Use of Plant Resources by Merosargus (Diptera, Stratiomyidae, Sarginae) Larvae

The genus Merosargus (Loew) has 142 described species. This great diversity in the genus could be explained by larvae resource-use specialization. However, information on larval habitats is still very scarce. In Merosargus species, adult males defend oviposition sites, and this territorial behavior may lead to interspecific competition and make even more important the specialization and niche partitioning to prevent competitive exclusion. This study identified substrate types used as a resource by Merosargus larvae and investigated the degree of specialization and overlap in resource use by different species at an Atlantic forest remnant in Minas Gerais, Brazil. Every potential resource, especially those with adults in the vicinity, was collected opportunistically from October 2001 to October 2004. In total, 292 individuals from 12 Merosargus species collected from 21 resource types and 15 plant species were reared in the laboratory. Plant species included herbs, vines, palms, and trees. Six Merosargus species were reared from only one resource type, and each resource type was used, on average, by less than two Merosargus species. Thus, Merosargus species exhibited a high degree of specialization and small overlap in larval resource selection, which could explain the high local and global diversity of the genus.


Introduction
The family Stratiomyidae (soldier flies) occurs in all warm temperate and tropical regions, especially in the wet tropics [1], and includes species with great morphological and habitat use diversity [2].
The genus Merosargus (Loew) is one of the largest in number of species in the family Stratiomyidae.Woodley [2] listed 142 species in the genus, and only two are not Neotropical.The previous revision of the genus is over 40 years old and listed only 109 species [13], thus the genus needs to be reviewed.
The biology of the genus is still poorly known.Woodley [2] reported the occurrence of Merosargus adults around various types of fruits and other rotting plant material on the forest floor, where males defend small territories, females oviposit, and mating occurs.He hypothesized that some species specialize on certain plant resources.Therefore, resource-use specialization by larvae could be an explanation for the great diversity in the genus.However, information on larval habitats is still very scarce and not species-specific [14][15][16][17][18].
In the Neotropical region, a wide variety of resources is available for phytophagous insects.Adult insects must recognize and select the best available oviposition substrates, and, for many species, resource quality has a major effect on their distribution and abundance [19].The choice of oviposition sites is crucial for larval development and adult survivorship and vigor.In fact, several factors such as Psyche nutritional quality, predation risk, and competition may act directly or indirectly in oviposition site selection [20].
Specialization in resource use is a way to maintain the coexistence of a large number of closely related insect species in tropical forests.However, recent studies have shown that different insect herbivore guilds exhibit different degrees of specialization and that some guilds such as chewers are predominantly generalists [21].In addition, scavengers such as litter arthropods are often assumed to be generalists because they harvest nutrients from dead plant material and litter-decomposing microbes rather than directly interacting with living plants [22].
Conversely, specialization could be more important in cases where territorial defense occurs.Oviposition site defense is typical of species that use ephemeral resources and where females mate multiple times [23].In these cases, last male sperm precedence is a common pattern [23,24], and this behavior has been demonstrated in one Merosargus species [25].Therefore, the defense of oviposition sites in Merosargus is very important to ensure mating just before oviposition.In fact, this behavior has already been observed in ten of the 18 Merosargus species found in our study site.Resident males avidly guard their territories repelling all intruders, including males of their own species or males and females of other species (Fontenelle et al. unpublished data).Therefore, strong interference competition could occur among Merosargus species using the same oviposition sites, which could lead to the exclusion of less combative species in the absence of resource use differentiation.
The understanding of oviposition site selection in Merosargus could not only explain the maintenance of high species diversity in the genus but also its origin, because assortative mating may occur in cases where males and females are attracted to a particular resource, which may have a prominent role in ecologically-driven sympatric speciation [26].
Thus, this study aimed to (i) identify the plant substrates used as a resource by Merosargus larvae, (ii) determine the period of the year when these substrates are used, (iii) investigate whether there are differences in the degree of specialization in resource use by different species, (iv) investigate whether there is an overlap in resource use by different species, and (v) determine which species use these resources similarly.Assuming that territorial defense leads to strong interspecific interference competition, we expect to find a high degree of specialization and low similarity in oviposition site selection among sympatric Merosargus species.[27].The PERD has an area of approximately 36,000 ha with a complex pattern of vegetation types [28].The climate is humid, tropical, and mesothermal, with the rainy season from October to March and dry season from April to September [29].
This study was conducted opportunistically from October 2001 to October 2004.Sampling was done simultaneously to field campaigns from other studies that used malaise traps and that investigated the territorial behavior of Merosargus species associated to Heliconia conducted during this period.Therefore, the sampling effort was not the same for all periods of the year or all visited areas.
Samples were collected at six sites composed of semideciduous forests at different regeneration levels.We also sampled substrates in a forest fragment near the PERD with the same vegetation type, but an impoverished vertebrate fauna.
Every substrate that could theoretically be used by Sarginae larvae was collected.The sighting of adults on substrates or around their vicinity was considered the strongest evidence of larval occurrence.After first spotting adults on a particular substrate type that substrate was collected at subsequent field trips, even in the absence of adults, to better determine the larval occurrence period.A sample was considered a particular substrate type (plant structure) collected in the same place at the same date, often corresponding to a large amount of substrates.
The substrates were taken to the laboratory and kept in sealed chambers at room temperature until adults emerged.This paper only reports on collected substrates containing Sarginae larvae that resulted in the emergence of adults.Adult Sarginae were identified to genus level using the Woodley [30] identification key.Merosargus adults were identified to species level using James and McFadden's [13] key.Several species reared in our study could not be identified with certainty because taxonomic revision is needed for the genus Merosargus.Therefore, we chose to use the species names that are still uncertain only as a reference for possible species or the closest species; these species were, respectively, indicated in the result tables with "cf." or "aff."before the specific epithet.Voucher specimens were deposited in the Laborat ório de Pesquisas Ambientais CODAAMB/IFMG-OP.

