Like other microgastrines, adult Glyptapanteles are free-living wasps that feed primarily on nectar, pollen, or secretions from scales and aphids (Landis et al. 2000), while larvae develop inside caterpillars. The female wasp penetrates the host cuticle with her ovipositor and oviposits eggs, which float freely in the hemolymph. Independent of the number of eggs deposited, in solitary parasitoid species only one larva completes development, whereas, in gregarious parasitoid species, more than one offspring successfully completes development (Hanson and Gauld 2006). All Glyptapanteles are endoparasitoid koinobionts, meaning that their hosts continue to develop after being attacked. Inside the host, the egg absorbs proteins through a specialized extra-embryonic membrane before hatching (Jervis et al. 2001, 2008). The wasp larvae develop by consuming only the hemolymph and fat body of the host (Shaw and Huddleston 1991). The host eventually dies, although usually not until the parasitoid larva or larvae have emerged through the host cuticle. At the end its second larval instar, each parasitoid larva emerges by burrowing through its host’s cuticle (Godfray 1994). Outside the host, the parasitoids molt to their third and final larval instar, then spin individual cocoons and pupate in the cocoon, on the outer surface of the host or nearby (Fig. 4; Calkins and Sutter 1976).

This ancient physiological interaction of microgastrine endoparasitoids with their hosts is in part mediated by a fascinating mutualistic association with polydnaviruses (PDVs), an alliance that arose 73±11 mya (Whitfield 2002, though an older origin, ~100 mya, is proposed by Murphy et al. 2008). These double-stranded DNA viruses are integrated into the genome (proviral DNA) of both male and female wasps. However, the largest fragment is excised from the female wasp genome only during her later pupal and adult stages. Thus, PDVs are only “free-living” during reproduction in wasp ovarian calyx tissue or after injection into host hemocoel’s (D’Amico and Slavicek 2012). Once in the larval host, PDVs, in concert with other injected material such as toxins and ovarian fluids, play a crucial role in the survival of the developing egg (Lapointe et al. 2007) by suppressing or misdirecting host immune systems and arresting host development.

Biologically, prior to the two inventories referenced here, Glyptapanteles was still poorly known in the Neotropics. Those host records were mainly restricted to Noctuidae, Geometridae, Pieridae, Notodontidae, and Megalopygidae. Noctuid hosts included Agrotis deprivata Walker feeding on Brassica oleracea [wild cabbage, Brassicaceae], Medicago sativa [Alfalfa, Fabaceae], Vicia villosa [hairy vetch, Fabaceae], and Zea mays [corn, Poaceae]; A. gypaetina Guenée and A. ipsilon (Hufnagel) feeding on Brassica oleracea, Daucus carota [wild carrot, Apiaceae], Helianthus annuus [sunflower, Asteraceae], Lactuca sativa [lettuce, Asteraceae], Medicago sativa, and Trifolium repens [white clover, Fabaceae]; Peridroma saucia (Hübner) feeding on Trifolium repens and Medicago sativa; Helicoverpa zea (Boddie) feeding on Zea mays (Corn, Poaceae); and Anticarsia gemmatalis (Hübner), Mythimna unipuncta (Haworth), Peridroma margaritasa (Haworth), Pseudoplusia includens (Walker), and Trichoplusia ni (Hübner) for which the food plants were not reported (Blanchard 1936, Whitfield et al. 2002a, Baudino 2005). Geometridae hosts included Thyrinteina leucocerae (Rindge) feeding on Psidium guajava [guava, Myrtaceae] and Eucalyptus grandisand [eucalyptus, Myrtaceae] (Grosman et al. 2008); Cyclomia mopsaria Guenée, Glena sp., and Physocleora sp. feeding on Erythroxylum microphyllum (Erythroxylaceae) (Marconato et al. 2008). Pieridae host was Dismorphia crisia lubina (Blater) feeding on Inga mortoniana (Fabacae, Koptur 1989). Notodontidae host includes Nystalea nyseus (Cramer) (Whitfield et al. 2002a). One host genus from Megalopygidae has been reported (Plodia Guenée, sp.?) feeding on Inga densiflora (Koptur 1989). Six additional lepidopteran host families also have been reported: Apatelodidae, Erebidae, Limacodidae, Nymphalidae, Pyralidae, and Saturniidae, although no species names were referenced (Whitfield et al. 2009).

Glyptapanteles was described in 1904 (Ashmead 1904), but its legitimacy as a distinct genus has only been accepted since 1981. Ashmead segregated it from Apanteles and erected the genus based on two females and one male collected from Manila, Philippines (Ashmead 1904). The description of Glyptapanteles coincided with an epoch of proliferation of descriptions of new genera within Microgastrinae in an attempt to subdivide the gigantic genus Apanteles. The advent of these newly described taxa only increased taxonomic confusion within the subfamily, due in large part to the fact that the worldwide fauna was both taxonomically not well-studied and the species descriptions were inadequate to enable taxonomists to categorize morphospecies among the plethora of new species encountered. Consequently, contemporary colleagues found it difficult to interpret those classifications and opted to ignore them.

In an attempt to stop this rapid increase in the production of generic names while also stabilizing the nomenclature, Muesebeck (1920, 1922) proposed a conservative approach and synonymized many genera, subsuming Glyptapanteles again into Apanteles (Whitfield et al. 2002b). Later, Nixon (1965, 1973) offered a reclassification of Microgastrinae, recognizing only one tribe (Microgasterini) with 19 genera, and Glyptapanteles was not considered to be a valid genus. Nixon kept a broad definition of Apanteles, dividing it into 44 species groups and suggesting that the genus ultimately should be split into more genera. Sixteen years later, Mason (1981) succeeded in subdividing the gigantic genus Apanteles and proposed a radical generic classification of the subfamily based on re-grouping of existing species groups, in which Glyptapanteles, one of the larger segregates of the Apanteles, was recognized as a valid genus in the tribe Cotesiini (Nixon’s Microgasterini therefore, was discarded). Mason divided Microgastrinae into five tribes (Apantelini, Cotesiini, Forniciini, Microgastrini, and Microplitini) based mainly on the association between female genitalic structures and host use. In his system, females that attack Microlepidoptera larvae possess long ovipositors, presumably an adaptation for reaching the small and cryptic Microlepidoptera caterpillars that feed in semi-concealed situations such as leaf rolls or silk webs. In contrast, females with short ovipositors attack Macrolepidoptera larvae, which are large caterpillars and live fully exposed on vegetation throughout their larval stages (Shaw and Huddleston 1991). In his reclassification, Mason assigned Glyptapanteles to the “Macrolepidoptera-attacking” Microgastrinae. However, host data now known for many genera have not substantiated Mason’s hypothesized associations, at least at the generic level; as a result, the tribal classification for the subfamily has been largely abandoned and the current classification is now based on genera and not tribes (Valerio et al. 2009, Whitfield et al. 2018).

Issues concerning Glyptapanteles taxonomy have partly contributed to the poor documentation of its natural history. Characteristics such as high incidence of morphological convergence and lack of obvious discrete morphological variation (character reduction) within both the genus and subfamily, together hinder or virtually preclude straightforward morphological identification of specimens (Shaw and Huddleston 1991). Other factors that possibly have contributed to the poor taxonomic state and its avoidance by taxonomists include its previously mentioned astonishing diversity, total lack of striking coloration (e.g., mainly black or dark brown), and the minute body size of its specimens (2–3 mm long). Taxonomic groups with small body sizes tend to be described much later than taxa with large body sizes (Jones et al. 2009). The poor understanding of Glyptapanteles diversity in the tropics could also be a consequence of other factors such as specialized ecological niches, relatively small population sizes, and the concealed parasitoid life history (Stireman et al. 2009).

