Molecular Phylogenetics and Evolution 64 (2012) 533–544

Contents lists available at SciVerse ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com/locate/ympev

Monophyly, divergence times, and evolution of host plant use inferred

from a revised phylogeny of the Drosophila repleta species group
Deodoro C.S.G. Oliveira a, Francisca C. Almeida b, Patrick M. O’Grady c, Miguel A. Armella d,
Rob DeSalle e, William J. Etges f,⇑
a
Departamento de Genética y Microbiología, Universidad Autonóma de Barcelona, Bellaterra BCN 08193, Spain
b
Departamento de Genètica, Universitat de Barcelona, Barcelona BCN 08071, Spain
c
Department of Environmental Science, Policy and Management, University of California, Berkeley, CA 94720, USA
d
Departamento de Biología, Universidad Autónoma Metropolitana-Iztapalapa, Av. Michoacán y la Purísma, Col. Vicentina, 09340 Mexico, D.F., Mexico
e
Division of Invertebrate Zoology, American Museum of Natural History, New York, NY 10024, USA
f
Program in Ecology and Evolutionary Biology, Department of Biological Sciences, SCEN 632, University of Arkansas, Fayetteville, AR 72701, USA

a r t i c l e i n f o a b s t r a c t

Article history: We present a revised molecular phylogeny of the Drosophila repleta group including 62 repleta group taxa
Received 2 January 2012 and nine outgroup species based on four mitochondrial and six nuclear DNA sequence fragments. With
Revised 12 May 2012 ca. 100 species endemic to the New World, the repleta species group represents one of the major species
Accepted 14 May 2012
radiations in the genus Drosophila. Most repleta group species are associated with cacti in arid or semiarid
Available online 24 May 2012
regions. Contrary to previous results, maximum likelihood and Bayesian phylogenies of the 10-gene data-
set strongly support the monophyly of the repleta group. Several previously described subdivisions in the
Keywords:
group were also recovered, despite poorly resolved relationships between these clades. Divergence time
Drosophila repleta species group
Host plants
estimates suggested that the repleta group split from its sister group about 21 million years ago (Mya),
Molecular phylogeny although diversification of the crown group began ca. 16 Mya. Character mapping of patterns of host
Molecular clock plant use showed that flat leaf Opuntia use is common throughout the phylogeny and that shifts in host
Cactus use from Opuntia to the more chemically complex columnar cacti occurred several times independently
Biogeography during the history of this group. Although some species retained the use of Opuntia after acquiring the use
of columnar cacti, there were multiple, phylogenetically independent instances of columnar cactus spe-
cialization with loss of Opuntia as a host. Concordant with our proposed timing of host use shifts, these
dates are consistent with the suggested times when the Opuntioideae originated in South America. We
discuss the generally accepted South American origin of the repleta group.
Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction ecological genetics and adaptation of host plant use (Ruiz and
Heed, 1988; Etges et al., 1999). Most species are cactophilic, using
The New World Drosophila repleta species group has proven to fermenting cactus tissues to carry out their life cycles in semiarid
be a valuable ecological and evolutionary model system as one of or arid environments (e.g. Ruiz and Heed, 1988; Ruiz et al.,
the largest species radiations in the genus (Patterson and Stone, 1990), but some species in the repleta group use a broad array of
1952; Throckmorton, 1975; Vilela, 1983; Wasserman, 1992; different resources and occupy habitats from wet, tropical forests
Markow and O’Grady, 2006). Study of particular species in the to temperate environments (Vilela, 1983; Pereira et al., 1983; Vela
group have revealed general insights into chromosome and gen- and Rafael, 2005; Acurio and Rafael, 2009). Therefore, an accurate
ome evolution (Cáceres et al., 1999; Negre et al., 2005), mecha- and well-supported phylogeny of the repleta group would help to
nisms of speciation (Coyne and Orr, 1997; Etges and Jackson, place many of these genetic, behavioral, ecological and evolution-
2001; Etges et al., 2010), sperm competition and evolution of ary problems into a broader phylogenetic perspective.
reproductive proteins (Wagstaff and Begun, 2005; Kelleher et al., Species identifications, taxonomy, and phylogenetic relation-
2007; Wagstaff and Begun, 2007; Almeida and DeSalle, 2008, ships within the repleta group have also proven to be interesting
2009), adaptation to temperature and desiccation stress (Gibbs challenges (e.g. Heed and Grimaldi, 1991; Etges et al., 2001; Diniz
and Matzkin, 2001; Gibbs et al., 2003), fly–cactus–yeast/bacteria and Sene, 2004). The precise number of species is unclear since
interactions (Barker and Starmer, 1982; Barker et al., 1990), and there are taxa that were proposed to be synonymies, and there
are also a number of cryptic species with poorly known species
⇑ Corresponding author. boundaries (e.g. Oliveira et al., 2005, 2008). Six species subgroups
E-mail address: wetges@uark.edu (W.J. Etges). have been described – mulleri, hydei, mercatorum, repleta, fasciola,

1055-7903/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ympev.2012.05.012
534 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544

and inca – and these subgroups have been further subdivided into four species were represented by two taxa each considered to be
species complexes, clusters and subclusters based on chromosome different subspecies: these are D. mojavensis baja, D. meridiana rio-
banding patterns, male genital morphology and/or ecological asso- ensis, D. fulvimacula flavorepleta, and D. mercatorum pararepleta.
ciations (Patterson, 1943; Wharton, 1944; Wasserman, 1962;
Rafael and Arcos, 1989). Recently, several molecular studies have 2.2. Molecular methods
addressed phylogenetic relationships of repleta species at different
phylogenetic levels (e.g. Rodriguez-Trelles et al., 2000; Oliveira Four mitochondrial and six nuclear primer pairs were used to
et al., 2005; Silva-Bernardi et al., 2006; Moran and Fontdevila, generate characters for phylogenetic analyses: primer sequences
2007; Robe et al., 2010). The most inclusive study of the repleta were previously published (see references below). One of the mito-
group so far is that of Durando et al. (2000) in which a phylogenetic chondrial regions includes partial sequences of both the small ribo-
hypothesis for 46 ingroup and six outgroup species was generated. somal RNA (srRNA) and large ribosomal RNA (lrRNA) and the
The overall basal relationships of this tree were poorly resolved complete tRNA-Val gene (srRNA–lrRNA; Oliveira et al., 2005). A sec-
and suggested paraphyly of the repleta and canalinea, dreyfusi and ond region is part of the mitochondrial Cytochrome c oxidase sub-
mesophragmatica species groups, although a few internal taxo- unit I (COI; Oliveira et al., 2005). A third region includes the
nomic groups were well resolved. complete mitochondrial Cytochrome c oxidase subunit II gene and
The ecology and biogeography of the repleta group has also pro- partial sequences of flanking tRNA-Leu and tRNA-Lys genes (COII;
vided insights into its origins and history. The adoption of cacti as Beckenbach et al., 1993). A fourth region is part of the mitochon-
breeding and feeding sites by many repleta group species is cer- drial NADH-ubiquinone oxidoreductase chain 2 (ND2; Oliveira
tainly one of the most extensive and successful ecological transi- et al., 2005). Partial sequences of the following six nuclear genes
tions in the genus, resulting in about 100 known species. The were also included: bride of sevenless (boss), sans fille (snf), Mito-
‘‘virilis-repleta’’ radiation forms a basal lineage within the subgenus chondrial assembly regulatory factor (Marf), seven in absentia (sina),
Drosophila appearing 25–36 million years ago (Mya; Throckmorton, fork head (fkh), and wee. Primer sequences for these nuclear regions
1975; Powell and DeSalle, 1995; Russo et al., 1995). Since the D. were reported in Bonacum et al. (2001).
repleta group is confined to the New World (apart from human Template DNA was extracted from 1 to 5 flies using a DNeasy
influences), it is likely the group arose in South America well after Extraction Kit (Qiagen, Valencia, CA), and loci of interest were ampli-
the origins of many of the major cactus groups when the interior fied using standard PCR protocols. Direct sequencing from purified
of the continent became warmer and drier due to the Andean uplift PCR products was performed on an ABI 3700 sequencer (PE Applied
ca. 17 Mya (Mauseth, 1990; Nyffeler, 2002). The centers of radiation Biosystems, Foster City, CA, USA). Sequences were corrected and
of the major cactus groups are located in Peru-Bolivia, the chaco compiled using Sequencer 4.7 (Gene Codes Corporation, Ann Arbor,
and caatinga of eastern South America, and possibly the Caribbean. MI, USA). All sequences generated for this study were deposited in
Based on the current distribution of the most generally ancestral GenBank under accession numbers JF736018–JF736503 (Table S1).
genera within the subfamilies Opuntoideae, Cactoideae, and Peres-
kioideae, the arid lands in Peru and Bolivia may be the centers of 2.3. Phylogenetic analyses
origin for all cacti (Edwards et al., 2005).
Here we present molecular phylogenetic analyses based on four Multiple sequence alignments were first adjusted by eye, and
mitochondrial and six nuclear gene regions from 62 ingroup and highly variable regions for which positional homology could not
nine outgroup taxa belonging to the virilis–repleta radiation be determined were manually excluded using MacClade 4.08
(Throckmorton, 1975; Tatarenkov and Ayala, 2001). Sampling in- (Maddison and Maddison, 2005). These regions were the introns
cluded five of the six proposed subgroups (only the inca subgroup of boss, snf, and Marf, a region of lrRNA, and most of the tRNA-Leu
was missing) and ca. 60% of the described species. We used the along with an intergenic spacer. The program Gblocks (Castresana,
resulting phylogenetic hypothesis to address several outstanding 2000) was used to further trim gapped regions, removing another
systematic and evolutionary problems: (1) with a monophyletic 5% of the nucleotide sites (parameters used were: minimum num-
repleta species group recovered, this phylogenetic hypothesis pro- ber of sequences for a conserved position = 36, maximum noncon-
vided groundwork for further systematic analysis of some sub- served positions = 8, and minimum length of a block = 10). This
groups and species complexes, (2) we present the first global matrix, 3957 bp in length and with 1204 parsimony-informative
dating of species divergence within the repleta group, (3) we characters (Table S2), was analyzed by maximum likelihood (ML)
mapped host cactus use onto the tree and show that there have based approaches for the estimation of the best data partitioning
been a number of phylogenetically independent host transitions scheme. Four alternative partition schemes were tested; (I) no par-
from Opuntia to the more chemically specialized columnar cacti, titioning of the data, (II) partitioning by gene (10 partitions), (III)
and (4) we mapped current species geographical locations onto partitioning codon positions 1 + 2 and 3 of mitochondrial and nu-
the tree but were unable to resolve a clear historical biogeography clear protein-coding regions separately, plus an additional partition
of these species. We discuss the evolution of host use and the geo- for the non-protein-coding mitochondrial sequences (5 partitions),
graphic origins of the repleta group. and (IV) the same as before but separating codon positions 1 and 2
in different partitions, nuclear and mitochondrial genes separately
(7 partitions). These analyses were performed with Treefinder (Jobb
2. Materials and methods et al., 2004), and the GTR+C substitution model was generally used
for all partitions. The best scheme according to both Akaike Infor-
2.1. Samples for molecular analyses mation Criterion (AIC) and Bayesian Information Criterion (BIC; as
in Sullivan and Joyce, 2005) was partition scheme IV (Table S3).
Data and specimens of ingroup and outgroup taxa included in Both ML and Bayesian (BI) analyses were accomplished using
this study (Table S1) were deposited in the Ambrose Monell Cryo the best partition scheme and partition-wise parameters estimated
Collection at the American Museum of Natural History, New York. from the data. ML searches were done in 20 independent runs
Most ingroup species were represented by one sample with a few using RAxML 7.2.6 (Stamatakis, 2006) and the GTR+C model was
exceptions. Drosophila canapalpa has been questioned as a valid used for each partition. Statistical support for nodes was obtained
species and its potential synonym, D. neorepleta, was also included with 200 bootstrap replicates and plotted on the best of the 20
in the ingroup species (Vilela, 1983; Wasserman, 1992). Further, trees obtained in the independent runs. BI searches were
 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544 535

