DISTRIBUTIONAL PATTERNS OF THE NEOTROPICAL FLY GENUS POLIETINA SCHNABL & DZIEDZICKI ( DIPTERA , MUSCIDAE ) : A PHYLOGENY-SUPPORTED ANALYSIS USING PANBIOGEOGRAPHIC TOOLS

Over the last decades, Neotropical region has been subdivided into smaller units (areas of endemism), yet these subdivisions were not necessarily based on an evolutionary perspective. Consequently, these areas of endemism may be biogeographic units that do not actually represent natural historical units. Here, the distributional patterns of the genus Polietina Schnabl & Dziedzicki, 1911 (Diptera, Muscidae; including 15 species) are analysed by applying panbiogeographic tools to recognise and propose primary homologous areas within the Neotropical region. The analysis and discussion of the results obtained here will be reconciled to the information provided by the phylogenetic hypothesis available for the genus.


INTRODUCTION
During the last few decades, the Neotropical Region has been studied using a classification perspective; that is, it has been divided into smaller historical units (areas of endemism).Morrone (2001b) found 33 studies classifying the Neotropics into regions, subregions and provinces through several different criteria, e.g.geographic, palaeontological, faunistic and floristic.However, most studies did not propose a classification based upon an evolutionary perspective and, therefore, many of these classifications have used or defined biogeographic units that do not represent natural units (Morrone, 2001b).We find three comprehensive works as the most important contributions on this question: Cracraft (1985), in analysing the avi-faunal distributional pattern, postulated 30 areas of endemism for South America (including west of the Andes and southernmost South America).Amorim & Pires (1996), using phylogenetic and biogeographic patterns of several animal groups, postulated that the Neotropics be divided into three main components: Caribbean, NW and SE components.The NW component comprised three smaller units: Andean-Mesoamerican, Southwest Amazonia and Northern Amazonia; whereas the SE component comprises Southeast Amazonia and the Atlantic Forest.Finally, Morrone (2001b) postulated that the region be divided into four subregions: Caribbean, Amazonian, Chacoan and Paraná.His proposal was based mostly on the panbiogeographic analysis of several animal and plant taxa.
Panbiogeography comprises a first step towards the recognition of primary biogeographical homologies, followed by a second step, which is the confirmation of those primary hypotheses as secondary homologies (Morrone, 2001a(Morrone, , 2004)).In this latter step, the primary homologies can be legitimate by the application of cladistic biogeographic methodology (the cladistic test, Morrone, 2001a).
Panbiogeographic tools can provide useful information through the generation of generalised tracks and nodes.Generalised tracks are obtained by the spatial overlapping of two or more individual tracks, which represent the spanning tree resulting from the minimal length connection of the known localities.The generalised tracks indicate the pre-existence of ancestral biota that became posteriorly fragmented by climatic and/or tectonic change (Craw et al., 1999;Morrone & Crisci, 1995).They may also indicate the existence of areas of endemism, because areas of endemism would be equivalent to smaller generalised tracks (Grehan, 1993;Morrone, 2001a).
Here, we analysed the distributional patterns of Polietina by applying panbiogeographic tools to recognise and propose primary homologous areas within the Neotropical Region.The results obtained will be discussed based on the information provided by the phylogenetic hypothesis of the genus.The discussion will also approach the results published in earlier biogeographic studies focusing Muscidae and the Neotropical region.