Data Analysis.
Plant species and resource type specificity were determined using a host specificity index (H) [31].This index is an estimate of the proportion of hosts used by a particular Merosargus species among all hosts available minus one where S F is the number of plant species (or resource types) used by a Merosargus species, and S T is the total number of plant species available.However, we modified the index replacing S T by S U , the total number of plants used by any Merosargus species.This was done because it was impossible to determine the actual availability of plant species as food Psyche 3 for Merosargus in the study area.However, this modified index underestimates specificity values compared to the original one.
We also used a measure of host specificity, calculated as because higher specificity values should be attributed to more specialist species.Conversely, MH represents the proportion of hosts not used by a particular species, among all hosts available minus one.The overlap in resource use was calculated using the Jaccard similarity index (J).We then constructed a cluster using the Jaccard distance (1 − J) as a measure of distance and the UPGMA amalgamation rule [32].
The Pearson correlation test was used [32] to test if the number of resource types used by each species was correlated with (1) the months when larvae were found, (2) the number of resources shared with other species, and (3) the average Jaccard similarity index (J avg ) calculated by averaging the similarity in resource use of each Merosargus species with all remaining species.

Results
A total of 45 samples with Merosargus larvae were collected from 15 plant species, including herbs, vines, palms, and trees and 21 resource types including stems and pseudostems, leaves, fruits, flowers, and inflorescences (Table 1).
Most resources with larvae were collected during a few months (M avg = 1.76 months) of the year.The resource most often sampled with larvae was H. episcopalis pseudostems (5 months).In addition, inflorescences of the same plant species were collected in three months.Thoracocarpus bissectus (Vell) stems were also sampled in three months.Several plant species containing larvae were collected in only one month.August was the month of the highest resource richness Psyche  1).Individual plant species hosted between 0 and 5 Merosargus species.A large proportion of plant species or resource types were used by only one Merosargus species.On average, only 1.81 Merosargus species were found per resource type (Table 1).The resources that were used by the largest number of species were H. episcopalis pseudostems (N = 5) and T. bissectus stems (N = 4).In addition, we never found more than two Merosargus species in any sample.
Most Merosargus species developed in vegetative organs such as stems, pseudostems, and leaves.However, M. bivittatus James, M. coxalis Lindner, and M. transversus McFadden used only reproductive organs (inflorescences, flowers, and fruits).Three species, M. cingulatus, M. varicrus, and M. aff.pallifrons, used both vegetative and reproductive organs, the latter represented by Musa sp. or Heliconia inflorescences.
Larvae of two other genera, Ptecticus sp. and Himantigera sp., also developed solely in reproductive organs of plants.The species M. aff.pallifrons and M. varicrus were also found in H. episcopalis pseudostems attacked by a lepidopteran borer (Table 2).
The plant samples resulted in the emergence of 361 Sarginae adults representing 15 species, including 292 individuals from 12 Merosargus species.The species with the highest number of adults were M. azureus (Enderlein) (N = 121) and M. aff.pallifrons (Curran) (N = 58) (Table 3).
The highest number of Merosargus species found in one month was five, in April and August (Table 3).In general, larvae of Merosargus species were collected during few months (M avg = 2.8).The species found for the highest number of months were M. azureus (N = 7), M. aff.pallifrons (N = 5), and M. varicrus (James) (N = 5).
In total, six Merosargus species were reared from only one resource type.On average, few plant species (S avg = 2.5) and   resource types (R avg = 3.1) were used by each Merosargus species.The species that used the greater number of resources were M. azureus (N = 10), M. aff.pallifrons (N = 7), and M. cingulatus Schiner (N = 5).In general, species that used more plant resources were also found over a greater number of months (r = 0.93; Figure 1).
The average plant species host-specificity measure was MH avg = 0.89 (H avg = 0.11).This means that a Merosargus species feeding on a particular plant species used an average of only 11% of other plant species used minus one.
Although over half of the plant species and resource types were used by a single Merosargus species, almost all species shared at least one plant species or resource type with another congener (Figure 2).The exceptions were M. coxalis and M. transversus, which used a single, exclusive resource type.Merosargus azureus used the largest number of exclusive plant species (N = 5) and resources types (N = 6).
As expected, generalist species were also the ones that shared the most resources with congeners (r = 0.83, P < 0.05; Figure 3).However, it should be noted that M. cingulatus exhibited greater resource overlap than expected from this correlation.
The small values of the Jaccard similarity index reflected the few overlaps in resource use between Merosargus species.The average similarity between a species and the remaining ones was below 25% for all species and below 15% for most species.There was no significant correlation between the number of resource types used and the average similarity indices (Figure 4).Most species exhibited similarity values between 10% and 15%, and this result was observed in more specialized species such as M. arcuatus James and M. opaliger Lindner as in relatively more generalist species such as M. aff.pallifrons and M. azureus.
