The invasion history of an exotic species often includes propagule introductions from multiple source populations, which can result in a wider range of trait values exhibited by invaders in the invaded range than might be expected from a single introduction (Crawford and Whitney Reference Crawford and Whitney2010). When an invader is introduced into a novel region via independent introductions, how does the existence of multiple founding populations affect its biology in ways that might affect the invasion process? Aspects of the biology that could affect invasion and that might well differ by source population include biomass allocation to reproduction (Lavergne et al. Reference Lavergne, Thompson, Garnier and Debussche2004; Moracová et al. Reference Moracová, Pyšek, Jarošik and Pergl2015), size at maturity (Weiner et al. Reference Weiner, Campbell, Pino and Echarte2009), and population dynamics (Iles et al. Reference Iles, Salguero-Gomez, Adler and Koons2016). Understanding the genetic history of introductions and how source population affects the life history and ecology of invaders may provide insights to improve management strategies and improve understanding of invasive characteristics of select species (Allendorf and Lundquist Reference Allendorf and Lundquist2003).
Box 1 Management Implications
The use of a single, easily obtained plant parameter for the prediction of multiple-plant biomass production and allocation has broad application in various disciplines of plant biology. The extrapolation of these biometric relationships to the management of exotic plants is also applicable, although these kinds of applications have received less attention in the scientific literature. For example, after comparing predictability of an assortment of variables, Rayamajhi and colleagues determined that diameter at breast height (DBH, 1.3 m) was an accurate predictor for biomass allocation of melaleuca [Melaleuca quinquenervia (Cav.) Blake] in Florida. DBH has subsequently been used to predict M. quinquenervia removal costs, carrying capacity for herbivores, and natural enemy–mediated alterations in plant partitioned biomass. In this study, stem diameter at the soil level is an effective predictor of Schinus terebinthifolius Raddi (Brazilian peppertree) biomass and, in contrast to DBH, offers the advantage of including all individuals regardless of their age, stature, or number of stems.
Schinus terebinthifolius Raddi (Anacardiaceae) is a large dioecious woody shrub, occasionally of small tree stature, growing to 7.5 m or more in height, often with multiple branches bearing bright red fruits that appear on female plants seasonally (Tobe et al. Reference Tobe, Craddock Burks, Cantrell, Garland, Sweeley, Hall, Wallace, Anglin, Nelson, Cooper, Bickner, Gilbert, Aymond, Greenwood and Raymond1998). Native to Brazil, Paraguay, and Argentina, the shrub has become an established problematic species globally around the subtropics including Florida (Cuda et al. Reference Cuda, Ferriter, Manrique and Medal2006; Ewel et al. Reference Ewel, Ojima, Karl and DeBusk1982; Morton Reference Morton1978). Schinus terebinthifolius was introduced to Florida in two separate introductions: one on the western coast near the town of Punta Gorda (Charlotte County) and one on the eastern coast near Miami (Dade County) (Williams et al. Reference Williams, Overholt, Cuda and Hughes2005). The source populations for these two introductions were from different regions of the native range of S. terebinthifolius (coastal southeastern Brazil and northeastern Brazil in the Bahia state, respectively) and, consequently, the genetic backgrounds of the two founding populations were distinct (Mukherjee et al. Reference Mukherjee, Williams, Wheeler, Cuda, Pal and Overholt2012; Williams et al. Reference Williams, Overholt, Cuda and Hughes2005). Today, there are three recognized genetic lineages: one originating from the western Punta Gorda introduction (the western genetic type), one originating from the eastern Miami introduction (the eastern genetic type), and a hybrid lineage resulting from intraspecific hybridization between individuals from the eastern and western introductions (the hybrid genetic type) (Mukherjee et al. Reference Mukherjee, Williams, Wheeler, Cuda, Pal and Overholt2012; Williams et al. Reference Williams, Overholt, Cuda and Hughes2005, Reference Williams, Muchugu, Overholt and Cuda2007). Previous common garden experiments have suggested that individuals from hybrid populations should have an advantage over individuals from either the eastern or western genetic types (Geiger et al. Reference Geiger, Pratt, Wheeler and Williams2011).
