Multifemale nests and social behavior in Euglossa melanotricha ( Hymenoptera , Apidae , Euglossini )

The nesting biology and social behavior of the euglossine bee species Euglossa melanotricha was analyzed based on the monitoring of eight nests found in man-made cavities and transferred to observation boxes. Euglossa melanotricha females usually construct their nests in cavities in the ground, in buildings, or in mounds. In this study, we present new data on the nesting biology of E. melanotricha. The process of reactivation of nests was commonly observed with one to three females participating in the reactivation. The duration of the process of reactivation ranged from 10 to 78 days (n = 31) and were longer during the rainy season. Time spent (in days) for provisioning, oviposition and closing a single cell was higher in reactivations that occurred during the dry period.151 emergences were observed (39 males and 112 females). 90 (80.3%) of the emerged females returned to the natal nest, but only 35 (38.9%) remained and actively participated in the construction and provisioning of cells. The other 55 abandoned the nests after several days without performing any work in the nest. Matrifilial nest structure was regulated by dominancesubordinate aggressive behavior among females, where the dominant female laid almost all eggs. Task allocation was recognized by behavioral characteristics, namely, agonism and oophagy in cells oviposited by other females. Euglossa melanotricha is multivoltine and its nesting is asynchronous with respect to season. Our observations suggest a primitively eusocial organization. These observations of E. melanotricha provide valuable information for comparison with other species of Euglossa in an evolutionary context.


introduction
The bees of the tribe Euglossini are the only members of the corbiculate bees that do not form large colonies with a typical queen and worker caste (Soucy et al. 2003). The genus Euglossa consists of 129 known species (Nemésio and Rasmussen 2011). These include solitary, communal and social species (Dressler 1982, Young 1985, Roberts and Dodson 1967, Garófalo et al. 1998, Soucy et al. 2003, Otero et al. 2008). The latter include those that form multi-female nests with a division of labor and overlapping generations (Garófalo 1985, Ramírez-Arriaga et al. 1996, Augusto and Garófalo 2004, Cocom Pech et al. 2008. A diversity of nesting behavior is observed in the different species of Euglossa. Some species construct aerial nests (Roubik andHanson 2004, Capaldi et al. 2007), while others exploit existing cavities found in both natural substrates, such as the soil (Bodkin 1918, Augusto andGarófalo 2007), termite mounds (Sakagami et al. 1967), bamboo stems (Garófalo et al. 1993), orchid roots (Roberts and Dodson 1967), and abandoned bees nests (Garófalo 1985(Garófalo , 1992, and man-made structures, including walls (Janvier 1955), bait boxes (Vázquez and Aguiar 1990), and abandoned hydraulic installations (Gonzales and Gaiani 1990).
Observations of the intranidal behavior of Euglossa carolina females (cited as E. cordata) during nest reactivation have clearly demonstrated that the mother or a sister becomes the reproductive female and the other females perform other nest related tasks (Garófalo 1985). This behavior is characterized by two components: reproductive, with the dominant female replacing subordinate's eggs (daughters' or younger sisters' eggs) with her own, and behavioral, with the dominant exhibiting agonistic behavior towards the subordinate. In addition, the dominant female rarely leaves the nest and becomes the main guard bee, while the subordinate females assume the tasks of collecting resin, constructing or reusing cells, and provisioning and ovipositing in them (Garófalo 1985, Augusto andGarófalo 2010).
Similar social organization have been reported by Ramírez-Arriaga et al. (1996) for Euglossa atroveneta, Augusto andGarófalo (2009) for E. fimbriata andCocom Pech et al. (2008) for E. viridissima. In contrast, in multifemale colonies of E. townsendi, all females are reproductively active. Behavioral interactions of dominance and subordination are lacking in this species and all the females were classified as either egg-laying or as forager/egg-laying. In this species reproductive dominance is displayed by an egglaying female that, after oophagy, replaces the forager/egg layer's eggs with her own. When more than one egg-laying female participates in a reactivation, multiple egg replacements in the same cell may occur (Augusto and Garófalo 2004).
Euglossa melanotricha Moure, 1967 is a medium-sized bee (body length 13 mm) commonly found in open areas of savanna habitat in Brazil and Bolivia. The species is abundant in the Serra do Espinhaço mountain range in the Brazilian states of Minas Gerais and Bahia, but is rare or absent in areas of low altitude or dense forest (Nemésio 2009). Two E. melanotricha nests have previously been found in the Brazilian Cerrado (savanna), although the behavior of the females was not monitored. The first was discov-ered by Sakagami et al. (1967) in an abandoned termite mound. The large number of cells (37) found in the nest, and the fact that two were being provisioned, indicated the reutilization of the site over successive generations. Augusto and Garófalo (2007) found a second nest in a cavity in the ground. Four closed cells and one live female were found in this nest. These observations obviously refer to nests in different phases of development.
This long-term study presents new data on the nesting behavior of Euglossa melanotricha, obtained from the monitoring of nests found in man-made cavities at a site in northeastern Brazil.