The cluster analysis also reflected the small overlap in resource use by Merosargus species (Figure 5).The exception was the similarity between M. arcuatus and M. opaliger: as both used the same single resource their distance was zero in the cluster.Other than these species, M. pictipes, M. gracilis,  and M. cingulatus clustered together, and so did M. pallifrons and M. varicrus.In addition, the two species that used only different, exclusive resources, M. transversus and M. coxalis, were completely isolated from each other and from all the other species in the cluster.

Discussion
In our study we identified several plant species and structures where larvae of a significant number of Merosargus species develop.We also showed that Merosargus uses a great diversity of substrates for larval development.
All plant species identified here, the genera Panicum, Thoracocarpus, Astrocaryum, Euterpe, Guarea, Urera, and Musa, and their respective families, except Arecaceae, shown here to be used by Merosargus, are new records.The plants of families Amaranthaceae and Cucurbitaceae whose genera were not identified are also new records for Merosargus.In fact, Merosargus larvae had only been previously observed in bracts of Heliconia (Heliconiaceae) inflorescences [14][15][16] and fallen flowers of Couratari stellata Smith (Lecythidaceae) [17].Moreover, Merosargus gracilis had been recorded in another Arecaceae species (Bactris gasipaes Kunth) in the pulp of fruits [18], and not using the trunk or leaves as observed here for the two species of this family: E. edulis and A. aculeatissimum.Oviposition occurs on substrates when they are recently damaged or already decaying.The predominance of herbaceous species as a resource for Merosargus could be explained by their soft tissues, which are more likely to be found damaged and more suitable for larval feeding.Damage to understory plants is especially likely to occur in the rainy season, when strong winds cause many trees to fall (Fontenelle pers.obs.).In addition, several animals can also cause plant damages.On several occasions, M. azureus males were observed approaching and establishing territories on damaged H. episcopalis pseudostems immediately after agoutis had gnawed on them (Fontenelle pers.obs.).In fact, some gnawed pseudostems were sampled and yielded several adults of this species.
At the forest fragment in the vicinity of PERD, however, the vertebrate fauna is scarce and there may be less damage to the plants and less availability of vegetative resources for Merosargus.In contrast, may be less frugivory and a large amount of rotting fruit is found on the forest floor, which could be exploited by fly species that use fruit as resources.
Our data reinforced the strong association between Merosargus and Heliconia.There are five native Heliconia species at PERD: H. episcopalis, H. spathocircinata (Aristeg.),H. aemygdiana, H. angusta (Vell.)[33], and H. x matenensis (Silva et al.) (H.episcopalis and H. spathocircinata hybrid).Samples from only one native Heliconia species did not result in the emergence of Merosargus adults (data not shown).There are also two exotic Heliconia species at PERD: H. rostrata (Ruiz & Pav ón) and H. psittacorum (L.f.), but they were not sampled.Nevertheless, all Heliconia species found at PERD are likely used by Merosargus species.
Previous studies have shown Merosargus species using only the inflorescences of Heliconia.Merosargus larvae are part of the phytotelmata fauna of some Heliconia inflorescences whose bracts are filled with water and where larvae feed on rotten flowers and other debris [14][15][16].
Bracts of Heliconia species found at PERD do not accumulate large amounts of water.Heliconia episcopalis, in particular, has small bracts that remain closed most of the time at an angle that prevents water accumulation, but the inflorescences have dozens of bracts where rotting flowers may be found (Fontenelle pers.obs.).Therefore, despite not collecting water, these inflorescences form an environment rich in rotten organic matter that is somewhat protected from predators and desiccation.
Pseudostems of the Heliconia species sampled are likely more important resources for Merosargus larvae than inflorescences.Pseudostems can be used in various forms and on several different occasions: when pseudostems of green plants are damaged or attacked by herbivores; when older leaves begin to rot; and when the entire pseudostem is in a more advanced state of decomposition.Bored pseudostems or with rotten sheaths have a strong bad smell.When the borer caterpillar chews on the central pseudostem shaft, it kills the inner leaf, which rots inside the pseudostem.Nevertheless, rotten sheaths are common in older plants and result from normal leaf senescence (Fontenelle pers.obs.).
Two Merosargus species, M. aff.pallifrons and M. varicrus (Figure 6), used both bored pseudostems and those with rotten sheaths (data not shown) and they were also found in inflorescences.In all instances larvae were found immersed in liquefied decaying organic matter subjected to heavy hypoxia.
In contrast M. azureus had a clear preference for fresh and newly damaged pseudostems (data not shown), which indicates that the volatile compounds that attract this species must be more associated to the plant itself than to its decomposition process.

