence from both the chloroplast haplotype data and microsatellite data (i.e., q ancestry values) that our type designations by site are appropriate for the remaining analyses.

Table 1

Table 1. Brazilian Peppertree Seed Source Locations, Type, and Genetic Typing Results.

Seed locations

In this experiment 211 seedlings were grown from the same seed lots as the experimental seedlings. The specific regions from where the seeds were acquired are listed in the table along with their longitude and latitude coordinates.

Type

The authors used the study by William et al. to classify the seed types as well as used their designations (i.e. Western, A; Eastern, B; and Hybrid). Statistical tests were run to determine if there was a true statistical difference in the results the authors say from the different seed types. The results indicated that there were significant differences among the three types Additional tests found significant differences between (1) western and hybrid seedlings (2) hybrid and eastern seedlings and (3) western and eastern seedlings.

Genetic typing

The authors looked at seedlings with the A or B chloroplast haplotype.

A haplotype is a specific group of genes or alleles that progeny inherited from one parent.


Statistical Analyses


One-way ANOVAs were used to quantify differences in the mean seed predation rate among sites and types. Bonferroni tests were used for pairwise post hoc comparisons. The arcsine square root transformation was used to meet assumptions of normality and homogeneity of variances. Due to the possible correlation in seedling performance within the same pot over time, repeated-measures ANOVA was used to gauge differences in the germinated seeds and dead seedlings across survey dates. To overcome the assumption of sphericity for the repeated-measures ANOVA, multivariate tests were performed with Wilks’s L being reported. To consider interactions among main effects, we analyzed differences between blocks, sites, and types for the number of germinated seeds, number of surviving seedlings (at the end of the 8-mo experiment), and dry weight of harvested seedlings using two-way ANOVAs. Bonferroni tests were used for pairwise post hoc comparisons. To explore the effect of density of individuals within pots by type on the variables of seedling survival and dry weight of seedlings, ANCOVAs were performed using density of seedlings at the seventh survey as a covariate. A linear regression was performed to investigate the effect of seedling density on seedling survival. SPSS 15.0 (SPSS, Chicago, IL) was used for the analyses. Data are presented as mean ± SE, unless noted otherwise, with α = 0.05.


RESULTS


Western plants had greater predation rates than hybrids, while the rates for eastern plants were intermediate. The proportion of fruit attacked by the seed predator was surprisingly low, although differences were observed among the 12 sites (F = 8.79, df = 11, P < 0.001) and the three types (F = 5.35, df = 2, P = 0.006; see fig. 2). Seed predation rates among sites ranged from 0% to 2.1% and among types from 0.1% to 0.8% (fig. 2).

Western and hybrid seeds were significantly heavier (by ~10%) than eastern seeds (F = 26.94, df = 2, P < 0.001; see fig. 3). There also were significant differences in the weight of seeds from the 12 sites (F = 18.25, df = 11, P < 0.001). Of the 60,000 seeds planted, 9817 (~16.4%) germinated over the 8-mo experiment. While attempts were made to remove exocarps from most individuals, many seeds germinated with their red exocarp still fully intact. Seeds began germinating by the third week, and nearly 90% germinated within a 4-wk timeframe, with a strong effect of time on germination rates (Wilks’s Λ = 0.11, F = 192.97, P < 0.001). The last observed seed germination was on October 21, 2008. Of the germinated seeds, 4209 (~43%) died over the 8-mo experiment. As with seed germination, seedling mortality was not uniform over the course of the experiment (Wilks’s Λ = 0.44, F = 28.77, P < 0.001); seedling mortality showed two peaks, one in April and the other in July– September. Hybrid seeds germinated at higher rates (18.4% ± 0.009%) than eastern seeds (14.6% ± 0:008%); however, no differences were found in the germination rate of the western seeds (16.0% ± 0.007%) and the other types (fig. 4). The proportion of germinated seeds was influenced by block (F = 37.74, df = 4, P < 0.001), type (F = 6.71, df = 2, P = 0.001), and the interaction between block and type (F = 2.45, df = 8, P = 0.013). For the two-way ANOVA on the proportion of germinated seeds, there were significant effects of block (F = 38.77, df = 4, P < 0.001) and site (F = 3.53, df = 11, P < 0.001), while the interaction of block and site was not significant (F = 1.28, df = 44, P = 0.11). Among the 12 sites, seed germination proportions ranged from 12.3% to 20.2%.