Material and methods
The present study took place within the urban area of the town of Campo Formoso (10°30'00"S, 40°19'00"W), Bahia, Brazil (MMA 2010) between October, 2008, and October, 2009. The local climate is classified as dry sub-humid tropical, with annual precipitation of 302-1935 mm, and a rainy season between March and August (PERH 2010).
A total of eight Euglossa melanotricha nests were found within the study area, including one inside an electrical installation, and seven in the holes in ceramic building blocks. The nests were transferred to wooden boxes (12 × 10 × 8 cm), each with a circular lateral opening of 1 cm in diameter, according to the methodology developed by Garófalo et al. (1993). The wooden boxes were then returned to the original locations of the nests and all females returned to their natal nests following transfer. The number of live females and closed cells per transferred nest varied from one to three females and eight to 22 cells respectively (Table 1). Each box was covered with a glass lid and connected to the exterior by an opening in the wall of blocks. The glass cover was removed for marking of each female on her mesonotum with permanent colored Opaque Color pens (a unique mark for each female). When necessary, females were remarked.
The nests were observed under red light between 07:00 h and 18:00 h, with eight hours of observation being conducted three days per week. Nocturnal observations were made once a week between 19:00 and 23:00 h. Quantitative behavioral data were collected using two complementary procedures. All-events records (Altmann 1974) were collected in 460 one-hour sessions, while individual focal samples were conducted at two-minute intervals in 884 one-hour sessions. Overall, 1344 hours of monitoring was conducted over 168 days of data collection.
Nest development was monitored for the collection of data on the following biological and behavioral parameters: (a) cell architecture; (b) nest reactivation and phenology; (c) female foraging behavior; (d) specific aspects of the activity of the females during construction, i.e. reuse of cells, supply, oviposition, cell closure, oophagy, cleaning the nest and the sealing of edges; (e) duration of the period of offspring development; (f ) physiological condition of the females (relative fecundity).
Where appropriate, the results were presented as the mean ± standard deviation. The relationship between the number of cells with eggs and the duration of the female activity period was evaluated using Pearson's correlation coefficient, while Mann- During the intervals between reactivations or during periods of inactivity, the females spent more time inside the nests without engaging in cell construction or provisioning. The number of oviposited cells significantly positively correlated with the duration (in days) of the activity period of the females (Table 2: r = 0.6231; p  duration of periods spent by dominant females in the nest guarding position was 13.2±6.13 min (range 6-30, n = 366).
Of the 31 processes reactivations observed, in 14 (45.1%) the dominant females disappeared, died or ceased ovipositing and were replaced by another female. In all cases, the substitute was another female that had emerged in the nest.
Following the emergence of a female and before the reutilization of a cell, the subordinate females cleaned the cell by removing the silk and pieces of the cell closure. This detritus was deposited on the bottom of the box. The mean duration of this behavior was 12.0±5.35 min (range: 2-27, n = 306).
Dominant females opened the closed cells in which subordinates had oviposited after an interval of between 31 and 240.3 min (i.e., more than four hours) after cell closure. Prior to reopening the cells, the dominant females behaved aggressively towards subordinates by biting and pulling them from the closed cell. Opening a cell took an average of 16.7±2.34 min (range: 12.6-21.68 min, n = 62). Following more than half (61% of 141 observed acts) of the subordinate ovipositions, the dominant female ingested the subordinate's egg. Oophagy took between 96 and 248 s (mean = 158.4±45.65 s, n = 86). Oophagy (86 events) occurred between 10:00h and 18:00h, and was most frequent (n = 61) between 14:00h and 18:00h (Fig. 3).

Duration of brood development
The length of brood development was compared between the rainy and dry seasons. The period was significantly longer during the rainy season (rainy season -males: 75.7 ± 3.55 days and females: 82.3 ± 1.92 days; dry season -males: 56.2 ± 0.86 days and females: 61.7 ± 2.44 days; Table 3: Z = 4.21; p < 0.0001, n = 26).

Relation between the physiological conditions of the number de females inseminated
The spermathecae of two dominant and three subordinate females were dissected for the analysis of possible differences related to social rank. The analysis revealed long ovarioles with mature or maturing oocytes and all females were inseminated.

Cell characteristics and arrangement
The exploitation of pre-existing cavities for the construction of nests observed in Euglossa melanotricha is a behavior typical of most Euglossa species (Garófalo 1985, Augusto and Garófalo 2004, with the exception of those that construct aerial nests, such as E. hyacinthina, E. championi (Eberhard 1988), E. turbinifex (Dressler 1982, Young 1985, and E. dodsoni (Riveros et al. 2009).

Reactivation and nest phenology
The replacement of the dominant female by a subordinate female is consistent with the hypothesis of an age-based dominance hierarchy, as occurs in other primitively eusocial bee species (Michener et al. 1971, Kumar 1975, Eickwort 1986, Yanega 1989, Schwarz and O'Keefe 1991, Schwarz and Woods 1994, Arneson and Wcislo 2003, Augusto and Garófalo 2009. The reactivation and abandoning of nests by Euglossa melanotricha followed an asynchronous pattern, which suggests a lack of any systematic relationship with environmental factors. An important aspect of this asynchrony in tropical bees and wasps is the continuous presence of males in the population. This allows the mating of potentially reproductive females throughout the year (Hunt 1999).
Larval provisioning requires large expenditures of time and energy for E. melanotricha. Besides high costs in time and energy, this amount of time away from the nest could increase the risk of brood parasitization or removal of pollen provisions by scavengers. The presence of parasites Anthrax spp. (Family Bombyliidae) and Hoplostelis bivittata (Megachilidae, Anthidiini) was observed only in nests with a single female in Euglossa viridissima (Cocom Pech et al. 2008). An adaptive benefit of multifemale nests may be protection against parasites (Roubik 1990). Further quantitative study may bring to light some of the mechanisms and risks of parasitization, and its potential role in the evolution of nesting behavior.

Social structure and female behavior
Specific aspects of the behavior of the females: oviposition, cell closure, "nest guarding", cell cleaning, hole sealing and oophagy In associations of Euglossa carolina, it has been observed that the oldest female assumes nest dominance (Garófalo 1985). Age and order of eclosion have been reported as determinant factors for task allocation in some species of primitively eusocial bees and wasps (Kumar 1975, Eickwort 1986, Yanega 1989, Schwarz and O'Keefe 1991, Schwarz and Woods 1994, Tsuji and Tsuji 2005, as in some Euglossa species. The high rates of return (80%) and effective reactivation (39%) recorded in Euglossa melanotricha were similar to those recorded in E. cordata, E. townsendi, and E. fimbriata (Garófalo 1985, Augusto and Garófalo 2004). In three-quarters of the reactivations, associations among females were observed, involving the overlap of generations and interactions of dominance and subordination. The dominant females engage in oophagy, oviposition and closure of the cells provisioned and oviposited in by subordinate females.
Because all nest-mates have developed ovaries, have mated, and do not differ in size, dominant and subordinate females are recognized by their behavioral characteristics. Dominant females exhibited agonistic behaviors towards subordinates and the intensities of these aggressive behaviors where the dominant female had already participated in a reactivation process.
The agonistic interactions observed in Euglossa melanotricha can be compared to the behavior of some groups of halictine bees (see Arneson and Wcislo 2003), in which all females are totipotent, as are the females of Euglossa, and the differentiation of dominance-subordination relationships is based on behavioral interactions among adults.
Although the agonistic behaviors displayed by dominant females do not prevent oviposition by subordinate females, reproductive dominance, reflected in the monopolization of offspring production, is achieved by the dominant female through the replacement of subordinate female eggs with her own. The monopolization of offspring production leads to the highest reproductive skew, as predicted by the concessionbased transactional skew model Keller 1995, 2001), such as that proposed for Euglossa cordata and E. fimbriata (Augusto and Garófalo 2009). Moreover, permitting oviposition by subordinate females and afterwards performing oophagy would be a prudent selfish strategy by dominant females to avoid group dispersal or lethal fighting among females for nest dominance, and is another prediction of transactional models of reproductive skew Keller 1995, 2001).
If the dominant female of Euglossa melanotricha replaces all the eggs laid by subordinates and she mates with only one male, as suggested by Zimmermann et al. (2009) for Euglossa species, then a high genetic relatedness between dominant and subordinate females must occur; this favors an optimum reproductive skew, as also predicted by the concession-based transactional skew model (Langer et al. 2004). This condition could help maintain social cohesion in multifemale nests and lead to long-lived colonies through successive reactivation (Augusto and Garófalo 2010), as reported by Garófalo (1987).
In Euglossa viridissima, no aggressive behavior was observed by dominant female towards her subordinates when they laid an egg, similar to the findings in E. townsendi (Augusto and Garófalo 2004). In contrast, in E. cordata (Augusto and Garófalo 1994) the dominant females impose their dominance over reproduction by showing antagonistic (aggressive) behavior towards subordinate females. However, in E. viridissima the dominant female showed threatening behavior when the subordinate females tried to touch a cell with her egg, which may be considered physical domination.
Oophagy of some dominant's eggs by subordinates was also observed in Euglossa viridissima, however, the dominant cannibalized such eggs and replaced them with her own, confirming her reproductive dominance. The behavior between the dominant and subordinates (associations between mother and daughters) in E. viridissima resembles that of a parasitic female that improves her own fitness on detriment of her daughters' reproduction (Stubblefield andCharnov 1986, Garófalo 2006). This seems to be a primitive case of dominant "mother" policing (coercion) on subordinates (her daughters) that may evolve as a result of the mother being twice as related to her offspring as to her daughters offspring (grandoffspring) (Ratnieks and Wenseleers 2007).
The oophagy of subordinate's eggs preceding oviposition by the dominant was also observed in reactivated nests of Euglossa cordata (Augusto and Garófalo 1994) and E. townsendi (Augusto and Garófalo 2004) but differed from that found in other Euglossa species like E. hyacinthina where only a communal association was established between non-related females and reproductive division of labor didn't occur (Soucy et al. 2003).
As emphasized by Zimmerman et al. (2009), detailed behavioral observations together with the genetic analysis of brood can help clarify the relationships among all females of an association and the real contribution of each one to the social context of the nest.
Oophagy may have a nutritional function (Crespi 1992), or it may be a reproductive strategy in nests containing more than one reproductive female. This is a characteristic of species with primitively eusocial behavior (Kukuk 1992).
The females of Euglossa melanotricha sealed the entrance to the nest during the night and when the weather was rainy. This behavior only occurred once all the females had returned from the field. In Lasioglossum (Evylaeus) villosulus, an essentially solitary species which will occasionally associate with conspecifics, the females seal the entrance to the nest in the absence of the other resident females (Plateaux-Quénu et al. 1989).