Herein, spatial variation in biomass allocation across these genetic types was assessed by destructively sampling a stratified random set of individuals of each type. Determining whether such differences in biomass allocation would result in differences in fitness is a question that can only be addressed by a longitudinal study of tagged individuals over time. In this context, a method for estimating aboveground biomass parameters from a set of nondestructive measurements (e.g., diameter of woody stems) is needed. Understanding biomass partitioning in S. terebinthifolius may also be useful in assessing the impact of herbivory by potential biological control agents on growth and fitness. Therefore, the specific objectives of this study were to (1) parameterize statistical models of the dependency of several performance metrics (biomass and numbers of fruits and leaves) on stem diameter (a nondestructive metric) and (2) quantify the influence of genetic type on biomass allocation to different functions.
Materials and Methods
Study Organism and Sites
As described in Tobe et al. (Reference Tobe, Craddock Burks, Cantrell, Garland, Sweeley, Hall, Wallace, Anglin, Nelson, Cooper, Bickner, Gilbert, Aymond, Greenwood and Raymond1998), S. terebinthifolius is a large shrub that can grow to a small size tree of up to 7.5-m high with multiple stems that are underground in some cases, arise separately from the ground, and give rise to a common crown. Each stem bears multiple branches along the stem axis, resulting in a multi-branching individual that can be as tall as it is wide.
Schinus terebinthifolius plants were destructively sampled at seven sites (Figure 1, Table 1) chosen to represent the range of introduction history (eastern genetic type: Chekika, Fort Pierce, and Holiday Park; hybrid genetic type: Big Cypress and Cape Canaveral; and western genetic type: Punta Gorda and Wild Turkey). Classification of sites into genetic types was based on a growing body of literature that examines the spatial pattern of Schinus terebinthifolius population genetics across Florida (Mukherjee et al. Reference Mukherjee, Williams, Wheeler, Cuda, Pal and Overholt2012; Williams et al. Reference Williams, Overholt, Cuda and Hughes2005, Reference Williams, Muchugu, Overholt and Cuda2007). Select stands were chosen based on landowner permission and access. We assumed that the individuals we harvested at each site are true to the genetic types as predicted by Williams et al.’s (2007) maps of genotypes.
Figure 1 Location of each of the seven study sites.
Table 1 Sites used to quantify the allometry of Schinus terebinthifolius in natural systems of Florida. GPS data are presented in decimal degrees and genetic type refers to predominate haplotype or hybrid as characterized by Williams et al. (Reference Williams, Muchugu, Overholt and Cuda2007).
Field Measures of Plant Size and Biomass
Field sampling took place during the winter months when seeds were present, from October 27 through February 10 over 2008 to 2011. The range of sizes was delineated at each site and we selected individuals within each portion of the size range to ensure representation of the full range of sizes present at a site. In the field, the diameter at the base of each stem (trunk) as well as the diameter at the base of each branch coming off from a main stem were measured using calipers for small individuals and DBH tapes for individuals with diameters too large to measure with calipers. For multistemmed individuals, each stem was measured separately. To determine whether nearby stems belonged to the same individual or were separate individuals, we dug down into the soil to determine whether the stems were arising from the same source below ground. Individual stems that lacked a common crown were treated as separate individuals. Individual stems were excised at the soil level and total plant fresh weight was measured while at the field site. Larger stems were subsampled by selecting approximately 50% of the branches and processing their constituent parts (leaves, fruits, and wood) at the USDA/ARS Invasive Plant Lab in Fort Lauderdale, FL. The fresh weights of the subsample components were recorded in the field. The subsamples and smaller individuals that were not subsampled were transported to the laboratory and dried to a constant weight at 60 C. During the drying period, each sample bag was weighed once a week, and the final dry weight of the sample was recorded when the weight of the sample reached a constant weight. The fresh weight to dry weight ratio obtained from the subsamples was used to calculate the total dry biomass of plants. Additionally, the total number of fruit and total number of leaves (which in S. terebinthifolius are compound with multiple leaflets) were counted per branch. Across all sites, 257 stems were destructively sampled, including 86 stems from sites with predominantly eastern genetic types, 109 stems from sites with predominantly hybrid genetic type, and 62 stems from sites with predominantly western genetic types (Table 1).
Simulation of Missing Branch-level Data for Larger Stems
Because only a subsample of branches from larger stems was sampled, the biomass fractions (branch dry weight, fruit dry weight, and leaf dry weight) and number of leaves and fruits for the remaining branches were simulated using the data from the measured branches. We first used linear regression to determine the relationship between branch basal diameter and each of the biomass allocation dependent variables. We then simulated each missing branch-level value by drawing a random value from a normal distribution with mean and variance determined from the regression based on branch diameter.