Fig.1.

Fig. 1. Mean proportion western ancestry (q) among the three types of Brazilian peppertree genotyped. Error bars show 1 standard error. Letters indicate significant differences (P < 0.05) among types based on specific post hoc comparisons (with Bonferroni correction).

Graph representation

This data represents the average percentage of Western ancestry among the three different groups.

The membership of each individual in a subpopulation (eastern or western) was estimated as (q), the ancestry coefficient, which varies on a scale from 0 to 1.0 and then plotted.

Graph interpretation

The authors found significant differences between all the pairs when compared. Therefore by looking at this microsatellite data (i.e., q ancestry values) the authors confirmed that their designations by site were appropriate for the remaining analyses.

 

The error bars on top of the graph indicates the greatest standard deviation in their calculations for the hybrid plants.  

Fig.2.

Fig. 2. Mean proportion of fruit with exit hole (i.e., seed predator damage) among the three Brazilian peppertree types. Error bars show 1 standard error. Letters indicate significant differences (P < 0.05) among types based on specific post hoc comparisons (with Bonferroni correction).

Graph representation

This data represents the average seed damage due to predators among the three different groups.

Graph interpretation

Western plants had greater rates of seed damage due to predators than hybrids, while the rates for eastern plants were intermediate. The overall proportion of fruit attacked by the seed predator was surprisingly low among sites and among types. The rates of attack among sites ranged from 0% to 2.1% and among types from 0.1% to 0.8%.


Seeds from the Kissimmee Prairie site had greater germination rates than those from the Pal-Mar and Chekika sites.

A greater proportion of hybrid seedlings survived than did western seedlings (63.6% vs. 53.2%; see fig. 5). Seedling survival proportions varied by block (F = 10.85, df = 4, P < 0.001) and type (F = 7.28, df = 2, P = 0.001), while the interaction was not significant (F = 1.66, df = 8, P = 0.105). There were differences in the seedling survival proportions among blocks (F = 11.47, df = 4, P < 0.001) and sites (F = 3.66, df = 11, P < 0.001), but there was no significant interaction between block and site (F = 1.28, df = 44, P = 0.111). Seedling survival proportions among the 12 sites ranged from 42.4% to 66.7%, with greater proportions of surviving seedlings from the Kissimmee Prairie, Lake Placid, Okeechobee, Cape Canaveral, Fisheating Creek, and Chekika sites than from the Punta Gorda site.

ANCOVAs were performed to consider the effect of density of seedlings within pots by type on the variables seedling survival and seedling dry weight. The ANCOVA on seedling survival by type could not be conducted as the interaction between type and seedling density was significant (F = 12.48, df = 2, P < 0.001). In other words, the ANCOVA test assumption of homogeneity of slopes was not met as the differences in seedling survival among the three types varied significantly as a function of the covariate density of seedlings within pots. The ANCOVA on seedling dry weight by type with density of seedlings within pots as the covariate passed the homogeneity-of-slopes test (F = 0.87, df = 2, P = 0.420). The result of the ANCOVA was significant (F = 7.70, df = 2, P = 0.001), and follow-up tests were conducted to evaluate pairwise differences among the three types using Holm’s sequential Bonferroni procedure. There were significant differences in the adjusted means only for the comparison of dry weight between the western and hybrid seedlings (F = 15.34, df = 1, P < 0.001).

A linear regression was performed to consider the effect of seedling density on seedling survival. A strong, positive relationship between the number of germinated seeds and number of surviving seedlings was observed (r2 = 0.591, P < 0.001). As the relationship was linear, the number of seedlings that survived was proportional to the number of seeds that germinated. In other words, there was little evidence of decreasing seedling survival with greater number of seedlings or greater seedling survival in low-density pots.

Dry weight of harvested seedlings (i.e., biomass) was influenced by type (F = 4.25, df = 2, P = 0.015), with hybrid seedlings obtaining greater dry weights than their western counterparts, 0.121 ± 0.006 versus 0.100 ± 0.005 g, respectively (fig. 6). The dry weight of harvested seedlings also was influenced by block (F = 43.95, df = 4, P < 0.001) and site (F = 3.51, df = 11, P < 0.001), while the interaction between block and site was not significant (F = 0.94, df = 43, P = 0.579). Seedlings harvested from the Fort Pierce site weighed, on average, significantly more than seedlings from all other sites except the Kissimmee Prairie, Pal-Mar, Fisheating Creek, and Okeechobee sites.

