Molecular phylogeny of the forensically important genus Cochliomyia (Diptera: Calliphoridae)

Abstract Cochliomyia Townsend includes several abundant and one of the most broadly distributed, blow flies in the Americas, and is of significant economic and forensic importance. For decades, Cochliomyia hominivorax (Coquerel) and Cochliomyia macellaria (Fabricius) have received attention as livestock parasites and primary indicator species in forensic entomology. However, Cochliomyia minima Shannon and Cochliomyia aldrichi Del Ponte have only been subject to basic taxonomy and faunistic studies. Here we present the first complete phylogeny of Cochliomyia including numerous specimens per species, collected from 13 localities in the Caribbean. Four genes, the mitochondrial COI and the nuclear EF-1α, 28S rRNA, and ITS2, were analyzed. While we found some differences among gene trees, a concatenated gene matrix recovered a robustly supported monophyletic Cochliomyia with Compsomyiops Townsend as its sister group and recovered the monophyly of Cochliomyia hominivorax, Cochliomyia macellaria and Cochliomyia minima. Our results support a close relationship between Cochliomyia minima and Cochliomyia aldrichi. However, we found Cochliomyia aldrichi containing Cochliomyia minima, indicating recent speciation, or issues with the taxonomy of the group. We provide basic information on habitat preference, distribution and feeding habits of Cochliomyia minima and Cochliomyia aldrichi that will be useful for future forensic studies in the Caribbean.


Introduction
Cochliomyia Townsend is endemic to the Americas and includes only four species: Cochliomyia minima Shannon, C. aldrichi Del Ponte, C. macellaria (Fabricius) and C. hominivorax (Coquerel). All of them are flesh eaters during their larval stage and are locally abundant. In particular, Cochliomyia macellaria is one of the most broadly distributed blow flies in the New World (Whitworth 2010). These species vary in habitat preference, feeding habits, dispersal abilities, and morphology among the species (Hall 1948, Whitworth 2010. For instance, C. aldrichi, C. minima and C. macellaria are primarily carrion feeders, while, C. hominivorax is an obligate parasite of mammals (Hall 1948, Stevens and Wallman 2006, Mc-Donagh and Stevens 2011. Cochliomyia hominivorax and C. macellaria have been intensely studied due to their commercial and forensic importance. Cochliomyia macellaria is one of the most forensically important species commonly found on decomposing remains. This species is considered important for post mortem interval estimations (Smith 1986, Byrd andCastner 2010) being among the first species to colonize corpses. In contrast, Cochliomyia hominivorax is an obligate parasite with its larvae producing myiasis and feeding on living tissue (Hall 1948, Guimaraes et al. 1983). This species is one of the most important insect pests of livestock in the Neotropics causing economic losses of billions of dollars every year (Vargas-Terán et al. 2005). Both species are common throughout the year in tropical, warm and humid areas (Hall 1948). Cochliomyia macellaria can be found in temperate climates from Canada to Argentina during the summer months (Whitworth 2010). Cochliomyia hominivorax initially ranged from southern United States to northern Argentina (Guimaraes et al. 1983) but has been eradicated from North America, Central America, Puerto Rico and the Virgin Islands (Vargas-Terán et al. 2005). It is worth noting that in 1988 this species was introduced in Libya and it was successfully eradicated in 1992 based on the sterile insect technique (SIT). This was the major international effort and avoid a major disaster for the livestock industry of Africa and Southern Europe (Lindquist et al. 1992). Despite those successfully eradications C. hominivorax continues to be an economically important pest in South America and parts of the Caribbean (Vargas-Terán et al. 2005).
The other two congeners, C. minima and C. aldrichi, are poorly known and research has been limited to descriptive morphology and faunistics (Hall 1948, Dear 1985, Whitworth 2010. These two species are restricted to the West Indies and C. aldrichi has been reported in the Florida Keys (Whitworth 2010). Dear (1985) listed C. minima for the Florida Keys, however Whitworth (2010) concluded that Dear mistakenly identified one C. aldrichi specimen as C. minima. Forensically important insects in the Caribbean are generally understudied and these two species have not played an important role in forensic entomology. Yet, due to their abundance and broad distribution in this region, including Cuba, Dominican Republic, Jamaica, Puerto Rico, Virgin Islands, Bahamas and Cayman Islands (Hall 1948, Dear 1985, Whitworth 2010 they have an enormous forensic potential. For example, recent studies conducted in Puerto Rico showed that C. minima is abundant and widely distributed on the island, and that adults are attracted to, and feed on, carrion (Yusseff-Vanegas 2014).
Although the adult morphology of the four species is well known (Hall 1948, Dear 1985, Whitworth 2010, studies on the relationship among Cochliomyia species have not been conducted yet. Morphological studies have provided synapomorphies of Cochliomyia that clearly diagnose it from all other Calliphoridae (Hall 1948;Dear 1985;Whitworth 2010). These include short and filiform palpus and phallus with extremely elongated paraphallus and a complex distiphallus (Dear 1985, Figs 37-44). Prior studies on the relationships among Calliphoridae (Stevens 2003, Harvey et al. 2008, McDonagh and Stevens 2011 and the subfamily Chrysomyinae (Singh and Wells 2011), including C. macellaria and C. hominivorax, supported Cochliomyia monophyly, and placed it as sister to Compsomyiops Townsend. However, the monophyly of the genus has not been formally tested with thorough sampling of all species, and the relationships among its species remain unknown. Furthermore, DNA-based methods can provide reliable identification of specimens by non-experts and will be particularly important for the identification of larval stages of C. minina and C. aldrichi that remain poorly known. For example, only the third instar of C. minima has been described (Yusseff-Vanegas 2014).
Here we provide a robust phylogenetic hypothesis of Cochliomyia based on four genes sequenced from 38 individuals collected throughout the Caribbean, including for the first time molecular data about C. minima and C. aldrichi. Our main goals are to test the monophyly of this genus and the validity of, and relationships among, its species.

