AOBPreview originally published online on August 15, 2005
Annals of Botany 2005 96(5):887-900; doi:10.1093/aob/mci241
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A Morphological Study of the Petunia integrifolia Complex (Solanaceae)
1 Faculty of Horticulture, Chiba University, 648 Matsudo, Matsudo City, Chiba 271-8510, Japan, 2 Center of Environment, Health and Field Science, Chiba University, 6-2-1 Kashiwanoha, Kashiwa City, Chiba 277-0882, Japan, 3 Graduate School of Science and Technology, Chiba University, 1-33 Yayoi-cho, Inage-ku, Chiba City, Chiba 263-8522, Japan, 4 Centro de Pesquisas de História Natural, 618 Rua Jaime Ribeiro Wright, Itaquera, São Paulo 08260-070, Brazil, 5 Facultad de Agronomia, Universidad de la República, Garzón 780, Montevideo, Uruguay and 6 Centro de Investigation de Recursos Naturales, INTA, Las Cabañas y Reseros s/n (1712), Castelar, Prov. de Buenos Aires, Argentina
* For correspondence. E-mail andot{at}faculty.chiba-u.jp
Received: 23 January 2005 Returned for revision: 4 April 2005 Accepted: 20 June 2005 Published electronically: 15 August 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGEMENTSLITERATURE CITED |
|---|
• Background and Aims Petunia inflata has been treated taxonomically in various ways: it has been described as an independent species, treated as a synonym of P. integrifolia, and also regarded as a subspecies of P. integrifolia. The present study was designed to resolve the ambiguity involving the P. integrifolia complex (P. integrifolia plus P. inflata).
• Methods Tentative identification (either integrifolia group or inflata group) was carried out in the field based on the observation of live specimens at the restricted type localities. The accuracy of the tentative identification was later tested with principal component and cluster analyses of data obtained by measuring 21 morphological characters on cultivated live specimens sourced from 113 natural populations of the P. integrifolia complex in Argentina, Brazil, Paraguay and Uruguay.
• Key Results There was a clear, statistically significant gap between the morphological measurements of the two groups, ensuring the accuracy of identification carried out in the field except for a probable hybrid swarm. Previously, the condition of the pedicel in the fruiting state was considered an important character distinguishing between these two groups; however, the condition of the pedicel was rather variable in the integrifolia group. The two groups were found to have geographically distinct distributions: the integrifolia group occurred in southern regions, whereas the inflata group occurred in northern regions.
• Conclusions Based on the available evidence, it is suggested that the two groups are allopatric species, P. integrifolia and P. inflata, in agreement with the opinion of Fries (1911).
Key words: Distribution, floral morphology, Petunia, Serra Geral, Solanaceae, South America, speciation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGEMENTSLITERATURE CITED |
|---|
The first garden petunia (Petunia x hybrida Vilm., Solanaceae) was reported to be an interspecific hybrid of P. axillaris (Lam.) Britton, Sterns & Poggenb. (=P. nyctaginiflora Juss.) and P. integrifolia (Hook.) Schinz & Thell. (=P. violacea Lindl.) (Paxton, 1836
). Since then, Petunia inflata R. E. Fr. and P. parodii Steere [=P. axillaris subsp. parodii (Steere) Cabrera] have also been proposed as the likely parents of modern garden petunias (Sink, 1984). Within these proposed ancestors, the taxonomic position of P. inflata has not yet been clearly defined. Fries (1911) originally described P. inflata as an independent species based on a comparison of morphological characters with P. integrifolia. Fries (1911) noted that one of the main features distinguishing these two taxa is the pedicel condition in the fruiting state. In P. integrifolia, the pedicels are deflexed, whereas in P. inflata they are inflexed. Subsequently, however, Smith and Downs (1966) regarded P. inflata as a synonym of P. integrifolia without providing any clear reason. Wijsman (1982) later studied specimens of P. integrifolia, P. inflata and P. occidentalis R. E. Fr. in several European herbaria and reported inter-gradation in pedicel condition and floral size between the three species over a geographical course, moving west from southern Brazil (P. integrifolia) through Paraguay and north-eastern Argentina (P. inflata) to north-western Argentina (P. occidentalis). In that study, the deflexus characters were reported to change to inflexus, and to coincide with a reduction in flower size along the geographical cline. Wijsman (1982) provided several subspecies names, such as subsp. integrifolia, subsp. inflata and subsp. occidentalis, for the ‘extremes’ of P. integrifolia. However, a previous study comparing morphology among his infraspecific taxa of P. integrifolia, as above, failed to confirm the inter-gradation reported by Wijsman (1982); instead, distinct morphological differences were found between the taxa concerned (Ando et al., 1995). Based on these studies and other evidence obtained from a cross-compatibility study, P. occidentalis was resurrected as an independent species (Tsukamoto et al., 1998). At that time, however, there was a hesitancy to come to a conclusion regarding the taxonomic positions of subsp. integrifolia and subsp. inflata because earlier experiments were conducted on a small scale. Accordingly, Wijsman's P. integrifolia subsp. integrifolia and subsp. inflata, which the authors describe as the ‘P. integrifolia complex’, were left for subsequent detailed studies.
Since Wijsman's (1982) work, Ando and Hashimoto (1993, 1994, 1995, 1996, 1998) described several new species of Petunia from uplands of southern Brazil based on the observation of native live plants. All the new species are unique in floral and other morphology and are readily distinguishable from previously known species including ‘P. integrifolia complex’. Fries (1911) recognized a subspecies of P. integrifolia (subsp. depauperata), which occurred along the Atlantic coast of Brazil. Smith and Downs (1966) treated this taxon as a variety of P. integrifolia, while Wijsman (1982) regarded this as a variant growing in nutrient-poor soil. However, this plant produces extremely long prostrated stems that bear sparse small flowers and linear leaves, and such features in morphology are never found in any taxa of Petunia with the exception of P. littoralis L. B. Sm. & Downs. A comparison of the morphology and the distribution range of this taxon with those of P. integrifolia will be reported elsewhere.
The objective of the present study was to evaluate Wijsman's Petunia integrifolia subsp. integrifolia and subsp. inflata using multivariate analyses of morphological characters. The first author has travelled >140 000 km in temperate and subtropical South America, and observed 448 natural populations of the P. integrifolia complex over a substantial part of the distribution range. In addition to being a candidate for the ancestor of garden petunias, as mentioned above, P. inflata is also an important experimental model system for studies of solanaceous self-incompatibility (Dowd et al., 2000). Therefore, there is a clear need to resolve some of the taxonomic ambiguity involving this taxon.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Observation at the restricted type locality
Hooker (1831)
described Petunia integrifolia (=Salpiglossis integrifolia) based on two specimens. The source of one of the specimens was uncertain, while J. Baird collected the other near the Río Negro, Uruguay. In order to observe live specimens of P. integrifolia at this location, the first author visited the Río Negro, Uruguay (on 28 Nov. 1989), and adjacent regions (Ando, 2003, 2004). For Petunia inflata, Fries (1911) established several syntypes. The first author also visited a number of locations in Argentina, including Bonpland in Misiones Province (on 27 Nov. 1990), one of the restricted type localities, and adjacent regions (Ando, 2003, 2004).
Plant materials
Seeds and herbarium specimens were collected between 1988 and 1999. Based on observations of live plants at the above locations, tentative identification of the material as either P. integrifolia or P. inflata was carried out for the native populations. Seeds of P. integrifolia (57 populations) and P. inflata (56) were collected over a substantial distribution range of the P. integrifolia complex. Additional samples were collected from Rio Grande do Sul (RS) in Brazil, since the distributions of the two taxa are very similar in that region.