Duration of brood development
The development period was similar for males and females, although seasonal variation was influenced by environmental factors, such as the temperature. Higher temperatures may contribute to increased metabolic rates, which may reduce development time considerably (Howe 1967).

Relation between the physiological conditions of the number de females inseminated
Presumably single females can establish a new nest. Their subordinate status is determined only by the presence of a dominant. These females may be in a state of "sit and waiting" (West-Eberhard 1978) in anticipation of eventually occupying the dominant reproductive position in the nest. As in Euglossa cordata (Garófalo 1985) and E. fimbriata (Augusto and Garófalo 2009), one of the E. melanotricha subordinates eventually replaced the dominant female. Danforth et al. (1996) have suggested that Gadagkar's (1990) hypothesis of assured fitness returns, i.e. indirect care of the offspring by non-dispersing individuals, may best explain the high frequency of nest reactivation.
The nesting behavior of Euglossa melanotricha presented in this study provides insights into the social organization of orchid bees. Further studies of the relatedness among individuals will provide data on reproductive partitioning in this species. West-Eberhard MJ (1978)  The bee tribe Augochlorini has an exclusively New World distribution, with maximal diversity in the Neotropics. This tribe is of particular interest because of the diversity of many of its biological traits, at the genus and species level, as well as within species. The social behavior in this group varies from solitary to primitively eusocial, with various degrees of sociality and transitions, including the origin of solitary behavior from eusocial ancestors (Eickwort 1969, Michener 1990, Danforth and Eickwort 1997, Wcislo and Danforth 1997, Engel 2000, Brady et al. 2006. The structure of the nests also presents ample variation within the tribe (Sakagami andMichener 1962, Eickwort andSakagami 1979). Although most species nest in the soil, some lineages have shifted to the use of decomposing wood as a nesting substrate. Such behavior has originated repeatedly within the tribe (Engel 2000), and is known in Augochlora, Megalopta, Xenochlora, and some species of Neocorynura (Brosi et al. 2006, Wcislo and Gonzalez 2006, Tierney et al. 2008a, b, Tierney et al. 2012. Augochlora is one of the more diverse genera within the tribe, with nearly 120 named species, classified in two subgenera, Oxystoglossella and Augochlora s. str. (Moure 2007). The genus ranges from southern Canada to northern Patagonia in Argentina, with most species inhabiting tropical areas. In Argentina the number of species strongly diminishes from north to south, being represented in the temperate Pampean region by only five species (Dalmazzo and Roig-Alsina 2011).
The two subgenera of Augochlora are considered as behaviorally divergent (Eickwort 1969, Michener 2007, Engel 2000. The subgenus Oxystoglossella includes species that nest in the soil and are primitively eusocial, with caste differentiation (Michener andLange 1958, Eickwort andEickwort 1972). Species of Augochlora s. str. have been considered solitary species that nest in soft wood (Eickwort 1969, Engel 2000. Their mandibles are modified, robust, with a lower preapical expansion and a well developed preapical tooth, suited for the substrate in which they dig. The behavioral characteristics of both subgenera have been inferred from what is known for a rather reduced number of species. In the case of Augochlora s. str., nest structure is known for A. pura (Say), A. hallinani Michener, A. sidaefoliae Cockerell, A. smaragdina Friese, A. esox (Vachal), A. isthmii Schwarz, and A. alexanderi Engel (Stockhammer 1966, Eickwort and Eickwort 1973, Zillikens et al. 2001, Wcislo et al. 2003, as well as some comments on a nesting site of A. amphitrite (Schrottky) (Sakagami and Moure 1967). The concept that the species of Augochlora s. str. are solitary has been challenged by Wcislo et al. (2003), who studied two nests of A. isthmii with more than one female. Limited data suggested that the nests might have been functioning as colonies, raising the question of whether the social behavior within the genus may be more variable than previously thought.
This contribution describes the structure of nests of A. amphitrite, and presents information on the nesting biology of the species. The data are compared to those known for other species of the subgenus.

Study site
The nests were studied in the reserve Refugio Natural Educativo "Ribera Norte" (34°28'10"S, 58°29'40"W), San Isidro, province of Buenos Aires, Argentina. This reserve is on the west margin of the Río de La Plata, and preserves a relict of gallery forest with typical riverine vegetation, including trees such as Ocotea acutifolia (Nees) Mez  (Cabrera and Willink 1973).