performed with MrBayes (Huelsenbeck and Ronquist, 2005), using we set their lower bound of the estimated time for the divergence
2 independent runs of 10,000,000 generations each, with trees of D. mettleri and D. mulleri (15.9 Mya) as a constraint for this split.
sampled every 1000 generations. Convergence was checked with The second internal point was set using information from Russo
Tracer v1.5 (Rambaut and Drummond, 2003) and with the online et al. (1995) and Matzkin and Eanes (2003) on the divergence of
program AWTY (Wilgenbusch et al., 2004). The first 8000 sampled D. mojavensis and D. arizonae, ca. 1.2–4.2 Mya (lower and upper
trees were discarded as the burn in. The difference between ML bounds, respectively). Additionally, this dating method requires a
and BI trees was evaluated with the SH test (Shimodaira and prior for the root of age. Since divergence between the virilis and
Hasegawa, 1999) as implemented in RAxML 7.2.6 (Stamatakis, the repleta species groups was not included in either of the most
2006). Maximum Parsimony (MP) analyses were done using PAUP cited Drosophila divergence time estimates (Russo et al., 1995;
4.0b10 with random, stepwise addition and 1000 replicates Tamura et al., 2004), we had to choose a date based on available
(Swofford, 2002). For parsimony trees, clade support was assessed evidence. Divergence between the repleta group and the Hawaiian
with bootstrap (1000 replicates, Felsenstein, 1985, 1988) and Drosophila was estimated to have occurred 32 Mya, suggesting that
decay indices (Bremer, 1988), which were calculated using the virilis-repleta split occurred most likely after that (Russo et al.,
TreeRot.v3 (Sorenson, 1999). Incongruence between the phyloge- 1995). Spicer and Bell (2002) estimated a divergence date between
netic signal of the mitochondrial and nuclear sequence sets was D. arizonae (repleta group) and the species of the virilis group of
tested using the Incongruence Length Difference test (Farris approximately 20 Mya. Combining this information and using
et al., 1995). D. virilis and the nannoptera species as outgroups, we set a prior
on the base of the tree of 26 ± 6 Mya.
2.4. Divergence time estimation and biogeographical analysis In order to place these divergence estimates into a biogeo-
graphic context, we categorized available species distribution data
A Bayesian approach was used to date the nodes of the repleta for all species in our ML phylogenetic reconstruction. We catego-
group phylogeny with Multidistribute (Thorne and Kishino, 2002). rized all species into North America, Caribbean, or South America
This method employs a relaxed clock allowing independent rates distributions plus all combinations of these locations and used
among branches, and lower and upper hard bounds for calibration DIVA (Dispersal Vicariance Analysis; Ronquist, 1996, 1997) to as-
dates. Following the manual by Rutschmann (2004), parameters for sess whether there was evidence supporting a North American or
the F84 sequence evolution model were estimated for the BI topol- South American origin of the D. repleta group. We used the ML tree
ogy with the program baseml from PAML v 4.4 (Yang, 2007). These with maxarea = 2. Consistent with recent phylogenetic and biogeo-
parameters and the BI topology were used for estimating the graphic analyses (Griffith and Porter, 2009; Nyffeler and Eggli,
branch lengths with estbranches and the node ages with multidiv- 2010), we hypothesized a priori that South American species
time, using 5 million generations sampled every 100 generations. should constitute the majority of taxa at the base of our tree.
Since there is no reliable time calibration point available within
the repleta group, we used previously inferred divergence times to 2.5. Fly host use
calibrate a molecular clock. Two internal calibration points were
defined as follows. First, assuming that the Russo et al. (1995) Published and unpublished records of Drosophila host use, i.e.
dates were underestimated as suggested by Tamura et al. (2004), breeding substrates, were gathered for the species used in

Fig. 1. Map of Mexico and the southern USA showing the sites sampled for Drosophila species in this study.
536 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544