MATERIALS AND METHODS
We applied Panbiogeographic methods (following Morrone &Crisci, 1995 andCraw et al., 1999) to analyse the distributional patterns of Polietina and to recognise spatial homologous areas.The known oc-currence localities of each species (Appendix) were plotted into maps and connected by their minimal geographical distance to obtain individual tracks.Those individual tracks were gathered and overlapped to obtain generalised tracks for the genus.Generalised tracks indicate early existence of ancestral biotas which had been posteriorly fragmented by climatic and/or tectonic changes (Morrone & Crisci, 1995;Morrone & Márquez, 2001).Also, generalised tracks can indicate the existence of areas of endemism, since areas of endemism would equate to smaller generalised tracks (Morrone, 2001a;Harold & Mooi, 1994).However, Craw et al. (1999) emphasise that none proposition with regard to the biogeographic processes explaining tracks congruence is implied.Therefore, generalised tracks could represent 1) a track of a ancestral biota posteriorly subdivided through vicariance events, 2) a concordant dispersal pathway used concomitantly by the taxa, 3) isolated events of dispersal, or 4) a combination among these scenarios (Craw et al., 1999).
Biogeographic nodes are complex areas that are recognised where occurs the meeting or overlapping of two or more generalised tracks (Craw et al., 1999;Crisci et al., 2003).Nodes serve as evidence that different ancestral biotic or geologic fragments are interrelated in space and time, resulting from terrain collision, docking or suturing, thereby indicating a composite area (Crisci et al., 2003).Also, nodes may be described as areas that represent geographic and phylogenetic boundaries for the taxa of interest (Heads, 1989).