The following notation was used for the accession codes representing the source of the collected samples: ‘A’, Argentina, ‘B’, Brazil, ‘P’, Paraguay, and ‘U’, Uruguay (see the Appendix). The location of each population was recorded using a global positioning system (GPS) receiver [Sony IPS/360 (Sony Co., Ltd, Shinagawa, Tokyo, Japan), Panasonic KX-G5550 (Matsushita Electric Industrial Co., Ltd., Kadoma, Osaka, Japan) or GPS III Plus (Garmin, Olathe, KS, USA)].
For all but 12 of the populations [B272, B522, B692, B724 (nine individuals), B688, B841, U128 (8), B809, U12 (7), B1206, U168 (6) and A166 (5)], ten individuals were raised from seed in a greenhouse, following the standard procedure for cultivating garden petunias. The plants were grown in 15-cm (diameter) pots, and the lateral stems were held vertical using plastic supports.
Measuring live plants
In a previous study, 33 morphological characters encompassing reproductive and other traits (X1–X33) were compared in each of the infraspecific taxa of Petunia integrifolia sensu Wijsman (1982) (Ando et al., 1995). For the present study, 21 of the morphological characters that had been found at that time to be useful for distinguishing subsp. integrifolia and subsp. inflata were chosen. The code numbers used for the characters follow those of Ando et al. (1995) (Table 1).
|
The morphological features were measured on all of the greenhouse-grown specimens between May and July 2001. One typical flower was chosen per plant and a mean value was calculated for each population. Characters were measured using digital calipers (Mitutoyo Co., Kawasaki, Kanagawa, Japan) when the plants reached full bloom. To measure the pedicel angle in the fruiting state and other morphological characteristics related to the capsule, selected flowers for each plant were crossed with pollen from a different individual belonging to the same population. The angle of the pedicel was measured when the capsule reached maturity.
Measuring herbarium specimens
Pedicel angles were also measured on voucher herbarium specimens collected from each of the native populations. Up to five pedicels with mature capsules were selected for the measurement. A capsule was considered mature when it dehisced or reached full size. The ratio of mature capsules with folded calyx lobes to those with straight (not folded) ones was also recorded for the voucher specimens. The first author visited four European herbaria [Herbarium, Botany Department, Natural History Museum, London (BM); Herbarium, Royal Botanic Gardens, Kew (K); Rijksherbarium, Leiden (L); and Herbarium, Institute of Systematic Botany, State of Utrecht, Utrecht (U)] to re-inspect the specimens that Wijsman (1982) studied.
Multivariate analysis
Fries (1911) studied herbarium specimens and regarded the pedicel condition in the fruiting stage as a constantly reliable key morphological character distinguishing P. inflata (inflexed) from P. integrifolia (deflexed). Therefore, character X33 (angle of pedicel to stem during fruiting) was not included in the multivariate analysis, in an attempt to find other characters differentiating the taxa.
The resulting data matrix contained 20 variables (means of the measurements for 20 morphological characters) and 113 cases (populations). This dataset was used for a principal component (PC) analysis to separate populations based on measured morphological characters, and to identify the morphological characters that differed between the groups. The PC analysis was carried out using the FACTOR procedure in the SPSS statistical package (SPSS Inc., Chicago, IL, USA).
Cluster analysis was carried out using a similar data matrix, except that it lacked the B812 population (discussed below). The PROXIMITY and CLUSTER procedures in SPSS were used to calculate a distance matrix and an amalgamation schedule, respectively. The distance matrix used squared Euclidian distances, and the amalgamation method used the un-weighted pair-group method on the arithmetic mean (UPGMA). The FACTOR procedure was also used on this data to obtain correlation coefficient (CC) matrices among variables for the inflata and integrifolia groups separately. The T-TEST procedure in SPSS was used to compare means between the two groups and between two subclusters of the integrifolia group.
Distribution map
A distribution map was produced using TNTlite (MicroImages Inc., Lincoln, NB, USA). Political boundaries and rivers were taken from the Digital Chart of the World (ESRI Inc., Redlands, CA, USA). The Rio Icamaquã and Rio Vacacaí, which were added to the map manually, were sourced from a 1:250 000 topographical map (Instituto Brasileiro de Geografia e Estatística, Rio de Janeiro, Rio de Janeiro State, Brazil). Contour intervals (200 m) were generated from GTOPO30 DEM data (USGS, Reston, VA, USA). Collection localities for each of the populations were overlaid as a database pinmap, using GPS co-ordinates taken during the field survey.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGEMENTSLITERATURE CITED |
|---|
PC and cluster analyses
When the PC analysis was performed on the data matrix consisting of 20 variables (morphological characters other than X33) and 113 cases (populations), four PCs with eigenvalues larger than 1·0 were obtained. These four groups explained 78·8 % of the total variance in the data matrix (Table 1). A two-dimensional scatter diagram for the first and second PC scores for the respective populations showed that two groups occupied separate regions, and one population (square, B812 population) was intermediate (Fig. 1).
|
The restricted type localities of P. integrifolia (U106, U168 and U239) were included in a group that was distributed mainly in the first and fourth quadrants of Fig. 1 (open and closed circles, 56 populations). The plants grown from all populations of this group were identified in the field as belonging to P. integrifolia. Hereafter, plants from these populations are referred to as the integrifolia group. The restricted type localities of P. inflata (A7, A119 and P1) belonged to another group distributed in the second and third quadrants (triangles, 56 populations). Specimens from all populations of this group were identified in the field as belonging to P. inflata. Hereafter, these populations are referred to as the inflata group.
Plants from the B812 population were identified in the field as P. integrifolia. When data obtained from ten individuals from the B812 population were added to the original data matrix and subjected to PC analysis, two and three individuals were distributed in the integrifolia and inflata groups in the scatter diagram, respectively. The remaining five individuals fell into the region intermediate between the two major groups (data not shown). In an attempt to clarify the morphological differences between the inflata and integrifolia groups, the B812 population was removed from subsequent analyses because it was considered a hybrid swarm.
The dendrogram produced from the modified (no B812 population) data matrix showed two distinct clusters (Fig. 2). The upper cluster of the dendrogram consisted of members of the inflata group and the lower cluster was made up of members of the integrifolia group.
|
Differences in the morphological characters
As assessed by the larger absolute values of the factor loading, characters related to the stamens (X5, X6 and their proportion X7), pistil (X8), stigma (X9 and X10), corolla (X21, X24 and X25), and X19 (the ratio of X6 : X16) were selected as the morphological characters that were most differentiated between members of the integrifolia and inflata groups (Table 1, shown in bold face). The mean values obtained for the morphological characters of the inflata group were significantly smaller than those of the integrifolia group, with the exception of character X4. This result was mostly confirmed with a t-test at the 0·1 % level of significance (Table 1), and was in agreement with a previous study conducted on a smaller scale (Ando et al., 1995
). The ranges of measurements of the inflata group for three characters (length of long stamens X5 < 16·42 mm, length of pistil X8 < 14·1 mm, and diameter of the corolla tube X24 < 8·5 mm) were clearly separate from those of the integrifolia group (X5 > 16·44 mm, X8 > 14·5 mm, and X24 > 9·0 mm). Considering the 95 % range (±1·96 x standard deviation), two more characters were also found to be useful for separating members of the inflata group (length of basal part of filament affixed to corolla tube X6 < 5·23 mm, and length along vertical axis of stigma X9 < 1·12 mm) from members of the integrifolia group (X6 > 5·78 mm, and X9 > 1·39 mm) (Fig. 3).
|
Correlation between flower size and length of the inner floral organs
Table 2 shows the correlation coefficients (CCs) between two characters related to flower size (X21, length of corolla tube; and X25, length of total corolla) and four characters related to the size of the inner floral organs (X5, length of long stamen; X6, length of basal part of filament affixed to corolla tube; X8, length of pistil; and X16, length of ovary). In the integrifolia group, the size of the inner floral organs (X5, X6, X8 and X16) resulted in higher CCs (1·0 % significant level) compared with characters related to flower size (X21 and X25). For the inflata group, character X6 did not correlate with characters X21 or X25, and character X8 did not correlate with character X25.
|
Identification in the field
Although a subset of morphological characters of the inflata group differed from those of the integrifolia group, as described above, field identification was based mainly on other morphological characters. Slight differences in the colour of the flower were one of the characters used for identification. The flowers of both taxa have been described as reddish-purple, but in the authors' experience, the tint of the inflata group is slightly brighter than that of the integrifolia group. This difference is evident from a distance of several metres, especially on cloudy days.
