Biological Invasions

, Volume 14, Issue 7, pp 1415–1430 | Cite as

Brazilian peppertree (Schinus terebinthifolius) in Florida and South America: evidence of a possible niche shift driven by hybridization

  • A. Mukherjee
  • D. A. Williams
  • G. S. Wheeler
  • J. P. Cuda
  • S. Pal
  • W. A. Overholt
Original Paper


Brazilian peppertree (Schinus terebinthifolius Raddi, Anacardiaceae) was introduced into Florida from South America in the 1800s and commercialized as an ornamental plant. Based on herbaria records and available literature, it began to escape cultivation and invade ruderal and natural habitats in the 1950s, and is now considered to be one of Florida’s most widespread and damaging invasive plants. Historical records and molecular evidence indicate that two genetic lineages of Brazilian peppertree were established in Florida, one in Miami on the east coast and a second near Punta Gorda on the west coast. Since arriving, the distributions of these two types have greatly expanded, and they have extensively hybridized. Principal component analysis and reciprocal niche fitting were used to test the equivalency of climatic niches of the Florida populations with the climatic niches of the two South American chloroplast haplotype groups which established in Florida. Both approaches indicated a significant shift in niches between the parental populations in the native range and the invasive populations in Florida. The models, however, closely predicted the areas of initial establishment. We hypothesize that (1) Brazilian peppertree was able to gain an initial foothold in Florida due to niche similarity and (2) the current dissimilarity in native and exotic niches is due to hybridization followed by rapid selection of genotypes adapted to Florida’s climate. In addition, to examine the potential consequence of the introduction of additional genetic diversity from the native range on invasion success, a niche model constructed with occurrences of all native genotypes was projected onto the continental United States. The result of this test indicated that under such an event, the potential invasive range would greatly expand to cover most of the southeastern USA. Our study suggests that multiple introductions from disjunct regions in the native range can facilitate invasion success.


Invasive species Niche conservation Hybridization Range expansion Lag period 


Ecological niche models (ENMs), also referred to as species distribution models (SDMs) (Franklin 2009), are often used to characterize the climate occupied by organisms in their native range, and to predict their distributions in areas outside of the native range (Peterson and Shaw 2003; Ebeling et al. 2008; Giovanelli et al. 2008; Elith and Leathwick 2009; Hinojosa-Díaz et al. 2009). When an organism has already invaded a new geographic area, ENMs constructed from native and introduced occurrences can be compared to test the conservatism of climatic niches across the two habitats (Broennimann et al. 2007; Pearman et al. 2008). Niche conservation in this sense can be defined as the tendency of the climatic niche of a species to remain unchanged over time or space (Wiens and Graham 2005). Dissimilarities in climatic niches between exotic and native ranges indicate that either the fundamental or the realized niche of the species has shifted. In the Grinnellian sense, the fundamental niche of a taxon includes all possible abiotic conditions where it can persist without immigration, whereas the realized niche is the set of environments actually occupied (Hutchinson 1957; Soberson and Nakamura 2009; Rödder and Engler 2011). In addition, the potential niche has been defined as the part of the fundamental niche that exists in a given geographic space and time (Jackson and Overpeck 2000; Soberson and Nakamura 2009). A shift in the realized niche may be due to release of an organism from competition or predation in the native range (Broennimann and Guisan 2008; Pearman et al. 2008; Jimenez-Valverde et al. 2011). On the other hand, genetic divergence in the invasive population as a result of rapid evolution following introduction (reviewed in Prentis et al. 2008) can cause a shift in the species’ fundamental niche. Multiple introductions of an organism from different native range source populations, followed by hybridization, may greatly increase genetic variability in the invaded range, and allow rapid adaptation to new niches (Novak and Mack 2005; Kolbe et al. 2007; Roman and Darling 2007; Williams et al. 2007; Suarez and Tsutsui 2008). Hybridization as a possible mechanism for increased invasiveness was first proposed by Ellstrand and Schierenbeck (2000), and has recently been reviewed by Schierenbeck and Ellstrand (2009). Williams et al. (2007) suggested that interspecific hybridization may provide an explanation for the aggressiveness of Brazilian peppertree in Florida.

Brazilian peppertree (Schinus terebinthifolius Raddi, Anacardiaceae), a South American native, is considered to be one of the most serious invasive plants in Florida (Schmitz et al. 1997), currently infesting an estimated 280,000 ha of terrestrial habitats in the state (Cuda et al. 2006). It invades disturbed sites such as highway right-of-ways, canals, fallow farmlands, pastures, and drained wetlands, and also natural communities including pinelands, hardwood hammocks, and mangrove forests (Cuda et al. 2006). Brazilian peppertree is a highly prolific seed producer and is able to reproduce after 3 years of growth (Ewel et al. 1982). Ecological studies in Florida have shown that Brazilian peppertree is tolerant of shade (Ewel 1986), fire (Doren et al. 1991), drought (Nilsen and Muller 1980a), and saline conditions (Ewe and Sternberg 2002, 2005; Ewe et al. 2007) and has allelopathic effects on neighboring plants (Gogue et al. 1974; Nilsen and Muller 1980b; Morgan and Overholt 2005; Donnelly et al. 2008). Although no cold tolerance studies have been conducted, the restricted distribution of Brazilian peppertree in Florida, along with its sub-tropical native distribution, strongly suggest that it has limited cold tolerance (Carhalho 1994; Gioeli and Langeland 2009).

The native range of Brazilian peppertree includes Brazil, Paraguay, Uruguay, and northern Argentina, and in these areas it is found in a variety of habitats from sea level to >1,300 m (Barkley 1944; Carhalho 1994; Wheeler unpublished). Historical accounts suggest that Brazilian peppertree was introduced into Florida at least three times. The first record is from Barkley (1944) where he mentions a specimen collected in Florida by Ferdinand Rugel sometime between 1842 and 1849. A second set of introductions occurred in 1898 and 1899 when shipments of seeds from Algeria, France, and an unknown location were sent to the USDA by the plant explorer, Walter T. Swingle. These seeds were forwarded to the Plant Introduction Station in Miami, and some seedlings were distributed locally (Morton 1978). A third introduction occurred in about 1926 when Dr. George Stone obtained seeds from ‘somewhere in Brazil’ and planted them at his home in Punta Gorda on the southwestern coast of Florida (Nehrling 1944). It was not until more than 100 years after the first introduction that Brazilian peppertree was recognized as an invader of native habitats in far southern Florida (Ewel et al. 1982; Alexander and Crook 1984). Brazilian peppertree is now widely naturalized in central and southern Florida, and further north along the coastal regions where the climate is moderated by proximity to the ocean (Wunderlin and Hansen 2004; Anonymous 2007a, b).

The molecular genetics of Brazilian peppertree in the native range and in Florida have recently been investigated (Williams et al. 2005, 2007). Two distinct chloroplast haplotypes were found in Florida, indicative of the establishment of plants from two source populations. The areas of introduction of the two haplotypes were most likely Miami and Punta Gorda, which agrees with the historical record. The origin of the western (Punta Gorda) A haplotype is southeastern Brazil (Williams et al. 2005), whereas the origin of the eastern (Miami) B haplotype is further north near Salvador in Bahia State (Williams, unpublished). Since arriving in Florida, the two haplotypes have extensively hybridized such that there is a clinal gradient in eastern/western ancestry along a southeast to northwest axis, as evidenced by variation at nuclear microsatellite markers (Williams et al. 2007). A recent common garden experiment showed that hybrids of these two introductions have higher survival, growth rates and biomass in Florida than the parentals (Geiger et al. 2011).