Specimens and DNA extraction
A total of 44 specimens were included in this study, 38 representing the ingroup plus six outgroup species [Chrysomya megacephala (Fabricius), C. rufifacies (Macquart), Hemilucilia sp., Lucilia cuprina (Wiedemann), Compsomyiops fulvicrura (Robineau-Desvoidy) and Compsomyiops callipes (Bigot)]. All sequences used here are new except for Compsomyiops fulviclura and C. callipes (Table 1). The specimens were collected in the Caribbean (Jamaica, Cuba, Dominican Republic, Puerto Rico, Saint Barts Martinique and Dominica) from 2011 to 2013 and in the following countries, Colombia (2014), Florida (2013) and Mexico (2010 and 2012) ( Table 1). All specimens were killed and preserved in 95% ethanol and stored at -20 °C. The adults were examined with a Leica MZ16 stereomicroscope and identified using the Whitworth (2010) keys. The DNA was isolated from thoracic muscle or two legs of each individual with the QIAGEN DNeasy Tissue Kit (Qiagen, Inc., Valencia, CA). Voucher specimens were deposited at the UVM Natural History Museum (in the Zadock Thompson Zoological Collections) and sequences were submitted to GenBank.

PCR amplification and sequencing
We amplified regions of three nuclear loci: the protein coding elongation factor-1 alpha (EF-1α), the ribosomal 28S, and internal transcribed spacer 2 (ITS2), plus the mitochondrial protein coding cytochrome oxidase I (COI). The primer sequences are listed in  (Friedrich and Tautz 1997). Amplified fragments were sequenced in both directions by University of Arizona Genetics Core. Sequences were interpreted from chromatograms using Phred (Green and Ewing 2002) and Phrap (Green 1999, Green andEwing 2002) using the Chromaseq module (Maddison and Maddison 2010a) in the evolutionary analysis program Mesquite 3.03 (Maddison and Maddison 2010b) with default parameters. The sequences were then proofread by examining chromatograms by eye. Alignments were done using MAFFT (Katoh et al. 2002) through the online portal EMBL-EBI. The gene matrices were then concatenated in Mesquite 3.03 (Maddison and Maddison 2010b) and the full aligned data set is 3368 bp.

Phylogenetic analysis
We partitioned each gene and codon position for a total of eight partitions that were exported from Mesquite for model choice and the appropriate models were chosen using jModeltest v2.1.4 (Posada and Crandall 1998), and the AIC criterion (Posada and Buckley 2004 (Zwickl 2006) using the same partitioning scheme and models.

Results
The phylogenetic analyses of the concatenated matrix, either using Bayesian or maximum likelihood approaches, recovered a generally well supported monophyletic Cochliomyia (Fig. 1). Cochliomyia macellaria, C. hominivorax and C. minima were recovered as monophyletic, while C. aldrichi was recovered as paraphyletic. Independent analyses of 28S and ITS2 supported the monophyly of Cochliomyia, while COI and EF-1α recovered it as a paraphyletic group (Suppl. material 1). At the species level, EF-1α and 28S had limited signal and did not distinguish between C. minima and C. aldrichi. COI recovered the monophyly of C. minima, but did not resolve relationships among C. aldrichi and C. macellaria. ITS2 fully resolved the relationships within Cochliomyia, and is the only gene that recovered the monophyly of C. aldrichi. Despite of the incongruence detected among the four gene trees, they all recovered monophyletic C. hominivorax and three of the four genes (COI, 28S and ITS2) strongly supported a monophyletic C. hominivorax as sister to the other three species. The concatenated dataset yielded a topology supporting a close relationship between C. minima and C. aldrichi which is congruent with the current taxonomy and indicates C. macellaria as the sister lineage of these two.