A significant difference in the diameter of the corolla tube (X24) (Table 1) was also apparent on visual inspection of the flower. The mouth of the corolla was larger in the integrifolia group than in the inflata group (see fig. 5 in Ando et al., 1995).
|
The colour and venation pattern on the inner-surface of the corolla tube was also a character that could reliably distinguish the inflata group from the integrifolia group. The colour of the inside of the corolla tube was much paler than the corolla limb in the inflata group, while in the integrifolia group the corolla tube was darker than the corolla limb. In the inflata group, darker coloured veins were restricted to a small area on the upper side of the inner corolla tube (Fig. 3B). The pattern of venation expanded widely inside the corolla tube in the integrifolia group, although this was not clearly visible, due to the darker ground colour (Fig. 3A). Using these characters, it was possible to distinguish individuals belonging to the inflata and integrifolia groups in the field.
Pedicel condition
Character X33 (angle of the pedicel relative to the stem during fruiting) was measured on plants cultivated from seed. For the inflata group, angles between 53 and 91° (mean 74°) were obtained. For the integrifolia group, the angle varied between 45 and 218° (mean 131°) (Fig. 4A). Among the herbarium specimens belonging to the inflata group, the range was much smaller (31–77°; mean 53°) (Fig. 4B). The pedicel angle of the integrifolia group recorded from herbarium specimens showed a narrower range (48–150°; mean 108°) compared with cultivated plants. The typical features of the pedicel characters are shown in Fig. 5A (integrifolia group) and B (inflata group).
|
Another character observable in herbarium specimens
The calyx lobes located next to the maturing capsules of specimens belonging to the integrifolia group were patent and out-curved, whereas those of the inflata group were closed and straight. This subtle difference changed the appearance of the calyx lobes, and was easily observed in the herbarium specimens (Fig. 5A and B). In most cases, at least one of the five calyx lobes of the integrifolia group was folded (see arrows, Fig. 5A). By contrast, all the calyx lobes of the inflata group were usually straight (Fig. 5B). The frequency distribution of the mean values is shown in Fig. 6.
|
Geographic distribution
Figure 7 shows a topographical map (200-m intervals) of the study region, indicating the location of populations of the integrifolia (open and solid circles) and inflata (triangles) groups, and population B812 (solid square). The distribution range of the inflata group clearly differed from that of the integrifolia group. The inflata group was located in northern areas, whereas the integrifolia group was located in southern areas. In Brazil, the inflata group was found in upland areas in northern RS and Santa Catarina (SC) states, whereas the integrifolia group was found in lowland areas, such as along the Rio Ibicuí flowing west, and the Rio Jacuí and its upper stream (the Rio Vacacaí), which both flow east (Fig. 7).
|
Outside Brazil, the inflata group was the sole taxa found in the Chaco, Corrientes, Formosa and Misiones Provinces of Argentina, and Paraguay while the integrifolia group was found exclusively in the Entre Ríos Province of Argentina, and Uruguay. Compared with Brazil, the range of the inflata group extended much further north in Paraguay and Argentina, up to 25°S (Fig. 7). In Corrientes Province, the range of the inflata group was restricted to the northern region, close to Paraguay. The range of the integrifolia group was concentrated along major rivers, such as the Río Uruguay and Río Negro (Fig. 7).
Sub-clusters of the integrifolia group
The dendrogram (Fig. 2) shows that amalgamation distances were smaller in the inflata group than in the integrifolia group. In the integrifolia cluster, a minor sub-cluster composed of five populations (B835, B841, B870, B949 and B971) was recognized as an outlier (Fig. 2, solid circles). Compared with the major sub-cluster, members of the minor sub-cluster were characterized by significantly larger flowers (X21 and X25), internal floral organs (X5, X6, X8, X9 and X10) and capsules (X30 and X31). This trend to larger organs was also evident for some other characters, such as X7 and X19 (Table 3). In addition, members of the minor sub-cluster had pedicel angles during fruiting (X34) of around 90° (Table 3).
|
Although members of the major sub-cluster of the integrifolia group always occurred in lowland habitats, this was not necessarily the case for members of the minor sub-cluster (Fig. 7, solid circles). Populations belonging to the minor sub-cluster were often located on steep sloping areas and ascending uplands (B870 and B949) and adjacent lowland (B835 and B971) in central RS and upland areas in south-western RS (B841) (Fig. 7).
Specimens inspected by Wijsman (1982)
The first author re-inspected two of the three specimens of the P. integrifolia complex that exhibited a pedicel condition described by Wijsman (1982) as ‘deflexed, but some aberrant’. The first specimen was found to be neither a member of the integrifolia nor inflata groups, but was a recently described species, Petunia altiplana T. Ando & Hashim. (Lindeman & de Haas 3621 deposited in U). The second specimen was not a species of Petunia, but was found to be a member of a sister genus, Calibrachoa linoides Sendtn. (Lindeman & de Haas 2649a in U). The third specimen, reported to have been collected from São Joaquim in SC, Brazil (Reitz s. n. in U) could not be found.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Identification of live plants
Although the taxonomic treatment of the integrifolia and inflata groups is complicated as mentioned in the Introduction, it was possible to distinguish the two groups in their natural habitats. The accuracy of the authors' tentative identification, based mainly on overall flower shape, colour and venation patterns, was validated by the clear statistical separation of two groups using PC and cluster analyses (Figs 1 and 2), with the exception of a probable hybrid swarm (B812 population).
Aside from the character of pedicel condition in the fruiting state, which was not included in the multivariate analyses (Figs 1 and 2), several morphological measurements distinguished members of the inflata and integrifolia groups. The inflata group was characterized by significantly shorter (or smaller) inner floral organs, such as stamens (X5), basal part of the filament fixed to the corolla tube (X6), pistil (X8), ovary (X16), calyx lobe (X12), calyx tube (X14) and stigma (X9 and X10) (Table 1). The ranges of three characters (X5, X8, and X24) for the inflata group did not overlap those of the integrifolia group.
In the integrifolia group, the size of the inner floral organs (X5, X6, X8 and X16) produced higher CCs with flower size [as assessed by the length of the corolla (X25) or corolla tube (X21)]. In the inflata group, the length of the basal part of the filament affixed to the corolla-tube (X6) and the length of pistil (X8) did not exhibit these features (Table 2), suggesting that characters X6 and X8 are stable in the inflata group irrespective of the flower size, which can vary between populations in different local environmental conditions. It is also noteworthy that two secondary characters, X7 [proportion of X6 to the length of long stamen (X5)] and X19 [ratio of X6 : the length of ovary (X16)], which are not directly related to flower size, also differed significantly between the inflata and integrifolia groups (Table 1).