In this study, the conservation of climatic niches between invasive and native populations of Brazilian peppertree was tested using both principal component analysis (PCA) and reciprocal niche fitting approaches. In addition, the potential consequence of the introduction of additional genetic diversity from the native range on invasion success was examined. We hypothesize that the evolution of novel Brazilian peppertree genotypes adapted to the climate in Florida has resulted from extensive hybridization between the two founding lineages originally introduced, and thus, we predict there will not be a conservation of niches between the native and introduced ranges.


Native range occurrence data

Field collections of Brazilian peppertree leaf tissue (n = 718) were made as part of explorative surveys in the native range to identify potential biological control agents of this invasive weed. The geoposition of each survey site was recorded. Collections in the Argentina range included samples from 13 trees collected between 2004 and 2009 in Misiones, Corrientes, and Entre Rios provinces. In Brazil, 696 trees were sampled between 2003 and 2009, and included all areas of the known range (JBRJ 2009, NYBG 2009, Tropicos 2009). Brazilian peppertree is reported to occur mostly along the Atlantic coast from Recife, Pernambuco (S 8.05°) south to Bagé, Rio Grande do Sul, Brazil (S 31.33°), then west to northeastern Argentina and eastern Paraguay. In Brazil, our surveys ranged from Natal, Rio Grande do Norte (S 5.79°), to Pelotas, Rio Grande do Sul (S 31.76°) at the northern and the southern extremes of Brazilian peppertree’s range, respectively. In Paraguay, nine samples were collected during surveys in 2003 and 2005.

Florida occurrence data

In total, 707 trees distributed throughout the invaded range in Florida were sampled from 2003 to 2006. These samples are the same as those previously genetically characterized in Williams et al. (2005, 2007). In both the native and invasive ranges, samples consisted of growing tips or young leaf tissue preserved in either 95% ethanol or silica gel.

Florida herbarium records

To reconstruct the history of the escape from cultivation of Brazilian peppertree in Florida, label information was collected from the three largest herbaria in Florida; the University of Florida Herbarium (FLAS) (, the Fairchild Botanical Garden Herbarium ( and the University of South Florida Herbarium ( Based on label data, Brazilian peppertree specimens were classified as being collected in cultivation, in a ruderal area (roadside, canal bank or other disturbed area), or in a relatively undisturbed natural area.

Climate data

We used the 2.5 arc minute resolution (~5 km2 at equator) current climate data, available freely from the WORLDCLIM database ( (Hijmans et al. 2005). Created by interpolating high resolution (30 arc second or ~1 km2 at equator) global climate data, the WORLDCLIM database provides monthly values of minimum/maximum temperature and precipitation, as well as other bioclimatic variables (Hijmans et al. 2005). Based on our working knowledge of the climatic requirements of Brazilian peppertree, six bioclimatic variables were used in our study to compare climatic niches between exotic and native habitats (see Table 1 for variable descriptions). Bioclimatic variables have been widely used in other niche based plant distribution modeling studies (Joyner et al. 2010; Mukherjee et al. 2011), and several authors have recommend that the selection of variables be based on ecological knowledge of the taxon being modelled (e.g. Broennimann et al. 2007; Elith and Leathwick 2009; Elith et al. 2010; Austin and Van Niel 2011). The worldwide gridded bioclimatic variables were clipped to match the spatial extent of the continental United States (US) and South America (SA) using DIVA GIS 5.4 (Hijmans et al. 2001).
Table 1

Loadings on three PCA axes of variables used to examine shift in climatic niche of Brazilian peppertree

Climate variables

Variable descriptions

Axis-1 (58%)

Axis-2 (21%)

Axis-3 (13%)

Bio 1

Annual mean temperature




Bio 4

Temperature seasonality




Bio 6

Min. temperature of coldest month




Bio 7

Temperature annual range




Bio 12

Annual precipitation




Bio 15

Precipitation seasonality




Values > 0 indicate a positive contribution, whereas, those <0 indicate a negative contribution to the axis. Values in parenthesis denote the percent of variability described by each axis, cumulatively explaining ~92% of variability

Genetic analyses

DNA extraction, polymerase chain reaction (PCR), and sequencing methods are given in detail in Williams et al. (2005). Briefly, we sequenced 716 bp of the trnS-trnG intergenic region of the chloroplast DNA (Hamilton 1999) in both forward and reverse directions for all 718 native samples used in this study. We directly sequenced 50 individuals in Florida and then determined the haplotypes of the rest of the individuals using length variation in a mononucleotide repeat in the trnS-trnG intergenic region (Williams et al. 2007). Sequences were trimmed and edited in SEQUENCHER v. 4.8 (Gene Codes Corporations) and aligned with Clustal W (Thompson et al. 1994). We used TCS to construct a parsimony network of all unique haplotypes (Clement et al. 2000). The network was then nested using the methods of Templeton et al. (1987). A full phylogeographic analysis of these clades in the native range will be presented elsewhere (Williams et al. unpublished data).

Tests of niche conservatism

The conservation of climatic niches of Brazilian peppertree between the native and invasive range was tested in environmental space (E-space, Rödder and Engler 2011) using Principal Component Analysis (PCA) and in geographic space (G-space, Rödder and Engler 2011) using reciprocal niche modeling. Specifically, we compared the native ranges of the two haplotype groups that contain the A and B haplotypes (SaGr2 and SaGr1) to the introduced A and B haplotype ranges in Florida (FlA and FlB) (see results for descriptions of the genetic groupings). To ensure validity of statistical analyses while testing conservatism of climatic niche, we only used a single occurrence point per grid cell (Joyner et al. 2010). In total, 194 point occurrences for FlA, 150 point occurrences for FlB, 98 point occurrences for SaGr1 and 94 point occurrences for SaGr2 were used for the analyses.

PCA based approach

Following Broennimann et al. (2007), bioclimatic space occupied by invasive and native haplotypes was examined by PCA analysis of a covariance matrix of the six climatic variables (giving equal weights to occurrences) using the ‘ade4’ package in R (, version 2.11.1). Statistical differences in climatic niches were verified using Monte-Carlo tests with 99 repeats (α = 0.05) with the inertia (or variance) percentages calculated between haplotype groups (Akhisar and Bener 2002; Broennimann et al. 2007). Niche shift was examined by drawing scatter plots showing positions of occurrence datasets on the climate space. Convex hulls were drawn around the climate space of each haplotype group to indicate the prevalence of the climate conditions encompassing 25, 50, 75 and 100% of occurrence points. Climatic variation within a dataset was illustrated with a 1.5 standard deviation inertia ellipse around the centroid of climate space. Correlation circles were constructed to indicate the contribution of each climatic variable to the PCA axes.

Niche modeling approach

Modeling parameters

The Maximum Entropy Species Distribution Model 3.3.3a (MaxEnt) (Phillips et al. 2006), was used for reciprocal-predictions of Brazilian peppertree’s distribution between invasive and native habitats. MaxEnt was used because earlier studies have shown that it often performs significantly better than other available models (Elith et al. 2006; Phillips et al. 2006; Rödder and Engler 2011, but see Peterson et al. 2007 for alternative view). MaxEnt was run using two approaches. First, each occurrence dataset (n = 4) was randomly divided into 10 cross-validation partitions (hereafter replications, named as Rep-1 to Rep-10), with 70% data for model training (training partitions) and the remaining 30% for extrinsic testing (test partitions) of predictive accuracy. For each training partition, MaxEnt was run with 100 replicates. The random seeded subsample procedure was used to generate intrinsic test data (=10% of training data). The average of replicated predictions was considered for statistical analysis. All other MaxEnt settings were set at default levels. In the second approach, full occurrence datasets (i.e. no extrinsic test points) were used (Lozier et al. 2009). The resultant predictions were used to measure the degree of overlap between predicted distributions (see "Model Evaluation" section below for methods).