Discussion
We present the first species complete phylogeny of the genus Cochliomyia including samples collected throughout the Caribbean from 13 different localities (Table 1). The concatenated matrix recovered a monophyletic Cochliomyia, partially resolved relationships among its species and recovered Compsomyiops as its sister group (Fig. 1), in congruence with prior studies (McDonagh andStevens 2011, Singh andWells 2011). Independent gene trees did not yield fully congruent relationships among species, unsurprising as genes have independent histories. Two nuclear genes, 28S and ITS2 (adjacent loci), strongly supported the monophyly of Cochliomyia while the other two genes, COI and EF-1α did not. These results differ from McDonagh (2009), where EF-1α and COI strongly supported the monophyly of Cochliomyia, while 28S recovered Cochliomyia as paraphyletic. However, McDonagh (2009) included only two of the species of Cochliomyia represented by one specimen each. The differences between the studies could be due to a variety of taxon sampling issues, where our sampling was designed specifically to test monophyly and relationships among Cochliomyia species.
The monophyly of C. hominivorax is supported in all analyses, however, independent gene trees were not congruent with regards to other species. The relatively slowly evolving nuclear genes EF-1α and 28S supported C. macellaria but failed to distinguish between C. minima and C. aldrichi. The relatively rapidly evolving COI "DNA barcode" was found suitable for species identification and delineation (Hebert et al. 2003). COI was the only gene that recovered the monophyly of C. minima, however, COI did not resolve relationships among specimens of C. aldrichi and C. macellaria. This is surprising as these species are clearly identifiable based on morphological characteristics (Hall 1948, Whitworth 2010. Other studies also reported similar results where COI failed to distinguish among some closely related calliphorids (Wallman and Donnellan 2001, Nelson et al. 2007, Whitworth et al. 2007, Harvey et al. 2008, DeBry et al. 2013, Whitworth 2014), a result that has been attributed to incomplete lineage sorting. Results from COI, EF-1α, and 28S combined suggested C. aldrichi as sister to C. macellaria, instead of to C. minima as we would expect based on morphological characteristics. Based on these results we opted to add the rapidly evolving nuclear marker, ITS2 to help resolve species level relationships (Nelson et al. 2007, Agnarsson 2010. ITS2 was the only gene that recovered C. aldrichi as a monophyletic group and supported C. minima as its sister lineage. Despite the incongruence detected between the four genes, a concatenate matrix recovered the monophyly of C. hominivorax, C. macellaria and C. minima, and supported the monophyly of C. minima plus C. aldrichi, mostly congruent with the current taxonomy. However, we found that C. minima is nested within C. aldrichi. That one species is paraphyletic with respect to another is not unexpected and does not necessarily refute their species status. The non-monophyly of C. aldrichi is surprising in that all specimens included in this study were collected from the tiny Mona Island (22 square miles). This indicates incomplete lineage sorting, or possibly recent speciation, rather than other processes like gene flow among species (given Mona's isolation, expectation of panmixia among C. aldrichi on the tiny island, and absence of C. aldrichi from other islands sampled). In contrast, C. macellaria and C. minima are present on most of the islands (Table 1) and the populations in different islands do not show any geographic structure (Fig. 1), indicating a constant gene flow among populations through migration.
The variability in feeding habits, habitat preference and morphology within Cochliomyia is considerable (Fig. 2). In feeding habits, C. aldrichi, C. minima and C. macellaria share similar behaviors. They are primarily carrion feeders, commonly found on decomposing cadavers. However, they are also capable of producing myiasis in open wounds as secondary facultative parasites under certain conditions or as primary facultative parasites as in the case of C. minima, (Hall 1948, Dear 1985. In contrast, C. hominivorax is an obligate parasite of mammals never found in decaying meats (Hall 1948, Stevens and Wallman 2006, McDonagh and Stevens 2011, but see Brody and Knipling 1943. Several authors have studied the evolution of parasitism within Calliphoridae and have concluded that the parasitic behavior in this family evolved independently several times (Stevens and Wallman 2006, McDonagh and Stevens 2011, Singh and Wells 2011. Within Cochliomyia, we conclude that parasitism evolved once in C. hominivorax, since the congeners are carrion feeders, as are members of the sister group, Compsomyiops (Fig. 2).
The habitat preferences of C. hominivorax and C. macellaria are largely known (Hall 1948, Greenberg 1971, Smith 1986, Wells and Greenberg 1992, Byrd and Butler 1996, Byrd and Castner 2010, Koller et al. 2011, however, little is known about C. minima and C. aldrichi. In recent studies of C. minima in Puerto Rico, Yusseff-Vanegas (2014) reported that C. minima prefer highly humid areas and can tolerate relatively cool conditions at altitudes >800m, while this species is absent from extremely dry and hot areas. Similar results were found in Dominican Republic and Cuba where C. minima was found abundantly in tropical and subtropical rain/moist forest even at altitudes >1300m, but absent from dry forest (unpublished data). These results supported the assumption that C. minima prefer humid cool areas, however, more studies are needed to understand its habitat preferences. In contrast, C. aldrichi seems to prefer hot dry areas, different from what we expected given the apparent recent divergence between C. minima and C. aldrichi. This is the case of recently divergent species like Lucilia sericata (Meigen) and L. cuprina Wiedemann, and L. coeruleiviridis Macquart and L. mexicana Macquart that have similar habitat preferences (Stevens and Wall 1996, Whitworth 2006, Byrd and Castner 2010, Whitworth 2010. However, C. aldrichi was found only on Mona Island, a subtropical dry forest with an average annual temperature of 27 °C (National Oceanic and Atmospheric Administration -NOAA) and low humidity through the year, strikingly different from C. minima. Yet, similar results have been reported before for closely related species like C. megacephala and C. pacifica (Singh et al. 2011) which are characterized by very different habitat preferences (Kurahashi 1981(Kurahashi , 1991. Despite we have extensively collected in Florida (Everglades and the Keys), Cuba, Puerto Rico and the Bahamas, where C. aldrichi was previously reported (Whitworth 2010), we did not find this species. This could be explained by sampling bias as we only collected during the summer when precipitation and relative humidity are very high in the Caribbean. It is possible, for example, that C. aldrichi may be seasonal, being present during the winter when conditions are generally drier and cooler in the Caribbean. Alternatively, our sampling might indicate the recent extinction of this species from areas outside Mona, nevertheless, further studies are necessary to test these alternative hypotheses.
Two of the four species, C. minima and C. aldrichi are Caribbean endemics while the other two are widespread (Figs. 1-2). It is difficult to assess the biogeographical history of widespread species, however, we can conclude from our data that divergence between C. minima and C. aldrichi probably occurred in the Caribbean after the area was colonized. Island colonization is sometimes accompanied by a reduction in dispersal abilities and such processes may have led to reduced gene flow among islands, and promoted the formation of the Caribbean endemics. Further phylogeographic/ phylogenomic studies including more taxa from the Caribbean and the continents are necessary to assess the colonization history of the genus and the possible secondary loss of dispersal ability in this group.