The difference in character X19 showed that the basal part of the filament affixed to the corolla tube (X6) is 1·98 (approx. 2) and 2·93 (approx. 3) times longer than the length of the ovary (X16) in the inflata and integrifolia groups, respectively (Table 1 and Fig. 3). From experience, it is suggested that character X19 is one of the most convenient floral characters for distinguishing the inflata and integrifolia groups. Unfortunately, many of the characters that differentiate the two groups are fragile and are not conserved in herbarium specimens.
Identification in the herbaria
In previous studies, Fries (1911) and Wijsman (1982) highlighted the pedicel condition in the fruiting state as the primary feature distinguishing the inflata and integrifolia groups. By contrast, it was found that this character (X33) is variable in the integrifolia group, as indicated by the large coefficient of variation (CV) value (Table 1). In fact, characters related to pedicel condition measured from cultivated members of the integrifolia group (Fig. 4A) differed markedly from those measured on voucher herbarium specimens (Fig. 4B). Lateral stems of both subspecies are usually ascending under natural conditions, but in the present study, lateral stems were maintained in a vertical position with the aid of plastic supports. Perhaps the manipulation of the lateral stems caused this discrepancy.
For herbarium specimens, the pedicel angle for the inflata group fell into a rather narrow range of <80° (Fig. 4B). Therefore, it is reasonable to conclude that the inflata group has inflexed pedicels at the fruiting stage. The pedicel angle for integrifolia was more variable and did not always exceed 90° (Fig. 4B). As a result, these cannot be considered as reliable taxonomic features. A member of the Petunia integrifolia complex can be identified as belonging to the integrifolia group when it has deflexed pedicels, since this characteristic does not occur in the inflata group (Fig. 4B). If a specimen has a slightly inflexed pedicel, however, other characters must be considered carefully before a clear identification can be made.
In this study, another morphological character was found that is readily observable in herbarium specimens and can be used to distinguish the two groups. For the integrifolia group at least one of the five calyx lobes is usually folded (Fig. 5A), whereas in the inflata group, all the calyx lobes are straight (Fig. 5B). It is worth noting that this character is stable in the integrifolia group, but more variable in the inflata group (Fig. 6), contrary to pedicel condition, as mentioned above (Fig. 4). A herbarium specimen of the Petunia integrifolia complex can be identified as belonging to the inflata group when most of the capsules have straight calyx lobes without folding, since this feature is rarely observed in the integrifolia group (Fig. 6).
The identification of the P. integrifolia complex, as described above, can be applied to mature plants; however, it is often much more difficult to identify younger plants and plants grown in shade. For mature plants, another index character can be used. The upper leaves of mature plants belonging to the inflata group tend to be very small and scale-like (see the stem on the right-hand side of Fig. 5B). This feature does not occur in the integrifolia group (Fig. 5A). Based on this morphological feature, the inflata and integrifolia groups can be distinguished without requiring the inspection of inner floral organs.
Distribution
In Brazil, the integrifolia and inflata groups were isolated geographically (Fig. 7). The inflata group occurred in northern upland areas, on a lava plateau called the Planalto. This region is characterized by rather flat, gentle hilly ground. The steep slope at the edge of Planalto is known as the Serra Geral. The elevation of the Planalto exceeds 1000 m at the eastern end close to the Atlantic coast, and the Serra Geral facing the east forms a perpendicular cliff. To the west, the elevation of the Planalto gradually decreases, and the Serra Geral facing south is strongly eroded to form a complex terrain. The limit between the Planalto and the lowland area, i.e. the border between the territories of the inflata and integrifolia groups, is evident in central RS, but becomes obscure in western RS. No geographical border, limiting the territories of the inflata and integrifolia groups seems to exist near the town of São Borja (indicated with a star in Fig. 7), although the territories of the two groups are close to each other at that point.
The region surrounding São Borja was visited four times (1993, 1994, 1996 and 1997) to study distribution (Ando, 2003, 2004). The smallest distance between populations of the integrifolia (B701, Mun. São Borja) and inflata (B699, Mun. São Borja, route BR285, 5 km west-south-west of Rio Icamaquã to São Borja, 28°57'47''S, 55°31'44''W) groups was 16 km. Population B812, which was considered to be a possible hybrid swarm, was situated within the territory of the integrifolia group (solid square in Fig. 7), but only 30 km from a population of the inflata group (B698, Mun. Santo Antônio das Missões). The Rio Icamaquã, which flows close to São Borja, appears as another border on the map (Fig. 7), although this river seems too small to function as an obstacle capable of physically separating the two groups.
Most of the research in this study was conducted in Brazil, and the number of populations assessed from other inland locations, such as Argentina and Paraguay, was limited. However, the inflata and integrifolia groups seem to have distinctly different distributions, and populations of the two groups were always geographically separated. Outside Brazil, the wetland areas between latitude 29° and 30°S seem likely to form another geographical barrier separating the two groups, although further research is necessary to better define the patterns of geographical distribution.
Petunia interior T. Ando & Hashim. is a species occurring on the uplands in interior SC and north-western RS close to SC (see fig. 4 in Ando and Hashimoto, 1996). Therefore, the distribution range of the inflata group (Fig. 7) is close to that of P. interior in the north-western RS. The inflata group seems to have the ability to survive as a weed. Several dozen hectares of harvested wheat fields, totally covered with purple flowers of the inflata group, have often been encountered there. However, P. interior has never been seen as a weed. It seems to adhere to undisturbed land. In the eastern upland regions of SC and RS, another species, P. altiplana, occurs (see fig. 5 in Ando and Hashimoto, 1993). Petunia altiplana seems to prefer much higher lands compared with the inflata group. There is no Petunia species whose distribution range is close to that of the integrifolia group except the inflata group. Petunia axillaris whose distribution range overlaps with those of the integrifolia and inflata groups has been excluded from the discussion thus far. This species has large white flowers and is considered to be reproductively isolated from the species with purple flowers like the P. integrifolia complex by an exclusive pollinator (Ando et al., 2001).
Comparison with the work of Wijsman
Wijsman (1982) described Petunia inflata and P. occidentalis as subspecies of P. integrifolia. He focused on the inflexus/deflexus position of the pedicels, a character observed in herbarium specimens. He also emphasized the inter-gradation of two other characters over a geographical course from east to west, where the deflexus character changed to inflexus. The classification system used in that study, based on the measurement of 22 herbarium specimens belonging to the P. integrifolia complex was as follows: (1) deflexed, (2) deflexed, but some aberrant, (3) inflexed, but some aberrant, and (4) inflexed (see table 2 in Wijsman, 1982). However, it is believed that this classification system is inappropriate, due to the high variability of pedicel condition among members of the integrifolia group, as shown in the present study (Fig. 4A and B).
Of the three specimens Wijsman (1982) classified as (2) ‘deflexed, but some aberrant’, it was found that two belonged to neither the integrifolia nor inflata groups, but were in fact Petunia altiplana and a species of Calibrachoa. The third specimen in this category (Reitz s. n. in U) was collected from São Joaquim, SC, but it was not possible to locate the specimen. Special attention has been given to this region of very rich vegetation and it has been visited 11 times (annually from 1988 to 2000, except 1994 and 1998; Ando, 2003, 2004). From the present observations, it can safely be concluded that Petunia altiplana is the sole species of the genus occurring in that region.