Model background

MaxEnt estimates the probability of species presence by contrasting the probability density of environmental variables across known occurrences of the species to that of randomly selected pseudo-absences from the target landscape or model background (Elith et al. 2011). As demonstrated by VanDerWal et al. (2009) and more recently by Barve et al. (2011), the geographical extent of modeling background may have significant influence on predicted distributions. To assess background related variability in the modeling output for each prediction, we used five background areas based on the ‘M’ region, defined as the geographic region accessible to a species through dispersal (Soberon and Peterson 2005; Barve et al. 2011). As it is not possible to determine the region that has been historically explored by Brazilian peppertree, we restricted the M region based on our survey data and South American records from the New York Botanical Garden (NYBG 2009). We first generated a convex hull around all known occurrences of Brazilian peppertree collected during our field surveys in each range. The convex hull was then buffered with 0, 10, 20 and 40% additional area to generate increasingly larger background areas. In addition, we used entire SA and US as background regions (referred to as country background) for native and invasive habitats, respectively. Supplementary Figs. S1 and S2 illustrate the background areas used for native and invasive ranges, respectively.

The increasingly larger backgrounds described above were restricted to either native or invasive ranges of BP. However, as BP now occurs in North and South America, the entire area is potentially part of the fundamental niche of this species. Considering this, we also generated a convex hull around all known occurrences (native + invasive) of BP and used this as background for generation of niche models (supplementary Fig. S3).

Model evaluation

Test of predictive accuracy: The area under the curve (AUC) of the receiver-operating characteristics (ROC) plot was used to test the accuracy of MaxEnt’s predictions (Fielding and Bell 1997). AUC is the summation of area under the ROC plot, constructed by plotting sensitivity (true positive rate, on y axis) against 1-specificity (false positive error rate, on x axis) (Franklin 2009). In order to confirm a shift in climatic niche, reciprocal comparisons of AUC were performed not only using the independent test datasets (30%, corresponding to each training partitions), but also using full datasets of other native and exotic occurrences. For presence only models, the value of AUC ranges between zero (counter prediction) to 1-a/2 (where a is the species prevalence or proportion of occupied sites in the landscape, which is generally unknown) (Elith et al. 2011). Following Thuiller et al. (2006), we used a conservative classification of AUC, where AUC < 0.8 was considered as poor or null; 0.8 < AUC < 0.9 as fair; 0.9 < AUC < 0.95 as good and 0.95 < AUC < 1.0 as very good. AUC was calculated using the ‘ROCR’ package of R. Scatter plots showing model performances (average AUC ± SD, n = 10 replications) were constructed for visual representation of accuracy of reciprocal predictions (Broennimann et al. 2007).

Test of niche overlap: The accuracies of reciprocal niche predictions also were tested using the niche overlap test proposed by Warren et al. (2008). We tested the level of overlap of predicted distributions generated by pairs of Brazilian peppertree occurrence datasets under the same background extent. The niche overlap test was implemented using the freely available software ENMTools 1.3 (Warren et al. 2008). We used the similarity statistic I, which ranges between 0 (no overlap) and 1 (identical prediction), to determine the levels of overlap between predicted distributions. To implement the niche overlap test, the continuous probability predictions generated by MaxEnt were converted to binary (presence/absence) predictions using the lowest presence threshold, defined as the minimum non-zero predictive value received by any known occurrence (Pearson et al. 2007). The lowest presence thresholds calculated for calibration data were replicated for binary conversion of projections. The observed values of I were compared to the null distribution using a one-tailed test (at α = 0.05) to determine statistical difference (Warren et al. 2008). For a pair of occurrence datasets, with nx and ny points, random sets of data (100 replicates) with nx and ny data points were generated from pooled set of nx + ny occurrences (Warren et al. 2008). Null distributions of I values were generated from niche models created using random datasets. For each background extent, null distributions for each pair of Brazilian peppertree occurrence datasets were generated using the Identity test tool available within ENMTools.

For preparation of final distribution maps, the results of the full dataset run were converted to binary maps using the lowest presence threshold. Predictions generated from either native or invasive occurrences were reclassified and color coded. Climates associated with centroids of Miami and Punta Gorda were considered to be representative of the climates at the locations of introduction. For visual reference, points representing putative locations of the first populations of Brazilian peppertree in Florida were included in the final prediction maps of invasive habitat.

Visualization of niche shift

To provide a visual representation of niche shift, the predicted niches of different haplotypes were plotted in bivariate space of annual mean temperature (Bio-1) and annual precipitation (Bio-12). These two variables were selected because they are uncorrelated and easy to interpret. For each occurrence dataset, the values of Bio-1 and Bio-12 associated with centroids of grid cells predicted to be suitable for Brazilian peppertree were extracted using the raster calculator within the Spatial Analyst extension of ArcMap 9.3.1 (ESRI, Redlands, CA). For visual clarity, 10% of the centroids were randomly selected for construction of niche space (Peterson and Shaw 2003). In addition, convex hulls, defined by extremes of Bio-1 and Bio-12, were drawn to clearly demarcate niche space. Convex hulls were calculated using the ‘chull’ function of R. The climatic positions of the introduction sites also were included in the scatter plots for visual reference.

Predicted distribution using all native range occurrences

To examine whether the introduction of additional genetic diversity from the native range would affect Brazilian peppertree’s distribution in the USA, we constructed a niche model with all available native occurrences (n = 718), irrespective of haplotype. As no statistical analysis was intended, all available data were used as the input. The model was generated using the default settings of MaxEnt with South America as the background and projected onto the continental USA. The lowest presence threshold was used to identify regions that were potentially suitable for the establishment Brazilian peppertree in the USA. A color coded binary (present/absent) prediction map was developed by classifying pixels with a suitability score lower than the minimum presence threshold as unsuitable and those with higher scores as suitable.


Genetic patterns

The A and B haplotypes that were introduced into Florida are nested in two separate clades in which haplotypes are separated by one to two mutations. In the clade containing the A haplotype (SaGr2), there are two haplotypes A and N separated by one mutation, with A being the most common haplotype (96% of individuals). In the clade containing the B haplotype (SaGr1), there are five haplotypes (B, O, K, L, M) separated by 1–2 mutations, with K being the dominant haplotype (90% of individuals) (Fig. 1). Both of these clades are nested separately from the other nine haplotypes in South America. SaGr1 haplotypes (haplotypes B, O, K, L, and M) were found along the northern coastal region of Brazil from S 7° to S 22° (n = 172 individuals) (Fig. 2). Haplotype B was only found close to Salvador, Bahia (S 12.9°). SaGr2 haplotypes (A and N) had a more southern distribution occurring predominantly from S 20° to S 32° (n = 187 individuals) (Fig. 2). Haplotype A was found throughout this southern range. These two groups (SaGr1 and SaGr2) overlap around Rio de Janeiro. The remaining 359 individuals belonged to nine other haplotypes found predominantly in the areas of Curitiba Brazil, northeastern Argentina, and Paraguay. In Florida, 389 individuals were haplotype A (FlA) and 318 were haplotype B (FlB).
Fig. 1