As currently delineated, Glyptapanteles is the result of the fusion of several Apanteles species groups. Mason (1981) transferred seven Apanteles species groups from Nixon’s classification (1965, 1973) to the newly established Glyptapanteles. Those species groups were: the octonarius group, siderion group, vitripennis group, and the four monotypics demeter group, fraternus group, pallipes group, and triangulator group. The first thing that stands out in this classification is the pattern of reliance on the external morphological characters used. Those species groups were characterized not by any exclusive feature, but by the combination of features. Thus, the high level of homoplasy makes Glyptapanteles a genus with boundaries poorly defined and taxonomically difficult among Microgastrinae. Additionally, these species groups were based mainly on the north-western European fauna instead of the Neotropical fauna. The traits-per-group proposed by Nixon are listed below.

The octonarius group was distinguished mainly by the weak and even curvature at the junction of the r and 2RS veins on fore wing (e.g., G. ronaldzunigai, Figs 193L, 194K), whereas the remaining groups have that junction distinctly angled (e.g., G. diegocamposi, Fig. 70K), it is distributed worldwide (Nixon 1973), and the vast majority of species reassigned by Mason to Glyptapanteles belong mainly to here. In the siderion group, the petiole shows a wide base and a very narrow apex, the propodeum is short and has a well-defined median longitudinal carina, the phragma of the scutellum is hidden, the inner side of the distal margin of the hind tibia has a dense silky fringe of setae that is entirely differentiated from the normal tibial setae, and it is distributed in Java and the Philippines (Nixon 1965). In the vitripennis group, the phragma of the scutellum is always visible, the metanotum is always without a lateral, forward-pointing projection, proximally the median area on T2 is extensively polished, otherwise usually with at least some minute striae or punctation laterally, the inner spur of the hind tibia is longer than the outer spur, reaching to middle of hind basitarsus, and it is distributed worldwide (Nixon 1965, 1973). Specimens belonging to the demeter group have a very short and thick antenna, only the proximal and the distal antennal flagellomeres are longer than wide, the phragma of the scutellum is concealed, the propodeum is coarsely rugose, and it is distributed in New Zealand (Nixon 1965). In the fraternus group, the median area on T2 is polished, the inner spur of the hind tibia is slightly longer than the outer one, not reaching beyond middle of hind basitarsus, and it is distributed widely in the northwest of Europe (Nixon 1973). In the pallipes group, the propodeum is rugose all over and possesses a strong median longitudinal carina, the pronotum lacks a dorsal furrow, the petiole is twice as long as wide, the phragma of the scutellum is widely visible, and it is distributed in the northwest of Europe and widely in North America (Nixon 1965, 1973). In the triangulator group, the median area on T2 can be polished or with minute punctation, the edges on the median area lose definition distally, but have weak longitudinal sculpture, the petiole has parallel sides, but is distally rounded, the phragma of scutellum is narrowly visible, the propodeum is strongly shining, almost unsculptured, except toward distal corners, both spurs of the hind tibia are short and subequal in length, the outer side of hind tibia has dense strong spines, and it is distributed in France, Germany, and England (Nixon 1973). As mentioned previously, it became clear that the overlap of external morphological characters among species group makes the identification of the Glyptapanteles species more doubtful and problematic.

Complications in the definition of Glyptapanteles also arise because a suite of external morphological characters that had been used for distinguishing them from the rest of Microgastrinae was later found to be shared with other genera. Currently, five genera within the subfamily have often been confused with Glyptapanteles; these are Cotesia Cameron, Distatrix Mason, Lathrapanteles Williams, Protapanteles Ashmead, and Sathon Mason. All five genera share: the hypopygium is evenly sclerotized from side to side and the fore wing with second r-m vein absent, so that the small areolet (second submarginal cell) is open distally (Mason 1981, Whitfield 1997). The less than fully diagnostic characters that complicate the separation of Glyptapanteles with those five genera are outlined here below.

Glyptapanteles and Cotesia. In Cotesia the petiole is virtually never narrower at its apex, the usual shape is a little longer than wide and broadened distally, but occasionally it can be wider than long or somewhat barrel-shaped (Mason 1981) or unusual narrowing at midlength (Gupta et al. 2016a); frequently the petiole on T1 is smooth proximally, but distally always with sculpture; the shape of the median area on T2 is variable, usually subrectangular, but it can be truncated trapezoidal or semicircular; and the propodeum is rugose, usually with a median longitudinal carina that may be partially obscured by rugosity and with an incomplete transverse carina laterally separating the rugose declivity from a smoother proximal area (Mason 1981). Cotesia tends to be a more dominant faunal component in temperate regions worldwide (Whitfield et al. 2009), but in the tropics, Cotesia is displaced ecologically by Glyptapanteles (Mason 1981). Some Neotropical Glyptapanteles collected at high elevations (> 1,000 m) seem to share morphological similarities present in Holarctic Cotesia [e.g., propodeum rugose (as in G. erictepei, Figs 80G, 81F; G. felipesotoi, Figs 82F, 83C) and petiole barrel-shape (as in G. marcpolleti, Fig. 151G, H)]. Some Glyptapanteles in the Australasian region exhibit coarsely rugose tergites, and sometimes they are confused with Cotesia. However, both genera can be separated by the shape of petiole on T1 and the median area on T2 (Austin and Dangerfield 1992).

Glyptapanteles and Distatrix. Distatrix is an unusual genus, with coloration partly xanthic (brownish yellow), large eyes (sometimes only in one sex), pedunculate cocoons, and is relatively rare in collections (Whitfield and Scaccia 1996, Whitfield et al. 2009). Distatrix possesses the following characteristics: the propodeum is smooth and weakly curved, sometimes with enlarged spiracles; the pronotum only with the ventral furrow; the inner spur on the hind tibia is much longer than half length of the hind basitarsus; the females of some species have an enlarged seta on the fore telotarsus (fifth tarsomere, as Protapanteles); the vannal lobe on the hind wing with the margin straight or concave; the petiole parallel-sided and rounded distally or weakly narrow distally, the petiole can be smooth or with weak sculpture; and the median area on T2 smooth with lateral grooves poorly defined distally. Glyptapanteles differs from Distatrix in having the petiole more strongly narrowed distally, the margins of the median area on T2 are better defined, and the ovipositor sheath exhibits normal setae distally (Whitfield et al. 2009). Again, these characteristics are also found in some Neotropical Glyptapanteles.

Glyptapanteles and Lathrapanteles. Lathrapanteles was separated from Sathon by Williams (1985), who placed four species of Sathon in this new genus: three from the Northeastern United States and one from South America. He was aware that Lathrapanteles was not a natural group and he realized the difficulties of separating both genera by one or a few external morphological characters. Some of the characteristics that define the genus, but are also shared with Glyptapanteles are: the pronotum with both dorsal and ventral furrows (e.g., G. andysuarezi, Fig. 19A, E), the metanotum reduced and without sublateral setose lobes, the phragma of the scutellum exposed (e.g., G. ianyarrowi, Fig. 107B, C), the propodeum weakly or strongly convex with median longitudinal carina or groove, the petiole evenly narrowed to apex or parallel-sided for the proximal 0.75 or slightly barrel-shaped, and the edges of the median area on T2 defined by grooves or obscured by rugosity (e.g., G. linghsiuae, Fig. 141D, G) (Williams 1985).