Field observations
A nesting site of A. amphitrite was discovered in March 2008, near the end of summer. Nests were observed during seven days (35 work hours), from March 12 to April 30, when the nests were excavated. Another nesting site was located the following year in February. It was observed during three days (15 work hours), from February 7 to 14, when the nests were excavated. Although nests were not found in spring, adults flying over flowers (September to November) were collected and kept for dissection.
The activity of the bees was recorded following the methods described by Michener et al. (1955). Nest entrances were marked individually. When possible, females entering and leaving the nests were marked with a two-color code on the mesoscutum using fingernail enamel. One color was used to indicate to which nest a female belonged, and the second one to discriminate between females of the same nest. The length of activity periods, incidence of sunlight, departures and arrivals, and the presence of pollen loads, were recorded.

Nest extraction and description
The methodology described by Sakagami and Michener (1962) was followed. Talcum powder was blown through the nest entrance to assist us to follow the nest as it was excavated with the aid of a knife and a sharp point. A caliper was used for field measurements. In the laboratory, observations and measurements were made with a stereomicroscope with an ocular micrometer. Measurements are given in centimeters, with mean values and standard errors. The contents of each cell were recorded. Voucher specimens are deposited in the collections of the Museo Argentino de Ciencias Naturales, Buenos Aires.

Dissections
The day of nest excavation, arriving bees as well as those found within the nest, were fixed in Kahle's solution. Presence of pollen loads, ovarian development, and presence of fat tissue, were recorded. Length of the body, maximum width of the eye and maximum width of the gena were taken. All measurement are in millimeters.
Three groups of females are recognized according to their ovarian development. The classification of Michener and Wille (1961) is followed, but simplified. Group A: ovaries large, well developed, usually with one or two eggs ready to be laid; posterior portions swollen forcing one or both ovaries to bend (Fig. 1). Group B: ovaries developed, but without eggs ready to be laid, so ovaries not as large as those of group A, and not bent (Fig. 2). Group C: ovaries not developed (Fig. 3).
The degree of wear of mandibles and wings is indicated in a scale from 0 (intact mandibles and wings) to 3 (much worn mandibles and tattered wings).

Nesting site
An aggregation of 18 nests was found in a fallen trunk of Salix sp. (Salicaceae) on March 12, 2008. The trunk, 3 m long and 0.8 m in diameter, was in an advanced state of decomposition, with soft wood colonized by fungi and various arthropods. Half of the trunk surface was covered by the plant Commelina diffusa Burm. f. (Commelinaceae), but the nests were on the uncovered surface, occupying an area of 0.60 m 2 on the upper and lateral parts. The nest entrances received sunlight from 11:30 to 15:00, being shaded by surrounding trees the rest of the day.
Three nests were found in railroad sleepers made from Schinopsis sp. ("quebracho colorado") (Anacardiaceae) on February 7, 2009. The sleepers (1.0 m long, 0.4 m wide, and 0.15 m thick) lay on the ground, forming the visitors trail in the wettest parts of the reserve. Schinopsis wood is well known for its hardness. The nest entrances were located in knots and cracks, where decomposition had begun to soften the wood. The entrances were on the upper and lateral surfaces, occupying an extension of 0.50 m 2 , and receiving sunlight from 11:00 to 15:00 hours.

Nest architecture
Nests on Salix and on Schinopsis differ considerably in their architecture, mainly in the distribution and arrangement of the cells.
Nest entrances on the trunk of Salix, separated by a minimum distance of 10 cm, presented a ring of compacted sawdust 0.75-1.00 cm in diameter (x -= 0.85 ± 0.08, n= 8) of the same color of the trunk surface. Active nests sometimes presented loose particles beyond the ring, which came from broken nest plugs. The tunnels, all unbranched, penetrated toward the interior of the trunk. They had a length of 7.00-15.00 cm (x -= 9.67 ± 2.56, n= 8), and a diameter of 0.45-0.50 cm (x -= 0.46 ± 0.03, n= 8); their smoothed walls were lined with substrate particles. Each tunnel led to a cluster of 2-10 cells (x -= 5 ± 2, n= 18), irregularly oriented, supported within a cavity by pillars. Two kinds of pillars were observed, those that were remaining parts of the substrate not excavated, and others, more frequent, made of compacted sawdust. The clusters were retrieved intact .
Nests on Schinopsis had shorter tunnels, 2.00-5.00 cm long (x -= 3.10 ± 1.36, n= 3). The soft material of the cracks was used for cell construction. The cells were in small groups or isolated, but without any pillars, and lying against the hard wood, with no surrounding cavity, taking advantage of masses of soft substrate within the crack (Fig. 6). Nests had 8-19 cells (x -= 13 ± 5, n= 3).
Cells of all nests were constructed with compacted particles of ground wood. The external surface was irregular, and the internal surface smooth and shiny, lined with a waxy substance. The cells were ovoid, with the lower surface slightly flattened (Fig. 7); the inner cavity was 0.80-1.45 cm long (x -= 1.07 ± 0.13, n= 72), 0.30-0.60 cm in diameter (x -= 0.46 ± 0.06, n= 72), and 0.25-0.45 cm in cell entrance diameter (x -= 0.36 ± 0.04, n= 72); the cell wall was 0.05-0.30 cm in thickness (x -= 0.09 ± 0.04, n= 72). The cell plugs were made with the same material as the cell walls, 0.15 cm in thickness, dish-shaped, with the outer surface concave. Table 1 summarizes the architectural characteristics of A. amphitrite, comparing them to other species of Augochlora s. str. known to date.

Cell contents
Nests collected from Salix in April had their cells filled with compacted sawdust; they had feces deposited on the posterior portion, oriented toward the bottom of the cell (Fig. 8).
Nests collected from Schinopsis in February were active, and the cell contents consisted of pollen masses with eggs, larvae (in various stages of development), pupae, and a few cells with feces and filled with sawdust ( Table 2). The pupae were all males. The pollen mass, placed near the bottom on the flattened surface of the cell, was slightly wider than long (0.40 × 0.35 cm), 0.43 cm high, and rather spherical, except for the flattened resting surface. The whitish egg was deposited on top of the mass, oriented along the longitudinal axis of the cell (Fig. 7).