Brazilian peppertree haplotype network. Size of labeled circles indicates relative proportions of each haplotype. Each line connecting the circles represents a single mutation and small unlabeled circles are inferred intermediates. The two haplotypes introduced into Florida are shaded grey and enclosed within boxes representing South American haplotype groups 1 and 2

Fig. 2

Reciprocal predictions and evaluations of niche models generated of invasive Florida Brazilian peppertree (a, b) into South America (upper, a, b) and native haplotype groups into Florida (lower, c, d). The models generated with country buffers (full USA or South America as background extent) are presented as an example. Note that for construction of evaluation scatter plots (b, d, mean AUC ± SD), the value of calibration AUC calculated from extrinsic test partitions was used for each dataset against corresponding projection AUCs. The arrows show the direction of projection. See supplementary Table S1 for AUC values used to construct scatter plots (b, d)

PCA analysis

The result of PCA analysis using all occurrences generated three significant axes, explaining 92% of variability within the occurrence dataset (Table 1). Graphical representation of climate space across the first two PCA axes (representing 79% of variability) clearly demonstrated a significant shift in climatic niche between invasive and native haplotypes of Brazilian peppertree (Fig. 3). The shift in climatic niche primarily occurred along Axis 1, with Axis 2 being informative primarily for SaGr2. The correlation circle of relative contributions of climatic variables to the PCA axes shows that niche shift along Axis 1 was driven primarily by temperature requirements (Table 1, Fig. 3). In contrast, water availability was the principal driver of niche shift along Axis 2 (Table 1, Fig. 3). Between-class analysis of variance using 99 Monte-Carlo randomizations further confirmed the shift in centroids of climatic niche between haplotypes. As indicated by the extensive overlap in climate envelope, there was no difference in realized niche between FlA and FlB (between-class inertia percentage = 0.13, P = 0.08) (Table 2).
Fig. 3

Envelopes describing climatic niches of invasive and native haplotypes of Brazilian peppertree plotted across the first two PCA axes. 1.5 inertia ellipses (=1.5 SD) were drawn around the centroids of climatic envelops occupied by each haplotype group. Convex hulls show prevalence (25, 50, 70 and 100%, respectively) of occurrence points across a climate envelope. The enclosed correlation circle describes the importance of individual bioclimatic variables along the two PCA axes. See Table 1 for description of climatic variables and contribution of individual variables to PCA axes

Table 2

Between-class inertia (or variance) percentages, generated by the between class analysis among invasive and native haplotype groups

















Analysis was performed using 99 Monte-Carlo repeats with α = 0.05

NSNot significant; * P < 0.05

Test of predictive accuracy

Reciprocal predictions of niches generated from invasive and native haplotype occurrences under different model backgrounds also supported a shift in realized niche between Brazil and Florida (Fig. 4a, b summarize the average AUC scores obtained for reciprocal predictions across buffers). Extrinsic tests of prediction accuracy using random test partitions (n = 10, see modeling parameters for methods) demonstrated that for all datasets and for all backgrounds, MaxEnt’s predictions were significantly better than random. As an example, we presented the reciprocal predictions generated under country buffer (entire South America/USA as model background). As illustrated in Fig. 2b, d, for all datasets, reciprocal predictions failed to correctly predict the current distribution of haplotypes in either Florida or Brazil. A similar trend also was observed for all other buffer extents (Fig. 4a, b). Overall, results were particularly poor for SaGr2 and FlA datasets. In contrast, relatively higher AUCs were recorded for cross-predictions with SaGR2 and FlB datasets (Fig. 4a, b). Despite the failure of models trained on South American occurrences to predict the current extent of the Florida infestation, projection of SaGr1 into Florida included the Miami area where haplotype B was first introduced, and the SaGr2 projection into Florida included the Punta Gorda area where haplotype A plants were first introduced (Fig. 2a, c, red stars).
Fig. 4

Comparison of AUC scores of predictions of South American (a) and USA (b) distributions, generated using pairs of Brazilian peppertree occurrences across all buffer extents. The lowest presence threshold was used to convert continuous predictions into binary. AUC was calculated using pseudo-absence points generated from the continental USA and South America for native and exotic habitats, respectively. Note that the results of the county buffer are omitted and is included in Fig. 2 (b, d)

Test of niche overlap

Similar to AUC test results, tests of niche overlap also showed low similarity between reciprocal predictions across all buffers. Figure 5a and b summarize the observed I values for South America and USA predictions, respectively. For all pairs of occurrence datasets, the observed I values were significantly different from the null distributions (P < 0.05, one tailed test) across all buffer extents. The results again demonstrated that significant niche differences exist between invasive and native haplotypes of Brazilian peppertree.
Fig. 5

Comparison of I values observed for predictions of South American (a) and USA (b) distributions generated using pairs of Brazillian peppertree occurrences across all buffer extents. See supplementary Figs. S1–S3 for illustration of buffer extents. Note the buffer Hull.all.Bp denotes the background extent demarcated by the convex hull drawn around all known occurrences of Brazilian peppertree (supplementary Fig. S3). For all buffer extents and pairs of occurrence dataset, the observed I values were significantly different from the null distributions (P < 0.05, one tailed test)

Visualization of predictive niche

Using predictions generated under country buffers as an example, visualization in bivariate climate space also explicitly showed differences in predictive niches generated using native and invasive haplotype groups (Fig. 6). Prediction using FlA occurrences has the largest environmental envelope, followed by that generated from FlB. In contrast, climatic envelopes representing niches calibrated using native occurrences were comparatively restricted, with that of SaGr1 being the smallest (Fig. 6). Graphical representation of niche space again confirmed the prediction of niche suitability at the locations of the first introductions of Brazilian peppertree into Florida (Fig. 6, red stars) by SaGr1 and SaGr2.
Fig. 6

Visualization of climatic niches of native and invasive Brazilian peppertree haplotype groups. The convex hulls represent the climatic envelop predicted for each haplotype group. The symbols (except red stars) represent 10% of randomly selected centroid of grid cells predicted suitable by each occurrence dataset. The red stars represent the climatic positions of the introduction sites in Florida. For visual symmetry, a similar color scheme to that of Fig. 2 was used

Florida herbaria records

In total, 244 records of Brazilian peppertree were found in databases of the three herbaria. Of these, 228 specimens had sufficient label information to determine whether the specimen was collected from a cultivated plant, a plant growing in a ruderal area (roadside, canal bank or other disturbed area) or a plant in a natural area. The first record was in 1929 from a cultivated specimen collected in Orlando, Florida (FLAS accession no. 16171), and all records up to 1951 were from cultivated trees. The first records in natural areas were from the Florida Keys in 1951 (Stock Island, hammock) and 1952 (Big Pine Key). The number of specimens in the three herbaria dramatically increased during the 1960–1979 period, and the majority collected during this period were from ruderal and natural areas (Fig. 7).
Fig. 7

Habitats in which herbaria specimens of Brazilian peppertree were collected in Florida from 1929 to present

Predicted distribution using all native range occurrences

The predicted distribution of Brazilian peppertree using all native range occurrences (Fig. 8) was much more expansive than the current known distribution in the USA, and covered much of the southeastern USA.
Fig. 8

Predicted invasive distribution of Brazilian peppertree using a model constructed with all available native range occurrences of Brazilian peppertree (irrespective of haplotype)