Glyptapanteles and Protapanteles. Protapanteles shares with Glyptapanteles a weakly sculptured propodeum and relatively trapezoidal median area on T2 (Whitfield et al. 2009). However, the most frequent characteristic used to distinguish the two genera is the petiole shape: parallel-sided (3/4 proximal or more) and then strongly narrowing at the apex, or sides gradually converging distally for Glyptapanteles, while for Protapanteles the petiole is parallel-sided throughout except for a strongly rounded apex (Mason 1981). Additional traits ostensibly exclusive to Protapanteles are: the pronotum with two distinctive furrows, one dorsal and other ventral; in females, the fore telotarsus usually with a conspicuous lateroventral curved seta and a weak distal excavation; the larval mandible with a row of 12 or fewer large teeth concentrated distally on the blade; and its distribution which is almost completely confined to the Holarctic Region (Mason 1981). However, Neotropical Glyptapanteles here described exhibit petioles with an extensive array of shapes ranging from barrel-shaped with apex rounded/truncate (e.g., G. phildevriesi, Fig. 187G, H; G. rafamanitioi, Fig. 190H, I) to petioles with broad base to a very narrow apex (e.g., G. pamitchellae, Fig. 178G, H; G. scottshawi, Fig. 199G, H). Some Glyptapanteles species also have a pronotum with both dorsal and ventral furrows (e.g., G. andywarreni, Fig. 20A, C, I; G. markshawi, Fig. 154A, E), and in some females a fore telotarsus with a curved seta can be found (e.g., G. boharti, Fig. 37A, E), although sometimes it is difficult to see the seta due to the small body size of some specimens. Recently, an earlier perspective has been resurrected that Glyptapanteles and Sathon should be part of an expanded Protapanteles (van Achterberg 2003, Fernández-Triana 2010). The three genera, Glyptapanteles, Protapanteles and Sathon, share the following traits: median area on T2 clearly delimited by a pair of submedial grooves, females with hypopygium evenly sclerotized and ovipositor sheath usually short (van Achterberg 2003), the anterior furrow of metanotum is glabrous and flattened (without sublateral lobe), and the phragma of scutellum exposed (Mason 1981). On the other hand, Protapanteles is in many respects intermediate morphologically between Cotesia and Glyptapanteles. It shares a quadrate petiole with Cotesia, and a weakly sculptured propodeum and more trapezoidal median area on T2 with Glyptapanteles (Whitfield et al. 2009).

Glyptapanteles and Sathon. As currently defined, Sathon resembles Glyptapanteles in nearly all features except ovipositor length: short for Glyptapanteles and long for Sathon. However, Neotropical Glyptapanteles exhibit ovipositors with a broad length spectrum, ranging from short (e.g., G. sydneycameronae, Fig. 212A, J, and G. victoriapookae, Fig. 219A, J) to long (e.g., G. alejandrovalerioi, Fig. 5A, G and G. alvarowillei, Fig. 10A, G, I). Additionally, Mason (1981) proposed Sathon as a new genus based upon the distinctive large external genitalia in males; yet again some Neotropical Glyptapanteles males also bear prominent genitalia (e.g., G. andybennetti, Fig. 15A, F, K and G. andydeansi, Fig. 17A, E, J). It has been suggested that Sathon probably should be subsumed within Glyptapanteles in the near future (Whitfield et al. 2009).

The two basic philosophical approaches to classification have always generated substantial controversy in taxonomy. The question is whether to divide (“split”) or to merge (“lump”) specific taxa. On the one hand, there are those who prefer a large number of small taxa, stressing diagnostic differences but on the other, there are those who support that it is better to recognize a relatively small number of taxa, emphasizing broader relationships. As it has been pointed out, the subfamily Microgastrinae has experienced taxonomic chaos during its history due to varied decisions taken in the past. Several genera within Microgastrinae are confused with Glyptapanteles, so it would be a premature decision to deal with potential synonyms here. A reclassification now would seem untimely and thus, a generic reclassification of Glyptapanteles must wait until more data have been accumulated. Currently, Hybrid Anchored Enrichment and Ultraconserved Elements are being used in order to revise the phylogeny of Microgastrinae. Those two approaches may hopefully provide a better understanding and more justified arguments to the synonymy of some genera in Microgastrinae.

The main objective of this paper is to describe for the first time a large array of Neotropical Glyptapanteles species based on the extensive material that is available from two large-scale rearing projects, one in Costa Rica and the other in Ecuador. This taxonomic revision incorporates morphology, DNA sequences (COI gene), and an extensive base of natural history knowledge. Additionally, a morphological image library and a dichotomous key are provided to facilitate species identification. This is not meant to be a full revision of the genus but instead is intended as a significant starting point for understanding their Neotropical biodiversity and as a guide for future research.

A robust intraspecific analysis benefits from a large quantity of specimens collected across its distributional range. The primary taxon sampling of Glyptapanteles for this study derives from two independent long-term rearing projects: the Caterpillar and Parasitoid Inventory of the Área de Conservación Guanacaste (ACG) in northwestern Costa Rica (http://www.acguanacaste.ac.cr, http://janzen.sas.upenn.edu) and the project Caterpillars and Parasitoids of the Eastern Andes (CAPEA) in Ecuador (www.caterpillars.org). These two Neotropical countries have high species richness and what appears to be endemism in the face of their neighboring countries being poorly studied for their Microgastrinae biodiversity.

The ACG project began in 1978; initially, samples were collected exclusively in dry forest on the small area of Santa Rosa National Park (SRNP). By the end of the 1980’s the sampling was expanded eastward and upward into the rain forest and cloud forest. Currently, the sampling covers a wide altitudinal range from 90 m to 2,000 m (Janzen et al. 2009, Janzen and Hallwachs 2016b). In 2003, the project incorporated DNA barcoding, a microgenomic identification system, by which species can be identified and usually discriminated through the analysis of a small segment of the genome: the mitochondrial gene cytochrome c oxidase I (COI). The CAPEA project was more recently started in 2001 and participants have collected and reared caterpillars at Yanayacu (black river in the Kichwa language), a biological station in the Quijos Valley, Napo Province, in the Andes of northeastern Ecuador (Miller and Dyer 2009). A variety of ecosystems have been sampled such as paramo, montane wet forest, cloud forest, and mid-elevation rain forest ranging from 3,800 m down to 100 m (Dyer et al. 2012).