constructing our repleta group phylogeny, all ingroup and outgroup large proportion of informative characters (Durando et al., 2000;
taxa, and a few additional species not included in the phylogeny, Bull et al., 2003; Bergsten, 2005). Despite using six nuclear genes,
for a total of 75 species/subspecies (Table S4). For North American 67% of the informative characters in our data came from mtDNA
species, we collected Drosophila in 33 locations throughout Mexico, (Table S2). MP inferences are more prone to be confounded by
southern California, and Arizona (Fig. 1). Adult flies were collected homoplasic molecular characters. The partition scheme and GTR+C
over fermenting bananas in the field; additionally, fermenting model used for ML and BI methods appears to have sufficiently cor-
cactus tissues, rots, were returned to the laboratory and all rected the saturation problem. The number of changes per distance
emerging imagoes were recorded and identified to species. Com- unit for the combined matrix did not show signs of a plateau for
parison of species that were baited with those emerging from rots either transitions or transversions (Fig. S1). Further, incongruence
allowed us to determine the degree to which cacti were being used between mitochondrial and nuclear data partitions was significant
as hosts by Drosophila at each site. There were no apparent differ- (ILD test, p = 0.004) and may also have accounted for the weak ba-
ences in species collected by baiting and those reared from sal support. However, this test was done only in a MP framework
fermenting cactus rots at most of our collecting sites, other than and so was probably influenced by saturation and long-branch
the presence of non-cactophilic species attracted to baits (Etges, attraction as suggested earlier.
unpubl. data).
Host use was mapped onto the repleta group phylogeny using 3.1. Monophyly of the repleta species group
MacClade ver. 4.08 (Maddison and Maddison, 2005). Character
states were unweighted and considered unordered. Host use was The branch leading to the repleta group was highly supported
coded as Opuntia (1), columnar cactus (2), ‘‘polymorphic’’ (1 and by model-based inference methods (ML bootstrap = 98%, BI poster-
2), soil (3), or ‘‘other’’ (4) to determine the number of ecological ior probability = 1), contrary to a previous MP phylogeny that
transitions over the course of their history. ‘‘Other’’ included fer- failed to recover a monophyletic repleta group and suggested para-
menting fruit, flowers, and sap. We did not specify individual spe- phyly in relation to the canalinea, mesophragmatica, and dreyfusi
cies of Opuntia due to difficulties with field identifications, known groups (Durando et al., 2000). In the MP tree presented here,
hybrids, and the decision that as a genus, species designations D. pegasa, a Mexican species assigned to the repleta group based
were likely to be less ecologically important to the flies than for on chromosomal inversions (Wasserman, 1992), clustered with
the more chemically complex columnar cacti. Enumerating the D. canalinea, the only representative canalinea group species we
various tribes and subtribes of columnar cacti added little more could access. The poor resolution obtained by Durando et al.
resolution to the patterns of host switching (results not shown). (2000) appeared to be caused by saturation of third positions of
We also used Mesquite, ver. 2.6 (Maddison and Maddison, 2009) mitochondrial protein-coding genes leading to long branch attrac-
to examine the evolution of host plant use within the repleta spe- tion. This problem is likely still affecting our MP tree, in spite of the
cies group. We traced host use to reconstruct ancestral character extended dataset. Our MP tree, nevertheless, provided much better
states in a maximum parsimony framework. We performed 1000 resolution and separated the repleta group from the mesophragmat-
randomizations using the following options: Analysis: New Bar ica and dreyfusi groups. Therefore, we conclude that the present
and Line Chart for Trees: Randomly modify current tree: reshuffle data provided strong corroboration for monophyly of the repleta
current taxa: steps in character. Attempts to assess the evolution of group (Fig. 2).
host cactus use in a maximum likelihood context failed because
polymorphic or missing host use data are not currently supported 3.2. Monophyly and relationships among subgroups and species
by categorical data likelihood calculations in Mesquite. complexes

The relatedness of several species of previously defined lineages

3. Results within the repleta group was consistently corroborated (Wasserman,
1992; Fig. 2). All analyses showed strong support (MP boot-
A dataset of four mitochondrial and six nuclear DNA sequences strap > 98%, ML bootstrap = 100%, BI posterior probability = 1) for
was generated to examine the phylogenetic relationships of 58 monophyly of three of the five subgroups investigated, i.e. the
species of the Drosophila repleta species group in greater detail. fasciola, hydei, and mercatorum subgroups. The repleta subgroup
In addition, our taxon sampling included four subspecies, as previ- was not well supported due to intermingling with the mercatorum
ously described (see Material and Methods), and nine outgroup subgroup. The affinity of the mercatorum and repleta subgroups
belonging to the virilis–repleta radiation (sensu Throckmorton, was consistent across analyses and well supported (MP bootstrap
1975) for a total of 71 terminal taxa. The phylogenetic inference 90%, ML bootstrap = 97%, BI posterior probability = 1), providing
methods (ML, BI, and MP) recovered similar topologies and were further evidence for a previously implied relationship between
congruent with respect to the most well-supported lineages recov- these two subgroups (Wasserman, 1982, 1992; Tatarenkov and
ered by all three methods (Fig. 2). BI and ML trees were not statis- Ayala, 2001). The large mulleri subgroup remained polyphyletic,
tically different (SH test, D(LH) = 17.3 ± 16.5). consistent with Wasserman’s (1982, 1992) conclusions.
A well-supported, monophyletic repleta group was recovered Species complexes in the mulleri subgroup were analyzed sepa-
from the ML and BI analyses, but not the MP analysis. Most major rately to refine their systematic relationships. A monophyletic mul-
repleta group lineages, subgroups, and species complexes previ- leri subsection, including the mulleri, longicornis, and buzzatii
ously defined by shared chromosomal inversions (Wasserman, complexes matched the results obtained by Durando et al.
1992) were also recovered. Consistent resolution and/or support (2000). In our trees, the meridiana complex joined that subsection
for basal nodes, i.e. the overall relationships amongst species with good support. Monophyly of the mulleri, buzzatii, and meridi-
complexes and subgroups of the repleta group, were less clear ana complexes was corroborated (MP bootstrap 100%, ML boot-
and were the primary source of topological incongruence among strap = 100%, BI posterior probability = 1), but not that of the
trees based on different inference methods (Fig. 2). longicornis complex (Oliveira et al., 2005). The anceps (ML boot-
The lack of support at some hypothesized basal nodes is a com- strap = 79%, BI posterior probability = 0.98) and the eremophila
mon observation in molecular phylogenies. Saturation in third base (MP bootstrap 90%, ML bootstrap = 97%, BI posterior probabil-
positions and consequent long-branch attraction are common ity = 1) complexes were placed outside this monophyletic mulleri
explanations, especially when mitochondrial genes account for a subsection. Two unaffiliated species, D. nigricruria and D. pegasa,
 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544 537

Fig. 2. Molecular phylogenetic hypotheses for the Drosophila repleta species group. Yellow boxes delimit complexes of the mulleri subgroup, while green boxes delimit other
repleta subgroups. (A) The tree obtained by ML searches, with the GTR+C substitution model, and parameters estimated for each of the 7 data partitions (see text and Table S2
for details). Numbers on the node: top left ML bootstrap values > 50%, top right BI posterior probability > 0.7, bottom left MP bootstrap > 50%, bottom right, Bremer decay
values > 1. (B) The tree obtained by Bayesian Inference. (C) Strict consensus of the nine most parsimonious trees (steps = 7628, CI = 0.30, RC = 0.16). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)
538 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544

had uncertain positions consistent with previous chromosomal methodological differences mentioned above, this result is actually
analyses (Wasserman, 1992). in agreement with morphological and chromosomal data (Vilela,
The phylogenetic relationships among these strongly supported 1983; Wasserman, 1992). The paucity of informative characters
repleta group lineages described above, as well as the order of early may be an indication of rapid diversification early on in the evolu-
branching in the repleta phylogeny, remain unresolved. Besides the tionary history of the repleta group (Throckmorton, 1975).

Fig. 3. Results of DIVA (Dispersal Vicariance Analysis) with current geographic locality (-ies) of D. repleta group species and outgroups mapped onto the ML tree. See the text
for details.
 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544 539

Fig. 4. Host use and divergence times for the D. repleta group plotted onto the BI tree. Numbers by the nodes are the time estimates and the bars represent their 95%
confidence intervals. Host substrates are color coded. ‘‘Soil’’ refers to cactus exudate-soaked soils, and ‘‘other’’ refers to other substrates, but not cactus. The pictures illustrate
typical Opuntia and columnar cactus growth forms.
540 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544

