Diversity and associations between Drosophilidae (Diptera) species and Basidiomycetes in a Neotropical forest

Drosophilidae is one of the most representative families of insects that occurs in fungal fruiting bodies of Basidiomycetes; however, the diversity and community structure of mycophagous Drosophilidae in the Neotropical region is poorly known. The aims of the present study were to describe the diversity of mycophagous Drosophilidae and to investigate its colonization of fungal hosts in a forest of southern Brazil. From 120 fungal samples (patches of mushrooms) of 17 Basidiomycetes genera, ﬂ ies were recorded emerging from 70 samples and collected in adult stages of 25 fungal samples, for a total of 4897 drosophilids belonging to 31 species and 5 genera. Drosophila Fallén was the most species-rich genus, whereas Hirtodrosophila Duda was the dominant genus. Studies performed in the Holarctic region indicate that mycophagous drosophilid have generalist habits; however, our results showed that most drosophilids use fewer than two fungal hosts, and most species of Hirtodrosophila and Leucophenga were restricted to abundant fungal species, suggesting a specialization for these resources. The most specialized fauna emerged from Auricularia , which was the most frequent fungal genus in our collection, and this result supports the assumption that specialization depends on the availability of fungal resources over time.


INTRODUCTION
A variety of insects colonize fruiting bodies of Basidiomycetes fungi; however, Drosophilidae is usually the most representative family because the fl ies use these bodies for feeding, mating and breeding sites (Bunyard 2003, Courtney et al. 1990, Shorrocks and Charlesworth 1980, Toda et al. 1999).Mycophagy in Drosophilidae is most likely derived from detritivorous feeding habits and may have arisen several times during their evolution (Throckmorton 1975).This habit is possibly synapomorphic in the genera Hirtodrosophila Duda, Mycodrosophila Oldenberg and Zygothrica Wiedemann (Grimaldi 1987) and FELIPE B. VALER et al. homoplasic in Drosophila Fallén, Leucophenga Mik and Scaptomyza Hardy (Courtney et al. 1990).Mycophagy was also suggested in the genera Paraliodrosophila Duda and Paramycodrosophila Duda (Vilela and Bächli 2007), which are closely related to Hirtodrosophila, Mycodrosophila and Zygothrica (Grimaldi 1990).
In certain species of Drosophilidae, fungal fruiting bodies are the only resource required throughout the life cycle, and such behavior is observed in Hirtodrosophila (except for rare instances, see Grimaldi 1987) and Mycodrosophila species, which are primarily mycophagous because they depend on the fungi for breeding and feeding sites (Courtney et al. 1990).In other drosophilids, mycophagy is more labile, and fungi represent only a fraction of the resources used.Such behavior can be observed in species of Zygothrica, which can be found in fungi and fl owers, or Drosophila, which may use a wide range of hosts, such as cacti, fl owers, sap, fruits and fungi (Carson 1971, Markow andO'Grady 2008).
Despite a broad larval niche, a number of Drosophila species were dominant in the fungi collected from the Holarctic region, where most of the ecological studies of mycophagous drosophilids have been conducted (Lacy 1984, Takahashi et al. 2005, Wertheim et al. 2000).The main Drosophila species groups that emerged from this resource were pinicola, quinaria, testacea and tripunctata, which all belong to the immigrans-tripunctata radiation (Markow andO'Grady 2006, Remsen andO'Grady 2002).However, in certain mycophagous communities, other genera stand out in species richness and abundance, such as Hirtodrosophila, Leucophenga and Mycodrosophila (Bunyard 2003, Toda et al. 1999, Tuno 1999).For instance, Leucophenga was the dominant genus and Hirtodrosophila was the most species-rich genus among the fl ies emerging from fungi in a largescale fi eld survey in Australia (van Klinken and Walter 2001).
In the Neotropical region, few studies have addressed mycophagous Drosophilidae.From the 1940s to 1960s, taxonomic studies described a number of mycophagous Hirtodrosophila and Zygothrica species but did not investigate their ecology (Burla 1956, Cordeiro 1952, Frota-Pessoa 1945, 1951, Mourão et al. 1965).Despite the frequency and diversity of Drosophilidae, few ecological studies have reported their feeding, mating or breeding habits on mushrooms (Grimaldi 1987, Heed 1957).In Brazil, such collections were conducted in only two biomes: the Atlantic Rainforest biome (Gottschalk et al. 2009, Val andKaneshiro 1988) and the Cerrado biome (Roque et al. 2006, Roque andTidon 2008).These studies expanded the knowledge of drosophilid species with mycophagous habits.
Although studies of Holarctic fauna have provided important contributions to our understanding of the ecology and evolution of mycophagous drosophilids, the characteristics of fauna in this region may differ from those of the Neotropical region as suggested by Courtney et al. (1990).Thus, the aim of this study was to describe the diversity of the drosophilid assemblage associated with fungal species of Basidiomycetes in a forest area of the Pampa biome located in southern Brazil.
The forest structure of HBITL shows at least three vegetation strata: arboreal, shrub and According to the Köppen classification system, the climate is Cfa (Kottek et al. 2006).Meteorological measurements performed between 1971 and 2000 reveal that the average annual temperature is 17.8°C and maximum and minimum monthly mean temperatures are 28.2°C and 8.6°C,

respectively (Agrometeorological Station of Pelotas 2014
).The rainfall is 1367 mm/year, there are approximately 120 rainy days, and the annual relative humidity is 80%.