In both projects, caterpillars were collected directly in the field and subsequently reared in “laboratory” conditions (partly enclosed rearing barn). Voucher specimens of the food plants also were collected for taxonomic identification. Plant vouchers for the CAPEA project were deposited at the Museo Ecuatoriano de Ciencias Naturales (Quito, Ecuador). Rearing took place individually for each host caterpillar in clear plastic bags, bottles, jars, or plastic cups in an open-air shelter with ambient temperature, humidity, and natural day length. Larvae were fed with fresh excised foliage of the food-plant species on which the caterpillar was collected and placed in containers as needed. Larvae were inspected daily to record stage of development, parasitoid emergence, or simply to remove frass. Each caterpillar of ACG was tagged with a voucher code which refers to the event-based record of finding the caterpillar and rearing it: yy-SRNP-xxxxxxx e.g., 90-SRNP-1146. The prefix refers to the last two digits of the year that caterpillar was discovered in the field. The acronym SRNP stands for Santa Rosa National Park, and the suffix is a unique number assigned within the year. When a solitary parasitoid emerged from its host, the same caterpillar voucher code was assigned at that time, but also a unique DNA wasp voucher code later was assigned for any further study of that specimen: DHJPARxxxxxxx (e.g., DHJPAR0001443, DHJ = Daniel Hunt Janzen and PAR = parasitoid) (Janzen et al. 2009, Janzen and Hallwachs 2016a, Janzen and Hallwachs 2016b). In gregarious samples only one wasp was selected to DNA barcoded and received a unique DHJPARxxxxxx code; however, the yy-SRNP-xxxxx code is retained by the unique one as well as all specimens reared from the same caterpillar sample. In the CAPEA project, the voucher code for each caterpillar collected was labeled as ECxxxx. EC stands for Ecuador and the suffix is a unique number assigned to each sample according to a list of consecutive numbers. The DNA wasp voucher code for each parasitoid wasp was presented as YY-Axxxxxxx (YY = Yanayacu, A = first author code).

The caterpillars collected in the field had already been parasitized (or not). Thus, parasitoid identification was based on adult wasps just after their emergence (Dyer et al. 2007) in the rearing containers, or by later study by taxonomists using all information available. In contrast, caterpillar identifications were done at the time of caterpillar collection based on larval morphology (using photographs and previous rearings), since the caterpillar host is destroyed when the parasitoid emerges. Alternatively, lepidopteran identifications were based on adult morphology when a presumably conspecific sample contained more than one caterpillar, including at least one survivor. After emerging, all parasitoid wasps were preserved in 95% to 100% ethanol, facilitating their future use in molecular systematic work. All the Microgastrinae samples were initially sent to the James B. Whitfield lab, [Systematics of Parasitic Hymenoptera at the University of Illinois, Urbana-Champaign (UIUC), Illinois, USA] and later to the Canadian National Collection (CNC) of Insects, Arachnids and Nematodes, Ottawa, Ontario, Canada or the Pontificia Universidad Católica del Ecuador, Colección Entomológica, Quito, Ecuador. Adults collected using Malaise traps were available (for Costa Rica) supplementing the reared material. Information and pictures about herbivore hosts (Lepidoptera) and food plants were provided by DH Janzen and LA Dyer and are accessible in searchable databases (http://janzen.sas.upenn.edu/ and http://caterpillars.unr.edu/lsacat/ecuador/).

In gregarious samples (some with more than 100 individuals) between six (three females and three males) to ten (five females and five males) specimens were selected for point mounting, while the remainder were kept in ethanol (100%) and refrigerated at -20 °C. Specimens selected to be point-mounted were previously treated with hexamethyldisilazane, [(CH₃)₃Si]₂NH –HMDS, to permit easy manipulation and avoid specimen fragmentation during handling (Heraty and Hawks 1998). Specimens were placed into a small vial (2 ml) permitting the evaporation of ethanol and then were completely soaked in HMDS. A small amount of reagent was used, not only due to the small size of the specimens, but also to avoid gas buildup in the vial. Containers were capped and left in a fume hood for 1 hour at room temperature. After this time, the excess HMDS was eliminated by transferring the specimens into a small Petri dish and permit to air dry for 30 minutes. Additionally, Petri dishes were covered with glass microscope slides to prevent loss of specimens. The impact of this treatment on DNA is unknown, but in the study, a specimen or a leg from a specimen was taken from an ethanol-preserved specimen for DNA barcoding, before being treated as above. If further DNA analysis is to be performed, the ethanol-preserved specimens will be used.

Wing slides were prepared for each species, and where feasible, for each sex. The right set of wings (fore and hind) was selected and placed between two glass microscope slides. Wings were detached from the body with the help of a #2 insect pin and thin forceps. Both wings were placed on the middle section of one of the two slides and soaked with ethanol, facilitating the spread of the wings on the slide. Wings were straightened out with the tip of a thin forceps through gentle movements to avoid tearing the wings. Afterward, excess ethanol was allowed to evaporate for a few seconds and only then was the second slide put on the top. For fastening the two slides together, cellophane tape was wrapped at both ends of the slide. Labels (identification and codes for sampling) were affixed to the right edge of the slide.

High-resolution images were obtained by two sources: scanning electron microscope (SEM) and digital photography. Each species description has images of full habitus, head, mesosoma, and metasoma in both lateral and dorsal views. In some cases, both male and female of each species were photographed. The species plates are presented in alphabetical order.

Scanning electron microscopy (SEM). All specimens used for SEM had their wings removed. No pre-cleaning procedure was done before SEM. The entire wasp was either directly affixed to or point-mounted on a metal stub with carbon adhesive tabs. Stub-mounted specimens were sputter coated using a Desk-1 TSC (Denton vacuum LLC, Moorestown, NJ, USA) with a gold-palladium alloy from at least three different angles while rotating the stub to ensure complete coverage. Then, images were taken with a Philips XL30 ESEM-FEG (FEI Company, Hillsboro, OR, USA) at the Imaging Technology Group (ITG) at Beckman Institute, UIUC. Each image was subsequently cropped and incorporated into the plates using Adobe Photoshop® CS v5 and saved as .jpg files.

Digital imaging. Digital photos were taken with a Leica® DFC425 digital microscope camera affixed to a Leica M205 stereomicroscope (Wetzlar®, Germany) with white LED (light-emitting diode) ring lights and dome. Specimens for imaging were held in place under the Leica by gray play dough mounted on a gray background. The LAS (Leica Application Suite) Multifocus module integrated within the Leica microscope was used to create a series of partially focused images. The acquisition of a composite focused image from Z-stack images at different focus positions was obtained with Zerene Stacker™ version 1.04 (http://zerenesystems.com/cms/stacker). The final image was post-processed with Adobe Photoshop CS v5 and saved as .jpg files.

All COI sequences of Costa Rican Glyptapanteles were generated at the Biodiversity Institute of Ontario (BIO, University of Guelph, Ontario, Canada), by methods described in other ACG inventory barcode inventory and taxonomic papers by Smith et al. (2007, 2008) and Fernández-Triana et al. (2014a), and deposited in the Barcode of Life Data System (BOLD, http://www.boldsystems.org; Ratnasingham and Hebert 2007). In contrast, the Ecuadorian sequences were generated in the facilities at UIUC, and then deposited in BOLD for analysis (https://doi.org/10.5883/DS-ASGLYP). At UIUC, an entire wasp (head and wings were the only body parts discarded) was selected for maceration in samples with multiple specimens (gregarious) while one hind leg of the pair was plucked in samples with a unique specimen (solitary), preserving the body for taxonomic work. In both situations, specimens were ground in a mini-pestle. Genomic DNA extractions were carried out using DNeasy tissue extraction kits (QIAGEN, Valencia, CA, USA). COI Primers used are listed in Table 1.