Hybrids had the highest relative fitness scores among the types for three out of the four life stages that were measured; hybrids also had the highest total fitness score (table 2). Total fitness of hybrids was more than 50% higher than either of their progenitors.


DISCUSSION

Overall, hybrids had significant advantages over both progenitors for several early life-history stages. The cumulative superiority of hybrids resulted in the establishment of nearly 45% more hybrid individuals than either parental type. In terms of seed germination, hybrids produced 20% more seedlings than eastern plants did. Hybrid seedlings had greater survival rates than western seedlings, with nearly 20% more living to the end of the experiment. The biomass of the resulting hybrid seedlings was also greater than that of the western seedlings; their average weight was 20% heavier.

Fig.3.


Fig. 3. Mean seed weight among the three Brazilian peppertree types. Error bars show 1 standard error. Letters indicate significant differences (P < 0.05) among types based on specific post hoc comparisons (with Bonferroni correction).

Graph representation

This data represents the average seed weight among the three different groups.

Graph interpretation

Western and hybrid seeds were significantly heavier (by ~10%) than eastern seeds. There also were significant differences in the weight of seeds from the 12 sites.

Fig.4.


Fig. 4. Mean proportion seed germination among the three Brazilian peppertree types. Error bars show 1 standard error. Letters indicate significant differences (P < 0.05) among types based on specific post hoc comparisons (with Bonferroni correction).


The calculation of total fitness across the four life stages measured revealed that hybrid total fitness was more than 50% greater than either of their progenitors.

The predispersal seed predation rates we observed were much lower than those previously recorded for Brazilian peppertree (Wheeler et al. 2001). Sampling across a similar geographic range, we found that the seed predator was absent from four of our 12 sites, while Wheeler et al. (2001) observed the seed predator at all 18 sites sampled. These incongruities in predation rates may be due to differences in methodology. We tallied seed predation by scoring exit holes in fruit, whereas Wheeler et al. (2001) dissected seed embryos from a subsample of fruit to consider diapausing insects yet to emerge. This difference also may be due to natural yearly variation in the populations of Megastigmus transvaalensis as Wheeler et al. (2001) found a significant effect of year on seed predation rates.

Our results on the phenology of seed germination are similar to other studies. Ewel et al. (1982) reported natural field germination to be concentrated from January to February, while their experimental seed outplantings showed peak germination during February and March, or 30 d after seed sowing. We observed nearly 90% of total germination within 4 wk of the first germination, occurring between March and April. There was a significant effect of block on germination and we feel this may be explained by shading from the border of tall trees on the western side of the planting area. We were unable to statistically account for this in subsequent tests, but relatively more seeds germinated in those blocks that were closest to the border of tall trees. While Ewel et al. (1982) had no germination after 6 mo, we noted germination, albeit a very small fraction (i.e., <0.01%) of total seeds sown, after 8 mo. Field seed germination reported by Panetta and McKee (1997) occurred more slowly, growing exponentially from 16–21 wk after seed was sown, and germination appeared to be associated with rainfall events. Although we provided supplemental water at the beginning of our experiment, we noted a marked increase in germination following the start of the rainy season in June followed by a decrease of the seed germination rate. Other researchers (Nilsen and Muller 1980; Ewel et al. 1982; Panetta and McKee 1997; Tassin et al. 2007) have shown that the exocarp contains seed germination inhibitors that can be removed by leaching in water. We suspect that this increase in seed germination at the start of the rainy season may have been the result of the inhibitors being leached away by water.