DATA COLLECTION
We searched for Basidiomycetes fungi along a 200 m trail, and we extended our search 10 m on either side of the trail.Fruiting bodies were found on the roots of plants, in leaf litter or on decaying wood.The collections were performed monthly between February and May 2011 and between February and June 2013, and each collection was conducted over a period of three hours in the morning.The sampling units of this study are patches of fungal bodies.We also captured adults on fruiting bodies with a net or entomological aspirator and fixed them in 70% ethanol in the fi eld.
All of the fruiting bodies found in the fi eld were collected, and when a large number of mushrooms was observed in the same location, only a fraction were collected.The collected fruiting bodies were transported in plastic bags to the laboratory, where they were then weighed to quantify the fungal mass, stored in glass vials with autoclaved sand and covered with synthetic mesh.To prevent dehydration, distilled water was added to the sand.Samples were maintained at 25 ± 1°C for four or fi ve weeks, and during this period, the emergent insects were aspirated daily and fi xed in 70% ethanol.

IDENTIFICATION OF BIOLOGICAL MATERIAL
We identified drosophilid males based on their external morphology and genitalia according to Frota-Pessoa (1945), Burla (1956), Wheeler and Takada (1971), Grimaldi (1987Grimaldi ( , 1990) ) and Vilela and Bächli (1990, 2004, 2007).We prepared the genitalia in accordance with Wheeler and Kambysellis (1966) and using Kaneshiro's (1969) modifi cations.The females were identifi ed based on the external morphology to the lowest taxonomic level possible.For analyses, we considered the abundance of females of cryptic species to be proportional to the number of emerged males from the same sample.
The fungus genera and/or species were identifi ed via photos of the fruiting bodies taken in the fi eld according to Lincoff (1981Lincoff ( , 2010)), Putzke and Putzke (1998), Polese (2005) and Laessoe and Lincoff (2010).Because of the advanced decay stage of certain fruiting bodies, we could not identify a number of samples, which were excluded from the analyses conducted at the species level.

ASSEMBLAGE CHARACTERIZATION
We assessed species dominance within the assemblage via a Whittaker plot that considered the relative abundance of species (pi, ratio between the absolute abundance of species i and total abundance) (Krebs 1999).Sampling adequacy was evaluated using emergence data from Basidiomycetes samples in a randomized species accumulation curve.The bootstrap method was used to support the sampling sufficiency (Epps andArnold 2010, Smith andvan Belle 1984) and performed with EstimateS v. 8.2.0 software with 500 randomizations (Colwell 2006).
The use of each fungal species by drosophilid species was evaluated according to the ratio between the number of colonized samples and total number of samples of each fungal species.To assess the species' temporality, the relative frequency of fungal species was calculated as the ratio between the number of collections in which the species were observed and total number of collections (eight collections).
We conducted Spearman's correlations between the total number of samples, total mass of fungal samples, relative frequency of each fungal species and total abundance and species richness of emerged drosophilid species.
To visualize the association between host fungal species and drosophilid species, we performed a correspondence analysis (CA) in the program Past v.2.17b (Hammer et al. 2001).The fungal species were considered as the independent variable, and drosophilid species abundance (with Log transformation) was considered as the dependent variable.To avoid the inclusion of subsampled species of drosophilid and fungus, three criteria were adopted for inclusion in the CA: (1) drosophilid species must have an absolute abundance ≥ 10 individuals; (2) drosophilid species must have been collected in at least three samples; and (3) fungal samples had to be determined to the species level.To test the significance of the groups obtained with the CA, an analysis of similarity (ANOSIM) using the Bray-Curtis index was conducted using the program Past v.2.17b.The Bonferroni correction for multiple comparisons was performed to adjust the signifi cance.