3.3. Divergence times and biogeography of the Drosophila repleta parts of their species range (Fellows and Heed, 1972; Heed,
group 1982; Ruiz and Heed, 1988). Others, such as D. buzzatii, use one
host predominantly, but have been found repeatedly using alter-
Our global divergence time estimates for the repleta group re- nate hosts in low frequency. Opuntia is the main host for D. buzzatii
vealed that the split between the ancestor of the repleta group throughout its range in South America and the world where it has
and those of related species groups was approximately 20.9 Mya, been introduced (Carson and Wasserman, 1965), yet a small
with the ‘‘crown’’ repleta group ca. 16.3 Mya old. Based on Bayesian (3–6%), but repeatable percentage of flies have been reared from
methodology with calibration points taken from the literature Echinopsis terschekii in Argentina (Table S4) where they are sym-
(Russo et al., 1995; Spicer and Bell, 2002; Matzkin and Eanes, patric with D. koepferae (Hasson et al., 1992; Fanara et al., 1999).
2003; Tamura et al., 2004), this is the first estimate made for diver- Including both Opuntia and columnar cacti as hosts for D. buzzatii
gence dating in this group. By the mid-Miocene (12–15 Mya), all produced the results presented in Fig. 4 with all South American
major repleta lineages, including the subgroups and species com- buzzatii complex species having gained the use of columnar cacti
plexes described above, had already emerged. More closely related only once, including four columnar specialists, D. borborema,
species, often morphocryptic, diverged during the Pleistocene, of- D. serido, D. gouveai and D. uniseta (Fig. 4, Table S4).
ten less than 1 Mya, e.g. the Caribbean species triad of D. mayagu- Since current phylogenetic consensus suggests that the Opun-
ana, D. straubae, and D. parisiena (Heed and Grimaldi, 1991). tioideae are monophyletic (Griffith and Porter, 2009) and are sister
Ancestral area construction revealed poor resolution of basal to the Cactoideae, including columnar cacti (Nyffeler, 2002; Griffith,
taxa in our ML tree (Fig. 3). We tried to resolve the locations of 2004), we hypothesized that the switch to Opuntia use is ancestral,
poorly known species, and categorized several species that are with the use of columnar cacti for breeding representing the de-
now human commensals, i.e. D. virilis, D. repleta, and D. hydei, or rived state, among extant cactus-breeding members of the repleta
have been transported around the world with their host plants, group. Mapping host use onto the BI phylogeny revealed that
i.e. D. mercatorum and D. buzzatii based on locations of natural pop- Opuntia is generally the ancestral host and columnar cactus use is
ulations and of their closest relatives. We then tried to use Mes- a derived condition based on available data (Fig. 4). Further, the
quite ver. 2.6 (Maddison and Maddison, 2009) with both the MP phylogeny revealed multiple independent transitions, at least 10,
and ML trees, but there no improvement in the resolution of ances- from Opuntia to columnar cactus (Fig. 4). Results of ancestral char-
tral localities (results not shown) because of the number of equiv- acter state reconstruction in Mesquite (Maddison and Maddison,
ocal nodes due to the widespread distributions of some species 2009) indicated no phylogenetic structure in host use (P > 0.05).
(Fig. 3). Also, incomplete sampling of South American (SA) species Therefore, there was no evidence that the Opuntia to columnar cac-
in our outgroups, e.g. one species each in the canalinea and meso- tus switch has a phylogenetic component.
phragmatica groups, and species at the base of this tree, e.g. in In some clades, however, many species retained the ancestral
the fasciola group, probably inhibited clearer biogeographic resolu- state of Opuntia use in parts of their species ranges or have not
tion. However, most of these groups are either restricted to SA (D. completely specialized to columnar cacti. Loss of Opuntia use has
pavani of the mesophragmatica group) or are distributed in both SA occurred six times in the repleta group. Outside of the repleta spe-
and North America (NA), or in the case of the 14 known species in cies group, D. pavani and the nannoptera group independently ac-
the canalinea group (Stensmyr et al., 2008), in all three areas. Thus, quired the trait of being strict columnar breeders (Ward et al.,
current locations of the representatives of the more basal lineages 1975; Heed, 1982; Pitnick and Heed, 1994; Etges et al., 1999; Table
in this MP tree suggest that dispersal between SA, NA, and the S4, Fig. 4).
Caribbean was widespread early in the diversification of the
D. repleta group.
4. Discussion
3.4. Host use and host shifts
4.1. Molecular systematics of the Drosophila repleta group
Together with our collections of wild Drosophila (Fig. 1) and
available published records, a detailed record of host use for most A previous combined molecular and chromosomal MP phylog-
of the repleta group and some related outgroup species was assem- eny generated for the repleta group was characterized by a wide
bled (Table S4). In total, host use data was compiled for 63 species basal polytomy (Durando et al., 2000). This assortment encom-
of repleta and 10 species belonging to other species groups. Five passed the anceps and eremophila complexes of the mulleri sub-
repleta group species, D. antonietae, D. desertorum, D. gouveai, group, the fasciola, hydei and mercatorum subgroups, and several
D. seriema, and D. nigrohydei, and one species of the mesophragmat- non-repleta species from three other species groups, i.e. the cana-
ica group, D. gaucha, were not available for molecular analysis (Ta- linea, dreyfusi, and mesophragmatica groups, suggesting that repleta
ble S4). Comparing Opuntia to columnar breeders, 33.3% (21/63) group was not monophyletic. They concluded that the lack of res-
used only Opuntia species, 17.5% (11/63) of these repleta group spe- olution was caused by saturation of third positions in mitochon-
cies used only columnar cacti as hosts, and at least 25.4% (16/63) of drial protein coding genes and suggested that better taxon
these species used both types of hosts. Therefore, 50.8% of repleta sampling, increased numbers of parsimony informative characters,
group species are host specialists, at least with respect to this and more characters sampled from slower evolving nuclear gene
broad ecological division between flat leaf Opuntia and columnar regions should improve phylogenetic resolution (Durando et al.,
cacti. The use of soil refers to oviposition in fermenting cactus exu- 2000). To overcome this problem, we (1) increased the number
date-soaked soils that has evolved in all members of the D. eremo- of parsimony informative characters from 501 to 1204, including
phila complex (3/63). The remaining repleta species used other the addition of six slow evolving nuclear genes, (2) increased sam-
substrates (4/63) or breeding sites are unknown (8/63). pling from 54 to 71 taxa, and (3) used model-based inference
Polymorphism in character states can increase the uncertainty methods.
of mapping host use evolution. A recurring problem in this analysis The extended molecular phylogeny (Fig. 2) is largely concordant
for some Drosophila species was deciding when to add additional with previous analyses of morphology, biogeography, chromosomal
hosts based on rearing records, particularly for those species that gene arrangements and molecular data for the repleta group (Vilela,
are relatively unstudied. Some species, e.g. D. mojavensis, are well 1983; Wasserman, 1992; Durando et al., 2000) suggesting that we
known to be oligophagous, using different host cacti in different are moving closer to estimating the phylogenetic relationships of
 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544 541

this large Drosophila radiation. Our phylogenetic hypothesis is bet- Nyffeler and Eggli, 2010). The use of forest fruits is still observed
ter resolved and supported (including the MP tree) than previous in species representing the immediate outgroups of the repleta