In Wijsman's (1982) map showing the geographical variation in pedicel condition for the P. integrifolia complex, three symbols representing inflexus condition (symbol, +) were marked in the Entre Ríos Province of Argentina, west of Uruguay (see fig. 2 in Wijsman, 1982). However, this is likely to have been a mistake, since no specimens were reported from this province (see table 2 in Wijsman, 1982). When the three symbols representing ‘deflexed, but some aberrant’ [symbol, (–)] and the three symbols in Entre Ríos are removed from consideration, his map also implies that the deflexus and inflexus conditions are distributed in southern and northern regions, respectively. Influenced by the instability of pedicel condition characters, Wijsman (1982) is likely to have drawn a conclusion that counters the true features observed in the native habitat.
Sub-cluster of the integrifolia group
The morphology of the integrifolia group seems more variable in comparison with the inflata group, as suggested by the larger amalgamation distances in the cluster analysis (Fig. 2). With respect to the morphological inter-gradation between the inflata and integrifolia groups, the existence of a minor sub-cluster within the integrifolia group should not be ignored, because members of the minor sub-cluster occur mostly along the border between territories of the inflata and integrifolia groups (closed circles in Fig. 7). In the dendrogram obtained from the cluster analysis (Fig. 2), a minor sub-cluster appears to be a bridge between the cluster of the inflata group and the major sub-cluster of the integrifolia group. However, this is not the case. The minor sub-cluster cannot be considered a morphological intermediate between the inflata group and the members of the major sub-cluster. Many of the floral organs of the minor sub-cluster members were larger than those of the major sub-cluster (Table 3) and distinct from those of the inflata group. As a possible explanation for the larger floral organs, hybrid effects cannot be discounted. Nevertheless, a hybrid swarm (B812) was situated in the intermediate region between the inflata and integrifolia groups (a square in Fig. 1), but the members of the minor sub-cluster appeared as outliers of the integrifolia group in the scatter diagram of the PC analysis (closed circles in Fig. 1). Analysis of the minor sub-cluster of the integrifolia group is left for future studies.
Conclusions
Wijsman (1982) emphasized an ‘apparent primary inter-gradation’ between Petunia integrifolia and P. inflata, and the involvement of a cline that ‘in going west, the deflexus character changes into inflexus’. However, no evidence of such an inter-gradation or cline between the integrifolia and inflata groups was found. Even omitting the pedicel condition, a clear gap existed in the set of morphological characteristics defining the two groups (Figs 1 and 2). Wijsman (1982) also noted that ‘Separation of P. inflata and P. integrifolia is very difficult’; however, based on the characters observed here, the authors believe that it is rather simple to distinguish the inflata and integrifolia groups in the field. Wijsman (1982) regarded P. integrifolia and P. inflata as ‘varieties of one biological species’, and stated, ‘The extremes can be given the following names, Petunia integrifolia subsp. integrifolia and subsp. inflata’. The present results disagree with these statements and suggest that the inflata group is not merely a morphological extreme of the integrifolia group, but rather an independent taxon with a clearly defined morphology and distribution. Furthermore, the treatment of Petunia inflata as a synonym of P. integrifolia (Smith and Downs, 1966) can also be safely ruled out.
Of the three different classification systems used for the inflata group (Fries, 1911; Smith and Downs, 1966; Wijsman, 1982), the present results support the opinion of Fries (1911) who referred to the inflata group as a natural taxon (i.e. Petunia inflata) that was independent of P. integrifolia. Of the various speciation mechanisms at work in the plant kingdom, at least three operate in the genus Petunia, including geographical separation (Ando et al., 2001). The inflata and integrifolia groups are inter-fertile (Tsukamoto et al., 1998), but interspecific genetic exchange seems to be prevented by their different distributions (Fig. 7). The inflata group is distinguished from the integrifolia group by at least eight site mutations in the analysis of restriction fragment length polymorphism of chloroplast DNA (Ando et al., 2005). Although it is necessary to consult more molecular information before conclusions can be drawn, the most plausible status of Petunia inflata should be an allopatric species, or ‘geographische Art’ (geographic species) in accordance with Fries (1911). At this time, suffice it to say that Petunia integrifolia and P. inflata are differentiated in a considerable number of morphological traits as represented in Table 4.
|
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGEMENTSLITERATURE CITED |
|---|
Voucher specimens of the materials used in the present study
Abbreviations of the herbaria followed Holmgren et al. (1990)
, except GHSP (Goro Hashimoto, Centro de Pesquisas de História Natural, São Paulo, Brazil) and Ando (temporary collection of Toshio Ando). An application has been made to the New York Botanical Garden to register GHSP in the herbaria database. It may appear in the next edition of the Index Herbariorum. Petunia integrifolia (Hook.) Schinz et Thell.
Argentina
Entre Ríos: Dept Colón: Rt. N14, Arroyo Urquiza 32°20'21''S 58°15'44''W, 28 Nov. 1998, T. Ando & J. Tsukahara A168 (BAB, Ando); Dept Colón: Rt. N14, Arroyo Concepción 31°44'27''S 58°18'12''W, 28 Nov. 1998, T. Ando & J. Tsukahara A171 (BAB, Ando); Dept Concordia: Colonia Ayuí, beside Embalse Salto Grande 31°12'02''S 57°58'43''W, 29 Nov. 1998, T. Ando & J. Tsukahara A175 (BAB, Ando); Dept Gualeguaychú: Rt. N136, Río Uruguay 33°04'04''S 58°15'20''W, 27 Nov. 1998, T. Ando & J. Tsukahara A163 (BAB, Ando); Dept Gualeguaychú: Gualeguaychú, beside Río Gualeguaychú 33°00'27''S 58°29'11''W, 27 Nov. 1998, T. Ando & J. Tsukahara A166 (BAB, Ando).