RESULTS
We collected 4897 drosophilids in 5 genera and 31 species.Of these, 4620 individuals in 26 species emerged from 70 samples, and 277 adult fl ies in 17 species were collected on 25 samples (Tables I and II, respectively).Although the drosophilid assemblage was not sampled to saturation, our sampling approach was suffi ciency (Fig. 2) because we recovered 90% of the 29 drosophilid species predicted by the bootstrap analysis.
To characterize our collection, the number of samples, mass (g) and relative frequency for each collected fungal species are shown in Table I.The number of fungus samples was positive correlated with the total abundance (rs = 0.683; p = 0.0003; df = 22) and species richness of the emerged fl ies (rs = 0.719; p = 0.0001; df = 22).The abundance of emerged fl ies was also correlated with the total mass of fungal samples (rs = 0.497; p = 0.015; df = 22), although species richness was not (rs = 0.349; p = 0.102; df = 22).In addition, the relative frequency of fungal species was positive correlated with drosophilid abundance (rs = 0.702; p = 0.0002; df = 22) and species richness (rs = 0.744; p = 0.00005; df = 22).
Out of 120 Basidiomycetes samples identifi ed at the genus level, 57 were positive for drosophilid emergence or adult collection events, whereas 63 were negative (see Appendix S1 in Supporting Information -Supplementary Material).Positive samples of Coprinus Persoon, Lycoperdon Persoon, Cantharellus Jussieu, Cortinarius (Persoon) Gray, Panus Fries and Clitocybe (Fries) Staude were not observed, although Lycoperdon was sampled on fi ve occasions.For most of the observed fungal species, less than 50% were colonized by drosophilids, and Agaricus Linnaeus, Auricularia Bulliard ex Jussieu, Polyporus P. Micheli ex Adanson and Melanoleuca Patouillard were the only genera that had colonized more than 50% of the samples.Of these last four genera, only Melanoleuca was sampled in a single collection, whereas Agaricus and Auricularia were the second-and third-most sampled genera (19 and 18 samples, respectively), indicating that they are important resources for the drosophilid assemblage.Furthermore, considering the relative frequency of fungal species as a measure of temporal availability, Auricularia auricula-judae (Bulliard) J. Schröter was the most common species at a relative frequency of 0.75, and it was followed by A. polytricha (Montagne) Saccardo and Polyporus sp.1, which both had a relative frequency of 0.40.
The dominant drosophilid species in our samples (pi ≥ 0.10) were Hirtodrosophila morgani aff., which had the highest relative abundance, H. mendeli (Mourão, Gallo and Bicudo), Zygothrica bilineata (Williston) and H. levigata (Burla) (Fig. 3).The intermediary species (0.10 > pi ≥ 0.01) were Z. ptilialis Burla, H. morgani (Mourão, Gallo and Bicudo)   These nine species accounted for approximately 90% of emerging individuals.The remaining 17 species were rare in our sample (pi < 0.01).Fig. 4 shows the number of fungal species from which the drosophilid species emerged, and it provides an overview of the resource range that these species can use, suggesting varying degrees of specialization.The drosophilid species that use other resources in addition fungi, such as D. paraguayensis and Z. ptilialis (Garcia et al. 2012)  Zygothrica sp.Z002], were only collected as adult fl ies and did not emerge from the fruiting bodies (Table II).
With the exception of H. levigata and H. morgani aff., all of the Hirtodrosophila species emerged from two species of Auricularia: A. auricula-judae and A. polytricha.Auricularia auriculajudae was the main resource of H. morgani aff., the dominant species in the assemblage, although only two fl ies of this species emerged from Polyporus sp.1.Other species that displayed a narrow range of fungal trophic resources were L. maculosa cf.
The increased availability of fungal species (measured by number of samples and relative frequency) led to increased abundance and species richness of the associated drosophilid.Döge et al. (2015) observed that the availability of resources is the main factor affecting the size of the populations of two species of drosophilids.We observed a positive correlation between for the mass of fungal species and drosophilid abundance, although a correlation was not observed between mass of fungal species and drosophilid species richness.Apparently, the drosophilid species were restricted to ovipositing or breeding on fungal species and could only colonize a limited number of species despite the high availability.
Regarding the diversity of the assemblage, our results are consistent with previous studies that assessed the emergence of drosophilid from Basidiomycetes fungi in the Neotropics, and we showed that the most species-rich genus was Drosophila and most abundant genus was Hirtodrosophila (Gottschalk et al. 2009, Roque et al. 2006).The dominance of Hirtodrosophila is most likely associated with the high colonization rates of Auricularia (Appendix S1) and high sample number of this fungus, which is the main trophic resource for Hirtodrosophila larvae in our study.