At UIUC, Polymerase chain reactions (PCR) for all primer pairs were carried out in 25 µl reaction volumes consisting of molecular biology grade H₂O=15.38 μl; 10Ex Taq buffer=2.5 μl; forward primer (10 μM)=1 μl; reverse primer (10 μM)=1μl; dNTPs mixture (10 mM)=2 μl; Takara Ex Taq DNA (5 U/µl)=0.125 μl and DNA template=3 μl. All amplifications were carried out using a Thermal Cycler (PCR Machine) C1000 Touch™ (Bio-Rad Laboratories, Hercules, CA, USA). The thermocycling program consisted of an initial denaturation step of 94 °C for 2 minutes, followed by 34 cycles [30 seconds at 94°, 25 seconds at 45 °C or 53 °C and 1 minute at 72 °C], and a final extension step of 72 °C for 4 minutes. A negative control was included in each round of amplifications that contained dH₂O instead of DNA template. All PCR products were visualized on 1.5% agarose gels (Fisher Scientific, Pittsburgh, PA, USA), stained with red dye (Phenix Research, Candler, NC, USA) and visualized using UV light to measure PCR success. Amplicons were purified with QIAquick PCR purification kits (QIAGEN, Valencia, CA, USA) according to the manufacturer’s protocol. Sequences were generated via Sanger cycle-sequencing using amplification primers (forward and reverse directions) and visualized on ABI 3730 Capillary Electrophoresis Genetic Analyzer (Applied BioSystems) at the W.M. Keck Center for Comparative and Functional Genomics, UIUC. The bioinformatics software Geneious ProTM 5.3.4 (Biomatters Ltd., Newark, NJ, USA) was used to visualize chromatograms, edit sequences, and assemble both forward and reverse sequences in contigs. Additionally, all the sequences were translated to amino acids (invertebrate mitochondrial or standard genetic code) to assist in manual adjustments and proof-reading.

Three different datasets (COI sequences, natural history (host records), and external morphological characters) were integrated in order to generate realistic discrimination of species. While provisional species hypotheses were associated with an approximate 2% sequence divergence (Jones et al. 2011, Smith et al. 2013, Ratnasingham and Hebert 2013), we do not consider this a strict rule or criterion. For example, if we observed clear and consistent morphological differences and/or obvious biological differences between specimens with high sequence similarity we “split” these genetically distinct units into two names based on the weight of evidence. Using the same weight of evidence criteria, we split morphologically similar individuals if they were characterized by both genetic and biological differences.

A tree of COI DNA sequences was constructed in MEGA6 (Tamura et al. 2013) using the Maximum Likelihood (ML) method based on the General Time Reversible model (+G, parameter = 0.3431) (Nei and Kumar 2000). Samples selected for this representative tree were the holotypes for each species except for 5 cases (indicated with an * on the tree) where the holotype was not successfully sequenced (G. boharti, G. alvarowillei, and G. alejandrovalerioi) or having sequence but with insufficient overlap to permit tree construction (G. mikeschauffi and G. sondrawardae). In these cases, we substituted other high-quality sequences from the same species. All COI sequences and their specimen information are available on BOLD: https://doi.org/10.5883/DS-ASGLYP

Descriptions are based on adult female/male holotypes. When additional specimens were available, notable intraspecific variation was reported. Base on a dataset of 126 characters and 484 character-states, a uniform format for species descriptions was generated with LucID 3.5 software (www.lucidcentral.com) using the Lucid3 Builder tool. The species descriptions are presented in alphabetical order.

Each examined sample that included type material has information about the country, province, region, sector, site, type of forest, elevation, latitude, longitude, collection date, collector, instar of caterpillar collected, date of formation of wasp cocoon (often), and emergence date of the adult parasitoid. The codes for sampling and DNA are also provided. For Costa Rican those codes are yy-SRNP-xxxxxx and DHJPARxxxxxx and for Ecuador EC-xxx and YY-Axxxx. Geographical coordinates are given in decimal degrees (DD). Latitude is expressed before longitude. Positive latitudes are north of the equator, negative latitudes are south of the equator. Positive longitudes are east of the prime meridian, negative longitudes are west of the prime meridian. The conversion of degree minutes seconds to decimal were obtained using the Federal Communication Commission (FCC) converter (http://transition.fcc.gov/mb/audio/bickel/DDDMMSS-decimal.html).

The total number of specimens examined as well as numbers of females and males are specified for each sample. At the beginning of each sample examined a series of numbers are presented [e.g., 8 (3♀, 3♂) (2♀, 0♂)]. The first number (8) indicates the total of specimens found in the sample followed by the number of female(s) and male(s) that were point-mounted (first parenthesis), and then by the quantity of females and males left in ethanol (second parentheses).

In the section of etymology, each species is named in honor of a person who has, during the past 55 years, helped Daniel H. Janzen, Winifred Hallwachs, and Lee A. Dyer, as well as many others identify and understand tropical fauna and flora. Those people constitute a diverse and far-flung team, without which this paper and many others like it would not exist. Mentors, colleagues, friends, and relatives of the first author are also included. Each person is mentioned after the word Etymology in each species account with a brief description of their interests.

Character sampling. Only characters derived from the external morphology were used. Most of the species descriptions are based on females; males were used only when females were absent.

Measurements. All specimens were examined using a Leica M125 stereomicroscope (Wetzlar, Germany). Holotype measurements were taken using a micrometer mounted in the microscope. Body length, antenna length, and fore wing length were taken in 2.0×, while remaining measurements in 10.0×. Body length was measured from the anterior margin of the head to the posterior margin of metasoma, excluding ovipositor and ovipositor sheath; and fore wing length from first axillary sclerite to the edge of the wing. All measurements are expressed in mm.

Taxonomic characters. All samples were identified to genus at UIUC using a key to New World genera of Microgastrinae (Whitfield 1997). Types for the previously described Neotropical species were examined: Blanchard collection, Buenos Aires, Argentina; Natural History Museum, London, UK (NHMUK); and United States National Museum USNM (now National Museum of Natural History, Smithsonian Institution, Washington, DC). Additional material examined: portions of Blanchard’s material, extensive reared material in USNM; Illinois Natural History Survey, Champaign, IL, USA; and Whitfield’s personal reared collection, Urbana, IL, USA. All illustrated and discussed in Whitfield et al. 2002a. Terminology for surface sculpturing follows Harris (1979), for wing venation follows Sharkey and Wharton (1997) and for morphology follows Mason (1981), Austin and Dangerfield (1992), Sharkey and Wharton (1997) and Whitfield (1997).

The antenna is described as it is resting above the body. The body coloration is defined as pale and dark. However, in the section’s coloration in adult wasps and species descriptions, the body coloration is treated in more detail. On the metasoma, the metasomal tergum 1 (T1) and the metasomal tergum 2 (T2) are divide into one mediotergite (medial chitinous portion) and two lateral tergites (two membranous lateral areas or laterotergites). Here, petiole (pe, Fig. 3C) on T1 and median area (ma, Fig. 3C) on T2 are the terms preferred over mediotergite on T1 and mediotergite on T2 respectively. Also, the sublateral area (sa, Fig. 3C) on T1 is the preferred terminology for the lateral tergite on T1. On T2, the terms adjacent area (ada, colored area next to the median area) and lateral end (le) are used to highlight the different colors of these two areas (Fig. 3C). Additionally, some morphological terms are used for the first time and refer to structures mainly in the scutellum and the metanotum. For the scutellum these are: axillary trough of scutellum (ATS) and medioposterior band of scutellum (BS). For the metanotum the terms are: anterior furrow of metanotum (AFM), the posterior furrow of metanotum (PFM), an axillary trough of metanotum (ATM), medioposterior band of metanotum (BM), and medioanterior pit of metanotum (MPM) (Fig. 2F). These external morphological terms, as well as others used in the descriptions, are illustrated in the Figures 2, 3.