Stem-level Biomass Variables
After missing branch-level values were filled in through simulation, all branch-level biomass variables (branch dry weight, fruit dry weight, and leaf weight) as well as the number of leaves and fruits per branch were summed by stem to arrive at the stem-level variables of interest: total dry weight, wood dry weight, fruit dry weight, leaf dry weight, number of leaves, and number of fruits). These variables all involved destructive sampling of the stems. Together, these stem-level biomass variables indicate how an individual allocates its aboveground biomass to different tissues (leaves, wood, and fruit).
Aboveground Biomass Allocation by History of Introduction
The proportion of aboveground biomass devoted to wood, leaves, and fruits was calculated for each stem by dividing the dry mass of wood, leaves, and fruits, respectively, by the total aboveground dry mass. To investigate the effect of haplotype on biomass allocation, we conducted two Kruskal-Wallis tests: one to determine whether the haplotypes differed with respect to the proportion of biomass allocated to fruit and a second to determine whether the haplotypes differed with respect to the proportion of biomass allocated to wood.
Aboveground Biomass Production by History of Introduction
To determine whether aboveground biomass production differed by history of introduction, we used Analysis of Covariance where we modeled each of the biomass and production variables (total dry weight, branch dry weight, fruit dry weight, leaf dry weight, number of fruit, and number of leaves) as a function of genetic type (eastern, hybrid, or western).
To determine the relationship between the nondestructive measure, stem size, and each of the biomass and production variables (total dry weight, branch dry weight, fruit dry weight, leaf dry weight, number of fruit, and number of leaves), we used linear regression of squared stem diameter (x 2 ) and each of the biomass and production variables of interest (y) by genetic type. A variety of transformations were explored, but squaring stem diameter consistently resulted in higher coefficients of determination so this transformation was used throughout. For each biomass allocation variable of interest, there is an equation of the following form:
$$y{\,\equals\,}m \cdot x^{2} $$
where the slope m varies depending on the genetic history of the individual stem. Lastly, a weighted average for each of the three genetic-type–specific slopes was calculated by weighting each slope by the proportion of individuals of that genetic type in our sample. In all the equations, the intercept was forced through zero.
Results and Discussion
Allocation of aboveground biomass to fruit differed statistically among the genetic types (Kruskal-Wallis test, χ2=25.975, df=2, p<0.00001), with eastern stems devoting the greatest proportion of biomass to fruit (Figure 2A). Allocation of aboveground biomass to wood also differed among the genetic types (Kruskal-Wallis test, χ2=7.2261, df=2, p=0.027): eastern and hybrid stems allocated similar proportions to wood, while western stems allocated the smallest proportion to wood (Figure 2B). The remaining biomass was allocated to leaves, with western stems having the highest allocation to leaves (Figure 2C).
Figure 2 Biomass allocation to (A) fruit, (B) wood, and (C) leaves by genetic type.
For all performance variables of interest (total dry weight, branch dry weight, number of fruits, fruit dry weight, leaf dry weight, and number of leaves), the slopes for the three genetic types were significantly different from each other (Table 2). The slope indicates how quickly each of the dependent variables increases with increasing stem size. For total dry weight, wood dry weight, and leaf dry weight, hybrid populations had the steepest slopes, while for number of fruits, fruit dry weight, and number of leaves, western populations had the steepest slopes (Table 2).
Table 2 Equation parameters for predicting how each biomass variable of interest (y) is related to basal stem diameter (x).
Individual stems from populations with different genetic histories varied in aboveground biomass allocation. Eastern individuals devoted more of their aboveground biomass to fruits than either of the other two genetic types, while western individuals tended to devote more of their aboveground biomass to leaves than either of the other two genetic types (Figure 2). The decision about whether to allocate biomass to fruit (reproduction) or leaves (photosynthesis) is a well-studied evolutionary tradeoff in biology (Obeso Reference Obeso2002; Reznick Reference Reznick1985; Smith Reference Smith1976). Previous studies have found that invaders tend to invest more biomass into tissue involved in carbon capture, allowing them to grow more quickly than in their native range (Grotkopp and Rejmanék Reference Grotkopp and Rejmanék2007; McDowell Reference McDowell2002; Tho et al. Reference Tho, Sorrell, Lambertini, Eller and Brix2016).