Drosophilids associated with fungi have been subject to extensive sampling in Japan (Toda et al. 1999), the United States (Lacy 1984) and Europe (Shorrocks and Charlesworth 1980), and these studies reported that a few species were highly dominant in relation to others, which is consistent with our results.The highest number of Drosophila species groups observed in our study belonged to the immigrans-tripunctata radiation, which is notable for its mycophagous habits (Courtney et al. 1990, Gottschalk et al. 2009, Lacy 1984, Morales-Hojas and Vieira 2012).Drosophila paraguayensis, which belongs to tripunctata species group, are frequently found on fruit but was the most abundant Drosophila species in our samples, suggesting that fungi are important breeding sites for this species, especially in summer and autumn.Saavedra et al. (1995) recorded a low emergence of D. paraguayensis from fruit in summer and autumn, although this species was observed breeding on fruit in spring and winter.In addition, Garcia et al. (2012), Hochmüller et al. (2010) and Poppe et al. (2012) sampled D. paraguayensis and observed low abundance in summer and autumn with banana-baited traps.Most likely, this versatile species changes its larval resources (i.e., fruit and fungi) seasonally, and this behavior is supported by the results of our study because D. paraguayensis was sampled in nine fungal species.
Drosophila melanogaster, D. simulans and D. willistoni cf., species, which breed on fruit, were collected occasionally in low abundance on fungus.The colonization of fungi by D. willistoni corroborates Roque et al. (2006), who collected this species emerging from Agaricales and Boletales.
van Klinken and Walter (2001) (Grimaldi 1987, Heed 1957) and Drosophila using other substrates as breeding resources (Carson 1971).However, for Leucophenga species in the Neotropics, fungal fruiting bodies are the only substrate in which they are known to rear their larvae (Gottschalk et al. 2009, Heed 1957, Roque et al. 2006).
Of the species collected on fruiting bodies, a subset of Zygothrica species did not emerge, which is consistent with Grimaldi (1987), who reported the use of fungal fruiting bodies as a courtship arena for Zygothrica.However, adult Leucophenga were not collected on fruiting bodies and were only collected as they emerged.Lachaise and Tsacas (1983) suggest that the feeding and breeding sites of Leucophenga species are distinct, which could explain our results.
Data in the literature suggest that most mycophagous drosophilid species have generalist habits because fruiting bodies are considered to be nutritionally homogeneous (Courtney et al. 1990, Hanski 1989).Thus, oligophagous or specialist species would be uncommon, and records reporting such habits would most likely be artefacts of insufficient sampling effort (Hanski 1989).Our results provide evidence that is inconsistent with this hypothesis because only 9 of the 26 drosophilid species were observed colonizing three or more fungal species, most of the recorded species were collected in fewer than two fungal species, and most species of Hirtodrosophila and Leucophenga were restricted to abundant fungal species (in collected sample mass), indicating specialization for these resources.In addition, we found species of drosophilid that were closely associated with Auricularia, Agaricus, Lepiota and Melanoleuca.The emerged fauna of Auricularia were the most particular and primarily represented by species of the hirticornis group of Hirtodrosophila.Auricularia was the most frequent and abundant in HBITL, and according to Courtney et al. (1990), these features are necessary to host a specialized fauna.The generalist habit described by Courtney et al. (1990) and Hanski (1989) in the Holarctic region may be a result of the absence of specialist species of Hirtodrosophila, Leucophenga and even Zygothrica because Drosophila is the dominant genus of mycophagous drosophilid assemblages in the Holarctic.
In conclusion, the studied drosophilid assemblage was dominated by Hirtodrosophila, and Drosophila was the most species-rich genus.The dominance of Hirtodrosophila should reflect a specialist habit, whereas the higher species richness of Drosophila may refl ect a generalist habit.The presence of drosophilids with specialist habits was most likely not a sampling artifact because the species richness did not increase with the mass of the fungal samples, suggesting a restriction imposed by fungal species on colonization by generalist drosophilids.The differences between the assemblage observed in our results and assemblages studied in the Holarctic region may have been caused by the lower species richness of specialized genera.

Figure 1 -
Figure 1 -Location of the Horto Botânico Irmão Teodoro Luís (HBITL).a. Map of South America with Brazil in gray and state of Rio Grande do Sul highlighted.b.Map of Rio Grande do Sul indicating the location of HBITL with a star, Pampa biome in black and Atlantic Rainforest biome in dark gray.c.Satellite photograph of the surveyed restinga forest area of HBITL (Source: Google Earth ® ).Scale bars: 100 km in b; 300 m in c.

Figure 5 -
Figure 5 -Axis 1 and 2 of the correspondence analysis (CA) relating the species of drosophilid and host fungal species from which they emerged.Samples of fungi are shown in bold.The number of samples used in the CA is shown in brackets in the legend.The numbered squares represent independent groups of drosophilid species according the analysis of similarity (ANOSIM).

Figure 4 -
Figure 4 -Number of Basidiomycetes fungal species used by each species of drosophilid.

TABLE II Absolute abundance of drosophilid species sampled as adults on Basidiomycetes fungal species in Horto Botânico Irmão Teodoro Luís, southern Brazil.
Figure 2 -Species accumulation curve for the drosophilid assemblage and species richness estimation by the bootstrap method.FELIPE B. VALER et al.