Southeastern Naturalist 173 A. Richardson Ahedor and W. Elisens 22001155 SOUTHEASTERN NATURALIST 1V4o(1l.) :1147,3 N–1o9. 61 Morphological Analyses of the Mecardonia acuminata (Plantaginaceae) Species Complex in the Southeastern USA Adjoa Richardson Ahedor1,* and Wayne Elisens2 Abstract - The Mecardonia acuminata complex is a classic example of a widespread endemic species in the southeastern US. Morphological variations in the complex resulted in the classification of at least three varieties or subspecies for the species by previous botanists. However, the distributions and diagnostic features of two of the subspecies, peninsularis and microphylla, are unclear due to shared morphological features and widespread distribution of the subspecies acuminata. The present study involved examination and biostatistical analyses of 3 vegetative and 5 reproductive characters that were known to serve as diagnostic features of 1 or more taxa of the species. Results of the study indicate that subspecies peninsularis can be distinguished by its ascending peduncle angle of suspension, diffuse basal branching habit (dendriform at base) of the shoot and small leaves, especially in the southern ranges of the complex. Subspecies microphylla can be distinguished based on its short (less than 20 mm) fruit peduncles and divaricate peduncle angle of suspension. Subspecies acuminata comprises individuals with divaricate peduncle angle of suspension and long fruit peduncles (>20 mm). Subspecies acuminata was also observed to comprise many individuals that were intermediates of 2 or 3 subspecies. Discriminant function analyses on longitude, latitude, and biogeography as well as Wilks’ lambda estimates suggest moderate to low clinal variation and a much broader historical range distribution of subspecies peninuslaris. The morphological variations in the species complex may be due to secondary contact with ongoing integration into subspecies acuminata. Introduction Mecardonia acuminata (Walter) Small (Axilflower) is a perennial herbaceous species that occurs mostly on the Coastal Plains of the southeastern USA, ranging from Maryland to Florida and eastern Texas, through the Mississippi Embayment to southeastern Missouri (Fig. 1; Brown et al. 2002, Pennell 1935). It is widespread along the southern ranges of its distribution but sparsely distributed on the extreme northern, western, and eastern ranges (Pennell 1922). The species is a wetland indicator commonly found in roadside ditches, on moist sandy or heavy loam soil that is acidic or sub-acidic, or along streams in pineland and deciduous woodland (Pennell 1935). It flowers mainly in the summer followed by the formation of fruits that persist through the fall (Pennell 1935, Rossow 1987), although individuals found in the peninsular Florida flower throughout the year (Pennell 1935).The flowers of M. acuminata can be distinguished from those of other species of Mecardonia by the white corolla with longitudinal purple veins on the posterior side of the throat 1Engineering and Science Division, Rose State College, Midwest City, OK 73110. 2Department of Microbiology and Plant Biology and Oklahoma Biological Survey, University of Oklahoma, Norman, OK 73019. *Corresponding author - aahedor@rose.edu. Manuscript Editor: Joey Shaw Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 174 (Pennell 1935, Rossow 1987, Wunderlin and Hensen 2003). It exhibits some degree of morphological variation; Pennell (1922) proposed the possibility of 3 to 5 subspecies. In a later classification of the species, Pennell (1935) affirmed the classification of the species into 3 main subspecies: acuminata, penisularis, and microphylla (Pennell 1935, Rossow 1987). His classification was based mainly on leaf shape and peduncle lengths. Rossow (1987) reaffirmed the taxonomy of the 3 subspecies based on the length of leaf and peduncle and branching habits of the stem. Wunderlin and Hensen (2003) confirmed the occurrence of all 3 subspecies in Florida; they distinguished the 3 subspecies based on leaf and peduncle length, sepal width, and branching patterns of the shoot. Based on the combined findings of previous botanists, subspecies acuminata is widespread and can be found almost throughout the entire distributional range of the species (Pennell 1935, Rossow 1987). Its leaves are greater than 25 mm (Pennell 1946, Rossow 1987) in length, peduncles range between 25–35 mm long, and lateral branches arise at a distance (mid-point) from the base of the stem (Rossow 1987, Wunderlin and Hensen 2003). Subspecies peninsularis occurs in central to southern Florida (Pennell 1922, 1935; Rossow 1987; Wunderlin and Hensen 2003). Its northern extent overlaps with the southern extent of subspecies acuminata in central Florida (Wunderlin and Hensen 2003). It can be distinguished from the 2 other subspecies based on its small leaves that are less than 25 mm long (Pennell 1935) and its diffuse basal branches (Pennell 1935, Rossow 1987, Wunderlin and Hensen 2003). Its peduncle is similar in length Figure 1. Map showing distribution range of Mecardonia acuminata in the southeastern USA and the 5 biogeographic regions within the range. Shaded region shows entire complex range and range of subspecies acuminata; the dark gray dotted line denotes the boundaries of each biogeographic region. Southeastern Naturalist 175 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 to that of subspecies acuminata. Subspecies microphylla, which is considered rare, occurs from northern Florida through southern Georgia to southeastern Louisiana and occurs in sympatry with subspecies acuminata (Pennell 1935, Rossow 1987, Wunderlin