attempts, and has allowed us to address several important evolu- group, i.e. D. camargoi, and D. canalinea; unfortunately, the ecol-
tionary problems for reconstructing a phylogeny of the repleta group ogy of the annulimana group remains poorly characterized. The
despite the persistent lack of support for some basal nodes. Thus, the overall ecology of fasciola subgroup species is not well character-
present study has strengthened support for monophyly of the repl- ized, but some have been reported to use fruits, flowers, and fungi
eta group due to the BI and ML inferred phylogenies that robustly (Wasserman, 1992). One member of this subgroup, D. onca, not
supported the repleta group separating it from its closest sister included here, is known to breed in Rhipsalis, a forest dwelling
groups. epiphytic cactus in Brazil (Pereira et al., 1983). Further, study of
The hydei and fasciola subgroups and the eremophila complex communities of yeasts responsible for tissue fermentation associ-
were recovered as the most basal repleta lineages (ML and BI trees; ated with fasciola group species D. carolinae, D. coroica, D. onca,
Fig. 2). The poorly studied fasciola subgroup was unfortunately and D. fascioloides revealed significant species differences from
underrepresented here due to difficulties in obtaining species. other forest dwelling drosophilids and more similar to yeast com-
Twenty-one species have been described for this subgroup from munities from known cactophilic Drosophila serido (Morais et al.,
tropical and cloud forests from Brazil to North America (reviewed 1995). Nyffeler and Eggli (2010) consider the 54 species in Rhip-
in Silva-Bernardi et al. (2006)). Wasserman (1992) proposed that salideae to be allied with the Cactoideae, so the humid-forest
the fasciola subgroup arose from a cytological ancestor of the mul- dweller condition of the fasciola subgroup could be either a de-
leri complex based on sharing of a single chromosome inversion. rived, reversed adaptation or the ancestral state for the entire
However, Diniz and Sene (2004) showed that this gene arrange- repleta group (Wasserman, 1962; Throckmorton, 1975; Vilela,
ment was a different inversion with breakpoints very close to those 1983; Diniz and Sene, 2004; Silva-Bernardi et al., 2006).
of the one found in the mulleri complex. Chromosome (Diniz and The use of Opuntia cactus species is common throughout the
Sene, 2004) and mtDNA COI gene-based (Silva-Bernardi et al., repleta phylogeny (Fig. 4), excluding the fasciola subgroup. Ecology
2006) phylogenies have resolved some of the relationships among of two of the three species in the eremophila complex, D. eremophila
smaller subsets of fasciola subgroup species but many of these taxa and D. micromettleri, is understudied, but they have been associated
are poorly known. Inclusion of the understudied and presumably with Opuntia exudate-soaked soils (Fogleman and Williams, 1987;
basal inca subgroup (Rafael and Arcos, 1989) and a larger sampling Heed, 1989; Table S4). Therefore, once the transition to the use of
of fasciola subgroup species may help to resolve the proper place- Opuntia cactus evolved, almost all species retained host breeding
ment of these basal repleta group lineages. affinities for cactus use. Since many repleta group species have been
The evolutionary association of the repleta and mercatorum sub- observed using seasonal cactus fruits for feeding and breeding
groups was apparent from chromosome inversions (Wasserman, (Etges and Heed, unpubl. data), this seems an obvious ecological
1982, 1992) and DNA phylogenies (Tatarenkov and Ayala, 2001). bridge from fruit use to fermenting pads and stem tissues. The hy-
All our analyses supported the clustering of the mercatorum and dei, mercatorum, and repleta subgroups species are restricted to
repleta subgroups (Fig. 2). Nevertheless, relationships within this Opuntia cactus with the exception of D. eohydei and D. repleta. As
clade were highly variable, including the placement of D. peninsu- D. repleta has become cosmopolitan and commensal with humans,
laris that has been considered to belong to either the mercatorum it has likely expanded its ancestral breeding site repertoire.
subgroup (Vilela, 1983) or the repleta subgroup (Wasserman, The use of the more chemically complex columnar cacti, as
1992). Therefore, merging the repleta and mercatorum subgroups reconstructed in our phylogeny, is derived relative to Opuntia. Most
may be phylogenetically justifiable. columnar cacti studied are characterized by significant concentra-
Wasserman (1982, 1992) considered the mulleri subgroup as a tions of a variety of different secondary chemicals including alka-
number of independent lineages with unclear evolutionary rela- loids, triterpene glycosides, fatty acids, and phytosterols while
tionships. According to the phylogenetic hypothesis presented Opuntia species typically lack appreciable amounts of these com-
here, a monophyletic mulleri subgroup would include four species pounds (Fogleman and Abril, 1990; Fogleman and Danielson,
complexes (Fig. 2). The meridiana complex is the basal lineage of 2001; Kircher, 1982). The transition to columnar cactus specializa-
this clade. The buzzatii complex is the next one to branch off, while tion has thus been a gradual one: the sister groups of almost all
the mulleri and the longicornis complexes are sister taxa and the columnar cactus specialists are either restricted to Opuntia or can
most recently diverged in the mulleri subgroup; the latter are pre- use both. Columnar cactus use has evolved in the mulleri-longicor-
dominantly North American with a few exceptions. The proposed nis lineage 6 times: in D. aldrichi, in the D. mojavensis cluster, in D.
longicornis complex does not appear to be monophyletic because longicornis, in D. huaylasi, in D. spenceri and D. hexastigma, and in
D. huckinsi and D. huichole clustered with the mulleri complex, in the mayaguana subcluster. The latter 3 cases have been accompa-
agreement with previous results (Oliveira et al., 2005; Fig. 2). nied by the loss of Opuntia use by D. huaylasi, D. parisiena, and
Two other mulleri species complexes, the eremophila and anceps the D. spenceri/D. hexastigma pair. Drosophila huaylasi has now been
complexes, and two additional species, D. nigricruria, and D. pegasa moved from the D. mojavensis cluster (Durando et al., 2000) to the
all occupy more basal relationships in our phylogeny with unclear mulleri cluster, and this species is known only from a few collec-
affinity to other mulleri complexes. tions in Peru reared only from smaller columnar-like cacti in the
genera Armatocereus and Neoraimondia (Table S4). In addition, pop-
4.2. Evolution of host use ulations of D. aldrichi in Texas and central-northern Mexico use
Opuntia exclusively (Patterson, 1943), but we discovered ‘‘D. aldri-
Use of fermenting cactus tissues as breeding substrates is the chi’’ using Myrtillocactus geometrizans in Tehuacan, Puebla and
hallmark of the repleta group (Carson, 1971, 2001; Heed, 1978). Pachycereus weberi in Cañón Zopilote, Guerrero, Mexico (Table
The evolution of the repleta group seems to be closely associated S4). In South America, ‘‘D. aldrichi’’ has also been reared from
with the transition from the use of fermenting fruits of non-cactus Armatocereus sp. (Suyo and Pilares, 1987). These disparate popula-
plants in moist forests to arid-adapted fleshy-stemmed desert tions of ’’D. aldrichi’’ are likely be different species given the degree
plants like Opuntia and other cacti. Interestingly, our divergence of reproductive isolation among some of them (Wasserman, 1992;
time dating suggested that the diversification of the main repleta Krebs and Barker, 1994), significant genetic differentiation of east-
group lineages occurred from 16 to 12 Mya (Fig. 4), which is close ern and western Mexican populations (Beckenbach et al., 2008;
to the estimate of the appearance of the Opuntioideae (15 Mya; Oliveira et al., 2008), and differentiation in patterns of host use.
542 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544