Predicted distributions of Brazilian peppertree in Florida, constructed from point occurrences of the two South American haplotype groups (SaGr1 and SaGr2) that invaded Florida, were much more limited than the current known distribution. The niche model for SaGr1 matched only a narrow strip of land along the far southern Atlantic coast extending from the Keys in the south to approximately West Palm Beach in the north. SaGr1 plants (haplotype B) were initially introduced into Miami (Williams et al. 2005), which falls within the prediction (Fig. 2). The niche model for SaGr2 covered a wider area of southern Florida, extending northward to Lake Okeechobee in the southern center of the state, and further north along the Atlantic and Gulf coasts where temperatures are moderated by the ocean. Punta Gorda, where haplotype A (SaGr2) is thought to have been introduced, is included in the area predicted (Fig. 2). Thus, niche models based on six bioclimatic variables, predicted that both Brazilian peppertree groups could gain a foothold in limited areas of southern Florida. As Brazilian peppertree was initially introduced, propagated and distributed as an ornamental plant, the probability of survival was likely increased by husbandry received in nurseries, and when planted in yards and other managed landscapes. The distributional expansion of Brazilian peppertree since its arrival in Florida has been towards the north, as was reflected in the results of the PCA analysis. The first axis accounted for about 60% of variability and was primarily related to temperature. The bivariate plot of annual temperature and rainfall (Fig. 6) also shows an expansion of Florida populations towards cooler areas than those occupied by the source populations.

There is often a lag period between the time of introduction of an exotic plant and the time that the plant is recognized as invasive. Lag times may simply reflect slow population growth during initial stages of exponential growth, a lack of detection in natural areas, or the time required for the selection of genotypes adapted to novel environments encountered in the introduced range (Mack et al. 2000; Sakai et al. 2001; Crooks 2005). Brazilian peppertree was introduced into Miami and Punta Gorda in the mid- to late 1800s (Barkley 1944; Morton 1978), but was not recognized as escaping from cultivation until the 1950s, suggesting a lag period of 50–100 years (Morton 1978, Fig. 7). This amount of time would be sufficient for the tree to complete 17–33 generations based on seed production after 3 years (Ewel et al. 1982). Although it is possible that non-detection may be partially responsible for the long lag time, it seems unlikely. Brazilian peppertree was not even mentioned in an extensive survey of vegetation of south Florida conducted in the early 1940s (Davis 1943), and our examination of herbaria records clearly suggests that the number of specimens collected outside of cultivation greatly increased after 1960. Slow population growth during early stages of exponential growth must be partially responsible for the delay in colonizing ruderal and natural areas, but is unlikely to provide a complete explanation, as Brazilian peppertree has a short generation time, is a highly prolific seed producer, and the seeds are readily consumed and spread by birds (Ewel et al. 1982, Panetta and McKee 1997).

We believe that the most plausible explanation for the extended lag period is genetic change. The distributions of the two haplotypes introduced into Florida have greatly expanded from their locations of introduction, and are now found sympatrically throughout much of peninsular Florida. Based on nuclear microsatellite DNA analyses, we know that the two parental types originally established have extensively hybridized (Williams et al. 2005, 2007), and a common garden experiment provided evidence of hybrid vigor with hybrids having greater biomass and higher survival than parentals in Florida (Geiger et al. 2011). The numerous genetic combinations produced through hybridization may have greatly facilitated adaptation to novel Florida environments, as has been shown in other systems (for a recent review see Schierenbeck and Ellstrand 2009).

The difference in realized niches of Brazilian peppertree in Florida and South America could be due to a shift in fundamental niche resulting from genetic changes following hybridization, as suggested above, but may also reflect a change in realized niches. Realized niches are typically smaller than fundamental niches due to competition and/or predation (Hutchinson 1957). In South America, distributions of haplotype groups may be governed in part by intraspecific competition between groups, whereby specific groups are best adapted to local conditions in their areas of endemism. In Florida, however, competition may be relaxed, allowing a more complete expression of the fundamental niche (Broennimann and Guisan 2008; Jimenez-Valverde et al. 2011). Reciprocal common garden experiments in Brazil will be needed to better understand the degree of local adaptation and niche boundaries of haplotype groupings of Brazilian peppertree. Herbivory also may influence the Brazilian peppertree distribution in South America, and surveys have identified a diverse insect herbivore community (Bennett et al. 1990; Bennett and Habeck 1991; McKay et al. 2009). Herbivory is presumably much higher in the native range than in Florida, where several insect herbivores have been reported on Brazilian peppertree, but damage is minimal (Cassani 1986; Cassani et al. 1989; Wheeler et al. 2001). Whether herbivory limits the distribution of Brazilian peppertree on a large scale in South America is unknown, but would seem unlikely. Examples of insect herbivory affecting plant distribution are few and limited to exclusion from some habitats within an area rather than exclusion from whole geographic zones (Galen 1990; Louda and Rodman 1996). In general, the influence of competition and predation on broad scale plant distributions is thought to be minimal (Pearson and Dawson 2003; Elith and Leathwick 2009; Rödder and Engler 2011), which is supported by the success of climatic niche models to accurately predict species distributions in many cases (see examples in Franklin 2009, but see Araújo and Luoto 2007 and Austin and Van Niel 2011 for alternate views).

Niche models used to predict invasive distributions of recently introduced species are typically constructed using all the native range occurrence data available (Thuiller et al. 2005; Giovanelli et al. 2008; Medley 2010), particularly when information about the origin of the source population is missing. This may have important implications on the geographic extent of modeled predictions. When we used all native range occurrences of Brazilian peppertree (including the 14 haplotypes which have not been introduced, see Fig. 1), the predicted exotic distribution (Fig. 8) was far more expansive than the current distribution (Fig. 2a), extending west to Texas and north to approximately the northern border of Georgia. The expanded prediction is due to the existence of genetic types of Brazilian peppertree in cooler areas of South America which have not been introduced into Florida. This points to a potential weakness in the use of niche models to predict the distributions of exotic species. Introductions typically consist of only a few individuals (or propagules) originating from a limited area in the native range, and thus may represent only a small fraction of the total genetic diversity of the species (Ward et al. 2008). Thompson et al. (2011) may have been the first to recognize the need to consider subspecific variability in the construction of niche models for invasive plant species. Similar to our study, they found that different sub-specific groups of the tree Acacia salinga occurred in different climatic zones in the native range, but that the native and invasive climatic niches were substantially different. Examination of intraspecific variability may be particularly important for genetically structured species with vast native ranges, such as Brazilian peppertree which extends in South America from S 7.1° to S 31.4°, or approximately 3,700 km. Individual populations may be locally adapted to prevailing climatic conditions in their areas of nativity. In such cases, models constructed using the full extent of a species native distribution may greatly exaggerate the predicted distribution if only a few individuals with limited genetic diversity were introduced. Our prediction based on all available native range data of Brazilian peppertree does provide useful information on the potential distribution, and suggests that the introduction of additional genetic diversity could greatly exacerbate the spatial extent of the invasion.

Niche modeling, combined with an understanding of both the introduction history and the native population genetic structure of invasive species, can enhance our understanding of how invasions become established and spread. Recent niche modeling studies of invasive species have found that niche-based distribution models are often poor predictors of the extent of established invasions (for example, Broennimann et al. 2007; Broennimann and Guisan 2008; Beaumont et al. 2009; Gallagher et al. 2010; Medley 2010; Thompson et al. 2011). The models, however, often predict the area of original introduction (for example, Broennimann et al. 2007; Medley 2010). By confining our modeling to the native source regions of the genetic types that were introduced into Florida, we also were able to detect this pattern in our study. Species may often gain an initial foothold in a new range by being introduced into a similar climate niche. Then, after a period of evolutionary change, they spread into new climatic zones (Kolbe et al. 2007; Medley 2010; Jimenez-Valverde et al. 2011; Thompson et al. 2011). Our study supports this scenario and suggests that multiple introductions from disjunct regions in the native range can facilitate and potentially greatly expand invasions through hybridization of the independent introductions.