Brazil
Rio Grande do Sul: Mun. Alegrete: Rt. RS566, Rio Ibicuí 29°18'36''S 56°02'52''W, 27 Nov. 1996, G. Hashimoto et al. B1198 (GHSP, Ando); Mun. Boqueirão do Leão: Rt. RS422, 4 km north-west of Boqueirão Leão to Barros Cassal 29°16'53''S 52°26'09''W, 26 Nov. 1994, G. Hashimoto et al. B870 (GHSP, Ando); Mun. Caçapava do Sul: Rt. BR290, 3·9 km north-east-east of the junction BR153/BR290 to Pântano Grande 30°21'03''S 53°19'51''W, 28 Nov. 1991, G. Hashimoto et al. B314 (BM, MBM, GHSP, Ando); Mun. Cacequi: Rt. RS640, Cacequi 29°51'59''S 54°49'33''W, 7 Dec. 1993, G. Hashimoto et al. B680 (GHSP, Ando); Mun. Cachoeira do Sul: Rt. BR153, 22·1 km north of the junction of route BR290 and BR153 30°04'37''S 52°52'32''W, 28 Nov. 1991, G. Hashimoto et al. B315 (GHSP, Ando); Mun. Cachoeira do Sul: Rt. BR290, 30 km west of the entrance to Cachoeira do Sul to Vila Nova 30°17'15''S 53°08'32''W, 3 Dec. 1993, G. Hashimoto et al. B651 (GHSP, Ando); Mun. Dom Pedrito: Rt. BR293, 12 km south-east-east of Rio Ibicuí da Armada to Dom Pedrito 30°52'06''S 54°55'03''W, 16 Nov. 1994, G. Hashimoto et al. B802 (BM, MBM, MVFA, US, GHSP, Ando); Mun. Encantado: Rt. RS332, Encantado 29°13'19''S 51°53'33''W, 27 Nov. 1995, G. Hashimoto et al. B971 (GHSP, Ando); Mun. Lavras do Sul: 29 km north-east of Dom Pedrito to Ibaré 30°51'01''S 54°28'19''W, 25 Nov. 1996, G. Hashimoto et al. B1190 (GHSP, Ando); Mun. Mata: 3 km south-west of Mata to BR287 29°33'53''S 54°28'16''W, 21 Nov. 1994, G. Hashimoto et al. B835 (GHSP, Ando); Mun. Pântano Grande: Rt. BR471, 10 km north of Pântano Grande to Rio Pardo 30°10'06''S 52°22'25''W, 25 Nov. 1994, G. Hashimoto et al. B867 (GHSP, Ando); Mun. Quaraí: Rt. BR293, 1·6 km north-west of Rio Cati to Quaraí 30°29'09''S 56°13'06''W, 27 Nov. 1991, G. Hashimoto et al. B300 (GHSP, Ando); Mun. Quaraí: Rt. BR293, 28·0 km south-east of Quaraí to Santana do Livramento 30°27'46''S 56°16'11''W, 27 Nov. 1991, G. Hashimoto et al. B301 (GHSP, Ando); Mun. Restinga Seca: Rt. RS509, 12 km west of Rio Jacuí to Santa Maria 29°43'26''S 53°24'03''W, 26 Nov. 1995, G. Hashimoto et al. B948 (GHSP, Ando); Mun. Rio Pardo: Rt. BR471, 3 km south of Rio Pardo to Pântano Grande 30°00'28''S 52°21'59''W, 25 Nov. 1994, G. Hashimoto et al. B868 (GHSP, Ando); Mun. Rosário do Sul: Rt. BR158, 10 km south-west of the junction BR158/BR290 to Santana do Livramento 30°20'06''S 55°00'35''W, 6 Dec. 1993, G. Hashimoto et al. B673 (GHSP, Ando); Mun. Rosário do Sul: Rt. RS640, 15 km north of the junction BR290/RS640 to Cacequi 30°07'18''S 54°49'28''W, 7 Dec. 1993, G. Hashimoto et al. B677 (GHSP, Ando); Mun. Rosário do Sul: 20 km north-west-west of Virador to São Leandro 30°22'43''S 55°25'19''W, 22 Nov. 1994, G. Hashimoto et al. B841 (GHSP, Ando); Mun. Santa Maria: Rt. BR392, 21·6 km south-east of the entrance of Santa Maria 29°53'25''S 53°43'53''W, 25 Nov. 1991, G. Hashimoto et al. B280 (GHSP, Ando); Mun. Santa Maria: Rt. BR287, 8 km east of Rio Ibicuí to Santa Maria 29°40'14''S 54°01'28''W, 18 Nov. 1994, G. Hashimoto et al. B818 (GHSP, Ando); Mun. Santana do Livramento: Rt. BR293, 15·5 km west of the junction BR293/BR158 30°49'33''S 55°20'15''W, 27 Nov. 1991, G. Hashimoto et al. B294 (MVFA, GHSP, Ando); Mun. Santiago: Rt. BR287, 22 km north-north-west of Jaguari to Santiago 29°21'05''S 54°45'29''W, 7 Dec. 1993, G. Hashimoto et al. B686 (GHSP, Ando); Mun. Santiago: Rt. BR287, 18 km south-east of Santiago to Jaguari 29°16'30''S 54°47'42''W, 7 Dec. 1993, G. Hashimoto et al. B687 (GHSP, Ando); Mun. Santiago: Rt. BR287, 2 km north of the south entrance of Santiago 29°11'49''S 54°50'56''W, 7 Dec. 1993, G. Hashimoto et al. B688 (GHSP, Ando); Mun. Santiago: Rt. BR287, 30 km south-east of Encruzilhada to Santiago 29°02'26''S 55°14'49''W, 8 Dec. 1993, G. Hashimoto et al. B705 (GHSP, Ando); Mun. Santiago: Rt. BR287, 42 km south-east of Encruzilhada to Santiago 29°03'05''S 55°07'36''W, 8 Dec. 1993, G. Hashimoto et al. B707 (GHSP, Ando); Mun. Santiago: Rt. BR287, 9 km north-west of the junction BR287/road to São Luíz Gonzaga 29°03'29''S 54°57'42''W, 8 Dec. 1993, G. Hashimoto et al. B708 (GHSP, Ando); Mun. Santiago: Rt. RS546, Santiago 29°12'14''S 54°53'02''W, 9 Nov. 1997, G. Hashimoto et al. B1304 (GHSP, Ando); Mun. São Borja: Rt. BR287, 7 km south-east of Nhuporã to Encruzilhada 28°49'53''S 55°43'42''W, 8 Dec. 1993, G. Hashimoto et al. B701 (GHSP, Ando); Mun. São Borja: Rt. BR287, 19 km south-east of Nhuporã to Encruzilhada 28°54'16''S 55°38'16''W, 8 Dec. 1993, G. Hashimoto et al. B702 (GHSP, Ando); Mun. São Borja: Rt. BR287, 11 km east of Encruzilhada to Santiago 28°58'46''S 55°23'45''W, 8 Dec. 1993, G. Hashimoto et al. B704 (GHSP, Ando); Mun. São Francisco de Assis: Rt. RS241, 8 km north-west of Rio Jaguarí to São Francisco de Assis 29°38'26''S 55°00'59''W, 26 Nov. 1996, G. Hashimoto et al. B1192 (GHSP, Ando); Mun. São Francisco de Assis: Rt. RS241, 8 km south-east of São Francisco de Assis to Rio Jaguarí 29°36'15''S 55°04'35''W, 26 Nov. 1996, G. Hashimoto et al. B1193 (GHSP, Ando); Mun. São Francisco de Assis: Rt. RS241, 16 km east of Manuel Viana to São Francisco de Assis 29°34'20''S 55°20'40''W, 26 Nov. 1996, G. Hashimoto et al. B1195 (GHSP, Ando); Mun. São Francisco de Assis: Rt. RS176, 31 km north of Manuel Viana to Rio Itú 29°19'13''S 55°26'49''W, 9 Nov. 1997, G. Hashimoto et al. B1314 (GHSP, Ando); Mun. São Gabriel: Rt. BR290, 28·8 km north-west of entrance of São Gabriel to Rosário do Sul 30°15'00''S 54°34'59''W, 28 Nov. 1991, G. Hashimoto et al. B311 (GHSP, Ando); Mun. São Gabriel: Rt. BR290, 9 km east of São Gabriel to Vila Nova 30°21'52''S 54°14'29''W, 23 Nov. 1994, G. Hashimoto et al. B848 (GHSP, Ando); Mun. São Pedro do Sul: Rt. BR287, 7 km west of entrance of São Pedro do Sul to São Vicente do Sul 29°38'03''S 54°15'42''W, 18 Nov. 1994, G. Hashimoto et al. B817 (GHSP, Ando); Mun. São Vicente do Sul: Rt. BR287, 5 km north of São Vicente do Sul to Jaguari 29°38'19''S 54°40'45''W, 7 Dec. 1993, G. Hashimoto et al. B682 (GHSP, Ando); Mun. Sobradinho: 15 km south-east of Sobradinho to Candelária 29°30'06''S 52°55'45''W, 26 Nov. 1995, G. Hashimoto et al. B949 (GHSP, Ando); Mun. Uruguaiana: Rt. BR472, 3 km south-west of João Arrengui to Uruguaiana 29°29'55''S 56°41'47''W, 17 Nov. 1994, G. Hashimoto et al. B809 (GHSP, Ando).