The following acronyms are used to denote the depositories:

CNC Canadian National Collection of Insects, Arachnids and Nematodes, Ottawa, Ontario, Canada.

PUCE Pontificia Universidad Católica del Ecuador, Colección Entomológica, Quito, Ecuador.

Morphological terms and their abbreviations used in the text and figures are:

ada adjacent area,

AFM anterior furrow of metanotum,

ATM axillary trough of metanotum,

ATS axillary trough of scutellum,

BM medioposterior band of metanotum,

BS medioposterior band of scutellum,

cl clypeus,

CR Costa Rica,

e eye,

EC Ecuador,

er epicnemial ridge,

f female,

fa face,

fc fore coxa,

fr frons,

G gregarious,

ge gena,

gusaneros parataxonomists or paraecologists who find and rear the caterpillars (Janzen and Hallwachs 2011),

hc hind coxa,

l lunule of scutellum,

la labrum,

le lateral end,

lp labial palp,

m male,

ma median area,

mc middle coxa,

md mandible,

me mesopleuron,

mls malar suture,

mp maxillary palp,

ms mesoscutum,

mtn metanotum,

MT Malaise trap,

MPM medioanterior pit of metanotum,

mtp metapleuron,

n nucha,

o ocellus,

OOL ocular ocellar line (the shortest distance between lateral ocellus and adjacent compound eye margin),

pe petiole,

pd pedicel,

pg precoxal groove,

ph phragma of scutellum;

pn pronotum (1, dorsal furrow; 2, central area; 3 ventral furrow),

POL posterior ocellar line (the shortest distance between the lateral ocelli),

pp propodeum,

ppl propleuron,

PFM posterior furrow of metanotum;

S metasomal sternum:

S4 sternum 4 or antepenultimate sternum,

S5 sternum 5 or penultimate sternum,

S6 sternum 6 or hypopygium,

sa sublateral area,

sc scape,

scl scutellum,

so solitary,

T metasomal tergum:

T1 tergum 1,

T2 tergum 2 and so on,

te temple,

v vertex,

YPT yellow-pan trap.

This Glyptapanteles species revision resulted in the recognition and description of 136 new species, none of them are shared between the two countries. Thus, 78 are from Costa Rica and the remaining 58 from Ecuador. The Costa Rican species are alejandrovalerioi, alexborisenkoi, alvarowillei, andybennetti, andydeansi, annettewalkerae, barneyburksi, billbrowni, bobhanneri, bobkulai, bobwhartoni, boharti, brianestjaquesae, carlhuffakeri, carlossarmientoi, carlrettenmeyeri, charlesmicheneri, charlesporteri, chrisdarlingi, chrisgrinteri, christerhanssoni, corriemoreauae, daveroubiki, daveschindeli, davesmithi, davidwahli, donquickei, eowilsoni, garygibsoni, gavinbroadi, gerarddelvarei, henrytownesi, howelldalyi, hugokonsi, iangauldi, ianyarrowi, ilarisaaksjarvi, jacklonginoi, jamesrobertsoni, jeremydewaardi, jesusugaldei, jjrodriguezae, johnburnsi, johnheratyi, johnlasallei, johnnoyesi, lubomasneri, malloryvanwyngaardenae, marjorietownesae, markshawi, meganmiltonae, mehrdadhajibabaei, mikegatesi, mikeschauffi, mikesharkeyi, nataliaivanovae, nealweberi, ninazitaniae, pamitchellae, paulhansoni, paulheberti, paulhurdi, philwardi, robbinthorpi, ronaldzunigai, roysnellingi, scottmilleri, scottshawi, shelbystedenfeldae, sondrawardae, stephaniecluttsae, stephaniekirkae, sujeevanratnasinghami, sureshnaiki, sydneycameronae, tanyadapkeyae, victoriapookae, and wonyoungchoi. The Ecuadorian species are alexwildi, andrewdebeveci, andysuarezi, andywarreni, ankitaguptae, betogarciai, carinachicaizae, celsoazevedoi, claudiamartinezae, diegocamposi, dorislagosae, edgardpalacioi, edwinnarvaezi, erictepei, felipesotoi, ferfernandezi, genorodriguezae, grantgentryi, gunnarbrehmi, haroldgreeneyi, helmuthaguirrei, henryhespenheidei, jaquioconnorae, jerrypowelli, jimmilleri, johnstiremani, josesimbanai, juanvargasi, jumamuturii, keithwillmotti, kevinjohnsoni, kyleparksi, linghsiuae, luchosalagajei, malleyneae, mamiae, marcelotavaresi, marcepsteini, marcpolleti, marshawheelerae, mayberenbaumae, michelleduennesae, mikepoguei, montywoodi, pachopinasi, petermarzi, phildevriesi, rafamanitioi, suniae, suzannegreenae, taniaariasae, thibautdelsinnei, thomaspapei, toluagunbiadeae, tomwallai, wilmersimbanai, yalizhangae, and yanayacuensis. Before this study, only six species had been described for the Neotropics. A total of 16,663 specimens was examined, of which 13,542 are preserved in 100% ethanol and 3,121 point-mounted.

The samples reviewed from Costa Rica are the result of 30 years of continuous collecting, from 1982 to 2012, and are accompanied by more than 100 undescribed sympatric species. In contrast, material examined from Ecuador covers a period of only five years (the oldest samples were caught in 2005 while the most recent in 2010). It is important to note that both inventories of the caterpillars and their food plants and parasitoids are still running and will continue for a number of years, which means even more species in this genus could be added to those described here and the 100+ undescribed species already collected. Indeed, material recently collected (2010–2019) for both projects, including reared specimens and Malaise-trapped specimens, was also available at the time of this revision, but it was not possible to include them. It is expected that those remaining species will be described in the near future.

Within Glyptapanteles COI, intraspecific variation is much less than interspecific variation (on average 0.09% vs. 10.1 % when estimated using the BOLD distance summary tool on sequences longer than 400 bp (i.e., overlapping)).

Glyptapanteles has a wide range of ecological distribution. These wasps can be found from 90 m to 2,800 m elevation. Elevation was not reported for only one species, G. toluagunbiadeae. Forty percent of the species (55 of 136 spp.) were reported above 1,500 m. All the species from Ecuador are found at or above an elevation of 1,000 m whereas Costa Rican species are found from 90 m to 1,460 m because that is the range available for sampling; there is no doubt that their distributions range down to sea level, and they will be found up to the highest available point (2,000 m in ACG).