Brazilian collections were conducted with the assistance of Dr. M. Vitorino, Universidade Regional de Blumenau, under the Instituto Brasileiro do Meio Ambiente permit 09BR003939/DF. Field assistance was generously provided by F. McKay, USDA/ARS/SABCL, Hurlingham, Argentina. We thank the Florida Department of Agriculture and Consumer Services, the Florida Fish and Wildlife Conservation Commission, South Florida Water Management District, and USDA/ARS for financial support for this work.

Supplementary material

10530_2011_168_MOESM1_ESM.doc (1.2 mb)
Supplementary material 1 (DOC 1185 kb)


  1. Akhisar I, Bener A (2002) Hierarchy analysis of three-way tables. Hacet J Math Stat 31:37–43Google Scholar
  2. Alexander TR, Crook AG (1984) Recent vegetational changes in southern Florida. In: Gleason PJ (ed) Environments of South Florida: present and past II. Miami Geological Soc., Coral Gables, Florida, p 210Google Scholar
  3. Anonymous (2007a) Other news: Brazilian pepper expands its range. Wildland Weeds 10:29Google Scholar
  4. Anonymous (2007b) Panhandlers beware! Wildland Weeds 11: 22Google Scholar
  5. Araújo MB, Luoto M (2007) The importance of biotic interactions for modelling species distributions under climate change. Glob Ecol Biogeogr 16:743–753CrossRefGoogle Scholar
  6. Austin MP, Van Niel KP (2011) Impact of landscape predictors on climate change modelling of species distributions: a case study with Eucalyptus fastigata in southern New South Wales, Australia. J Biogeogr 38:9–19CrossRefGoogle Scholar
  7. Barkley FA (1944) Schinus L. Brittonia 5:160–198CrossRefGoogle Scholar
  8. Barve N, Barve V, Jimenez-Valverde A, Lira-Noriega A, Maher SP, Peterson AT, Soberon J, Villalobos F (2011) The crucial role of the accessible area in ecological niche modeling and species distribution modeling. Ecol Model 222:1810–1819CrossRefGoogle Scholar
  9. Beaumont LJ, Gallagher RV, Thuiller W, Downey PO, Leishman MR, Hughes L (2009) Different climatic envelopes among invasive populations may lead to underestimations of current and future biological invasions. Divers Distrib 15:409–420CrossRefGoogle Scholar
  10. Bennett FD, Habeck DH (1991) Brazilian peppertree: prospects for biological control in Florida. In: Center TD, Doren RF (eds) Proceedings of the symposium of exotic pest plants, pp 23–33, 2–4 November 1988. Miami, FLGoogle Scholar
  11. Bennett FD, Crestana L, Habeck DH, Berti-Filho E (1990) Brazilian peppertree: prospects for biological control. In: Delfosse ES (ed) Proceedings VII. International symposium on biological control of weeds, pp 293–297, 6–11 March 1988, Rome, Italy. Ministero dell’Agriculture e delle Foreste, Rome/CSIRO, Melbourne, AusraliaGoogle Scholar
  12. Broennimann O, Guisan I (2008) Predicting current and future biological invasions: both native and invaded ranges matter. Biol Lett 4:585–589PubMedCrossRefGoogle Scholar
  13. Broennimann O, Treier UA, Muller-Scharer H, Thuiller W, Peterson AT, Guisan A (2007) Evidence of climatic niche shift during biological invasion. Ecol Lett 10:701–709PubMedCrossRefGoogle Scholar
  14. Carhalho PER (1994) Especés Florestais Brasileiras Recomendações Silviculturais, Potencialidades e Uso da Madeira. Embrapa, Colombo, Parana, BrazilGoogle Scholar
  15. Cassani JJ (1986) Arthropods on Brazilian peppertree, Schinus terebinthifolius (Anacardiaceae), in south Florida. Fl Entomol 69:184–196CrossRefGoogle Scholar
  16. Cassani JJ, Maloney DR, Habeck DH, Bennett FD (1989) New insect records on Brazilian peppertree, Schinus terebinthifolius (Anacardiaceae), in south Florida. Fl Entomol 72:714–716CrossRefGoogle Scholar
  17. Clement M, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9:1657–1659PubMedCrossRefGoogle Scholar
  18. Crooks JA (2005) Lag times and exotic species: the ecology and management of biological invasions in slow-motion. Ecoscience 12:316–329CrossRefGoogle Scholar
  19. Cuda JP, Ferriter AP, Manrique V, Medal JC (2006) Florida’s Brazilian peppertree management plan. Recommendations from the Brazilian peppertree task force and Florida exotic pest plant council. Accessed 19 January 2011
  20. Davis JH (1943) The natural features of southern Florida. Fl Geol Sur Bull 25:1–311Google Scholar
  21. Donnelly MJ, Green DM, Walters LJ (2008) Allelopathic effects of fruits of the Brazilian peppertree Schinus terebinthifolius on growth, leaf production and biomass of seedlings of the red mangrove Rhizophora mangle and the black mangrovie Avicennia germinans. J Exp Mar Biol Ecol 357:149–156CrossRefGoogle Scholar
  22. Doren RF, Whiteaker LD, Larosa AM (1991) Evaluation of fire as a management tool for controlling Schinus terebinthifolius as secondary successional growth on abandoned agricultural land. Environ Manage 15:121–129CrossRefGoogle Scholar
  23. Ebeling SK, Welk E, Auge H, Bruelheide H (2008) Predicting the spread of an invasive plant: combining experiments and ecological niche model. Ecography 31:709–719CrossRefGoogle Scholar
  24. Elith J, Leathwick JR (2009) Species distribution models: ecological explanation and prediction across space and time. Annu Rev Ecol Evol Syst 40:677–697CrossRefGoogle Scholar
  25. Elith J, Graham CH, Anderson RP, Dudik M, Ferrier S, Guisan A, Hijmans RJ, Huettmann F, Leathwick JR, Lehmann A, Li J, Lohmann LG, Loiselle BA, Manion G, Moritz C, Nakamura M, Nakazawa Y, Overton JM, Peterson AT, Phillips SJ, Richardson K, Scachetti-Pereira R, Schapire RE, Soberon J, Williams S, Wisz MS, Zimmermann NE (2006) Novel methods improve prediction of species’ distributions from occurrence data. Ecography 29:129–151CrossRefGoogle Scholar
  26. Elith J, Kearny M, Phillips S (2010) The art of modelling range-shifting species. Methods Ecol Evol 1:330–342CrossRefGoogle Scholar
  27. Elith J, Phillips SJ, Hastie T, Dudı′k M, Chee YE, Yates CJ (2011) A statistical explanation of MaxEnt for ecologists. Divers Distrib 17:43–57CrossRefGoogle Scholar
  28. Ellstrand NC, Schierenbeck KA (2000) Hybridization as a stimulus for the evolution of invasiveness in plants? Proc Natl Acad Sci USA 97:7043–7050PubMedCrossRefGoogle Scholar
  29. Ewe SML, Sternberg L (2002) Seasonal water-use by the invasive exotic, Schinus terebinthifolius, in native and disturbed communities. Oecologia 133:441–448CrossRefGoogle Scholar
  30. Ewe SML, Sternberg LSL (2005) Growth and gas exchange responses of Brazilian pepper (Schinus terebinthifolius) and native South Florida species to salinity. Trees Struct Funct 19:119–128CrossRefGoogle Scholar