Conclusions
We provide the first complete phylogeny of Cochliomyia, supporting its monophyly and placement within the subfamily Chrysomyinae. Given incongruence among gene trees and low level of information at the species level for slowly evolving genes, the resolution of the outstanding questions in Cochliomyia phylogeny will require more data rich approaches, such as those offered by NGS methods. Nevertheless, we advance knowledge on the phylogeny, distribution, and life history of these species that should prove useful in future research and in realizing the potential of these species as forensic insects.
America (2013) and Colombia (2014). We would also like to thanks Fabián García Espinoza from Universidad Antonio Narro Unidad Laguna for supplying specimens from Mexico and to Molly Mactaggart for collecting specimens from Bahamas (2015). We are especially grateful to the following for help with organizing fieldwork Alexander Sanchez (Cuba), Lauren Esposito, Gabriel de los Santos, Solanlly Carrero, and Kelvin Guerrero (Dominican Republic), Lauren Esposito (Jamaica, Colombia and the Lesser Antilles). Our sincere thanks to all our CarBio collaborators for participation in these fieldtrips and research (see islandbiogeography.org). Many current and graduated members of the Agnarsson and the Binford labs were also instrumental in organizing and executing fieldwork including Lisa Chamberland, Federico Lopez-Osorio, Carol Yablonsky, Laura Caicedo-Quiroga, Jose Sanchez, Angela Alicea, Trevor Bloom, Ian Petersen, Alex Nishita, Katy Loubet-Senear, Angela Chuang, Anne McHugh, Micah Machina and many more. Thanks to Sean Kelly, Rebecca Rivera, Ricardo Burgos and to members of the Agnarsson laboratory for comments that improved this manuscript, Laura May-Collado, Lisa Chamberland, Laura Caicedo, Federico López-Osorio, Jie Lui, Muhammad Kala and Gabriel Melo Alves dos Santos. We also would like to thank the undergraduate students, Cole Rachman and Omar Neyra who performed some of the DNA extraction and help in the sorting and identification process. All material was collected under appropriate collection permits and approved guidelines. Funding for this work comes from National Science Foundation (DEB-1314749 and DEB-1050253) to I. Agnarsson and G. Binford. Development of this project was further supported by a UVM APLE grant to Omar Neyra. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.