RESULTS AND DISCUSSION
Based on geographical data available in the literature, the individual tracks for 12 species of Polietina (Figs. 2-3) generated 17 generalised tracks (Fig. 4) from the spatial congruence (overlapping) of the individual tracks.Species composition and the nature of each generalised track will be discussed and reconciled with the phylogenetic hypothesis proposed by Nihei (2004b) (Fig. 1).
A number of generalised tracks obtained for Polietina show congruence with the phylogenetic pattern (Fig. 4).Tracks 1-5 and 7 comprise the sister-species P. nigra and P. prima, while tracks 1 and 2 are supported exclusively by that clade.The polytomic clade P. minor + P. bicolor + P. univittata is present in the tracks 5, 6 and 9, although none of them is composed simultaneously by the three species.And, tracks 13 and 16 are supported by the clade P. flavithorax + P. major.Along with these tracks, track 5 is important because it contains nearly all species in clade B (Fig. 1), except for P. bicolor and P. flavithorax.However, these two species are present in other tracks contiguous with track 5: P. bicolor in track 6 (contiguous with track 5) and P. flavithorax in track 7 (contiguous with and partly overlapping track 5).That area, supported by tracks 5-7, is spatially coincident with the "Atl component" of Camargo & Pedro (2003) based on distributional patterns of Meliponine bees, and could represent an important area of endemism for the diversification of clade B. Similarly, the remaining tracks discussed above may represent areas of endemism for the species in question, especially tracks comprising closely related species.Generalised tracks might also indicate coincident or isolated dispersal events (Craw et al., 1999), however, if phylogeny-supported, a track can indicate an area of endemism or the pre-existence of ancestral biota for the concerned species.For example, generalised tracks formed by the sister-species P. prima and P. nigra probably indicate that the geographical area along which exists congruence in the distribution of these two contemporaneous species is a historically area, with importance to their diversification context.
Most generalised tracks were based on the distribution of species comprising clade B, and are concentrated in the SE component of Amorim & Pires (1996).This corroborates the suggestion that diversification of clade B is associated with that component (Nihei, 2004b).Clade A, on the other hand, was represented in several tracks only by P. orbitalis (widespread over SE component) (Fig. 4).With regard to the remainder of clade A, P. rubella, P. concinna and P. wulpi occur from southern Central America northward through Mexico to the southern USA, whereas P. flavidicincta is restricted to eastern Amazonian forests (see Figs. 2-3).The scarce geographical data for the species of clade A was the major limitation for generalised tracks in the Mesoamerican areas.Nevertheless, further collecting efforts on these areas could provide a rather sound and consistent analysis with regard to species of clade A and to Mesoamerica.
With respect to the subregions postulated by Morrone (2001b), some generalised tracks support some of his subregions.Tracks 3-12 are coincident with the Paraná subregion; and the individual tracks of P. bicolor (Fig. 2A), P. minor (Fig. 2F), and P. univittata (Fig. 3E), three closely related species that form a clade (Fig. 1), are exclusive to that subregion.The Amazonian subregion is supported by tracks 1 and 15-17 although they are restricted to the south of the Amazon River, and there is only one exclusive individual track, that of P. flavidicincta (Fig. 2C).
We compared the distributional patterns of Polietina with the patterns of the Muscidae genera studied by Carvalho et al. (2003).From that comparision, we intended to recognise eventual similarities between the patterns observed in Polietina and taxa belonging to the same family.In that study, those authors analysed the tracks of three muscid genera: Cyrtoneurina Giglio-Tos, 1893, Cyrtoneuropsis Malloch, 1925and Bithoracochaeta Stein, 1911 (Figs. 6A-B, 7).Track 5 of Polietina is spatially coincident with track "g" of Cyrtoneurina (Fig. 6A), tracks "u" and "v" of Cyrtoneuropsis (Fig. 6B), and track "d" of Bithoracochaeta (Fig. 7).Track 1 of Polietina is coincident with tracks "r" and "n" of Cyrtoneuropsis, while track "f " of Cyrtoneurina superimposes to track 4 and partly to track 3 of Polietina.Of the tracks of Cyrtoneurina, Cyrtoneuropsis and Bithoracochaeta that are congruent with Polietina, only   one was supported by the phylogenetic pattern available for each genus (Pamplona, 1999;Couri & Motta, 2000).Track "g" of Cyrtoneurina comprised four species, three of which (C. geminata, C. costata, C. crispaseta) are closely related taxa (Pamplona, 1999).Also, track "d" of Bithoracochaeta comprised two closely related, but not sister, species.
In this analysis, ten nodes were identified for Polietina (Fig. 5).Node "h" demarcates the southern distribution of clade P. nigra + P. prima, whereas node "i" delimits the northern distribution of both clades P. minor + P. bicolour + P. univittata and P. flavithorax + P. major (see more on node "i" below, in comparison with nodes identified for other muscid genera).Hence, the geographical area between nodes "h" and "i" represents a sympatric zone among these three clades and also with P. steini, the most basal species in clade B. The latter species has its distribution delimited by the nodes "c" and "j" (nodes as 'distribution margins' sensu Heads, 2004).Some interesting information on vicariance, not recognised when analysing the generalised tracks and nodes, was possible to obtain only when considering the individual tracks isolatedly (Figs 8-9).Individual tracks of sister-species have indicated the possible location of areas of ancient vicariance among these spe-cies.Figure 8 shows the individual tracks of the clade P. concinna + P. rubella + P. wulpi + P. orbitalis connected altogether, and the probable area of vicariance of that clade and the most basal species of clade A, P. flavidicincta.And Figure 9 shows the individual tracks of part of clade B (Fig. 1 depicts clade B with a basal polytomy, but Nihei, 2004, also supported a resolved solution with a sister-group relationship between FIGURE 6. Generalised tracks and biogeographical nodes of Cyrtoneurina Giglio-Tos (A) and Cyrtoneuropsis Malloch (B) (modified from Carvalho et al., 2003).