and Hensen 2003). Its leaf size is similar to that of subspecies acuminata, its peduncles are normally less than 12 mm long (Pennell 1935), and its sepals are at least 2 mm wide (Wunderlin and Hensen 2003). Despite the established morphological characteristics of the 3 subspecies, variations of the characters are obvious in each subspecies and therefore pose problems to subspecies identification on the field. Furthermore, due to the sympatric association of subspecies microphylla and acuminata, identification of the former is ambiguous. Similarities between the 2 are evident particularly in the leaf shape, size, and branching pattern (habit), which are supposed to be major delimiting features of subspecies peninsularis and acuminata (Pennell 1922, 1935; Rossow 1987; Wunderlin and Hensen 2003). The objectives of this study were to (1) assess the morphological variations of the species complex and test the current taxonomic circumscription of the 3 subspecies, (2) assess any clinal variation among characters, and (3) determine the geographical range distribution of the 3 subspecies and any biogeographic implications. Field-Site Description Within the range of the species complex in the southeastern US, there are 5 phylogeographic regions that can be hypothesized based on climate and topographic features that could represent physiographic barriers (Soltis et al. 2006). These barriers are the subtropical climate in southern Florida, the Apalachicola and Mississippi rivers, and the high elevation of the Fall Line (Soltis et al. 2006). The regions are (1) Florida Peninsular, (2) Atlantic Coastal Plains, (3) East Gulf Coastal Plains (between Apalachicola and Mississippi rivers), (4) West Gulf Coastal Plain (west of the Mississippi river), and (5) northwestern region of the Fall Line (Fig. 1). We sampled a total of 402 specimens from 13 southeastern states (Appendix 1). Three hundred and thirty were herbarium specimens obtained from the Missouri Botanical Garden (MO), the Botanical Research Institute of Texas (BRIT), the University of Florida (FLAS), the University of Georgia, and the University of Oklahoma (OKL). The remaining 72 were fresh field specimens collected from 7 states: Florida, Georgia, Alabama, Mississippi, Tennessee, Louisiana, and Texas. Methods We evaluated 8 vegetative and reproductive characters: habit, leaf size, leaf shape, floral peduncle length, fructiferous peduncle length, peduncle angle of suspension, fruit length, and sepal width (Table 1). Some of the reproductive characters were not measured for all specimens due to the varying reproductive stages of the plants at the time of sampling. Habit, a vegetative character, was also not recorded for all herbarium specimens since the roots had been removed from some specimens. We recorded length and shape of leaves Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 176 for all specimens and assessed plant habit to be either dendriform at base, dendriform at mid-point/region, or intermediate (dendriform at mid-region but also with reduced basal branching) (Fig. 2). We measured length of leaf from the tip of the leaf to the base for leaves located in the mid-section of the plant (mature leaves). Leaf shape for each plant was determined to be linear, elliptic, ovate, lanceolate, or oblanceolate (Pennell 1935). We made separate measurements of the length of floral and fructiferous peduncles for plants that had both flowers and fruits. We measured sepal width at the broadest mid-portion of the outermost sepal and fruit length from the tip of the fruit to the point of attachment of the fruit to the peduncle. Leaf size, peduncle length, fructiferous peduncle length, fruit length, and sepal width were measured in millimeters. We scored peduncle angle of suspension as 1 = divaricate and 2 = ascending; habits as branching/dendriform at mid-point = 1, at base = 2, and intermediate = 3; and leaf shapes as 1 = oblanceolate, 2 = elliptic, 3 = linear, 4 = ovate, and 5 = lanceolate. Table 1. Variation among quantitative morphological characters in the Mecardonia acuminata complex. n = samples size, S.D. = standard deviation, F = Fischer’s statistics, ANOVA = one-way analyses of variance conducted for each taxon. Fisher’s statistic was used to determine if variances are equal. Letters a and b denote subsets of taxa based on characters as obtained from Duncan’s multiple range test. *** = P ≤ 0.001, ** = P ≤ 0.01, * = P ≤ 0.05 Quantitative character/ Taxon n Mean S.D. F ANOVA Leaf length (mm) M. acuminata 406 23.50 7.05 108.95*** ssp. acuminata 346 25.11b 5.77 4.20*** ssp. peninsularis 48 11.29a 2.13 1.74 ssp. microphylla 16 22.53b 8.02 2.69 Floral peduncle length (mm) M. acuminata 298 17.01 5.40 13.23*** ssp. acuminata 251 17.20b 5.22 2.58** ssp. peninsularis 40 17.13b 5.89 1.16 ssp. microphylla 10 9.0a 3.62 1.49 Fruiting peduncle length (mm) M. acuminata 338 22.73 4.94 42.34*** ssp. acuminata 285 23.17b 5.21 2.66** ssp. peninsularis 41 22.46b 6.12 1.88 ssp. microphylla 16 12.00a 1.50 4.64* Fruit length (mm) M. acuminata 331 6.50 0.84 7.56*** ssp. acuminata 280 6.56b 0.87 0.83 ssp. peninsularis 39 6.0a 0.80 0.88 ssp. microphylla 15 6.13ab 0.83 0.63 Sepal width (mm) M. acuminata 387 1.60 0.36 7.16*** ssp. acuminata 328 1.62a 0.35 2.03** ssp. peninsularis 47 1.42a 0.32 2.50** ssp. microphylla 15 1.61a 0.49) 0.59 Southeastern Naturalist 177 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 The data generated from the measurements and scores were statistically analyzed using SPSS ver. 15.0 for windows (SPSS, Inc. 2006), and graphs were plotted using SigmaPlot ver. 10.0 and Excel (Microsoft Corp. 2003). We performed descriptive