Psyche
Couturier et al. [18] found larvae of Ptecticus and Merosargus in Lecythidaceae flowers.They suggested that flower use is atypical in Stratiomyidae and occurs because of the fetid-scented flowers and the bat floral syndrome of the species of plants used by Stratiomyidae larvae.However, we found Merosargus coxalis using the flowers of Lecythis lurida, whose flowers are not particularly fetid and are beepollinated.Thus, we believe that flower morphology is more important than smell as an explanation for oviposition site selection by Merosargus.The morphology of Lecythidaceae flowers is quite distinctive, with stamens arranged in a ring protected by a hood-like, dense cluster of petals [34], forming a wet and safe environment for the larvae.Furthermore, the use of flowers by Merosargus larvae is not unusual, as they are very common in Heliconia flowers, and is not restricted to a specific floral syndrome, as the Heliconia species found at PERD are pollinated by hummingbirds [35].
In fact, several Merosargus species are attracted to the smell of rotten plant resources, although this is not true for all species identified in our study.We believe that different species are attracted to different volatile compounds, and the study of these compounds may help elucidate some of the mechanisms responsible for the resource preferences exhibited by different Merosargus species.
Additionally, resource preference can have a strong influence on the spatial distribution of populations.Other studies that sampled populations using more standardized methods showed that Merosargus species are concentrated in sites where large quantities of preferred resources are found.Sampling with malaise traps in areas with large H. episcopalis aggregations resulted in a large number of M. azureus, M. varicrus, M. gowdeyi (Curran), and M. gracilis (Williston) adults (Fontenelle et al. unpublished data), which further emphasizes the importance of this plant species as a resource for these insect species.
Because sampling was not uniform little can be said about the distribution of plant resources and species throughout the year.Few plant resources are found in only one season, but resource availability seems to increase during the rainy season.
Therefore, many resources are seasonally available, particularly plant reproductive organs, but vegetative parts may also be more available at certain times of the year.For example, trees fall mostly after the first rains, while borer caterpillars and agouti predation are more common in the dry season (Fontenelle et al. unpublished data).
Merosargus adults are also seasonal.However, more abundant species can be sampled throughout the year (Fontenelle et al. unpublished data), and their larvae and pupae are likely to be even more frequent than adults because the duration of these stages must be greater than that of the adult phase.Moreover, species that use less seasonal resources or a greater variety of resources available at different times could ensure the occurrence of larvae during several months of the year.Therefore, the relationship between the number of resources used and the number of months in which larvae were found should be interpreted with caution.In fact, this pattern may be explained by a correlation between both variables with the species abundance.most abundant Merosargus species are likely to be found in more months and may have more known resources.
Merosargus species assemblages in individual plant species were species poor, as they only included up to five species.However, these figures are still higher than those observed for fruit flies (1-3) [31].
The average host specificity of fruit fly species relative to all plant species used as a resource found by Novotny et al. [31] was H avg = 0.02.Using our modified index, the recalculated fruit fly H was H avg = 0.03 for the genus Bactrocera, which used 25 plant species (1.67 plant species on average), and H avg = 0.13 for the genus Euphranta, which used five plant species (1.50 plant species on average), a more similar result to the one found for Merosargus in our study.
We can assume that resource selection for larvae of the genus Merosargus is quite specialized.Moreover, we found few substrate types that were used by more than two species, leading to a slight overlap in resource use.Both specialization and little overlap in resource use may promote stable coexistence of Merosargus species.In fact, niche differentiation, and particularly resource partitioning, may promote the coexistence of closely related species (e.g., [36]).
Specialization in resource use is a possible explanation for the large number of closely related insect species found in tropical forests.However, some insects groups and guilds are predominantly generalists [21].
Most scavengers such as litter arthropods are often assumed to be generalists because they harvest nutrients from dead plant material and litter-decomposing microbes rather than directly interacting with living plants [22].However, feeding niche differentiation was demonstrated and considered responsible for the high species richness and diversity of soldierless termites in neotropical rainforests [37].
When competition between members of different species occurs extreme conflicts are expected and the less successful competitor will either be driven to extinction or, more commonly, be forced to modify its feeding pattern [38].Interspecific interactions among Merosargus adults include direct interference in resource use through territorial defense and monopolization (Fontenelle et al. unpublished data).Therefore, resource partitioning may be even more important in this genus as it is so diverse locally, even though males have such intense territorial behavior.Other mechanisms of resource partitioning may occur for species that were clustered together in the cluster analysis such as slight differences in adult activity time or area [38].
Genner et al. [39] studying territorial cichlid fish suggest that ecological generalism reduces the intensity of interspecific competition while specializations for the same resources increase it.We found that one of the most aggressive species M. azureus (Fontenelle et al. unpublished data) is the most generalist in resource use.Further studies on Merosargus territorial behavior, measuring the intensity of aggression intra-and interspecifically, and degree of resource monopolization will be essential to better understand the selective pressures leading to resource selection of this genus.