Most Neotropical Glyptapanteles (89%) are between 2 and 3.4 mm in length, whereas a low percentage is below (5%) or above (6%) that range. The seven species with body length less than 2 mm are G. daveroubiki (1.67), G. carlossarmientoi and G. philwardi (each with 1.81), G. ronaldzunigai (1.86), G. carlrettenmeyeri and G. wonyoungchoi (each with 1.91), and G. chrisgrinteri (1.96). The eight species more than 3.5 mm are: G. charlesporteri (3.53), G. pachopinasi (3.58), G. sureshnaiki (3.63), G. ferfernandezi (3.68), G. ninazitaniae (3.78), G. alvarowillei (3.81), G. andydeansi (3.85), and G. malleyneae (3.88).

In most species here described, the body coloration is generally dark, ranging from dark brown to black. Pale coloration (light or dark yellow) is mostly limited to legs, first segments of the metasoma, and, rarely, the antenna and the mesosoma. In species with a yellow metasoma, that pale coloration is limited exclusively to the first terga and first sterna when the specimen is seen in lateral view. However, twelve species displayed an uncommon pattern: the yellow coloration extends beyond including all the terga and all the sterna. Thus, in lateral view the metasoma appears to be completely yellow in females: G. andybennetti (Fig. 14A), G. andydeansi (Fig. 16A), G. annettewalkerae (Fig. 23A, K), G. billbrowni (Fig. 29A), G. bobhanneri (Fig. 31A), G. brianestjaquesae (Fig. 39A), G. claudiamartinezae (Fig. 58A, J), G. gavinbroadi (Fig. 88A), G. mayberenbaumae (Fig. 157A, J), G. paulhurdi (Fig. 184A, J), G. suzannegreenae (Fig. 211A, K), and G. sydneycameronae (Fig. 212A, J).

The six Neotropical Glyptapanteles species described previously to this study show bodies exclusively tinted with dark colors. Here, G. stephaniecluttsae (Figs 204–205) is the first Neotropical species with extensive yellow coloration on the mesosoma although all the mesoscutum, small portions on the scutellum, and the metanotum are dark brown. The head in this species is black. A closer look at the metasoma shows that in lateral view, the predominant coloration is pale (yellow) though dark brown areas are present and limited by a small area in the dorsal part of the T4 and beyond. It must be noted, however, that in dorsal view the dark areas are more extensive than pale areas. Thus, the T3 and the remaining terga are completely brown. The yellow coloration is limited to 3/4 proximally of the petiole and sublateral areas on T1, the median area but not in the male (Fig. 205D), and the lateral ends on T2 (Fig. 204H, I). This Costa Rican gregarious species was collected at 730 m in rain forest. Pale coloration or “xanthic coloration” along with enlarged compound eyes are features generally associated with species that are nocturnally active (Whitfield and Scaccia 1996).

Taking advantage of the large number of new species here described, the variation in the body coloration has been stressed in detail. These minute wasps are not wholly black as might be thought initially. The color variation in those species is focused on the antenna (five types), the propleuron (two types), the petiole (three types), the median area on T2 (three types), and the legs (13 types). In the legs, the greatest color variation occurs in the hind ones and includes mainly the segments of the coxa, the femora, and the tibia. It is worth mentioning that coloration was not taken into consideration in the key for species separation.

Antennal coloration. At least five types of antennal coloration were observed: all the antennal flagellomeres have the same color in both sides, only the dorsal part of all the antennal flagellomeres or only the dorsal part of the first proximal flagellomeres or only the last distal antennal flagellomeres are light-colored with the remaining areas dark-colored, and the antenna tricolored.

All the antennal flagellomeres have the same color throughout. This is the most common color pattern. The antenna is completely dark brown throughout and occurs on most Glyptapanteles species (83 spp.). However, there are two variations: all the flagellomeres are stained darker with a black tint, this shade is present in only three species (G. michelleduennesae, G. montywoodi, and G. petermarzi) or all the flagellomeres can be completely yellow-brown, which was recorded in only one species (G. johnburnsi).

The dorsal part of all the antennal flagellomeres is lighter than the ventral part. In this category, the dorsal coloration of the antenna (resting above the body) can be yellow-brown (only in G. donquickei) or light brown (as in G. alejandrovalerioi, G. eowilsoni, G. marcelotavaresi, G. paulhansoni, G. phildevriesi, and G. sydneycameronae). In both cases, the ventral coloration is (dark) brown.

Only the dorsal part of the most proximal flagellomeres is lighter than the ventral part. In 27 species, the coloration of the dorsal part of the most proximal flagellomeres is light brown. The species that repeat this pattern are G. daveschindeli, G. garygibsoni, G. gavinbroadi, G. hugokonsi, G. ilarisaaksjarvi, G. johnlasallei, G. keithwillmotti, G. kyleparksi, G. luchosalagajei, G. markshawi, G. meganmiltonae, G. mehrdadhajibabaei, G. mikeschauffi, G. nataliaivanovae, G. robbinthorpi, G. roysnellingi, G. scottmilleri, G. shelbystedenfeldae, G. sondrawardae, G. sujeevanratnasinghami, G. suniae, G. tanyadapkeyae, G. thibautdelsinnei, G. thomaspapei, G. toluagunbiadeae, G. yalizhangae, and G. yanayacuensis.

In other species, the dorsal coloration of the most proximal flagellomeres is yellow-brown and it was recorded in ten species: G. charlesmicheneri, G. christerhanssoni, G. davidwahli, G. howelldalyi, G. jamesrobertsoni, G. jeremydewaardi, G. jesusugaldei, G. jjrodriguezae, G. malloryvanwyngaardenae, and G. marjorietownesae. In one species, G. davesmithi, the first five proximal flagellomeres are dorsally yellow. In all instances, the ventral coloration of those proximal flagellomeres is dark brown. However, there is a slight modification to this pattern: in G. scottshawi, the first seven-eight proximal antennal flagellomeres are yellow on both sides. On all the examples cited, the remaining flagellomeres are brown or dark brown.

The last distal antennal flagellomeres are lighter than the remaining flagellomeres. Only four species fit in this category. The last distal antennal flagellomeres can be completely yellow (G. ninazitaniae and G. sureshnaiki only in females), yellow-brown (G. bobwhartoni) or light brown (G. mikegatesi). In all these species, the remaining flagellomeres are brown or dark brown.

Antenna tricolored. Only one species, G. wonyoungchoi, has the antenna with three colors: the first four proximal antennal flagellomeres are completely yellow, the following five to seven are totally yellow-brown, and the remaining flagellomeres are brown on both sides.

Propleuron coloration. With respect to propleuron two types of coloration were observed: propleuron matches the mesosoma coloration and propleuron coloration differs partially or completely from mesosoma coloration.

Propleuron matches the mesosoma coloration. In the vast majority of the species (102 spp.) the propleuron is dark as the mesosoma. Only in one species, G. stephaniecluttsae, the propleuron is as pale (yellow) as mesosoma.