Behavioral observations
Females observed leaving and entering nests in March-April did not carry pollen loads. Activity began soon after the sunlight hit the trunk; before that, the entrances were covered with closed tumuli. Flights were inconstant, and up to three females were seen leaving and entering the same nest. The females spent 10-15 minutes perching on the surrounding vegetation, where flying males were also observed. Returning females had erratic flights, and inspected cracks and small holes in the trunk. The three nests collected in February were active. Foraging activity began 15-20 minutes after the sunlight hit the entrances (around 11:20). A female pushed the plug of sawdust with its hind legs, scattering the particles 2-3 cm around the tumulus. After that the female remained at the entrance, with only its head visible, for 3-5 minutes before departing. After 7-10 minutes the same female came back to the nest laden with pollen. Usually, as soon as a female left the nest, another one showed its head at the entrance. When disturbed, the female turned around, plugging the hole with its metasomal terga. Activity continued for approximately 4 hours until no more sunlight bathed the nests (around 15:00). Nest 1 had five females, four of which were captured when returning to the nest (two with, and two without pollen loads); the fifth female was never observed outside the nest and was captured when it was extracted. Another nest had two females; only one of them was observed collecting pollen. The third nest had a single female. Recently emerged males were found in two nests, and males were seen flying in the surroundings of the nesting area and on flowers of Ludwigia (Onagraceae), 50 cm away from the nests.

Discussion and conclusions
The nests of A. amphitrite presented two types of nest architecture according to the substrate where they were built. Common features to both types were the entrance surrounded by a ring of compacted sawdust, and the unbranched tunnels leading to the cells. Cells of all nests had the same structure, and similar proportions to those of other species of Augochlora s. str. (Table 1).
Nests constructed in the thick trunk of Salix, with a large mass soft wood, had the cells grouped in clusters surrounded by a gallery of similar diameter to that of the tunnel, and supported by a varying number of pillars. Nests constructed in the decomposing parts of the cracks and knots of the hard wood of Schinopsis had the cells toward the end of the tunnel, constructed against the hard walls, without any pillars or surrounding cavity. In both cases the number of cells was variable and the orientation of the cells irregularly radiated.
Cluster nests are known for A. pura and A. sidaefoliae, while studied nests of A. isthmii, A. alexanderi, A. hallinani and A. smaragdina had tunnel nests with sessile cells distributed along the tunnel, and A. essox had nests with grouped cells, but not forming clusters (Table 1) (Stockhammer 1966, Eickwort and Eickwort 1973, Zillikens et al. 2001, Wcislo et al. 2003. A nest with a cluster of cells supported by pillars within a cavity is the predominant and probably plesiomorphic type of nest within the tribe Augochlorini (Eickwort and Sakagami 1979, Danforth and Eickwort 1997, Engel 2000, and it is the plesiomorphic condition for the monophyletic Augochlora genus-group (Engel 2000, Coelho 2004), indicating that departures from the cluster type of nest are derived conditions within Augochlora. Although the nests of few species of Augochlora s. str. have been studied, their structure is highly variable. The two types of nests found in the present study indicate that this variation can be intraspecific, coincidently with the variation found by Stockhammer (1966) for A. pura, which had tunnel, planiform, and cluster nests, with some intermediate forms. This variation observed within Augochlora s. str., greater than in other genera of Augochlorini, would result from the irregularity of the nesting substrate. Species that nest in the soil, may be less constrained by the substrate, and can fully express their behavioral capabilities. Species of Augochlora nest on the substrate offered by diverse plants, usually trees, but also bromeliads (Zillikens et al. 2001), which offer a heterogeneous supply regarding the size, shape and degree of decomposition of the nesting sites. Probably all species of Augochlora s. str. can construct well defined clusters when an unconstrained substrate is available, as is the case in A. pura and A. amphitrite.
The daily activity pattern of the females was limited by the forest environment where the bees were studied. Females left the nests to collect pollen during a period of 3.5-4 hours, while the sunlight hit the nesting site.
The annual cycle in the study area was typical of the cycle of most halictids in temperate regions, although the winters in the study area are mild and the temperatures in July are rarely freezing. Activity begins in spring (September-October), when posthibernating females begin to visit flowers. Females captured at this time showed well developed ovaries and would be the foundresses of the first nests. The activity continues until mid March, when the nests become inactive and females of the last generation are looking for hibernacula. Females captured at this time had undeveloped ovaries and abundant fat tissue.
Nests studied in summer (February) contained larvae in various stages of development, male adults, male pupae, and adult females, which correspond at least to the first brood of the foundress (Table 2). Although possible, we are not certain whether more generations are bred between spring and mid-summer. A further brood is produced by the end of summer, so at least two broods are produced during the activity cycle. Although most females collected in February had developed ovaries, one female, from nest 1, had the ovaries undeveloped and actively collected pollen. It also had worn mandibles and wings, all characteristics of the worker caste in social halictids (Michener et al. 1955). Also, the presence within the excavated nests of females that were never observed outside the nest, is indicative of a social division of tasks.
Values taken from the fixed females show size variation among females with enlarged ovaries. The female that was never observed outside the nest in multi-female nest 1 was distinctly larger than the others in the same nest. It also had an allometrically enlarged head, with a broad gena. Females of A. amphitrite have distinct cephalic polymorphism (Dalmazzo and Roig-Alsina 2011), which can be indicated by the maximum width of the eye -maximum width of the gena coefficient. A few other females of nests 1-3 had moderately enlarged heads.
Although the number of studied nests is very low, the information recovered is suggestive of social behavior in A. amphitrite. Wcislo et al. (2003) reached similar conclusions for A. isthmii, pointing out that social behavior within Augochlora s.str. is more variable than previously thought, since members of the subgenus had been considered as solitary and derived from an eusocial ancestor (Eickwort 1969, Michener 1990, Danforth and Eickwort 1997, Wcislo and Danforth 1997, Engel 2000, Brady et al. 2006. Further studies are needed, both in the field and laboratory, to understand the degree of sociality in Augochlora, and whether its occurrence is widespread in the subgenus. Eickwort GC, Eickwort KR (1972)   2006). The subfamily has a worldwide distribution and members are found in most terrestrial habitats. The history of higher classification of the Agathidinae was summarized by Sharkey (1992) who also proposed a tribal level classification based on ground-plan coding. Sharkey et al. (2006) conducted phylogenetic analyses based on morphology and the D2-3 regions of 28S rDNA. The Oriental fauna of Agathidinae was first revised by Bhat and Gupta (1977) and they provided a detailed history of taxonomic research for the area. Sharkey et al. (2009)  Species concepts are based on morphological data and cytochrome c oxidase (COI) data. Phenetic and phylogenetic trees, using 558 base pairs of COI data, were constructed using neighbor-joining (NJ), maximum parsimony (MP) and Bayesian methods. MP was performed using TNT (Goloboff et. al, 2008) [traditional search with 100 random addition sequences followed by branch-swapping, saving 100 trees per replication; 1000 bootstrap replications were used to estimate branch reliability]. The Bayesian analysis was performed using MrBayes v3.1.2 (Ronquist and Huelsenbeck 2003). Best-fitting DNA substitution models were determined using MrModeltest2.2 (Nylander 2004). The general time reversible model of evolution with a parameter for invariant sites and rate heterogeneity modeled under a gamma distribution (GTR+I+G) was determined as the best-fitting model. The Bayesian analysis consisted of two independent Bayesian MCMC runs initiated from different random starting trees. The analysis ran for 2,000,000 generations, reaching a topological similarity criterion of 0.01; trees were sampled every 500 generations. 25% of the trees from each run were removed as burn-in upon topological convergence. The NJ tree was produced from PAUP* (Swofford 2002) using default settings. Figure  2 presents the NJ tree, which was much more resolved than were the phylogenetic trees produced by MP and Bayesian analyses. We mapped the support values of the Bayesian and MP analyses on the NJ tree. Branches without values are those that collapsed in the phylogenetic analyses.
Morphological terms follow Sharkey and Wharton (1997) except for the following: measurements are given for the length and apical width of the first metasomal mediotergite (MT1). Measurement of the apical width is straightforward, however since the base of the tergite is usually hidden from view it is difficult to measure the total length. Instead we measure from the apex of the large tendon that emanates from the propodeum and inserts near the base of the median tergite. Abbreviations used in text: S1, S2, S3; metasomal mediosternite 1, 2, 3; MT1, MT2, MT3: metasomal mediotergite 1, 2, 3; LT1, LT2, LT3: metasomal laterotergite 1, 2, 3.
Morphological terms used in this revision were matched to the Hymenoptera Anatomy Ontology (HAO, Yoder et al. 2010) (Appendix 4). Identifiers (URIs) in the format http://purl.obolibrary.org/obo/HAO_XXXXXXX represent anatomical concepts in HAO version http://purl.obolibrary.org/obo/hao/2011-05-18/hao.owl. They are provided to enable readers to confirm their understanding of the anatomical structures being referenced. To find out more about a given structure, including, images, references, and other metadata, use the identifier as a web-link, or use the HAO:XXXXXXX (note colon replaces underscore) as a search term at http://glossary.hymao.org.
All 19 species are treated with a diagnosis and distributional data. They are illustrated with color photos using a JVC digital camera mounted on a Leica MZ16 microscope and Automontage® stacking software. Distributional data are listed for all species and a Google map via Berkeley Mapper is included for all species. The descriptions are of the holotype and variation is given in parentheses.
The source files for the keys, descriptions, illustrations, DNA sequence and distributional data are all freely available to future researchers who may wish to build on this beginning. DNA trace files and primer information are available through the Barcode of Life Data system (BOLD) [Ratnasingham and Hebert 2007] at http:// www.boldsystems.org. Sixteen of the twenty five Zelodia COI sequences were generated by BOLD (project ASTRK Revisions of Thai Agathidinae Braconids), the remaining nine were generated in the Sharkey lab. All sequences have been deposited in GenBank database (JQ763436-JQ763460). All twenty five Zelodia COI sequences are characterized by a -1 frameshift mutation. A majority of Agathidinae CO1 sequences are distinguished by a series of 1 bp deletions which are not restricted to one portion of the barcode region. Codon composition in sequences with 1bp deletions remains highly biased towards AT and substitutions remain biased towards 3 rd codon position (M. Alex Smith pers. comm.). It is suggested that the genes are correctly decoded by a programmed frameshift during translation (Beckenbach et al. 2005) and are functional.
Distribution data, pdf's of non-copyright references, images, notes, and host and type information can be found by searching TaxaBank (a combined specimen and taxonomic database; http://purl.org/taxabank). Codes beginning with an "H" and followed by numbers are unique identifiers used for specimens in the Sharkey lab at the University of Kentucky, and in the specimen database TaxaBank (e.g., H647).
Abbreviations used for specimen depositories are as follows:

BMNH
The Natural History Museum, London, England.

Results
Refer to the tree in Sharkey and Clutts (2010) for generic level placement. The host lepidopterans of the genus are unknown. The neighbor joining branching diagram in Figure 2, based on COI mtDNA, was used to help in determining species limits although we did not devise any cutoff threshold to delimit species. We conducted Bayesian and parsimony analyses and where these agreed with the NJ tree the support values are given in Figure 2. An examination of this figure shows that the NJ diagram and the phylogenetic analyses were in close agreement. We also used morphological differences to delimit species. For example, Z. saksiti and Z. charoeni are very similar (1/558 bp difference) in COI sequences but are very different morphologically. Contrastingly, Zelodia wangi is a widespread species and COI sequences show some variation (2/558 bp difference), however we could find no morphological differences. The complex may represent several species. taxonomy

Key to Thai species of Zelodia
Distribution. Widespread in western Malaysia and likely present in southern Thailand. Distribution map can be found at http://purl.org/thaimap/idrisi
Distribution. Recorded from southeastern and northwestern Thailand. Distribution map can be found at http://purl.org/thaimap/nopadoli Etymology. Dedicated to Mr. Nopadol Nachin, chief of Tad Tone National Park.   Fig. 11 Diagnosis. Mesoscutum mostly melanic except margins pale; hind tibia mostly pale except apex and extreme base black; hind femur melanic; apex of fore wing hyaline with white setae.
Distribution. Known only from the type locality in western Malaysia but likely to be found in peninsular Thailand. Distribution map can be found at http://purl.org/ thaimap/pahangensis Etymology.    Fig. 13 Diagnosis. Mesoscutum mostly melanic except margins pale; hind tibia all melanic.
Molecular data. TaxaBank#/BOLD Process ID/Genbank Accession: H381/ ATRMK205-11/JQ763448. Distribution. Known only from the type locality in northwestern Thailand. Distribution map can be found at http://purl.org/thaimap/toyae Etymology. Dedicated to Ms. Chayanit (Toy) Satatha. Toy was the sorter of Diptera for the TIGER project and is currently a technician at QSBG. The species name reflects her nick-name, Toy.
Molecular data.  , km33/helipad, 735m, 12.836°N, 99.345°E, MT, 11-18.v.2009. Paratype. ♀, Thailand, Kaeng Krachan NP, Panernthung/km27, 950m, 12.822°N, 99.371°E, MT, 8-15.vi.2009. The genus Aphelinus (Hymenoptera: Aphelinidae) comprises 84 recognized species (Noyes 2011), all of which are internal parasitoids of aphids. There are several complexes of closely related species in the genus, and identification of species within these complexes has been problematic (Heraty et al. 2007), leading to a confused literature on host specificity. The Aphelinus mali complex comprises 11 described species (Ashmead 1888; Evans et al. 1995;Gahan 1924;Girault 1913;Haldeman 1851;Hayat 1998;Prinsloo and Neser 1994;Timberlake 1924;Yasnosh 1963;Zehavi and Rosen 1988), and there are 6 other closely related species (Carver 1980;Hayat 1998;Howard 1917;Kurdjumov 1913;Walker 1839;Yasnosh 1963) that differ from the members of the complex in one or two traits ( Table 1). The species within the complex have diverged little in morphology so the taxonomy within the complex has been confused, and many specimens have been identified as A. mali (Haldemann) or A. gossypii Timberlake that are likely different species based on differences in the aphid host species and geographical regions from which they were collected. We provide a key to identification of species in the A. mali complex based on 19 traits coded primarily from species descriptions. We describe three new species that were collected in China and Korea during exploration for natural enemies of the soybean aphid, Aphis glycines Matsumura, all of which are candidates for biological control of this important pest.
The genus Aphelinus consists of several complexes of cryptic species including the mali complex, the varipes complex (Heraty et al. 2007), the asychis complex (Kazmer et al. 1995(Kazmer et al. , 1996, the perpallidus complex (unpublished data), and possibly others. Cryptic species are closely related species that differ little in the morphological features used for taxonomy, but differ critically in physiological, behavioral and ecological traits, such as climatic adaptation and host range (Darling and Werren 1990;DeBach 1969). Recent evidence from molecular studies suggests that cryptic species of hymenopteran parasitoids may be far more common than previously realized (Campbell et al. 1993;Clarke and Walter 1995;Darling and Werren 1990;Kankare et al. 2005a;Kankare et al. 2005b;Kazmer et al. 1996;Molbo et al. 2003;Rincon et al. 2006;Stouthamer et al. 2000;Stouthamer et al. 1999). The success of biological control programs depends on accurate species-level identifications of hosts and natural enemies, but choosing the best parasitoids for biological control programs is complicated by cryptic species (Rosen 1986;Wharton et al. 1990). Because cryptic species are difficult to recognize, studies on host ranges of parasitoids have often confounded more than one parasitoid species (Clarke and Walter 1995;Hopper et al. 1993), making analysis and prediction of host range difficult. Heightened concern about potential impacts of introduced parasitoids on non-target species makes accurate prediction of host range crucial to biological control introductions.