Psyche 9
Considering the small sample size, the large number of single records, and especially the high plant diversity in forests of the study area, (over 1100 species) ( [33,40,41]; see also [42] for a phytosociological study), we believe that only a small fraction of the potential resources available to Merosargus flies was sampled in this study.
Expanding the sampling effort to new plant species may reduce resource specialization estimates when additional resources are included.Conversely, it could also increase the estimates if additional Merosargus species with restricted resource ranges are included.
A taxonomic revision of the group, particularly using molecular techniques, may reveal host-specific cryptic species eventually masking the specialization as has occurred with other groups in Diptera [43,44].In addition, it remains to be studied if different genotypes, within a Merosargus species, prefer different resource types, an evidence of assortative mating that can act either increasing diversity, through sympatric speciation [26], as reducing the niche breadth [45].

Figure 1 :
Figure 1: Relationship between the number of plant resource types used by Merosargus species and the number of months when larvae were found in samples collected at the Parque Estadual do Rio Doce (PERD), Minas Gerais, Brazil, between October 2001 and October 2004.

Figure 2 :Figure 3 :
Figure 2: Number of plant species (a) and resource types (b) used by each Merosargus species found in samples collected at the Parque Estadual do Rio Doce (PERD), Minas Gerais, Brazil, between October 2001 and October 2004.Black bars represent plant species/resource types used by a single species, and open bars those shared with at least one species.

Figure 5 :
Figure 5: Cluster analysis of Merosargus species based on the similarity in resource use by the larvae from samples collected at the Parque Estadual do Rio Doce (PERD), Minas Gerais, Brazil, between October 2001 and October 2004.The cluster was constructed using the Jaccard distance (1−J) and the UPGMA amalgamation rule.

Table 1 :
Plant species and structure used as a resource by Merosargus larvae, month of occurrence, and number of Merosargus species found in samples collected at the Parque Estadual do Rio Doce (PERD), Minas Gerais, Brazil, between October 2001 and October 2004.

Table 2 :
Merosargus species with respective plant species and substrate types found in samples collected at the Parque Estadual do Rio Doce (PERD), Minas Gerais, Brazil, between October 2001 and October 2004.

Table 3 :
Merosargus species, month of occurrence, number of adults reared in the laboratory, number of plant species and resources types (in brackets), and resource specificity of each species found in samples collected at the Parque Estadual do Rio Doce (PERD), Minas Gerais, Brazil, between October 2001 and October 2004.
P avg : average number of plant species found; MH avg : measure of specificity (proportion of other plants species not used).