statistics for each character and constructed box plots to depict patterns of variation in characters (Pereira et al. 2007). We conducted multiple analyses of variance (MANOVA) to determine differences among group means (Woods et al. 2005, Zar 1996) and calculated Wilk’s lambda values (U-statistics) to test the null hypothesis. F-values (group mean squares/error mean square) with levels of significance (P-values) were estimated from these analyses. We excluded from each analysis samples with missing values and also performed Spearman’s rank-order correlation analysis to determine relationship among pairs of characters. Correlations were tested for significance with two-tailed tests. We performed multiple regression analyses to determine clinal variations for each character (Henderson 2005) and calculated regression coefficients for each character, with longitude and latitude as the independent variables. The curves corresponding to the highest regression coefficient of 5 relationships (linear, logarithmic, exponential, power, and quadratic) were chosen (Schmalzel et al. 2004). Regression lines and coefficients were calculated using the Least-Squares method in Excel (Microsoft Corp. 2003). Since most of the herbarium specimens were not identified to the subspecies level, we performed the multivariate analysis discriminate function (DF) analyses to determine if samples grouped into clusters without a priori assumption of the 3 subspecies (Boonkerd et al. 2002, Woods et al. 2005). Individual specimens served as operational taxonomic units (OTUs), and missing data were excluded from the analyses (Crawford 2003). In these evaluations, we conducted 3 separate DF analyses based on latitude, longitude, and the 5 phylogeographic regions that were predetermined based on physiogeographic barriers in the southeastern US (Soltis Figure 2. Three types of habit corresponding to branching/dendriform patterns observed in Mecardonia acuminata species complex. Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 178 et al. 2006). To assess the distinctness of taxa in these DF analyses, we visually inspected scatter plots for partial or distinct clusters. The data matrix was then partitioned into subspecies based on known morphological characteristics (Pennell 1922, 1935; Rossow 1987; Wunderlin and Hensen 2003). Thus, a priori decision of subspecies was imposed in these subsequent analyses. Univariate analysis of variance (ANOVA) was performed on quantitative characters and chi-square analyses were performed on qualitative characters to determine if a statistically significant difference existed among subspecies. We conducted regression analyses to assess clinal variations in each subspecies and DF analyses to evaluate subspecies delimitations. In order to test the taxonomic significance of branching patterns (habit), we further partitioned the original data matrix into the 3 observed habits: dendriform at mid-point, dendriform at base, and intermediate. We conducted discriminate function analyses to determine if the different subspecies would segregate into distinct or partial clusters based on habit. Diffuse branching at the base (dendriform at base) is believed to be one of the major diagnostic features of subspecies peninsularis (Pennell 1922, 1935; Rossow 1987; Wunderlin and Hensen 2003). Results Overall patterns of morphological variations across subspecies Descriptive statistics. All specimens were collected from 25° to 38° North, and -76° to -96° West. Leaf lengths ranged from 7.00 mm to 45.00 mm with a mean of 23.50 mm. Fruit lengths ranged from 4.00 mm to 9.00 mm with a mean of 6.50 mm. Sepal widths ranged from 1.00 mm to 2.50 mm with a mean of 1.60 mm (Table 1). Floral peduncle lengths ranged from 5.00 mm to 32.00 mm with a mean of 17.01 mm, whereas fructiferous peduncles ranged from 11.00 mm to 38.00 mm with a mean of 22.70 mm (Table 1, Fig. 3). Leaves were mainly oblanceolate (55.3%) in shape, followed by elliptic (41.7%), linear (2.5%), ovate (0.2%) or lanceolate (0.2%). Habits of specimens were observed to be 57.6% dendriform from mid-point, 23.7 % dendriform from base, and 15.6% intermediate (Table 2). Intermediate habit has not been reported in previous studies, but was observed to be prevalent in some populations, especially in Tennessee and northern Alabama. Peduncle angle was divaricated in 65.1% of the specimens and ascending in 32.2% of the specimens. All characters were found to be statistically significant across latitude (P less than 0.01) and longitude (P < 0.05) except for sepal width, which was significant only across longitude at P < 0.1. Estimates of Fisher’s F statistics (F -value) suggest that leaf length (F = 108.95, P < 0.001) and fructiferous peduncle length (F = 42.34, P less than 0.001) (Table 1) were the two most variable quantitative characters in the species, whereas leaf shape (χ2 = 479.1, P< 0.001) and habit (χ2 = 131.14, P < 0.001) were the two most variable qualitative characters. Clinal variation in characters Pearson’s correlations and linear regressions. All characters were correlated with one or more other characters. Eleven of the 28 pairwise character correlations were Southeastern Naturalist 179 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 significantly correlated at P ≤ 0.00, P ≤ 0.01, or P ≤ 0.05 (Appendix 2). The strength of significant associations ranged from |r| = -0.24 (leaf length and peduncle angle) to 0.66 (floral peduncle and fructiferous peduncle). Leaf length and peduncle angle exhibited significant or moderate associations with both latitude and longitude (Fig. 4). Fruit length and sepal width were also correlated with latitude, whereas floral and fructiferous peduncles were correlated with longitude. Moderate to weak clinal Figure 3. Box plots illustrating variations in five morphological characters in the three subspecies. Means denoted by vertical bars (|), shaded boxes indicate 50% of variation ranging from 25th to 75th percentile. Horizontal bars delimit the 10th and 90th percentiles, dots denote outliers. Numbering on Y-axis: 1 = ssp. acuminata, 2 = ssp. peninsularis, and 3 = ssp. microphylla. Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 180 variations were therefore observed for most characters from south to north (latitude) and from east to west (longitude). Leaf length exhibited the strongest clinal variation along latitude (R2 = 0.282, P < 0.001) and longitude (R2 = 0.1, P < 0.001) (Fig. 4). These analyses indicate low but statistically significant associations between leaf length and both latitude and longitude. Multiple analyses of variance (MANOVA). Multivariate ANOVA (MANOVA) for each character along latitude indicates statistically significant results (P ≤ 0.01) for 7 out of the 8 characters measured (Table 3). The results indicate similarities within most characters except for leaf length and peduncle angle. No significant difference was observed for leaf shape along latitude (Wilk’s Lambda = 0.944, P = 0.43). No significant effects of latitude were detected for habit, leaf length, peduncle angle, and sepal width (Table 3). Canonical discriminate function analyses (DFA) and biogeography. The two DF analyses on latitude and longitude did not reveal distinct but partial clusters. The first DF analysis based on latitude revealed partial separation of samples occurring along latitude 26° to 28° in southern Florida, where most subspecies peninsularis occur (Fig. 5). Samples of latitude 29° (northern Florida Peninsular) were intermediate between subspecies peninsularis (pure) and the remaining M. acuminata cluster, but closer to the latter group. This finding indicates that plants in latitude 29° are morphologically similar to subspecies acuminata even Figure 4 (following page). Linear regression analysis of all Mecardonia acuminata specimens showing latitudinal and longitudinal associations of characters. Table 2. Variation among qualitative morphological characters in the Mecardonia acuminata complex (n = sample size). Significance values: * = 0.05, ** = 0.01, *** = 0.001. % observed dendriform branching at Charatcer Taxon n Basal Mid-Point Intermediate χ2 Habit M. acuminata 395 23.7 57.6 15.6 121.4*** ssp. acuminata 340 18.2 69.2 18.8 131.14*** ssp. peninsularis 45 71.1 28.9 0.0 8.02* ssp. microphylla 15 26.7 73.3 0.0 3.27 % observed Character Taxon n Divaricate Ascending χ2 Peduncle angle M. acuminata 405 61.5 32.0 47.184*** ssp. acuminata 339 72.9 28.0 65.490*** ssp. peninsularis 48 25.0 75.0 12.000*** ssp. microphylla 15 80.0 20.0 5.400* % observed Character Taxon n Ovate Elliptic Lanceolate Linear Oblanceolate χ2 Leaf shape M. acuminata 405 0.2 41.7 0.2 0.2 55.3 556.171*** ssp. acuminata 345 0.3 43.2 2.3 2.3 55.9 479.10*** ssp. peninsularis 48 0.0 33.3 0.0 0.0 66.7 5.33* ssp. microphylla 15 0.0 40.0 0.0 13.3 46.7 2.80 Southeastern Naturalist 181 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 182 Figure 5. Two-dimensional scatter plot of canonical discriminate function analysis of morphological characters evaluated in this study versus latitude. Symbols correspond to Mecardonia acuminata entire operational taxonomic units (OTUs) from latitudes. Table 3. Results of multivariate analyses (MANOVA) of Mecardonia acuminata complex showing the extent to which morphological characters differ across latitude/longitude and among biogeographic regions and subspecies. Significance values: * < 0.05, ** < 0.01, *** < 0.001. Degrees of freedom for number of groups per analyses: Latitude = 13, Longitude = 20, Biogeography = 4, Subspecies = 2; degrees of freedom for total number of samples evaluated = 208. Leaf Leaf Peduncle Floral Fruit Fruit Sepal Statistic Habit length shape angle peduncle peduncle length width Wilks’ lambda Latitude 0.875 0.636 0.944 0.741 0.876 0.887 0.888 0.865 Longitude 0.830 0.705 0.913 0.788 0.901 0.889 0.885 0.842 Biogeography 0.905 0.716 0.983 0.873 0.952 0.948 0.984 0.953 Subspecies 0.975 0.668 0.998 0.871 0.903 0.790 0.946 0.974 F-Value Latitude 2.483** 9.919*** 1.021 6.051*** 2.454** 2.215** 2.185** 2.705** Longitude 2.052** 4.185*** 0.947 2.697*** 1.101 1.215 1.302 1.876** Biogeography 5.671*** 21.467*** 0.941 7.824*** 2.714* 2.953* 0.873 2.687** Susbspecies 2.709 53.265*** 0.230 15.832*** 11.431*** 28.494*** 6.129** 2.816 Southeastern Naturalist 183 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 though they have basal branches and small leaves. The second DF analysis based on longitude did not reveal any significant clusters, although samples from longitude -76° to -82°, along the Atlantic Gulf Coast, were slightly separated from samples from the remaining specimens (Fig. 6). Both latitude and longitude DF analyses therefore support the occurrence of subspecies peninsularis in southern Florida, but suggest a much smaller range of distribution than previously reported (Pennell 1922, 1935; Rossow 1987; Wunderlin and Hensen 2003). The DF analyses did not reveal a partial segregation of samples occurring along latitude 30° to 31° or longitude -83° to -90°, where subspecies microphylla is known to occur. This result suggests that subspecies microphylla is morphologically similar to subspecies acuminata and it is embedded in the range of the latter, confirming the sympatric distribution of the two subspecies. Results of DF analysis to test the effect of biogeographic barriers on the distribution of the species showed a consistent and marked partial separation similar to that Figure 6. Two-dimensional scatter plot of canonical discriminate function analysis of morphological characters evaluated in this study versus longitude. Symbols correspond to Mecardonia acuminata entire OTUs from longitudes. Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 184 obtained for latitude (Fig. 7). Region 1 (southern Florida) was partially separated from the remaining 4 regions (2–5) which did not show any apparent biogeographic pattern or separation. The lack of a biogeographic pattern for regions 2–5 is consistent with species that exhibit high levels of outbreeding and physiological tolerance to fluctuating environmental conditions (Fritsch and Lucas 2000) . Tests of subspecies Canonical discriminate function, univariate, and regression analyses. Discriminate function analysis to test subspecies delimitation based on previously reported diagnostic features (Pennell 1922, 1935; Rossow 1987; Wunderlin and Hensen 2003) clearly separated subspecies into 3 distinct clusters (Fig. 8). Seven out of the 8 characters analyzed showed significant variations within each subspecies (Tables 1, 2). Fruit length was less variable in subspecies acuminata and peninsularis (Fig. 3). Subspecies acuminata demonstrated the most variants for all characters, with most outliers within the range of the other 2 subspecies. Discriminate function analyses, conducted to test the reliability of habit in distinguishing subspecies, did not resolve the data into 3 subgroups (Fig. 9). Diffuse branching at the base habit Figure 7. Two-dimensional scatter plot of canonical discriminate function analysis of morphological characters evaluated in this study versus Biogeographic Regions. Group centroids correspond to Mecardonia acuminata specimens obtained from 1 = Florida Peninsular, 2 = Atlantic Coastal Plains, 3 = East Gulf Coastal Plains, 4 = West Gulf Coastal Plains, and 5 = Northwestern Region of the Fall Line. Symbols: ×’s represent OTUs of Florida Peninsular, circles represent OTUs of the Atlantic Coastal Plains, triangles represent OTUs of the East Gulf Coastal Plains, pentagons represents OTUs of the West Gulf Coastal Plains, and squares represent OTUs of the Northwestern Region of the Fa ll Line. Southeastern Naturalist 185 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 (dendriform at base), a character used in identifying subspecies peninsularis was found to be present among 18.2% of subspecies acuminata and 26.7% of subspecies microphylla, mostly occurring north of central Florida (Table 2, Fig. 9). The remaining 6 characters were less variable in each subspecies and therefore were reliable for delimiting subspecies. Duncan’s multiple range test (Table 1) separated leaf length into 2 subsets and indicated a statistically significant difference between subspecies peninsularis and the other 2 subspecies (P = 0.06). Two subsets were also obtained for both floral and fructiferous peduncle lengths. In each of these 2 characters, subspecies microphylla was separated from the other 2 subspecies. Fruit length/size of subspecies peninsularis was smallest with a mean of 6.00 mm, and that of subspecies acuminata was large with a mean of 6.56 mm. The lack of statistically significant subsets for some characters indicates overlap of ranges among subspecies (Fig. 3). Linear regression analyses for each subspecies suggest moderate clinal variations for most characters of subspecies microphylla (Fig. 10). Clinal variations in leaf and peduncle length were also observed for subspecies peninsularis (Fig. 11), but no significant clinal variation was observed for any character of subspecies acuminata (Fig. 12). Figure 8. Two-dimensional scatter plot of canonical discriminate function analysis of morphological characters evaluated in this study versus subspecies. Group centroids correspond to 1 = subspecies acuminata, 2 = subspecies peninsularis, and 3= subspecies microphylla. Symbols: triangles represent OTUs of subspecies acuminata, circles represent OTUs of subspecies peninsularis, and squares represent OTUs of subspecies microphylla. Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 186 Discussion Morphological variations and taxonomic circumscription Subspecies acuminata. The results confirm the widespread distribution of subspecies acuminata, even at the lower latitudes (26°) where subspecies peninsularis is predominant. Subspecies acuminata exhibits a wide range of variation in all characters examined with no significant clinal variations of characters. Fruit length was similar in all 3 subspecies, and subspecies acuminata and microphylla share similar leaf length and sepal width. Subspecies acuminata also shares similar floral and fructiferous peduncle length with peninsularis. Morphological characteristics shared between subspecies acuminata and the other subspecies indicates that some individuals may be either morphological intermediates or hybrid s. Subspecies peninsularis. Evidence from this study revealed that subspecies peninsularis can be identified by small leaf length and sepal width and that it has a Figure 9. Two-dimensional scatter plot of canonical discriminate function analysis of morphological characters evaluated in this study versus habit. Group centroids correspond to Mecardonia acuminata specimens with 1 = branching at mid-point, 2 = branching at base, and 3 = intermediate branching. Symbols represent OTUs separated out based on branching pattern. Figure 10 (following page). Linear regression analyses of subspecies microphylla specimens showing latitudinal and longitudinal associations of characters. Southeastern Naturalist 187 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 188 Figure 11. Linear regression analyses of subspecies peninsularis specimens showing latitudinal and longitudinal associations of characters. Southeastern Naturalist 189 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 Figure 12. Linear regression analyses of subspecies acuminata specimens showing latitudinal and longitudinal associations of characters. Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 190 sympatric distribution with the other subspecies north of Florida. It was identified in the states of Georgia, Tennessee, Maryland, and Texas (Fig. 9, Appendix 1). In northern Georgia, it was found in the same county (Catoosa) as subspecies microphylla (Appendix 1). The diagnostic features found in these areas suggest that subspecies peninsularis may have had a broader distribution range than previously presumed by botanists. The current southern populations of peninsularis may represent relicts that were restricted to a glacial refugium in southern Florida (Pennell 1935). It has been reported that Southern Florida served as a refugium for many plants and animals during the Pleistocene (Soltis et al. 2006). The small leaf lengths/sizes of these populations may be due to climatic or ecological factors and not taxonomy. These climatic or ecological effects are evident in its prolonged flowering season in southern Florida (Pennell 1935) and in the fact that most members of subspecies acuminata and microphylla sampled from southern Florida had leaf lengths ranging from 8.00 mm to 20.00 mm. Subspecies microphylla. It is apparent from our observations that the major diagnostic characters of subspecies microphylla are shorter floral and fructiferous peduncles. Although short floral peduncles were observed in some members of subspecies acuminata, their corresponding fructiferous peduncles were relatively long (>20.00 mm). Because of the many shared characteristics of the two subspecies and their sympatric distribution, identification of subspecies microphylla is problematic. Nevertheless, DF analyses of all 8 characters combined revealed 3 distinct subspecies, thereby supporting the taxonomic circumscription of subspecies microphylla. Figure 13. Map showing distribution of Mecardonia acuminata ssp. peninsularis and M. acuminata ssp. microphylla within the range distribution of the species complex. Dotted line denotes western and northern limits of M. acuminata ssp. acuminata. Southeastern Naturalist 191 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 The current investigations revealed that subspecies microphylla is fragmented but widespread within the range of the complex, particularly in the western portion (Fig. 13). Our results inferred that the subspecies can be found in at least 10 states, as opposed to 5 as earlier reported (Pennell 1935, Rossow 1987). Eastern Texas and western Louisiana, where most microphylla populations were detected, was once a Pleistocene refugium (Remington 1968, Swenson and Howard 2005). Thus, the observed distribution of microphylla in the southeastern US suggests a partial discontinuous distribution separated by the Mississippii River, which is similar to the distribution of Pinus taeda L. (Loblolly Pine; Al-Rabab’ah and Williams 2002, Soltis et al. 2006). Few samples of subspecies microphylla were observed in southern Florida (Charlotte County) where subspecies peninsularis is prevalent (Fig 10). Therefore, subspecies microphylla may not be as rare as originally thought, but may be dispersed within the range of the other 2 subspecies as low-density populations. Remington (1968) identified 4 major and minor suture-zones in the eastern US where movement and fragmentation of species ranges were influenced by Pleistocene glacial retreats and advances. In these suture-zones, species and subspecies of plants and animals came into contact, interacted, and even hybridized. One of the major suture-zones, Northern Florida Suture-Zone, extends westwards along the coast of Alabama and Mississippi in region 1 (Fig. 1). Two minor suture-zones, Louisiana-East Texas Suture-Zone and Southern Appalachian-Ozark Suture-Zone, are located in region 3 and region 5. These suture-zones are concordant with most of the sympatric distribution of subspecies microphylla. Clinal variations in the species complex Clinal variations were not observed for characters examined in subspecies accuminata; however, as latitude increases, leaf length (size) for subspecies peninsularis and microphylla increases. This latitudinal association with leaf length has been observed in some endemic southeastern US taxa such as the Halesia carolina L. (Carolina Silverbell) (Styracaceae) complex (Fritsch and Lucas 2000). The leaf-length latitudinal association in the 2 subspecies implies that when they occur north of their core range, they are not distinguishable from subspecies acuminata due to their larger leaf sizes (except for the short peduncle lengths in subspecies microphylla; Table 1, Fig. 3). Larger leaves are therefore not shared taxonomic characters of subspecies acuminata and microphylla alone, but present in subspecies peninsularis occurring at higher latitudes. Clinal variation was observed in subspecies microphylla and peninsularis particularly across latitude. Many characters tend to increase with latitude, which explains why microphylla is hard to identify in the northern range of subspecies accuminata. With the exception of its short peduncles, all other characters fall within the range of subspecies acuminata. Although subspecies peninsularis is not readily identified in the northern ranges of the species complex, its habit (dendriform at base or diffuse basal branching) and ascending peduncle angle were observed in some members of subspecies acuminata. Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 192 Longitudinal clinal variation was minimal for most characters except for floral and fructiferous peduncle lengths. In subspecies microphylla, there was a significant increase in floral peduncle length across longitude along with a slight increase in the fructiferous peduncle. The opposite was observed for subspecies peninsularis in that its peduncles, particularly floral, decreased across longitude. Thus, at the more western longitudes of its range, subspecies peninsularis’ peduncles were similar to those of subspecies acuminata, which obscures its detection regardless of its relatively small leaves. Species distributed over one or more climatic belts spanning latitudes often possess clines for physiological characteristics and their associated vegetative characters (Stebbins 1950). These clines result from adaptive responses to the environmental conditions prevailing in the different parts of the species range (Fritsch and Lucas 2000, Spurr and Barnes 1980). Thus, the partial separation of peninsularis occurring in southern Florida may be due to the subtropical climatic effect on leaf size and other physiological characteristics that were not evaluated in this study. Biogeography of the species complex The separation of specimens from southern Florida (most of the Florida Peninsular region) from the rest of the specimens (Figs. 5, 7) confirms the climatic, biogeographic, or ecological impact of that region on the morphology of the species. This biogeographic pattern has been observed in a few angiosperms in the southeastern US including Liriodendron tulipifera (Parks et al. 1994, Sewal et al. 1996). Lack of a biogeographic pattern on the distribution of subspecies acuminata and microphylla north of the Florida Peninsular region, indicates sympatry and effective dispersal mechanism(s) of the species irrespective of physiogeographic barriers. No clear geographic patterns have been found in some plant species occurring in that region including Liquidambar styraciflua L. (Sweetgum; Soltis et al. 2006), Prunus (Shaw and Small 2005), and Arabidopsis thaliana (L.) Heynh. (Mouse-ear Cress; Jorgensen and Mauricio 2004). Although the dispersal mechanisms of M. acuminata species complex were not examined in this study, its pollen is known to be dispersed by bees (Ahedor 2007). The small seed sizes of the species (less than 0.50 mm) may be easily dispersed by wind and water (A.R. Ahedor, pers. observ.). Physiogeographic barriers in the southeastern US, such as rivers and high elevations, may not pose barriers to species dispersal. Clustering of peninsularis specimens from latitude 29° in the Central Florida region (Fig. 5) with northern specimens (north of latitude 29°) of subspecies microphylla and acuminata suggest these are probably hybrids of peninsularis and 1 or 2 other subspecies. This region is the Northern Florida Suture-Zone (Remington 1968) where individuals of subspecies peninsularis and 1 or 2 other subspecies may have hybridized. These individuals of subspecies peninsularis exhibit diffuse basal branching or are dendriform at base. It has been documented that extensive hybridization of species and genera may have occurred in this Florida Peninsular refugium during the Pleistocene when many taxa were forced into close proximity (Edwards et al. 2006, Soltis et al. 2006). Southeastern Naturalist 193 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 The current widespread distribution of the species in the southeastern US may be due to secondary contact following the Pleistocene and Quarternary glaciation events in the eastern US. The lack of vegetative (leaf length) clinal variation and overlap of characters such as habit, fruit length, and sepal width suggest secondary contact of previously differentiated entities that are currently mixing freely. Furthermore, the presence in other subspecies of characters reported as diagnostic of subspecies peninsularis, the sympatric distribution of subspecies microphylla, and the identification of a third intermediate habit in the complex suggest hybridization and backcrossing of subspecies. It is likely that microphylla individuals may be backcrossing more into acuminata parentals than peninsularis thereby obscuring their field identification. Interactions of penisularis and acuminata in the Northern Florida Suture-Zone coupled with ecological factors also resulted in fewer identifiable peninsularis in those ranges. Therefore, results of this study depict complex evolutionary processes in the M. acuminata complex that are masked by distinct but inconsistent morphological features. Acknowledgments We thank the following herbaria: Botanical Research Institute of Texas (BRIT), University of Florida (FLAS), University of Georgia (GA), Missouri Botanic Garden (MO), Vanderbilt University (VDB), and University of Oklahoma (OK) for specimen loans. We thank Ms. Amy Buthod for providing assistance with specimen loan requests from herbaria. We thank Brent Berger and Dwayne Estes for assisting with field collection, Cal Lemke for assistance in the Greenhouse, and Susan Pittman for assisting with graphics. Literature Cited Ahedor, A. R. 2007. Systematics of the Mecardonia acuminata (tribe Gratioleae, Plantaginaceae) complex of southeastern USA. Ph.D. Dissertation. University of Oklahoma, Norman, OK. Al-Rabab’ah, M.A., and C.G. Williams. 