Whereas the phenology of seed germination exhibited a unimodal distribution with an early peak followed by a long tail, seedling mortality showed a bimodal distribution over time. Seedling mortality closely tracked seed germination from March to April and then monotonically decreased to a low point in June, only to begin increasing again with a second peak in September. This is in contrast to the findings of Ewel et al. (1982) who noted that seedling mortality was not concentrated during specific times but rather happened throughout the year. We agree with these authors that a combination of density-dependent (e.g., intraspecific competition) and density-independent (e.g., environmental conditions, specifically, too little or too much water) factors are the cause of seedling mortality. However, we cannot use submergence in summer floodwaters as they did to explain seedling death because the common garden occupied a well-drained site that did not flood. Rather, water stress may have been a more important driver of seedling mortality. The first peak of seedling mortality occurred at the end of the dry season, when we were experiencing exceptionally low precipitation. The second peak of seedling mortality began during the transition from the wet to the dry season in southern Florida, when conditions become increasingly drier. We attribute the significant effect of block on seedling survival to be due to shading from the border of tall trees on the western border of the planting area. As with the significant effect of block on seed germination, we were unable to statistically account for this in subsequent testing, but there was relatively greater seedling survival in blocks closer to the shade of the tall tree border.

Fig.5.


Fig. 5Mean proportion of seedlings surviving to the end of the 8-mo experiment among the three Brazilian peppertree types. Error bars show 1 standard error. Letters indicate significant differences (P < 0.05) among types based on specific post hoc comparisons (with Bonferroni correction).

 

Fig.6.


Fig. 6. Mean individual dry weight of harvested seedlings at the end of the 8-mo experiment among the three Brazilian peppertree types. Error bars show 1 standard error. Letters indicate significant differences (P < 0.05) among types based on specific post hoc comparisons (with Bonferroni correction).

 

Our results indicate that hybrid seeds germinated at a greater rate than eastern seeds. The difference between these two groups, at 20% more germinants for the hybrid seeds, represents ~5000 more germinated seeds from a medium-sized female tree that measures 200 cm tall with four stems (P. Pratt, unpublished data). The overall seed germination rate, ~16.4%, was similar to those reported by other authors for soil-sown experiments (15%, Nilsen and Muller 1980; <1% to >30%, Ewel et al. 1982; 17.7%, Panetta and McKee 1997). Our experimental protocol, however, allowed us to test for differences among the three types found in Florida. The result of reproductive superiority of hybrid individuals (i.e., most often the F1 generation) to their progenitors has been found in other species (for reviews, see Arnold and Hodges 1995; Gaskin and Schaal 2002; Ainouche et al. 2003; Rhode and Cruzan 2005; Campbell et al. 2006; Rieseberg et al. 2007; Ross and Auge 2008) and can contribute to the hybrid’s greater fitness. The germination of a greater number of seeds provides the filtering process of more fit genotypes via natural selection, additional raw material (i.e., seedlings). As the western and eastern invasion fronts made contact, this led to the production of intraspecific hybrids (Williams et al. 2007) and the eventual establishment of many novel genotypes (Williams et al. 2005) across a large geographic range in southern Florida.

In terms of seedling survival, we found hybrid seedlings to have an advantage over western seedlings; ~20% more hybrid than western seedlings had survived by the end of the 8-mo common garden experiment. Other researchers have reported similar cases of greater survival of hybrids versus their progenitors in common garden experiments (Rhode and Cruzan 2005; Campbell et al. 2006). These results carry significant implications for invasive species. Greater survival of novel hybrid genotypes, especially as isolated pioneering populations, may increase the rate of spread of the invasion front via the colonization of seedlings (sensu Moody and Mack 1988). The cumulative impact may be even greater as an advantage during the seed to seedling transition becomes magnified by additional advantages at subsequent life-history stages over time. For Brazilian peppertree, scaling up the hybrid advantage of both seed germination and seedling survival led to the establishment of nearly 45% more hybrid seedlings than either progenitor over the 8-mo experiment. Hybrids ultimately had a numerical advantage over both eastern and western seedlings.

At the densities achieved in our experiment, it appears that there is little evidence of greater survival of seedlings at lower densities. This does not necessarily mean that density-dependent factors (e.g., water stress, shading) were unimportant in seedling mortality compared to density-independent factors (e.g., flooding), as suggested by Ewel et al. (1982) in their study. Additional experimental evidence would be necessary to fully evaluate the effects of density on seedling survival. While our methodology was different from Ewel et al. (1982), we routinely encountered localized densities of seeds in the field that were much greater than we used in our experiment. Brazilian peppertree seeds are consumed by a wide variety of vertebrates (e.g., birds, raccoons, deer, and bears), and large numbers of seeds are deposited in the resulting scat, often several thousand seeds in an area equivalent to our individual pots (J. Geiger, personal observation). Also, ‘‘halos’’ consisting of large quantities of fallen fruit often surround the periphery of individual female trees in the field (J. Geiger, personal observation).