  31. Ewe SML, Sternberg LSL, Childers DL (2007) Seasonal plant water uptake patterns in the saline southeast Everglades ecotone. Oecologia 152:607–616PubMedCrossRefGoogle Scholar
  32. Ewel JJ (1986) Invasibility: lessons from south Florida. In: Mooney H, Drake J (eds) Ecology of biological invasions of North America and Hawaii. Springer, New York, pp 214–230CrossRefGoogle Scholar
  33. Ewel JJ, Ojima DA, Karl DA, DeBusk WF (1982) Schinus in successional ecosystems of Everglades National Park. South Florida Research Center Report T-676. Everglades National Park, p 141Google Scholar
  34. Fielding AH, Bell JF (1997) A review of methods for the assessment of prediction errors in conservation presence/absence models. Environ Conserv 24:38–49CrossRefGoogle Scholar
  35. Franklin J (2009) Mapping species distributions: spatial inference and prediction. Cambridge University Press, CambridgeGoogle Scholar
  36. Galen C (1990) Limits to the distributions of alpine tundra plants: herbivores and the alpine skypilot, Polemonium viscosum. Oikos 59:355–358CrossRefGoogle Scholar
  37. Gallagher RV, Beaumont LJ, Hughes L, Leishman MR (2010) Evidence for climatic niche and biome shifts between native and novel ranges in plant species introduced to Australia. J Ecol 98:790–799CrossRefGoogle Scholar
  38. Geiger JH, Pratt PD, Wheeler GS, Williams DA (2011) Hybrid vigor for the invasive exotic Brazilian peppertree (Schinus terebinthifolius Raddi., Anacardiaceae) in Florida. Int J Plant Sci 172:655–663CrossRefGoogle Scholar
  39. Gioeli K, Langeland K (2009). Brazilian pepper-tree control. University of Florida, Cooperative Extension Service. Institute of Food and Agricultural Sciences, SS-AGR-17. Accessed 19 January 2011
  40. Giovanelli JGR, Haddad CFB, Alexandrino J (2008) Predicting the potential distribution of the alien invasive American bullfrog (Lithobates catesbeianus) in Brazil. Biol Invasions 10:585–590CrossRefGoogle Scholar
  41. Gogue GJ, Hurst C, Bancroft L (1974) Growth inhibition by Schinus terebinthifolius. HortSci 9:301Google Scholar
  42. Hamilton M (1999) Four primer pairs for the amplification of chloroplast intergenic regions with intraspecific variation. Mol Ecol 8:521–523PubMedGoogle Scholar
  43. Hijmans RJ, Guarino L, Cruz M, Rojas E (2001) Computer tools for spatial analysis of plant genetic resources data: 1. DIVA-GIS. Plant Genetic Resources Newsletter pp 15–19Google Scholar
  44. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate surfaces for global land areas. Int J Climatol 25:1965–1978CrossRefGoogle Scholar
  45. Hinojosa-Díaz IA, Feria Arroyo TP, Engel MS (2009) Potential distribution of orchid bees outside their native range: the cases of Eulaema polychroma (Mocsáry) and Euglossa viridissima Friese in the USA (Hymenoptera: Apidae). Divers Distrib 15:421–428CrossRefGoogle Scholar
  46. Hutchinson GE (1957) Population studies—animal ecology and demography—concluding remarks. Cold Spring Harb Sym 22:415–427CrossRefGoogle Scholar
  47. Jackson ST, Overpeck JT (2000) Response of plant populations and communities to environmental changes of the late quaternary. Paleobiology 26(Suppl):194–220CrossRefGoogle Scholar
  48. JBRJ (2009) Instituto de Pesquisas Jardim Botânico do Rio de Janeiro. Jabot—Banco de Dados da Flora Brasileira. Accessed 1 October 2010
  49. Jimenez-Valverde A, Peterson AT, Soberon J, Overton JM, Aragon P, Lobo JM (2011) Use of niche models in invasive species risk assessments. Biol Invasions 13:2785–2797CrossRefGoogle Scholar
  50. Joyner TA, Lukhnova L, Pazilov Y, Temiralyeva G, Hugh-Jones ME, Aikimbayev A, Blackburn JK (2010) Modeling the potential distribution of Bacillus anthracis under multiple climate change scenarios for Kazakhstan. PLoS One 5: 1–15 e9596Google Scholar
  51. Kolbe JJ, Glor RE, Schettino LR, Lara ADAC, Larson A, Losos JB (2007) Multiple sources, admixture, and genetic variation in introduced Anolis lizard populations. Conserv Biol 21:1612–1625PubMedCrossRefGoogle Scholar
  52. Louda SM, Rodman JE (1996) Insect herbivory as a major factor in the shade distribution of a native crucifer (Cardamine cordifolia A. Gray, bittercress). J Ecol 84:229–237CrossRefGoogle Scholar
  53. Lozier JD, Aniello P, Hickerson MJ (2009) Predicting the distribution of Sasquatch in western North America: anything goes with ecological niche modelling. J Biogeogr 36:1623–1627CrossRefGoogle Scholar
  54. Mack RN, Simberloff D, Mark Lonsdale W, Evans H, Clout M, Bazzaz FA (2000) Biotic invasions: causes, epidemiology, global consequences, and control. Ecol Appl 10:689–710CrossRefGoogle Scholar
  55. McKay F, Oleiro M, Walsh GC, Gandolfo D, Cuda JP, Wheeler GS (2009) Natural enemies of Brazilian peppertree (Schinus terebinthifolius: Anacardiaceae) from Argentina: their possible use for biological control in the USA. Fl Entomol 92:292–303CrossRefGoogle Scholar
  56. Medley KA (2010) Niche shifts during the global invasion of the Asian tiger mosquito, Aedes albopictus Skuse (Culicidae), revealed by reciprocal distribution models. Glob Ecol Biogeogr 19:122–133CrossRefGoogle Scholar
  57. Morgan EC, Overholt WA (2005) Potential allelopathic effects of Brazilian pepper (Schinus terebinthifolius Raddi, Anacardiaceae) aqueous extract on germination and growth of selected Florida native plants. J Torrey Bot Soc 132:11–15CrossRefGoogle Scholar
  58. Morton JF (1978) Brazilian peppertree: its impact on people, animals and the environment. Econ Bot 32:353–359CrossRefGoogle Scholar
  59. Mukherjee A, Christman MC, Overholt WA, Cuda JP (2011) Prioritizing areas in the native range of hygrophila for surveys to collect biological control agents. Biol Control 56:254–262CrossRefGoogle Scholar
  60. Nehrling H (1944) My garden in Florida. American Eagle, EsteroGoogle Scholar
  61. Nilsen ET, Muller WH (1980a) A comparison of the relative naturalizing ability of two Schinus species (Anacardiaceae) in southern California. II: Seedling establishment. Bull Torrey Bot Club 107:232–237CrossRefGoogle Scholar
  62. Nilsen ET, Muller WH (1980b) A comparison of the relative naturalization ability of two Schinus species in southern CaliforniaI. Seed germination. Bull Torrey Bot Club 107:51–56CrossRefGoogle Scholar