Methods
Three new species in the A. mali complex were collected from Aphis glycines in the Peoples Republic of China near Beijing and Xiuyan (Liaoning Province) and in the Republic of South Korea near Miryang (Gyeongsangnam Province) and maintained as laboratory cultures at the Beneficial Insects Introduction Research Unit, USDA-ARS, Newark, DE. All of the specimens described below were taken from lab cultures, killed in 95% ethanol, and most were critical-point-dried and card-mounted. Selected specimens were then slide-mounted in Canada balsam. Specimens photographed for coloration (15)(16)(17)(18)(19)(20)(29)(30)(31)(32)(33)(34) were killed in ethanol and photographed as soon as possible, by placing specimens on a layer of KY® jelly in a small watch glass, submerging the specimen in ethanol, and photographing using a Leica MZ 16 stereomicroscope, fiber optic illumination, a Zeiss Axiomat MRc5 camera, and Helicon Pro image-stacking software. Slide-mounted specimens were photographed using differential interference contrast optics (DIC) with an Olympus BH2 compound microscope, and the same camera and software. Final modifications to images were made using Adobe Photoshop, Adobe Lightroom, and Adobe InDe- Carver 1980 sharpae 3 Hayat 1998 brunneus 4 Yasnosh 1963 daucicola 4 Kurdjumov 1913 lapisligni 4 Howard 1917 1 new species described in this paper 2 insufficient description to be included in tree or key 3 difference from mali complex: more than 1 line of setae in delta region 4 difference from mali complex: posterior femur dark sign. Type material and other specimens examined have been deposited as indicated in the species descriptions. The label data for each specimen has been digitized and all specimens bear individual accession numbers for Texas A&M University Insect Collection (e.g. TAMU x0616203), as well as a machine-readable bar-code. In the verbatim label data provided for holotypes, a single | symbol indicates a new line on a label, and the || symbol indicates a second or third label. Vouchers are maintained at -20 o C in molecular grade ethanol at the Beneficial Insect Introduction Research Unit, Newark, Delaware, and at the Department of Entomology, Texas A&M University, College Station, Texas. We tabulated and coded 19 traits for species in the A. mali complex, using the original species descriptions for the most part. These traits included color of scape, pedicel, club, coxae, femora, tibiae, tarsi, and metasoma, as well as shape of third funicle and club (length:width) and length of ovipositor relative to mesotibia. For some traits, males and females differed (e.g., F3 shape, procoxae color) and the values were scored separately. When trait data were lacking from original descriptions, we used data from later descriptions. Trait values for the new species in the complex were taken from specimens freshly killed in ethanol and slide-mounted specimens. These traits were used to construct an on-line, interactive, multiple entry identification key to the mali complex which is available on request. Of the 19 traits, 12 proved to be most consistent and useful in distinguishing species, and these are presented in Table 2. Table 3 is a list of anatomical terms used in the paper followed by URI values (uniform resource identifiers), that will link the terms to precise definitions and illustrations in the Hymenoptera Anatomy Ontology project (see http://portal.hymao.org and http://hymao.org for more information on this initiative). Additional information on morphological terminology in Chalcidoidea is available in Gibson (1997) and http://www.canacoll.org/Hym/Staff/Gibson/apss/chglintr.htm.
The ventral surface of the antennal scape refers to the surface that is ventral when the antennae are deployed, or anterior when the antennae are folded on the face. F1, F2 and F3 refer to the first, second and third segments of the funicle of the antennal flagellum, respectively. T1, T2 etc. refer to metasomal terga. We use the term ovipositor to refer to the anatomical cluster consisting of the first valvula, second valvula, third valvula, first valvifer and second valvifer. Length of the ovipositor is the measurement (generally of a slide-mounted specimen) from the anterior margin of the second valvifer to the posterior (distal) end of the third valvula.