Uruguay
Flores: Rt. 3, 1·9 km north-west of Andresito 33°09'19''S 57°08'31''W, 13 Nov. 1990, T. Ando & K. Buto U168 (MVFA, SI, GHSP, Ando).
Río Negro: Rt. 2, 1·6 km north-west of the bridge over Río Negro to Fray Bentos 33°13'28''S 58°01'31''W, 28 Nov. 1989, T. Ando & H. Watanabe U106 (MVFA, S, SI, GHSP, Ando); Parque Puerto Viejo, N of San Javier 32°30'13''S 58°08'50''W, 12 Nov. 1990, T. Ando & K. Buto U161 (BM, MVFA, S, SI, US, GHSP, Ando); Rt. 3, 1 km north-west of the bridge over Río Negro 33°07'46''S 57°10'19''W, 4 Nov. 1991, T. Ando & S. Iida U239 (MVFA, Ando); Las Canas 33°10'13''S 58°21'18''W, 14 Dec. 1992, T. Ando & K. Shibata U263 (BM, MVFA, S, US, Ando).
Rivera: Rt. 5, 7·2 km south of junction Rt. 5/30 to Tacuarembó 31°18'27''S 55°40'22''W, 18 Nov. 1988, T. Ando & H. Kokubun U12 (BM, MVFA, Ando).
Salto: Rt. 3, entrance of Termas de Arapey 30°54'27''S 57°41'37''W, 3 Nov. 1991, T. Ando & S. Iida U231 (MVFA, Ando); 5 km south-west of Salto 31°23'49''S 57°59'07''W, 4 Nov. 1991, T. Ando & S. Iida U233 (BM, MVFA, Ando).
Tacuarembó: San Gregorio de Polanco 32°37'29''S 55°49'54''W, 23 Nov. 1988, T. Ando & H. Kokubun U46 (MVFA, Ando); Gruta de los Cuelvos, north-west of Tacuarembó, 30 Nov. 1989, T. Ando & H. Watanabe U128 (MVFA, Ando).
Petunia inflata R. E. Fr.
Argentina
Chaco: Dept Quitilipi: Rt. P4, 7·4 km south-south-west of Pampa Verde to Quitilipi 26°31'16''S 60°04'29''W, 20 Nov. 1999, T. Ando et al. A260 (BAB, Ando).
Corrientes: Dept Empedrado: Rt. N12, 8·2 km north-north-west of Arroyo San Lorenzo to Empedrado 28°02'43''S 58°47'57''W, 21 Nov. 1999, T. Ando et al. A270 (BAB, Ando).
Formosa: Dept Pirané: Rt. N81, 11·0 km north-west of entrance of Pirané to Palo Santo 25°38'07''S 59°09'04''W, 17 Nov. 1999, T. Ando et al. A249 (BAB, Ando).
Misiones: Dept Candelaria: Rt. N4, Parque Central, Bonpland 27°29'05''S 55°28'41''W, 27 Nov. 1990, T. Ando & K. Buto A7 (SI, GHSP, Ando); Dept San Ignacio: Rt. RN12, 6·2 km north-east of Santo Pipó to Jardin América 27°06'51''S 55°21'32''W, 10 Nov. 1995, T. Ando & T. Tsukamoto A119 (BAB, Ando).
Brazil
Rio Grande do Sul: Mun. Arvorezinha: 10 km west of Morangueira to Soledade 28°41'54''S 52°14'46''W, 21 Nov. 1996, G. Hashimoto et al. B1163 (GHSP, Ando); Mun. Barros Cassal: 17 km south-west of Barros Cassal to Segredo 29°08'33''S 52°42'38''W, 26 Nov. 1995, G. Hashimoto et al. B956 (GHSP, Ando); Mun. Casca: Rt. RS324, 11 km west of Casca to Marau 28°33'04''S 52°04'14''W, 28 Nov. 1993, G. Hashimoto et al. B587 (GHSP, Ando); Mun. Ciríaco: Rt. BR285, 27 km west of Lagoa Vermelha to Cruz Altinha 28°14'57''S 51°44'32''W, 28 Nov. 1993, G. Hashimoto et al. B581 (GHSP, Ando); Mun. Cruz Alta: Rt. RS342, Cruz Alta 28°38'37''S 53°37'51''W, 25 Nov. 1993, G. Hashimoto et al. B540 (GHSP, Ando); Mun. Erexim: Rt. BR153, 1 km south of the junction BR153/RS331 27°37'30''S 52°14'07''W, 1 Dec. 1996, G. Hashimoto et al. B1222 (GHSP, Ando); Mun. Getúrio Vargas: Rt. RS135, 9·7 km north of north entrance of Getúlio Vargas to Erexim 27°49'21''S 52°17'32''W, 23 Nov. 1991, G. Hashimoto et al. B261 (GHSP, Ando); Mun. Giruá: Rt. RS544, 13 km south-south-west of Santa Rosa to Guarani das Missões 27°58'49''S 54°32'35''W, 29 Nov. 1996, G. Hashimoto et al. B1211 (GHSP, Ando); Mun. Guarani das Missões: Guarani das Missões 28°09'01''S 54°33'02''W, 29 Nov. 1996, G. Hashimoto et al. B1209 (GHSP, Ando); Mun. Guarani das Missões: Rt. RS544, 3 km north of Guarani das Missões to Santa Rosa 28°07'43''S 54°33'43''W, 29 Nov. 1996, G. Hashimoto et al. B1210 (GHSP, Ando); Mun. Ibiaçá: Rt. RS126, entrance to Ibiaçá 28°01'37''S 51°46'33''W, 9 Dec. 1993, G. Hashimoto et al. B724 (GHSP, Ando); Mun. Ijuí: Rt. BR285, 6·7 km west of Rio Conceição to Santo Ângelo 28°22'40''S 54°05'46''W, 24 Nov. 1991, G. Hashimoto et al. B272 (GHSP, Ando); Mun. Júlio de Castilhos: 39 km south-east-east of Júlio de Castilhos to Pinhal Grande 29°21'06''S 53°24'30''W, 25 Nov. 1993, G. Hashimoto et al. B544 (MBM, GHSP, Ando); Mun. Júlio de Castilhos: 25 km north-east of Pinhal Grande to Barragem de Itaúba 29°15'34''S 53°14'52''W, 25 Nov. 1993, G. Hashimoto et al. B549 (GHSP, Ando); Mun. Machadinho: 2 km north of Machadinho to Capinzal (SC) 27°32'55''S 51°39'52''W, 27 Nov. 1993, G. Hashimoto et al. B574 (GHSP, Ando); Mun. Nonoai: Rt. RS406, 14 km south-west of Nonoai to Trinidade 27°26'36''S 52°53'06''W, 26 Nov. 1993, G. Hashimoto et al. B564 (GHSP, Ando); Mun. Palmeira das Missões: Rt. BR468, 15 km north-west of Palmeira das Missões to Coronel Bicaco 27°49'14''S 53°25'02''W, 24 Nov. 1993, G. Hashimoto et al. B522 (GHSP, Ando); Mun. Palmeira das Missões: Rt. RS569, 8 km east of Palmeira das Missões to Sarandi 27°53'37''S 53°14'20''W, 7 Nov. 1997, G. Hashimoto et al. B1270 (GHSP, Ando); Mun. Panambi: Rt. BR285, 6·1 km west of the junction BR285/RS158 to Ijuí 28°21'43''S 53°36'58''W, 24 Nov. 1991, G. Hashimoto et al. B269 (MVFA, GHSP, Ando); Mun. Santa Rosa: Rt. RS544, 4 km south-west of Santa