The transition to columnar cacti has also occurred in the buzzatii and the mostly North American mulleri complexes was dated to
complex and in the anceps complex. Subsequent switches away approximately 11.3 Mya. Furthermore, the phylogeny presented
from Opuntia use has occurred twice in the South American buzz- here suggests that the islands of the Caribbean were colonized inde-
atti complex, in D. uniseta within the northern martensis species pendently by repleta species at least five times as evidenced by the
cluster and in D. borborema, D. serido, and D. gouveai (Table S4) in geographic distributions of D. mulleri, the closely related triad of
the southern buzzatii cluster. Loss of Opuntia use has also occurred D. mayaguana, D. straubae and D. parisiena, D. stalkeri and D. richard-
in the D. anceps/D. nigrospiracula species pair (Heed, 1982). Overall, soni, D. peninsularis, and D. micromettleri. This history is, however,
columnar specialization has evolved at least seven times involving incomplete because we did not have access to the complete D. repl-
11 of the 65 species in the repleta group: D. huaylasi, D. parisiena, eta fauna in the Caribbean, e.g., D. paraguttata of the fasciola sub-
D. spenceri, D. hexastigma, D. borborema, D. serido, D. gouveai, group, a species known only from a single strain from Jamaica
D. uniseta, D. nigrospiracula, D. anceps, and D. mettleri (the latter (Wasserman, 1992). This distribution of sister species suggests that
species is included because it is associated with columnar cacti, dispersal into the Caribbean may have originated from both North
although it breeds in fermented cactus exudate-soaked soils at (e.g. D. mayaguana, D. straubae, and D. parisiena) and South America
the base of the plants). Therefore, ecological transitions to colum- (e.g. the stalkeri subcluster and the canalinea group), followed by
nar cacti are common, having occurred multiple times in different local inter- and intra-island diversification (Heed and Grimaldi,
clades from North and South America within the D. repleta group. 1991).
The origin of the repleta group cactus hosts and the discovery of
4.3. Origin and dispersal patterns of the Drosophila repleta group a broader South American drosophilid species diversity suggest a
South American origin for the repleta group. Taken together, the
Throckmorton (1975) proposed that Drosophila colonized the mass of historical and biogeographical data suggest that during
New World from Asia and went through a series of major species the mid-Miocene, in an isolated and drier South America, the repl-
radiations during the Miocene. In general, rapid species radiations eta group originated and quickly radiated along with its cactus
should occur soon after the origin of a group, likely related to eco- hosts. As South America moved northward, diversification of both
logical innovation and invasion of new niches (Kocher, 2004; Rokas cacti and the D. repleta group allowed colonization of North
et al., 2005; Hallstrom and Janke, 2008). The Mexican Trans-Volca- America and invasion of the Caribbean islands. During this process,
nic Region had been considered the center of diversification for the breeding site shifts from tropical, ephemeral fruits and flowers to
repleta group (Patterson and Stone, 1952; Throckmorton, 1975), epiphytic cacti and the widespread and abundant flat leaf Opuntia
where most D. repleta species were known due to earlier, more allowed further host plant shifts and specialization in many repleta
intensive collecting efforts in the USA and Mexico starting in the species to the more chemically complex columnar cacti.
1940s. Patterson and Stone (1952) pointed out that central and
southern Mexico (app. 19–22° N. latitude) ‘‘is of most interest for Acknowledgments
Drosophila distribution, for here are to be found 89 of the 391 spe-
cies known to occur in the Americas and Neotropical regions’’. The We are grateful to A. Alverson, A. Acurio, and D. Vela and two
D. repleta group has about half its known members in this zone’’. anonymous reviewers for comments on the manuscript. Financial
However, the diversity of D. repleta group species and relatives assistance was provided by grants from the National Science Foun-
has become apparent in South America with increased collecting dation DEB 01-29105 to R. DeSalle and P.M. O’Grady, INT-9724790
efforts (e.g. Brncic, 1957; Pereira et al., 1983; Vilela, 1983; Ruiz to W.J. Etges and W.B. Heed, and CONACyT and the Universidad
et al., 2000; Vela and Rafael, 2005; Acurio and Rafael, 2010). More Autónoma Metropolitana-Iztapalapa to M.A. Armella. F.C. Almeida
species continue to be described, e.g. the inca subgroup with three was funded by a Juan de la Cierva fellowship, Ministerio de Ciencia
described (Rafael and Arcos, 1989; Rafael and Vela, 2003; Vela and y Innovación, Spain. Molecular data were generated while D.C. Oli-
Rafael, 2005) and additional undescribed species from Peru and veira was an Ambrose Monell Research Fellow at the AMNH. This
Ecuador (Andrea Acurio pers. comm.). paper is dedicated to Bill Heed who inspired the study of repleta
Similarly, host plant information has also been heavily biased group breeding sites, provided unpublished host use data, and
towards samples from USA and Mexico so far preventing definitive helped identify species for this study.
biogeographical analysis of host use. Nevertheless, recent phyloge-
netic and systematic analyses of cacti in the Opuntioideae have
suggested a South American origin and that only late-diverging lin- Appendix A. Supplementary material
eages are represented in North America (Griffith and Porter, 2009;
Supplementary data associated with this article can be found,
Nyffeler and Eggli, 2010). Given the elevated species diversity and
endemism of Mexican columnar cacti, particularly in southern in the online version, at http://dx.doi.org/10.1016/j.ympev.
2012.05.012.
Mexico, this region likely represents a more recent center of radi-
ation of endemic columar cactus species, such as those in the gen-
era Carnegiea, Escontria, Lophocereus, Myrtillocactus, Pachycereus, References
Polaskia, and Stenocereus, all of which are used by various Drosoph-
Acurio, A.E., Rafael, V.L., 2009. Inventario taxonómico de drosophilidae (Diptera) en
ila species. el Parque Nacional Yasuni, Amazonia Ecuatoriana. Acta Amazon. 39, 713–718.
Despite the lack of a clear historical biogeography, the repleta Acurio, A.E., Rafael, V.L., 2010. Diversity and geographical distribution of Drosophila
group seems to be marked by repeated interchanges between South (Diptera, Drosophilidae) in Ecuador. Dros. Inform. Serv. 92, 20–25.
Almeida, F.C., DeSalle, R., 2008. Evidence of adaptive evolution of accessory gland
and North America, since most subgroups have representative spe- proteins in closely related species of the Drosophila repleta group. Mol. Biol.
cies in both continents (Fig. 3). Mapping species geographic distri- Evol. 25, 2043–2053.
butions on our phylogeny suggested multiple dispersal events Almeida, F., DeSalle, R., 2009. Orthology, function, and evolution of accessory gland
proteins in the Drosophila repleta group. Genetics 181, 235–245.
across the Isthmus of Panama in both directions (Fig. 3). At least
Barker, J.S.F., Starmer, W.T., 1982. Ecological Genetics and Evolution: The Cactus–
some of these intercontinental dispersal events may have occurred Yeast–Drosophila Model. Academic Press, New York.
long before the Panama Isthmus joined North and South America Barker, J.S.F., Starmer, W.T., MacIntyre, R.I., 1990. Ecological and Evolutionary
ca. 3 Mya, creating a land bridge for the Great American Biotic Inter- Genetics of Drosophila. Plenum Press, New York.
Beckenbach, A.T., Wei, W., Liu, H., 1993. Relationships in the Drosophila obscura
change (Stehli and Webb, 1985; Marshal, 1988). In one case, the species group, inferred from mitochondrial cytochrome oxidase II sequences.
split between the predominately South American buzzatii complex Mol. Biol. Evol. 10, 619–634.
 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544 543