P. prima + P. nigra and P. flavithorax + P. major).The vicariance spots in the Amazon domain (Fig. 9) were not recognised in the nodes analysis (Fig. 5), on the other hand, the large area of vicariance recognised within the Atlantic Forest is partially represented by the nodes "h" and "i" (Fig. 5).Furthermore, the vicariance spot located more centrally in the Amazon domain is partially represented by the generalised track 1 (Fig. 4), whereas, the vicariance spot in Atlantic Forest is represented by several tracks (3,4,5,6,7).
We identified the nodes for the genera studied in Carvalho et al. (2003) (Figs.6A-B, 7) and compared with Polietina.There is only one node congruent among Polietina (node "i"), Cyrtoneurina, Cyrtoneuropsis and Bithoracochaeta on the Atlantic Forest domain.Unlike the other genera, with no recognized nodes in Amazonian Forest or Central America, Cyrtoneuropsis has five nodes in those areas.For Polietina, this can be explained because most of the species are distributed southward to the Amazon River (Figs. 2-3), yet in Cyrtoneuropsis most species occur in northeastern South America and Mesoamerica (see fig. 4 of Carvalho et al., 2003).Morrone (2003) identified four generalised tracks in the Neotropics on the basis of the distributional patterns of freshwater decapods of the family Trichodactylidae.The generalised tracks (named Caribbean, Amazonian, Chacoan and Paraná) strongly support the previously proposed subregions of Morrone (2001b).Morrone (2003) also identified three nodes: 1) in the southwest Colombia, between the Caribbean and Amazonian tracks; 2) in Bolivia, between the Amazonian and Chacoan tracks; and 3) in the border region of Argentina-Paraguay-Brazil, between Carvalho et al., 2003).
The nodes identified and discussed here may represent areas of great importance for the historical context of the diversification of these taxa, especially if one regards nodes as biogeographic boundaries of relict fragments of different ancestral biota coming into contact in the present day (Crisci et al., 2003).Biogeographic nodes may also serve as the basis for the selection of priority areas for the proposition of conservation units (Grehan, 1993;Luna et al., 2000), particularly if identification of the nodes is supported by phylogeny, i.e. they are meaningful within an evolutionary context.It means that the generalised tracks, as well as the nodes, are not randomly composed by two or more unrelated species; they are comprised by species with a common history on Earth.To identify the nature of biogeographic nodes is a very complex task.Some studies have been concerned to present and discuss the nature of biogeographic nodes on theoretical grounds (Heads, 1989(Heads, , 2004)), however, in practice, it is very complicate to 'give' a node a specific nature and a scientific explanation.In the present study, we provided some examples where a biogeographic node could indicate the existence of an area of ancient vicariance (see node "a" of Fig. 5, and the area of vicariance of Fig. 8; and nodes "h" and "i" and the area of vicariance on Atlantic Forest indicated in Fig. 9).Morrone (2001a, also Morrone & Crisci, 1995) stressed the importance of an integrative view in the biogeographic practice, and recommended the reconciliation of different methodologies, each one being used for a specific aim within the study.In the present study, we did not apply any additional methodology other than the panbiogeographic tools used to recognise and support the nodes, however, we have sustained our discussion on the basis of the phylogenetic relationships among the species of Polietina.Be- cause sister-taxa are contemporaneous in the time scale and share a common history on Earth, the generalised tracks formed by them are historically more important than tracks comprised randomly by taxa derived from distant lineages.The same interpretation can be extended to biogeographic nodes, i.e., where a given generalised track comprised by clade X meets another track comprised by clade Y forming a node and knowing that clades X and Y are sister-groups.
We believe that our results, in addition to earlier works, provides an accumulative basis for further studies focusing the study of biogeographic homologies within the Neotropical region.For example, the nature of nodes was not fully demonstrated herein, although we hope subsequent studies can advance on this subject and identify possible ancient geological events responsible for the location of the nodes recognised here.

FIGURE 4 .
FIGURE 4. Generalised tracks obtained for Polietina, each one numbered and indicated its composing species.

FIGURE 5 .
FIGURE 5. Biogeographical nodes identified from the generalised tracks of Polietina.
the Chacoan and Paraná tracks.In the nodes of Trichodactylidae and Polietina, two nodes are nearly coincident.Node "a" of Polietina is congruent with the Bolivian node of Trichodactylidae, and node "e" is nearly congruent with the Argentina-Paraguay-Brazil node.These two nodes are clearly within and between the boundaries of the Amazonian and Chacoan, and the Chacoan and Paraná subregions, showing that the distributional patterns of Trichodactylidae and Polietina are congruent with that classification proposal of Morrone (2001b).

FIGURE 8 .
FIGURE 8. Clade A and the individual tracks of its species.The gray circle representing the single locality known for P. rubella; the dashed gray line representing the connection between the northern track (concinna + rubella + wulpi) and P. orbitalis track; the dashed lined circle in black colour indicating area of ancient vicariance.

FIGURE 9 .
FIGURE 9. Part of clade B and the individual tracks of its species.The dashed lined areas (circles and elipse) in black colour indicating areas of ancient vicariance between the clades P. prima + P. nigra and P. flavithorax + P. major.