Rosa to Guarani das Missões 27°54'44''S 54°30'29''W, 29 Nov. 1996, G. Hashimoto et al. B1212 (GHSP, Ando); Mun. Santo Ângelo: São Miguel das Missões 28°32'55''S 54°33'36''W, 24 Nov. 1991, G. Hashimoto et al. B274 (GHSP, Ando); Mun. Santo Ângelo: Rt. RS344, 14 km north-north-west of Santo Ângelo to Giruá 28°11'45''S 54°19'21''W, 24 Nov. 1993, G. Hashimoto et al. B538 (GHSP, Ando); Mun. Santo Antônio das Missões: Rt. BR285, 6 km west of the entrance of Santo Antônio das Missões to Itaroquém 28°29'33''S 55°16'53''W, 8 Dec. 1993, G. Hashimoto et al. B692 (GHSP, Ando); Mun. Santo Antônio das Missões: Rt. BR285, 7 km west of Santo Antônio das Missões to São Borja 28°29'36''S 55°17'42''W, 8 Dec. 1993, G. Hashimoto et al. B694 (GHSP, Ando); Mun. Santo Antônio das Missões: Rt. BR285, 19 km NEE of Rio Icamaquã to Itaroquém 28°33'03''S 55°34'00''W, 8 Dec. 1993, G. Hashimoto et al. B697 (GHSP, Ando); Mun. Santo Antônio das Missões: Rt. BR285, 10 km north-east-east of Rio Icamaquã to Itaroquém 28°35'36''S 55°38'41''W, 8 Dec. 1993, G. Hashimoto et al. B698 (GHSP, Ando); Mun. Santo Antônio das Missões: Rio Piratinim, between São Nicolau and Garruchos 28°12'29''S 55°19'10''W, 28 Nov. 1996, G. Hashimoto et al. B1205 (GHSP, Ando); Mun. Sarandi: Rt. RS404, 7 km north of Sarandi to Rondinha 27°53'20''S 52°54'50''W, 26 Nov. 1993, G. Hashimoto et al. B560 (GHSP, Ando); Mun. São Borja: 8 km north-east of Rincão do Meio to Garruchos 28°26'46''S 55°31'22''W, 28 Nov. 1996, G. Hashimoto et al. B1200 (GHSP, Ando); Mun. São Borja: 24 km south-east of Garruchos to Rincão do Meio 28°17'53''S 55°31'49''W, 28 Nov. 1996, G. Hashimoto et al. B1202 (GHSP, Ando); Mun. São Borja: Garruchos 28°10'56''S 55°38'31''W, 28 Nov. 1996, G. Hashimoto et al. B1203 (GHSP, Ando); Mun. São Borja: 24 km south-east of Garruchos to Santo Antônio das Missões 28°15'18''S 55°26'25''W, 28 Nov. 1996, G. Hashimoto et al. B1204 (GHSP, Ando); Mun. São Luíz Gonzaga: Rt. BR285, 3 km east of São Luíz Gonzaga to Santo Ângelo 28°25'01''S 54°55'53''W, 8 Dec. 1993, G. Hashimoto et al. B709 (GHSP, Ando); Mun. São Luíz Gonzaga: Rt. BR285, 28 km E of São Luíz Gonzaga to Santo Ângelo 28°24'14''S 54°40'54''W, 8 Dec. 1993, G. Hashimoto et al. B710 (GHSP, Ando); Mun. São Luíz Gonzaga: Rt. RS544, 3 km north-east of Rolador to Cerro Largo 28°14'34''S 54°47'36''W, 29 Nov. 1996, G. Hashimoto et al. B1208 (GHSP, Ando); Mun. São Nicolau: 8 km south-east of São Nicolau to São Luíz Gonzaga 28°13'07''S 55°12'19''W, 28 Nov. 1996, G. Hashimoto et al. B1206 (GHSP, Ando); Mun. São Valentim: Rt. RS480, 5 km north-west of São Valentim to Erval Grande 27°31'45''S 52°32'20''W, 1 Dec. 1996, G. Hashimoto et al. B1219 (GHSP, Ando); Mun. Segredo: Lagoão 29°14'09''S 52°47'46''W, 26 Nov. 1995, G. Hashimoto et al. B954 (GHSP, Ando); Mun. Selbach: Rt. RS153, 12 km east of Ibirubá to Tapera 28°38'39''S 52°58'10''W, 9 Dec. 1993, G. Hashimoto et al. B719 (GHSP, Ando); Mun. Tapejara: 8 km north-east of Tapejara to Ibiaçá 28°01'28''S 51°57'27''W, 9 Dec. 1993, G. Hashimoto et al. B723 (GHSP, Ando); Mun. Três de Maio: Rt. RS210, 11 km west of Três de Maio to Santa Rosa 27°46'23''S 54°20'05''W, 24 Nov. 1993, G. Hashimoto et al. B531 (GHSP, Ando); Mun. Tupanciretã: Rt. BR158, 10 km north of entrance of Tupanciretã to Cruz Alta 28°56'44''S 53°38'49''W, 7 Nov. 1997, G. Hashimoto et al. B1287 (GHSP, Ando); Mun. Tupanciretã: 8 km east of Tupanciretã to Rt. BR158 29°03'16''S 53°46'01''W, 8 Nov. 1997, G. Hashimoto et al. B1288 (GHSP, Ando); Mun. Tupanciretã: Rt. BR377, 78 km west of Tupanciretã to Santiago 29°01'01''S 54°23'35''W, 8 Nov. 1997, G. Hashimoto et al. B1302 (GHSP, Ando); Mun. Tuparendi: Rt. RS344, 4 km north-west of Tuparendi to Porto Mauá 27°44'32''S 54°30'31''W, 24 Nov. 1993, G. Hashimoto et al. B533 (GHSP, Ando); Mun. Victor Graeff: Rt. BR386, 12 km north-north-west of Rio Ijuí to Carazinho 28°30'35''S 52°38'22''W, 26 Nov. 1993, G. Hashimoto et al. B556 (GHSP, Ando).
Santa Catarina: Mun. Anita Garibaldi: Rt. SC458, 16 km north-west of Cerro Negro to Anita Garibaldi 27°45'16''S 51°00'25''W, 30 Nov. 1993, G. Hashimoto et al. B628 (GHSP, Ando); Mun. Campos Novos: Rt. SC456, 5·8 km north of junction BR470/SC456 to Monte Carlo 27°17'31''S 50°58'33''W, 3 Dec. 1991, G. Hashimoto et al. B350 (GHSP, Ando); Mun. Itapiranga: Itapiranga 27°10'52''S 53°43'10''W, 25 Nov. 1992, G. Hashimoto et al. B377 (GHSP, Ando); Mun. São José do Cerrito: Rt. BR282, 8 km south-east of Rio Canoas to São José do Cerrito 27°34'29''S 50°48'15''W, 29 Nov. 1992, G. Hashimoto et al. B439 (GHSP, Ando).
Paraguay
Central: Rt. 1, Itá, 28 Jul 1990, T. Ikemizu & T. Ando P1 (Ando).
Intermediate form between P. integrifolia and P. inflata.
Brazil
Rio Grande do Sul: Mun. São Borja: 10 km north of Capitão Porto Alegre to São José 28°51'29''S 55°32'12''W, 18 Nov. 1994, G. Hashimoto et al. B812 (GHSP, Ando).
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGEMENTSLITERATURE CITED |
|---|
The authors th