Of those seedlings that survived, hybrids attained greater average biomass (i.e., 20% heavier dry weight per individual) than western seedlings. This advantage is impressive considering that on average there were nearly 20% more hybrid seedlings per pot than there were western seedlings. In other words, the hybrid seedlings not only had greater survival rates (i.e., reached higher per-pot densities), but they also attained greater biomass than the western seedlings. Several recent studies on interspecific hybrids have shown greater vegetative growth of hybrids compared to parental genotypes (Vila and D’Antonio 1998; Erfmeier and Bruelheide 2005; Ross and Auge 2008). The ultimate fitness consequence of greater early vegetative growth is difficult to gauge as fitness for perennial organisms relies on measuring lifetime reproductive success (Rhode and Cruzan 2005). Still, for genotypes grown in a common garden setting, size comparisons made early in life can be a valid predictor of reproductive output as the incorporation of future survival and reproduction within the reproductive output value (sensu Caswell 1989) may be a good proxy for ultimate fitness. The next crucial questions to answer are whether the advantage hybrids have over the western seedlings in vegetative growth translates into greater survival to maturity and greater reproductive output. This knowledge would be valuable to land managers in formulating strategies to deal with this invasive plant.

While it is beyond the scope of this study to conclude that the hybridization of the eastern and western Brazilian peppertrees was the driver of their invasion in Florida, our results do show that the hybrids are superior to their progenitors in several early life-stage transitions (i.e., seed germination, seedling survival, and growth). Other evidence confirms that the current amount of genetic variation found within Florida populations is equivalent to that found in populations from its native range in South America (Williams et al. 2005). This agrees with a recent report that multiple introductions of invasive species, as with Brazilian peppertree in Florida, generally result in increasing levels of genetic variation over time (Dlugosch and Parker 2008). Due to hybridization of distant source populations, hybrid Brazilian peppertree in Florida represent novel genotypes not found in their native range, and they most likely retain high evolutionary potential.

Table 2

Table 2. Average Relative Fitness for the Three Types by Life Stage and Total Fitness.


These two facts may complicate management of this invasive. First, imported biological control agents may be maladapted to the novel genotypes that now dominate much of Florida. Second, the possibility and reality of rapid evolution in invasive species stresses the importance of addressing the evolutionary potential of introduced species in terms of risk assessment and management (Dlugosch and Parker 2008; Whitney and Gabler 2008). For exotic species, the connections between founder effects, multiple introductions, and successful invasion are paramount in comprehending how evolutionary concerns may be integrated into the management strategies for invasives. Future research should incorporate these important features that define the present state of Brazilian peppertree invasion in Florida.

 


ACKNOWLEDGEMENTS


We thank J. Burch, D. George, M. O’Quinn, L. Rodgers, and J. Taylor for their assistance with fruit collection logistics and D. Lieurance, L. Witek, B. Mattison, C. Belnavis, and H. Aguilar for their assistance with the experiment.


REFERENCES AND NOTES

 