Results and discussion
Following the work of Zehavi and Rosen (1988), we consider the following traits to be diagnostic for the A. mali complex: (1) head and body dark except for parts of the metasoma; (2) metafemur pale, (3) a single complete row of setae proximal to the linea calva of the fore wing, with a few additional setae in the angle between this row and the marginal vein; (4) linea calva open (no setae at its posterior edge); (5) mesoand metacoxae dark; (6) metatibia more or less dark. The A. mali complex consists of eleven described species, and there are six similar species with either a dark metafemur or more than one line of setae proximal to the linea calva (Table 1). Species within the complex have been distinguished by color and shape of antennal segments (particularly the third funicular segment), color of legs and metasoma, and relative length of ovipositor versus mesotibia (Ashmead 1888; Evans et al. 1995;Gahan 1924;Girault 1913;Haldeman 1851;Hayat 1998;Prinsloo and Neser 1994;Timberlake 1924;Yasnosh 1963;Zehavi and Rosen 1988). Diagnosis. Female. Head and mesosoma dark brown to black; legs with procoxa yellowish white, meso-and metacoxae dark brown to black, femora yellowish white, protibia yellowish white, mesotibia yellowish white with center greyish, metatibia dark grey to black with base pale; metasoma with base, apex, and venter yellow, remainder brown; antenna white to yellowish white; F3 1.3-1.7 times as long as broad; club 3.2 times as long as broad. Male similar except procoxa grey; pro-and mesofemur sometimes with darkened center; metasoma brown with base and apex yellow; scape dark greyish brown with greyish yellow distal tip, swollen in center, maximum width 3× distal end width, with three to five volcano-shaped secretory pores in a single line on ventral surface, pedicel greyish yellow, third funicle more than 2 times as long as broad, club 3.9 times as long as broad. Description. Female (Figs 2,4,6,8,10,11,12,13). Body length. 0.77-0.93 (Holotype 0.90 mm).
Hosts. In the field, Aphis glycines is the only known host. In laboratory experiments, A. glycinis parasitizes A. glycines and closely related species in the genus Aphis.
Etymology. This species is named for the host from which it was collected. The species epithet is a noun in genitive case.
Relationships. Aphelinus glycinis is closest to A. engaeus and A. ficusae Prinsloo and Neser based on our matrix of traits (Table 2). Aphelinus glycinis differs from A. engaeus in having elongated third funicle segments in males and females, and it differs from A. ficusae in having an ovipositor more than 1.2× as long as the mesotibia and grey procoxa in males. It also differs from these species in its aphid hosts and geographical distribution. Aphelinus glycinis is a specialist on Aphis species close to Aphis glycines, but A. engaeus is reported from Schizaphis graminum (Rondani) and Sitobion ochnearum (Eastop) and A. ficusae was reared from an undetermined aphid on Ficus sycomorus (Prinsloo and Neser 1994 Diagnosis. Females. Head and mesosoma dark brown to black; legs with coxae dark brown to black, profemur dark grey with pale apex, mesofemur dark grey to black, metafemur white, protibia white with pale greyish base, mesotibia dark grey to black with pale base and apex, and metatibia dark grey to black with pale base; metasoma yellowish brown with base and apex yellow; antennae yellow with basal half of scape and pedicel sometimes greyish; F3 quadrate; club 2.8 times as long as broad. Males similar except scape swollen in middle, 3× broader in middle than at distal end, with 2 or 3 volcano-shaped secretory pores; scape dark yellowish grey, pedicel pale greyish yellow; club 3.3 times as long as broad. Description. Female (Figs 16,18,20,22,24,25,26,27). Body length. 0.75-0.94 (Holotype 0.87 mm).
Head. (Figs 16, 22) Head 1.2× as broad as high in frontal view, about as broad as mesosoma; frontovertex 0.4× head width and as broad as scape length; posterior ocelli 0.5× their diameter from eye margin, 3.0× their diameter from one another, and 0.33× their diameter from occipital margin; mandible with 2 acute teeth and a broad truncate surface below the teeth; antennae as in Fig. 21 with scape 4.8 longer than broad, pedicel 1.8× as long as broad, F1 anneliform, F2 1.5× as broad as long, F3 quadrate, club 2.8× as long as broad and 3.3× times longer than F3, with 4-6 longitudinal sensilla.
Host. In the field, Aphis glycines is the only known host. In laboratory experiments, A. rhamni parasitizes A. glycines and closely related species in the genus Aphis, and rarely Rhopalosiphum padi L. and Schizaphis graminum.
Etymology. This species is named for the primary host plant of the aphid species from which it was collected. The species epithet is a noun in genitive case.
Relationships. Aphelinus campestris and Aphelinus gossypii are the closest described species to A. rhamni based on our matrix of traits (Table 2). Aphelinus rhamni differs from both species in having a more elongate club and in coloration of the metatibia. Aphelinus rhamni has a much narrower host range than A. gossypii, which is reported from at least 18 species of aphids in 10 genera and two tribes, including species which A. rhamni does not parasitize in laboratory experiments.
Aphelinus coreae sp. n. urn:lsid:zoobank.org:act:F4B3A880-2136-474C-815C-13406F2A48A0 http://species-id.net/Aphelinus_coreae Figs 29-42 Diagnosis. Females. Head and thorax dark brown to black; legs with coxae dark brown to black, profemur dark grey with distal half pale, mesofemur dark grey to black, metafemur pale yellowish white, protibia pale yellowish white to somewhat fuscous, mesotibia dark grey to black with distal half pale, and metatibia dark grey to black with pale base; metasoma dark brown with base and apex yellow; antennae yellow; F3 quadrate. Males similar except scape swollen in middle, 2.0× as broad in middle than at distal end, with two or occasionally three circular secretory pores in the middle of a shallow depression on ventral surface, scape dark yellowish grey with distal half yellow, pedicel greyish yellow.
Head. (Figs 30, 36) Head 1.3× as broad as high in frontal view, about as broad as mesosoma; frontovertex 0.4× head width and as broad as scape length; posterior ocelli approximately their own diameter from eye margin, 5× their diameter from one another, and 0.5× their diameter from occipital margin; mandible with two acute teeth and a broad truncate surface below the teeth, ventral tooth sometimes not distinct; antennae as in Figs 30 and 36 with scape 4.0× as long as broad, pedicel 1.6× as long as broad, F1 anneliform, F2 1.4× as broad as long, F3 subquadrate or very slightly broader than long, club 3.75× as long as broad and 3.5× longer than F3, with 7-8 linear sensilla.
Mesosoma. (Figs 32, 34, 41) Mesosoma and scutellum with fine reticulate sculpture, longest diameter of reticulations approximately 2-3× diameter of scutellar sensilla; interior of reticulations with fine, granulate surface (visible only in slide-mounts under high magnification), mid-lobe of mesoscutum with 2 pairs of long setae and about 40-60 short setae, side lobes each with 2 long and 1-2 short setae; scutellum with 2 pairs of long setae; pair of scutellar sensilla approximately equidistant from anterior and posterior pairs of long setae; mesotibial spur 1.1× mesobasitarsus; metatibial spur 0.6× metabasitarsus. and mesotibia. Like A. rhamni, A. coreae has a much narrower host range than A. gossypii. Aphelinus coreae is very close to A. rhamni, but male A. coreae have shorter clubs and, as noted in the key, the two species differ in coloration of scape and mesotibia. Although difficult to distinguish, these species are reproductively isolated in laboratory crosses. Their DNA differs by 2130 fixed substitutions and 293 indels across 1.8 megabases of homologous DNA sequence. They also differ in host specificity: A. coreae parasitizes species of Aphis, e.g. A. nerii Boyer de Fonscolombe and A. rumicis L., not parasitized by A. rhamni in laboratory experiments.