  63. Novak SJ, Mack RN (2005) Genetic bottlenecks in alien plant species: influence of mating systems and introduction dynamics. In: Sax DF, Gaines SD, Stachpwicz JJ (eds) Species invasions: insights into ecology, evolution, and biogeography. Sinauer, Sunderland, pp 201–228Google Scholar
  64. NYBG (2009) New York Botanical Garden. Accessed 1 October 2009
  65. Panetta FD, McKee J (1997) Recruitment of the invasive ornamental, Schinus terebinthifolius, is dependent upon frugivores. Aust J Ecol 22:432–438CrossRefGoogle Scholar
  66. Pearman PB, Guisan A, Broennimann O, Randin CF (2008) Niche dynamics in space and time. Trends Ecol Evol 23:149–158PubMedCrossRefGoogle Scholar
  67. Pearson RG, Dawson TP (2003) Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Glob Ecol Biogeogr 12:361–371CrossRefGoogle Scholar
  68. Pearson RG, Raxworthy CJ, Nakamura M, Peterson AT (2007) Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. J Biogeogr 34:102–117CrossRefGoogle Scholar
  69. Peterson AT, Shaw J (2003) Lutzomyia vectors for cutaneous leishmaniasis in Southern Brazil: ecological niche models, predicted geographic distributions, and climate change effects. Int J Parasitol 33:919–931PubMedCrossRefGoogle Scholar
  70. Peterson AT, Papes M, Eaton M (2007) Transferability and model evaluation in ecological niche modeling: a comparison of GARP and Maxent. Ecography 30:550–560Google Scholar
  71. Phillips SJ, Anderson RP, Schapire RE (2006) Maximum entropy modeling of species geographic distributions. Ecol Model 190:231–259CrossRefGoogle Scholar
  72. Prentis PJ, Wilson JRU, Dormontt EE, Richardson DM, Lowe AJ (2008) Adaptive evolution in invasive species. Trends Plant Sci 13:288–294PubMedCrossRefGoogle Scholar
  73. Rödder D, Engler JO (2011) Quantitative metrics of overlaps in Grinnellian niches; advances and possible drawbacks. Glob Ecol Biogeogr 20:915–927CrossRefGoogle Scholar
  74. Roman J, Darling JA (2007) Paradox lost: genetic diversity and the success of aquatic invasions. Trends Ecol Evol 22:454–464PubMedCrossRefGoogle Scholar
  75. Sakai AK, Allendorf FW, Holt JS, Lodge DM, Molofsky J, With KA, Baughman S, Cabin RJ, Cohen JE, Ellstrand NC, McCauley DE, O’Neil P, Parker IM, Thompson JN, Weller SG (2001) The population biology of invasive species. Ann Rev Ecol Syst 32:305–332CrossRefGoogle Scholar
  76. Schierenbeck KA, Ellstrand NC (2009) Hybridization and the evolution of invasiveness in plants and other organisms. Biol Invasions 11:1093–1105CrossRefGoogle Scholar
  77. Schmitz DC, Simberloff D, Hofstetter RH, Haller W, Sutton D (1997) The ecological impact of nonindigenous plants. Island Press, Washington, pp 9–61Google Scholar
  78. Soberon J, Peterson AT (2005) Interpretation of models of fundamental ecological niches and species’ distributional areas. Biodiv Informatics 2:1–10Google Scholar
  79. Soberson J, Nakamura M (2009) Niches and distributional areas: concepts, methods, and assumptions. Proc Natl Acad Sci USA 106:19644–19650CrossRefGoogle Scholar
  80. Suarez AV, Tsutsui ND (2008) The evolutionary consequences of biological invasions. Mol Ecol 17:351–360PubMedCrossRefGoogle Scholar
  81. Templeton AR, Boerwinkle E, Sing CF (1987) A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. I. Basic theory and an analysis of alcohol dehydrogenase activity in Drosophila. Genetics 117:343PubMedGoogle Scholar
  82. Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22:4673PubMedCrossRefGoogle Scholar
  83. Thompson GD, Robertson MP, Webber BL, Richardson DM, Le Roux JJ, Wilson JRU (2011) Predicting the subspecific identity of invasive species using distribution models: Acacia saligna as an example. Divers Distrib 17:1001–1014CrossRefGoogle Scholar
  84. Thuiller W, Richardson D, PYŠEK P, Midgley G, Hughes G, Rouget M (2005) Niche-based modelling as a tool for predicting the risk of alien plant invasions at a global scale. Glob Change Biol 11:2234–2250CrossRefGoogle Scholar
  85. Thuiller W, Broennimann O, Hughesw G, Alkemade JRM, Midgley GF, Corsi F (2006) Vulnerability of African mammals to anthropogenic climate change under conservative land transformation assumptions. Glob Change Biol 12:424–440CrossRefGoogle Scholar
  86. (2009) Missouri botanical garden. Accessed 1 October 2009
  87. VanDerWal J, Shoo LP, Johnson CN, Williams SE (2009) Abundance and the environmental niche: environmental suitability estimated from niche models predicts the upper limit of local abundance. Am Nat 174:282–291PubMedCrossRefGoogle Scholar
  88. Ward SM, Gaskin JF, Wilson LM (2008) Ecological genetics of plant invasion: what do we know? Inv Plant Sci Manage 1:98–109CrossRefGoogle Scholar
  89. Warren DL, Glor RE, Turelli M (2008) Environmental niche equivalency versus conservatism: quantitative approaches to niche evolution. Evolution 62:2868–2883PubMedCrossRefGoogle Scholar
  90. Wheeler GS, Massey L, Endries M (2001) The Brazilian peppertree drupe feeder Megastigmus transvaalensis (Hymenoptera: Torymidae): Florida distribution and impact. Biol Control 22:139–148CrossRefGoogle Scholar
  91. Wiens JJ, Graham CH (2005) Niche conservatism: integration of evolution, ecology and conservation biology. Ann Rev Ecol Syst 36:519–539CrossRefGoogle Scholar
  92. Williams DA, Overholt WA, Cuda JP, Hughes CR (2005) Chloroplast and microsatellite DNA diversities reveal the introduction history of Brazilian peppertree (Schinus terebinthifolius) in Florida. Mol Ecol 14:3643–3656PubMedCrossRefGoogle Scholar
  93. Williams DA, Muchugu E, Overholt WA, Cuda JP (2007) Colonization patterns of the invasive Brazilian peppertree, Schinus terebinthifolius, in Florida. Heredity 98:284–293PubMedCrossRefGoogle Scholar
  94. Wunderlin RP, Hansen BF (2004) Atlas of Florida vascular plants. Accessed 1 October 2010. Institute for Systematic Botany, University of South Florida, Tampa, FL

Copyright information

© Springer Science+Business Media B.V. 2011

Authors and Affiliations

  • A. Mukherjee
    • 1
  • D. A. Williams
    • 2
  • G. S. Wheeler
    • 3
  • J. P. Cuda
    • 1
  • S. Pal
    • 4
  • W. A. Overholt
    • 5
  1. 1.Department of Entomology and NematologyUniversity of FloridaGainesvilleUSA
  2. 2.Department of BiologyTexas Christian UniversityFort WorthUSA
  3. 3.Invasive Plant Research LaboratoryUSDA/ARSFort LauderdaleUSA
  4. 4.Department of StatisticsUniversity of FloridaGainesvilleUSA
  5. 5.Indian River Research and Education CenterUniversity of FloridaFort PierceUSA

Personalised recommendations