  1. M.L. Ainouche, A. Baumel, A. Salmon, G. Yannic, New Phytol 161, 165–172 (2003).
  2. F.W. Allendorf, L.L. Lundquist, Conserv Biol 17, 24– 30 (2003).
  3. M.L. Arnold, S.A. Hodges, Trends Ecol Evol 10, 67–71 (1995).
  4. S.C.H. Barrett, B.C. Husband, The genetics of plant migration and colonization. Pages 254–278 in A.H.D. Brown, M.T. Clegg, A.L. Kahler, B.S. Weir, eds (Sinauer, Sunderland, MA, 1990).
  5. B. Blossey, R. Notzold, J Ecol 83, 887–889 (1995).
  6. L.G. Campbell, A.A. Snow, C.E. Ridley, Ecol Lett 9, 1198–1209 (2006).
  7. H. Caswell, Matrix population models. (Sinauer, Sunderland, MA, 1989). 
  8. C. Darwin, On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life. (J Murray, London, 1859) 
  9. K.M. Dlugosch, I.M. Parker, Mol Ecol 17, 431–449 (2008).
  10. N.C. Ellstrand, K.A. Schierenbeck, Proc Natl Acad Sci USA 97, 7043–7050 (2000).
  11. A. Erfmeier, H. Bruelheide, Ecography 28, 417–428 (2005).
  12. J.J. Ewel, eds. Ecology of biological invasions of North America and Hawaii. pgs 214-230 (Springer, New York, 1986).
  13. J.J. Ewel, D.S. Ojima, D.A. Karl, W.F. DeBusk, Schinus in successional ecosystems of Everglades National Park. Report T-676. (South Florida Research Center, National Park Service, Everglades National Park, Homestead, FL, 1982)
  14. B. Facon, P. Jarne, J.P. Pointier, P. David, J Evol Biol 18, 524–535 (2005).
  15. A. Ferriter, ed Brazilian peppertree management plan for Florida. (Brazilian Peppertree Task Force, Florida Exotic Pest Plant Council, 1997).
  16. J.F. Gaskin, B.A. Schaal, Proc Natl Acad Sci USA 99, 11256–11259 (2002).
  17. D.H. Habeck, F.D. Bennett, E.E. Grissell, Fla Entomol 72, 378–379 (1989).
  18. M.B. Hamilton, Mol Ecol 8, 521–523 (1999).
  19. N.A. Hardiman, T.M. Culley, Am J Bot 97, 1698–1706 (2010).
  20. J.L. Hierro, J.L. Maron, R.M. Callaway, J Ecol 93, 5–15 (2005).
  21. C.S. Kolar, D.M. Lodge, Trends Ecol Evol 16, 199–204 (2001).
  22. J.J. Kolbe, R.E. Glor, L. Rodriguez Schettino, A. Chamizo Lara, J.B. Losos, Nature 431, 177–181 (2004).
  23. J.G. Lambrinos, Ecology 85, 2061– 2070 (2004).
  24. C.E. Lee, Trends Ecol Evol 17, 386–391 (2002).
  25. M.E. Moody, R.N. Mack, J Appl Ecol 25, 1009– 1021 (1988).
  26. J.F. Morton, Econ Bot 32, 353–359 (1978).
  27. E.T. Nilsen, W.H. Muller, Bull Torrey Bot Club 107, 51–56 (1980).
  28. S.J. Novak, R.N. Mack, eds. Species invasions: insights into ecology, evolution, and biogeography. pgs 201-228 (Sinauer, Sunderland, MA, 2005). 
  29. F.D. Panetta, J. McKee, Aust J Ecol 22, 432–438 (1997).
  30. J.K. Pritchard, M. Stephens, P. Donnelly, Genetics 155, 945–959 (2000).
  31. J.K. Pritchard, W. Wen, Documentation for Structure software. Version 2, (2003). 
  32. J.M. Rhode, M.B. Cruzan, Am Nat 166, 124–139 (2005).
  33. L.H. Rieseberg, S.C. Kim, R.A. Randell, K.D. Whitney, B.L. Gross, et al. Genetica 129, 149–165 (2007).
  34. C.A. Ross, H. Auge, Plant Ecol 199, 21–31 (2008).
  35. D.C. Schmitz, D. Simberloff, R.H. Hofstetter, W. Haller, D. Sutton, eds. Strangers in paradise: impact and management of nonindigenous species in Florida. pg 961 (Island, Washington, DC, 1997). 
  36. C.A. Stockwell, A.P. Hendry, M.T. Kinnison, Trends Ecol Evol 18, 94–101 (2003).
  37. J. Tassin, J.N. Riviere, P. Clergeau, Restor Ecol 15, 412–419 (2007).
  38. M. Vila, C.M. D’Antonio, Ecol Appl 8, 1196–1205 (1998).
  39. G.S. Wheeler, L.M. Massey, M. Endries, Biol Control 22, 139–148 (2001).
  40. K.D. Whitney, C.A. Gabler, Divers Distrib 14, 569–580 (2008).
  41. D.A. Williams, E. Muchugu, W.A. Overholt, J.P. Cuda, Heredity 98, 284–293 (2007).
  42. D.A, Williams, W.A. Overholt, J.P. Cuda, C.R. Hughes, Mol Ecol 14, 3643–3656 (2005).