Beckenbach, A.T., Heed, W.B., Etges, W.J., 2008. A mitochondrial DNA analysis of Heed, W.B., 1989. Origin of Drosophila of the Sonoran Desert revisited: In search of a
vicariant speciation in two lineages in the Drosophila mulleri subgroup. Evol. founder event and the description of a new species in the eremophila complex.
Ecol. Res. 10, 475–492. In: Giddings, L.V., Kaneshiro, K.Y., Anderson, W.W. (Eds.), Genetics, Speciation
Bergsten, J., 2005. A review of long-branch attraction. Cladistics 21, 163–193. and the Founder Principle. Oxford Univ. Press, pp. 253–278.
Bonacum, J., DeSalle, R., O’Grady, P., Oliveira, D., Wintermute, J., Zilversmit, M., 2001. Heed, W.B., Grimaldi, D.A., 1991. Revision of the morphocryptic, Caribbean
New nuclear and mitochondrial primers for systematics and comparative mayaguana species subcluster in the Drosophila repleta group (Diptera:
genomics in Drosophilidae. Dros. Inform. Serv. 84, 201–204. Drosophilidae). Am. Mus. Novit. 2999, 1–10.
Bremer, K., 1988. The limits of amino acid sequence data in angiosperm Huelsenbeck, J.P., Ronquist, F., 2005. Bayesian analysis of molecular evolution using
phylogenetic reconstruction. Evolution 42, 795–803. MrBayes. In: Nielsen, R. (Ed.), Statistical Methods in Molecular, Evolution, pp.
Brncic, D., 1957. Las especies Chilenas de Drosophilidae. Colec. Monografias Biol. 183–232.
Univ. Chile 8, 1–136. Jobb, G., von Haeseler, A., Strimmer, K., 2004. TREEFINDER: a powerful graphical
Bull, N.J., Schwarz, M.P., Cooper, S.J.B., 2003. Phylogenetic divergence of the analysis environment for molecular phylogenetics. BMC Evol. Biol. 4, 18.
Australian allodapine bees (Hymenoptera, Apidae). Mol. Phylogenet. Evol. 27, Kelleher, E.S., Swanson, W.J., Markow, T.A., 2007. Gene duplication and adaptive
212–222. evolution of digestive proteases in Drosophila arizonae female reproductive
Cáceres, M., Ranz, J.M., Barbadilla, A., Long, M., Ruiz, A., 1999. Generation of a tracts. PLoS Genet. 3 (8), e148.
widespread Drosophila inversion by a transposable element. Science 285, 415– Kircher, H.W., 1982. Chemical composition of cacti and its relationship to Sonoran
418. Desert Drosophila. In: Barker, J.S.F., Starmer, W.T. (Eds.), Ecological Genetics and
Carson, H.L., 1971. The ecology of Drosophila breeding sites. Harold L. Lyon Evolution. The Cactus–Yeast–Drosophila Model System. Academic Press,
Arboretum Lecture Number Two. University of Hawaii, Honolulu, pp. 1–27. Sydney, pp. 143–158.
Carson, H.L., 2001. Danko Brncic and the flower flies. Rev. Chil. Hist. Nat. 74, 33–40. Kocher, T.D., 2004. Adaptive evolution and explosive speciation: the cichlid fish
Carson, H.L., Wasserman, M., 1965. A widespread chromosomal polymorphism in a model. Nat. Rev. Genet. 5, 288–298.
widespread species, Drosophila buzzatii. Am. Nat. 99, 111–115. Krebs, R.A., Barker, J.S.F., 1994. Crosses between D. aldrichi from Australia and
Castresana, J., 2000. Selection of conserved blocks from multiple alignments for northwest Mexico suggest that they are different species. Dros. Inform. Serv. 75,
their use in phylogenetic analysis. Mol. Biol. Evol. 17, 540–552. 133–134.
Coyne, J.A., Orr, H.A., 1997. ‘‘Patterns of speciation in Drosophila’’ revisited. Maddison, D.R., Maddison, W.P., 2005. MacClade 4: Analysis of Phylogeny and
Evolution 51, 295–303. Character Evolution, Ver. 4.08. Sinauer, Sunderland, MA.
Diniz, N.M., Sene, F.M., 2004. Chromosomal phylogeny of the Drosophila fasciola Maddison, W.P., Maddison, D.R., 2009. Mesquite: A Modular System for
species subgroup revisited (Diptera, Drosophilidae). Genet. Mol. Biol. 27, 561– Evolutionary Analysis. Version 2.6.
566. Markow, T.A., O’Grady, P., 2006. Drosophila: A Guide to Species Identification and
Durando, C.M., Baker, R.H., Etges, W.J., Heed, W.B., Wasserman, M., DeSalle, R., 2000. Use. Academic Press, New York.
Phylogenetic analysis of the repleta species group of the genus Drosophila using Marshal, L.G., 1988. Land mammals and the Great American Interchange. Am. Sci.
multiple sources of characters. Mol. Phylogenet. Evol. 16, 296–307. 76, 380–388.
Edwards, E.J., Nyffeler, R., Donoghue, M.J., 2005. Basal cactus phylogeny: Matzkin, L.M., Eanes, W.F., 2003. Sequence variation of alcohol dehydrogenase (Adh)
implications of Pereskia (Cactaceae) paraphyly for the transition to the cactus paralogs in cactophilic Drosophila. Genetics 163, 181–194.
life form. Am. J. Bot. 92, 1177–1188. Mauseth, J.D., 1990. Continental drift, climate and the evolution of cacti. Cact. Succ.
Etges, W.J., Jackson, L.L., 2001. Epicuticular hydrocarbon variation in Drosophila J. 62, 302–308.
mojavensis cluster species. J. Chem. Ecol. 27, 2125–2149. Morais, P.B., Rosa, C.A., Hagler, A.N., Mendonca-Hagler, L.C., 1995. Yeast
Etges, W.J., Johnson, W.R., Duncan, G.A., Huckins, G., Heed, W.B., 1999. Ecological communities as descriptors of habitat use by the Drosophila fasciola subgroup
genetics of cactophilic Drosophila. In: Robichaux, R. (Ed.), Ecology of Sonoran (repleta group) in Atlantic rain forests. Oecologia 104, 45–51.
Desert Plants and Plant Communities. University of Arizona Press, Tucson, pp. Moran, T., Fontdevila, A., 2007. On the phylogeny of the Drosophila hydei subgroup:
164–214. new insights from combined analyses of nuclear and mitochondrial data. Mol.
Etges, W.J., Armella, M.A., O’Grady, P., Heed, W.B., 2001. Two new species of Phylogenet. Evol. 43, 1198–1205.
Drosophila (Diptera: Drosophilidae) in the repleta group from Mexico. Ann. Negre, B., Casillas, S., Suzanne, M., Sánchez-Herrero, E., Akam, M., Nefedov, M.,
Entomol. Soc. Am. 94, 16–20. Barbadilla, A., de Jong, P., Ruiz, A., 2005. Conservation of regulatory sequences
Etges, W.J., de Oliveira, C.C., Noor, M.A.F., Ritchie, M.G., 2010. Genetics of incipient and gene expression patterns in the disintegrating Drosophila Hox gene
speciation in Drosophila mojavensis. III. Life history divergence and reproductive complex. Genome Res. 15, 692–700.
isolation. Evolution 64, 3549–3569. Nyffeler, R., 2002. Phylogenetic relationships in the cactus family (Cactaceae) based
Fanara, J.J., Fontdevila, A., Hasson, E., 1999. Oviposition preference and life history on evidence from trnK/matK and trnL-trnF sequences. Am. J. Bot. 89, 312–326.
traits in cactophilic Drosophila koepferae and D. buzzatti in association with their Nyffeler, R., Eggli, U., 2010. A farewell to dated ideas and concepts – molecular
natural hosts. Evol. Ecol. 13, 173–190. phylogenetics and a revised suprageneric classification of the family Cactaceae.
Farris, J.S., Kallersjo, M., Kluge, A.G., Bult, C., 1995. Constructing a significance test Schumannia 6, 109–149.
for incongruence. Syst. Biol. 44, 570–572. Oliveira, D.C.S.G., O’Grady, P.M., Etges, W.J., Heed, W.B., DeSalle, R., 2005. Molecular
Fellows, D.P., Heed, W.B., 1972. Factors affecting host plant selection in desert- systematics and geographical distribution of the Drosophila longicornis species
adapted cactiphilic Drosophila. Ecology 53, 850–858. complex (Diptera: Drosophilidae). Zootaxa 1069, 1–32.
Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the Oliveira, D.C.S.G., Leonidas, M., Etges, W.J., O’Grady, P.M., DeSalle, R., 2008. Species
bootstrap. Evolution 39, 783–791. delimitation in the Drosophila aldrichi subcluster (Diptera: Drosophilidae) using
Felsenstein, J., 1988. Phylogenies from molecular sequences: inference and DNA sequences. Zootaxa 1725, 37–47.
reliability. Annu. Rev. Genet. 22, 521–565. Patterson, J.T., 1943. The Drosophilidae of Southwest. Univ. Tex. Publ. 4314, 7–203.
Fogleman, J.C., Abril, J.R., 1990. Ecological and evolutionary importance of host plant Patterson, J.T., Stone, W.S., 1952. Evolution in the Genus Drosophila. MacMillan, New
chemistry. In: Barker, J.S.F., Starmer, W.T., MacIntyre, R.J. (Eds.), Ecological and York.
Evolutionary Genetics of Drosophila. Plenum, New York, pp. 121–141. Pereira, M.A.Q.R., Vilela, C.R., Sene, F.M., 1983. Notes on breeding and feeding sites
Fogleman, J.C., Danielson, P.B., 2001. Chemical interactions in the Cactus– of some species of the repleta group of the genus Drosophila (Diptera,
Microorganism–Drosophila model system of the Sonoran Desert. Am. Zool. 41, Drosophilidae). Cienc. Cult. (São Paulo) 35, 1313–1319.
877–889. Pitnick, S., Heed, W.B., 1994. New species of cactus-breeding Drosophila (Diptera:
Fogleman, J., Williams, J., 1987. Oviposition site preference of cactophilic Drosophila Drosophilidae) in the nannoptera species group. Ann. Entomol. Soc. Am. 87,
in the eremophila complex. Dros. Inform. Serv. 66, 51–52. 307–310.
Gibbs, A.G., Matzkin, L.M., 2001. Evolution of water balance in the genus Drosophila. Powell, J.R., DeSalle, R., 1995. Drosophila molecular phylogenies and their uses. Evol.
J. Exp. Biol. 204, 2331–2338. Biol. 28, 87–138.
Gibbs, A.G., Perkins, M.C., Markow, T.A., 2003. No place to hide: microclimates of Rafael, V., Arcos, G., 1989. Subgrupo inca, un nuevo subgrupo del grupo repleta, con
Sonoran Desert Drosophila. J. Therm. Biol. 28, 353–362. descripción de Drosophila huancavilcae n. sp (Diptera, Drosophilidae). Evol. Biol.
Griffith, M.P., 2004. What did the first cactus look like? An attempt to reconcile the 3, 233–243.
morphological and molecular evidence. Taxon 53, 493–499. Rafael, V., Vela, D., 2003. Drosophila yangana sp. nov. un nuevo miembro del grupo
Griffith, M.P., Porter, J.M., 2009. Phylogeny of Opuntioideae (Cactaceae). Int. J. Plant repleta, subgrupo inca (Diptera: Drosophilidae). Rev. Pontificia Univ. Católica
Sci. 170, 107–116. Ecuador 71, 129–139.
Hallstrom, B.M., Janke, A., 2008. Resolution among major placental mammal Rambaut, A., Drummond, A.J., 2003. Tracer: MCMC Trace Analysis Package.
interordinal relationships with genome data imply that speciation influenced Robe, L.J., Loreto, E.L.S., Valente, V.L.S., 2010. Radiation of the ‘‘Drosophila’’ subgenus
their earliest radiations. BMC Evol. Biol. 8, 162. (Drosophilidae, Diptera) in the Neotropics. J. Zool. Syst. Evol. Res. 48, 310–321.
Hasson, E., Naviera, H., Fontdevila, A., 1992. The breeding sites of the Argentinian Rodriguez-Trelles, F., Alarcon, L., Fontdevila, A., 2000. Molecular evolution and
species of the Drosophila mulleri complex (subgenus Drosophila repleta group). phylogeny of the buzzatii complex (Drosophila repleta group): a maximum-
Rev. Chil. Hist. Nat. 65, 319–326. likelihood approach. Mol. Biol. Evol. 17, 1112–1122.
Heed, W.B., 1978. Ecology and genetics of Sonoran Desert Drosophila. In: Brussard, Rokas, A., Kruger, D., Carroll, S.B., 2005. Animal evolution and the molecular
P.F. (Ed.), Ecological Genetics: The Interface. Springer-Verlag, New York, pp. signature of radiations compressed in time. Science 310, 1933–1938.
109–126. Ronquist, F., 1996. DIVA 1.1. Computer Program and Manual Available by
Heed, W.B., 1982. The origin of Drosophila in the Sonoran Desert. In: Barker, J.S.F., Anonymous FTP from Uppsala University (ftp.sysbot.uu.se).
Starmer, W.T. (Eds.), Ecological Genetics and Evolution: The Cactus–Yeast– Ronquist, F., 1997. Dispersal–Vicariance Analysis: a new approach to the
Drosophila Model System. Academic Press, Sydney, pp. 65–80. quantification of historical biogeography. Syst. Biol. 46, 195–203.
544 D.C.S.G. Oliveira et al. / Molecular Phylogenetics and Evolution 64 (2012) 533–544