2002. Population dynamics of Pinus taeda L. based on nuclear microsatellites. Forest Ecology and Management 163:263–271. Boonkerd, T., S. Saengmanee, and B.R. Baum. 2002. The varieties of Bauhinia pottsii G. Don in Thailand (Leguminosae-Caesalpinoideae). Plant Systematics and Evolution 232:51–62. Brown, L.E., K. Hillhouse, B.R. MacRoberts, and M.H. MacRoberts. 2002. The vascular flora of Windham Prairie, Polk County, East Texas. Texas Journal of Science 54(3):227–240. Crawford, P.T. 2003. Biosystematics of North American species of Nuttallanthus (Lamiales). Ph.D. Dissertation. University of Oklahoma, Norman, OK. Edwards, C., D.E. Soltis, and P.S. Soltis. 2006. Molecular phylogeny of Conradina and other related mints (Lamiaceae) from the southeastern USA: Evidence for hybridization in Pleistocene refugia? Systematic Botany 31(1):193–207. Fritsch, P.W., and S.D. Lucas. 2000. Clinal variation in the Halesia Carolina complex (Styracaceae). Systematic Botany 25(2):197–210. Henderson, A. 2005. A multivariate study of Calyptrogyne (Palmae). Systematic Botany 30(1):60–85. Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 194 Jorgensen, S. and R. Mauricio. 2004. Neutral genetic variation among wild North American populations of the weedy plant Arabidopsis thaliana is not geographically structured. Molecular Ecology 13:3403–3414. Microsoft Corp. 2003. Microsoft Excel. Richmond, WA. Parks, C.R., J.F. Wendel, M.M. Sewell, and Y-L. Qui. 1994. The significance of allozyme variation and introgression in the Liriodendron tulipifera complex (Magnoliaceae). American Journal of Botany 81:878–862. Pennell, F.W. 1922. Some overlooked Scrophulariaceae of Rafinesque. Torreya 22:77–84. Pennell, F.W. 1935. The Scrophulariaceae of eastern temperate N. America. Proceedings of the Academy of Natural Sciences of Philadelphia Monograph 1:137–159 . Pereira, M.P., G.E. Perez, and E.S. Balbuena. 2007. European sweet vernal grasses (Anthoxanthum: Poaceae, Pooideae, Aveneae): A morphometric taxonomical approach. Systematic Botany 32(1):43–59. Remington, C.L. 1968. Suture-zones of hybrid interaction between recently joined biotas. Evolutionary Biology 2:321–428. Rossow, R.A. 1987. Revision del genero Mecardonia. Candollea 42:431–474. Schmalzel, R.J., R.T. Nixon, A.L. Best, and J.A. Tress, Jr. 2004. Morphometric variation in Coryphantha robustispina (Cactaceae). Systematic Botany 29(3):553–568. Sewell. M.M., C.R. Parks, and M.W. Chase. 1996. Intraspecific chloroplast DNA variation and biogeography of North American Liriodendron L. Mongoliaceae. Evolution 50:1147–1154. Shaw, J., and R.L. Small. 2005. Chloroplast DNA phylogeny and phylogeography of North American plums (Prunus subgenus Prunus section Prunocersus, Rosaceae). American Journal of Botany 92:2011–2030. Soltis, D.E., A.B. Morris, J.S. McLachlan, P.S. Manos, and P.A. Soltis. 2006. Comparative phylogeofraphy of unglaciated eastern North America. Molecular Ecology 15:4261– 4293. Spurr, S.H., and B.V. Barnes. 1980. Forest Ecology. John Wiley and Sons, New York, NY. SPSS, Inc. 2006. SPSS for Windows, version 14.0. Chicago, IL. Stebbins, G.L., Jr. 1950. Variation and evolution in plants. Columbia University Press, New York, NY. Swenson, N.G., and D.J. Howard. 2005. Clustering of contact zones, hybrid zones, and phylogeographic breaks in North America. American Naturalist 166:581–591. Woods, K., K.W. Hilu, J.H. Wiersema, and T. Borsch. 2005. Pattern of variation and systematics of Nymphaea odorata: I. Evidence from morphological and inter-simple sequence repeats (ISSRs). Systematic Botany 30(3):471–480. Wunderlin, R.P., and B.F. Hansen. 2003. Guide to the vascular plants of Florida. University Press of Florida, Gainesville, FL. Zar, J.H. 1996. Biostatistical Analysis 3rd Edition. Prentice Hall, Upper Saddle River, NJ. Southeastern Naturalist 195 A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 Appendix 1. Table showing states and counties where subspecies peninsularis and microphylla were identified; * denotes counties where both subspecies were i dentified. Subspecies peninsularis Subspecies microphylla State County State County Florida Broward Florida Calhoun Charlotte* Charlotte* Citrus Jackson Dade Washington Desoto Alabama Geneva Dixie Lee Gilchrist St. Clair Hardee Mississippi George Hendry Lamar Hernando Evangeline Lousiana Beauregard Grant Highlands Texas Newton Indian River Orange Lake Polk Lee Red River Levy San Jacinto Manatee Titus Martin Tyler Marion Virginia King William Okeechobe North Carolina Catham Palm Beach Georgia Charlton Pasco Catoosa* Pinellas Dekalb Polk Bradley Seminole Calhoun St. Lucie Lafayette Arkansas Drew Oklahoma McCurtain Taylor Volusia Georgia Catoosa* Alabama Wilcox (Maryland) D.C. Tennessee Rutherford Texas Ushur Southeastern Naturalist A. Richardson Ahedor and W. Elisens 2015 Vol. 14, No. 1 196 Appendix 2. Correlation among morphological characters to latitude and longitude in the Mecardonia acuminata complex based on Pearson’s rank-order correlation analysis. Significance values: * = 0.05, ** = 0.01, *** = 0.001 Leaf Leaf Peduncle Floral Fruit Fruit Sepal Character Habit length shape angle peduncle length penduncle width Latitude Longitude Habit 1.000 -0.087 0.122* 0.154** -0.062 -0.030 0.065 0.050 -0.001 0.090 Leaf length 1.000 -0.046 -0.240** -0.010 0.099 0.333** 0.248** 0.531*** 0.307** Leaf shape 1.000 0.001 0.076 0.022 -0.154** -0.002 0.044 0.037 Peduncle angle 1.000 116.000* 0.069 -0.075 -0.013 -0.219** -0.342** Floral pedencle 1.000 0.660*** -0.041 -0.132* -0.049 -0.148* Fruit peduncle 1.000 0.153** 0.018 0.003 -0.186** Fruit length 1.000 0.259** 0.151** 0.006 Sepal width 1.000 0.196** 0.064 Latitude 1.000 0.249** Longitude 1.000