Western individuals produced the most fruits (552 fruits per cm of diameter), followed by hybrid individuals (425 fruits per cm of diameter), while eastern individuals produced the fewest fruits (400 fruits per cm of diameter; Figure 3). Overall, this means that a reproductive female with a basal stem diameter of 3.05 cm (the median stem diameter of all stems in our study) would be expected to produce anywhere between 1,220 and 1,684 fruits per year depending on its genetic type. However, a previous common-garden experiment suggests that hybrid seeds have a higher survival rate than western seeds (Geiger et al. Reference Geiger, Pratt, Wheeler and Williams2011), so despite producing more fruit, recruitment of new western individuals might indeed be lower than hybrid individuals. We have shown here that populations that differ in introduction history differ in resource allocation; what is yet to be determined are the fitness consequences of such differences. In a companion study, to address the fitness consequences, we will be following populations of marked individuals from these different sources. The demographic data will be used to estimate population growth rates associated with the different allocation strategies.
Figure 3 Scatterplot showing the relationship between squared stem diameter and number of fruits. Dash-dot lines represent the regression equations for individuals of eastern genetic type, dashed lines represent the regression equations for individuals of hybrid genetic type, dotted lines represent the regression equations for individuals of western genetic type, and solid lines represent the weighted average.
Individuals from the hybrid populations exhibited the highest overall biomass per cm of stem diameter, dry weight of leaves per cm of stem diameter, dry weight of wood per cm of stem diameter, and the smallest fruit dry weight per cm of stem diameter. The rates of increase in biomass per unit size of hybrid stems were intermediate to the rates of eastern and western populations in regard to both the number of fruits and the number of leaves. An earlier common-garden experiment showed that hybrid seedlings had a higher biomass than seedlings of the western genetic type (Geiger et al. Reference Geiger, Pratt, Wheeler and Williams2011). Our results suggest that hybrids maintain this advantage in total biomass throughout larger life stages, which may contribute to life-long increases in performance of the hybrid individuals, a hypothesis known as hybrid vigor.
These data indicate that there is strong evidence that genetic lineages influence biomass allocation. However, caution should be used when interpreting these results. While we assumed that all plants in each location shared the same history of introduction, it is likely that there is some overlap, with individuals of different genetic lineages living in the same population. Indeed, the proportion of eastern ancestry (in contrast to western ancestry) is predicted to fall between 0.8 to 1 for Chekika; 0.6 to 0.8 for Big Cypress, Ft. Pierce, and Holiday Park; 0.4 to 0.6 for Cape Canaveral, 0.2 to 0.4 for Wild Turkey; and between 0 and 0.2 for Punta Gorda (Figure 7 in Williams et al. Reference Williams, Muchugu, Overholt and Cuda2007).
Because S. terebinthifolius is a tropical shrub lacking annual tree rings, it is difficult to reliably know the ages of individuals, so it is possible that the ages of individuals differed between the sites. It is also likely that spatial variation in environmental factors (e.g., temperature, precipitation, habitat type, soil type) across the study region may have contributed to the different patterns of biomass allocation we observed. However, the general trends we found and equations for estimating biomass components and the production of fruits and leaves should still be scalable to populations with similar genetic compositions. Also, we did not measure belowground biomass in this study; however, underground competition and alteration of the soil microbiome by S. terebinthifolius are likely to be important mechanisms facilitating the continued invasion and success of S. terebinthifolius individuals in new habitats (Dawkins and Esiobu Reference Dawkins and Esiobu2016; Nickerson and Flory Reference Nickerson and Flory2015).
In conclusion, we found that there was spatial variation in the biomass allocation of S. terebinthifolius stems, revealing a signature of their introduction history. Hybrid individuals tended to have a greater biomass than individuals from either of the source populations, although they tended to devote a greater proportion of their biomass to leaves and woody material than to fruit. We developed a method for estimating important plant size metrics via nondestructive sampling that will allow us to estimate biomass production and allocation of individuals of different sizes in long-term demographic plots.
Acknowledgements
The authors wish to thank J. Geiger, L. Kasarjian, E. Mattison, J. Ledi, R. Moscat, L. Witek, and H. Aguilar for assistance with field work. The authors are also indebted to collaborators who helped identify S. terebinthifolius study populations and facilitated site access: L. Rodgers from the South Florida Water Management District, H. Cooley from Everglades National Park, D. Hall of the USDA’s U.S. Horticulture Research Laboratory in Fort Pierce, D. George at the Cape Canaveral Airforce Station, J. Burch at Big Cypress National Preserve, C. Olson from Lee County Conservation 20/20, and the Charlotte Harbor Preserve State Park staff. This research was funded in part by the USDA’s Invasive Plant Research Laboratory. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. USDA is an equal opportunity employer and provider.