Ruiz, A., Heed, W.B., 1988. Host plant specificity in the cactophilic Drosophila mulleri Wilgenbusch, J.C., Warren, D.L., Swofford, D.L., 2004. AWTY: A system for graphical
species complex. J. Anim. Ecol. 57, 237–249. exploration of MCMC convergence in Bayesian phylogenetic inference. http://
Ruiz, A., Heed, W.B., Wasserman, M., 1990. Evolution of the mojavensis cluster of ceb.csit.fsu.edu/awty.
cactophilic Drosophila with descriptions of two new species. J. Hered. 81, 30– Yang, Z., 2007. PAML 4: phylogenetic analysis by maximum likelihood. Mol. Biol.
42. Evol. 24, 1586–1591.
Ruiz, A., Cansian, A.M., Kuhn, G.C.S., Alves, M.A.R., Sene, F.M., 2000. The Drosophila
serido speciation puzzle: putting new pieces together. Genetica 108, 217–227.
Russo, C.A., Takezaki, N., Nei, M., 1995. Molecular phylogeny and divergence times
Further reading
of Drosophilid species. Mol. Biol. Evol. 12, 391–404.
Rutschmann, F., 2004. Bayesian Molecular Dating using PAML/Multidivtime. A Step- Benado, M., 1989. Competitive release in the cactophilic fly, Drosophila venezolana.
by-step Manual. University of Zurich, Zurich. Ecotropicos 2, 45–48.
Shimodaira, H., Hasegawa, M., 1999. Multiple comparisons of log-likelihoods with Benado, M., Montero, C., 1988. Competition between cactophilic species Drosophila
applications to phylogenetic inference. Mol. Biol. Evol. 16, 1114–1116. starmeri and D. uniseta. Rev. Chil. Hist. Nat. 61, 187–190.
Silva-Bernardi, E.C.C., Morales, A.C., Sene, F.M., Manfrin, M.H., 2006. Phylogenetic Escalante, A., Benado, M., 1990. Predation on the cactophilic fly, Drosophila starmeri,
relationships in the Drosophila fasciola species subgroup (Diptera, in the columnar cactus, Pilosocereus lanuginosus. Biotropica 22, 48–50.
Drosophilidae) inferred from partial sequences of the mitochondrial Fogleman, J.C., Hackbarth, K.R., Heed, W.B., 1981. Behavioral differentiation
cytochrome oxidase subunit I (COI) gene. Genet. Mol. Biol. 29, 566–571. between two species of cactophilic Drosophila. III. Oviposition site preference.
Sorenson, M.D., 1999. TreeRot, Version 2. Boston University, Boston, MA. Am. Nat. 118, 541–548.
Spicer, G.S., Bell, C.D., 2002. Molecular Phylogeny of the Drosophila virilis Species Fontdevila, A., Pla, C., Hasson, E., Wasserman, M., Sanchez, A., Naveira, H., Ruiz, A.,
Group (Diptera: Drosophilidae) Inferred from Mitochondrial 12S and 16S 1988. Drosophila koepferae: A new member of the Drosophila serido (Diptera:
Ribosomal RNA Genes. Ann. Entomol. Soc. Am. 95, 156–161. Drosophilidae) superspecies taxon. Ann. Entomol. Soc. Am. 81, 380–385.
Stamatakis, A., 2006. RAxML-VI-HPC: maximum likelihood-based phylogenetic Fontdevila, A., Wasserman, M., Pla, C., Pilares, L., Armengol, R.d., Suyo, M.d.P.,
analyses with thousands of taxa and mixed models. Bioinformatics 22, 2688– Sanchez, A., Vasquez, J., Ruiz, A., Garcia, J.L., 1990. Description and evolutionary
2690. relationships of two species of the Drosophila mulleri cluster (Diptera:
Stehli, F.G., Webb, S.D., 1985. The Great American Biotic Interchange. Plenum Press, Drosophilidae). Ann. Entomol. Soc. Am. 83, 444–452.
New York. Gibson, A.C., Horak, K.E., 1978. Systematic anatomy and phylogeny of Mexican
Stensmyr, M.C., Stieber, R., Hansson, B.S., 2008. The Cayman crab fly revisited – columnar cacti. Ann. Miss. Bot. Gard. 65, 999–1057.
phylogeny and biology of Drosophila endobranchia. PLoS One 3, e1942. Goñi, B., Martinez, M.E., Valente, V.L.S., Vilela, C.R., 1998. Preliminary data on the
Sullivan, J., Joyce, P., 2005. Model selection in phylogenetics. Annu. Rev. Ecol. Evol. Drosophila species (Diptera, Drosophilidae) from Uruguay. Rev. Bras. Entomol.
Syst. 36, 445–466. 42, 131–140.
Suyo, M.P.P., Pilares, L., 1987. Drosophila novemaristata: descripcion, taxonomia y Heed, W.B., 1957. Ecological and distributional notes on the Drosophilidae (Diptera)
nuevas localidades peruanas. Rev. Peru. Entomol. 30, 61–64. of El Salvador. Univ. Tex. Publ. 5721, 62–78.
Swofford, D.L., 2002. PAUP: Phylogenetic Analysis Using Parsimony (and other Heed, W.B., 1977. A new cactus-feeding but soil-breeding species of Drosophila
methods), Version 4.0 beta. Smithsonian Institution, Washington DC. (Diptera: Drosophilidae). Proc. Entomol. Soc. Wash. 79, 649–654.
Tamura, K., Subramanian, S., Kumar, S., 2004. Temporal patterns of fruit fly Heed, W.B., Castrezana, S., 2008. Drosophila sonorae (Diptera, Drosophilidae), a new
(Drosophila) evolution revealed by mutation clocks. Mol. Biol. Evol. 21, 36–44. species in the repleta species group from Mexico. Zootaxa 1725, 27–36.
Tatarenkov, A., Ayala, F.J., 2001. Phylogenetic relationships among species groups of Heed, W.B., Heed, S.R., 1972. Ecology, weather, and dispersal of Drosophila on an
the virilis–repleta radiation of Drosophila. Mol. Phylogenet. Evol. 21, 327–331. island mountain. Dros. Inform. Serv. 48, 100–101.
Thorne, J.L., Kishino, H., 2002. Divergence time and evolutionary rate estimation Kircher, H.W., Heed, W.B., Russell, J.S., Grove, J., 1967. Senita cactus alkaloids: their
with multilocus data. Syst. Biol. 51, 689–702. significance to Sonoran Desert Drosophila ecology. J. Insect Physiol. 13, 1869–
Throckmorton, L., 1975. The phylogeny, ecology, and geography of Drosophila. In: 1874.
King, R. (Ed.), Handbook of Genetics. Plenum, New York, pp. 421–469. Manfrin, M.H., Sene, F.M., 2006. Cactophilic Drosophila in South America: a model
Vela, D., Rafael, V., 2005. Catorce nuevas especies del género Drosophila (Diptera, for evolutionary studies. Genetica 126, 57–75.
Drosophilidae) en el Bosque húmedo montano del volcán Pasochoa, Pichincha. Manríquez, G., Benado, M., 1994. Echinopsis chilensis (Friedrich et Rowland): an
Ecuador. Rev. Ecuat. Med. Cienc. Biol. 27, 28–41. endemic breeding site for Drosophila pavani Brncic 1957. Rev. Chil. Entomol. 21,
Vilela, C.R., 1983. A revision of the Drosophila repleta species group (Diptera, 185–186.
Drosophilidae). Rev. Bras. Entomol. 27, 1–114. Morais, P.B., Rosa, C.A., Hagler, A.N., Mendoca-Hagler, L.C., 1994. Yeast communities
Wagstaff, B.J., Begun, D.J., 2005. Molecular population genetics of accessory gland of the cactus Pilosocereus arrabidae as resources for larval and adult stages of
protein genes and testis-expressed genes in Drosophila mojavensis and D. Drosophila serido. Antonie van Leeuw. 66, 313–317.
arizonae. Genetics 171, 1083–1101. Morais, P.B., Lachance, M.-A., Rosa, C.A., 2005. Saturnispora hagleri sp. nov., a yeast
Wagstaff, B.J., Begun, D.J., 2007. Adaptive evolution of recently duplicated accessory species isolated from Drosophila flies in Atlantic rainforest in Brazil. Int. J. Syst.
gland protein genes in desert Drosophila. Genetics 177, 1023–1030. Evol. Microbiol. 55, 1725–1727.
Ward, B.L., Starmer, W.T., Russell, J.S., Heed, W.B., 1975. The correlation of climate Pipkin, S.B., 1965. The influence of adult and larval food habits on population size of
and host plant morphology with a gradient of an inversion polymorphism in neotropical ground-feeding Drosophila. Am. Midl. Nat. 74, 1–27.
Drosophila pachea. Evolution 28, 565–575. Richardson, R.H., Smouse, P.E., Richardson, M.E., 1977. Patterns of molecular
Wasserman, M., 1962. Cytological studies of the repleta group of the genus variation. II. Associations of electrophoretic mobility and larval substrate with
Drosophila. IV. The hydei subgroup. Univ. Tex. Publ. 6205, 73–84. species of the Drosophila mulleri complex. Genetics 85, 141–154.
Wasserman, M., 1982. Evolution and speciation in selected species groups. Ruiz, A., Fontdevila, A., Wasserman, M., 1982. The evolutionary history of Drosophila
Evolution of the repleta group. In: Ashburner, M., Carson, H.L., Thompson, J.N., buzzatii III. Cytogenetic relationships between two sibling species of the buzzatii
Jr. (Eds.), The Genetics and biology of Drosophila. Academic Press, London, pp. cluster. Genetics 101, 503–518.
61–139. Valadão, H., Hay, J.D.V., Tidon, R., 2010. Temporal dynamics and resource
Wasserman, M., 1992. Cytological evolution of the Drosophila repleta species group. availability for drosophilid fruit flies (Insecta, Diptera) in a gallery forest in
In: Krimbas, C.B., Powell, J.R. (Eds.), Drosophila Inversion Polymorphism. CRC the Brazilian savanna. Int. J. Ecol. 2010, 1–7.
Press, Boca Raton, FL, pp. 455–552. Vilela, C.R., Sene, F.M., 1977. Two new Neotropical species of the repleta group of the
Wharton, L.T., 1944. Interspecific hybridzation in the repleta group. Univ. Tex. Publ. genus Drosophila (Diptera: Drosophilidae). Pap. Avulsos de Zool. (São Paulo) 30,
4445, 175–193. 295–297.