Propleuron coloration differs partially or completely from mesosoma coloration. In five species the propleuron coloration (entirely yellow or entirely light brown) deviates from mesosoma coloration (brown or brown-black): G. alexwildi (Fig. 9H), G. barneyburksi (Fig. 25A), G. bobhanneri (Fig. 31A), G. jaquioconnorae (Fig. 115I), and G. mikesharkeyi (Fig. 168A). In 28 species, the propleuron has some pale areas (yellow-brown or light brown) that contrasts with the dark color of the mesosoma: G. andrewdebeveci (Fig. 12J), G. annettewalkerae (Fig. 23J), G. brianestjaquesae (Fig. 39A), G. charlesporteri (Fig. 50A), G. daveschindeli (Fig. 64A), G. grantgentryi (Fig. 94I), G. gunnarbrehmi (Fig. 95J), G. haroldgreeneyi, G. helmuthaguirrei (Fig. 97I), G. jesusugaldei, G. josesimbanai, G. mayberenbaumae, G. meganmiltonae (Fig. 158A), G. nataliaivanovae (Fig. 171I), G. nealweberi (Fig. 173G), G. ninazitaniae (Fig. 175I), G. pamitchellae (Fig. 178I), G. paulheberti, G. paulhurdi (Fig. 184A), G. scottmilleri, G. scottshawi (Fig. 199I), G. shelbystedenfeldae, G. sondrawardae (Fig. 203I), G. stephaniekirkae (Fig. 206C), G. sujeevanratnasinghami (Fig. 208I), G. sureshnaiki (Fig. 210I), G. sydneycameronae (Fig. 212I), and G. yanayacuensis (Fig. 223I).

Petiole coloration. Three types of coloration were observed: petiole entirely dark, petiole entirely pale, and petiole with two colors.

Petiole entirely dark. This is the category that contains the largest number of species. In 107 species the petiole is dark brown or brown-black (same coloration as metasoma) and the contours of the petiole can or can not be darkened.

Petiole entirely pale. This is an unusual case. Only in G. stephaniecluttsae, the petiole is yellow with contours yellow-brown.

Petiole with two colors. In two species, the pale coloration dominates over the dark one. Glyptapanteles alexwildi (Fig. 9H) has the petiole yellow-brown with the entire inner edge dark brown. Glyptapanteles suzannegreenae (Fig. 211H) has it dark yellow; however, the lateral parts on the distal half have light yellow-brown tints. In both species the petiole contours are darkened. In some species, the pale coloration on the petiole covers more area than the dark one. Thus, in six species 3/4 of the proximal part is pale and the distal 1/4 is dark. The pale coloration varies from yellow, yellow-brown, brown-orange, and reddish brown while the dark coloration is dark brown or brown-black. Those species are: G. charlesmicheneri, G. corriemoreauae, G. davidwahli, G. ilarisaaksjarvi (Fig. 109D), G. malloryvanwyngaardenae (Fig. 146F), and G. pamitchellae (Fig. 178G). In one species, G. eowilsoni, proximal 2/3 is reddish brown and the distal 1/3 is black. In some species the dark coloration on the petiole dominates the pale coloration. Thus, in three species 3/4 of the distal part is dark and the proximal 1/4 is pale. The pale coloration changes from yellow-brown and reddish and the dark coloration from dark brown and black. These species are G. mehrdadhajibabaei, G. scottshawi (Fig. 199G), and G. sondrawardae (Fig. 203G). In seven species the dark and the pale colorations cover the petiole in equal proportions. The pale coloration is variegated, and can be yellow, brown-orange, brown-red, reddish or light brown. The dark coloration can be brown or dark brown. Those species are: G. daveschindeli, G. jamesrobertsoni, G. jesusugaldei, G. jumamuturii, G. kyleparksi, G. mayberenbaumae, and G. tanyadapkeyae (Fig. 214G).

In seven species the petiole is mostly dark but with a central pale area. The pale coloration can be yellow, yellow-brown or light brown, and the dark coloration fluctuates between brown and brown-black. These species are G. donquickei, G. gerarddelvarei, G. henrytownesi, G. howelldalyi, G. jeremydewaardi, G. johnheratyi, and G. robbinthorpi.

Petiole with three colors. In two species the petiole coloration intensifies from proximal to distal, in both species their contours are darkened. In Glyptapanteles marjorietownesae proximally the petiole is yellow, medially reddish/yellow-brown and distally brown and in G. markshawi the petiole proximally is yellow-brown, medially light brown and distally dark brown.

Coloration in the median area on T2. In most of the species here described (109 spp.) the median area (chitinous portion) is dark (dark brown or brown-black) and next to it there is an obviously defined colored area (membranous portion), the adjacent area (ada, Fig. 3C), which is also dark, and the lateral ends pale.

In twelve species the median area is dark and the lateral ends are pale; this means the adjacent area is missing on T2: G. alexborisenkoi (Fig. 8G), G. bobhanneri, G. brianestjaquesae, G. carlhuffakeri (Fig. 42G), G. christerhanssoni, G. daveroubiki, G. daveschindeli, G. gavinbroadi, G. howelldalyi, G. marjorietownesae, G. mikeschauffi (Fig. 166F), and G. paulhurdi (Fig. 184H).

In twelve species both the median area and lateral ends are dark (the adjacent area is absent): G. alejandrovalerioi, G. ankitaguptae (Fig. 22H), G. carlossarmientoi, G. celsoazevedoi (Fig. 47H), G. haroldgreeneyi (Fig. 96H), G. jerrypowelli (Fig. 118H), G. johnburnsi, G. johnnoyesi (Fig. 131F), G. marcelotavaresi (Fig. 149G), G. marcepsteini (Fig. 150H), G. marcpolleti (Fig. 151G), and G. toluagunbiadeae (Fig. 217G).

Four species show unusual color patterns. First, both the median area and the lateral ends are pale, and the adjacent area is dark: G. stephaniecluttsae (Fig. 204H) and G. stephaniekirkae (Fig. 206G). Second, the median area as well as the lateral ends are pale, the adjacent area is missing, G. wonyoungchoi (Fig. 221G). Third, the median area has two colors (proximal 1/3 yellow, distal 2/3 brown), the lateral ends are yellow, and the adjacent area is absent in G. suzannegreenae (Fig. 211H).

Coxal coloration. Taking into account the coloration of all the coxae compared with the body coloration, six types were recorded: the body and all the coxae yellow; the body dark and all the coxae yellow; the body dark, the fore and the middle coxae yellow, and the hind coxae with two colors; the body and the hind coxae dark, and the fore and the middle coxae yellow; the body, the middle and the hind coxae dark, and the fore coxae yellow; and the body and all the coxae dark.

The body and all the coxae yellow (Figs 204A, 205A). This is an unusual pattern of coloration, only was reported in one species, G. stephaniecluttsae.

The body dark and all the coxae yellow. Only four species have the hind coxae completely yellow [G. alexwildi (Fig. 9K), G. andrewdebeveci (Fig. 12K), G. mayberenbaumae (Fig. 157J), and G. yanayacuensis (Fig. 223J)].

The body dark, the fore and the middle coxae yellow, and the hind coxae with two colors. It is worth mentioning that of the three pairs of coxae are the hind ones which can exhibit at the same time two colors which vary in quantity and location. In this category, the fore and the middle coxae always are completely yellow, thus the differences are based on hind coxae coloration. This is the largest size category with a total of 51 species. First, it will start with the species whose hind coxae coloration is mostly yellow with a little of brown or brown-black and it will end with the species showing the opposite pattern, brown or black coloration dominates over the yellow coloration.

In one species (G. meganmiltonae, Fig. 158A) the hind coxae are yellow, but proximally with an irregular area brown, although in the male they are completely light brown (Fig. 159A); in one species (G. ankitaguptae, Fig. 22K) the yellow hind coxae have a tiny black spot both in proximal and distal edges; and in one species (G. marcelotavaresi, Fig. 149J) has the hind coxae both proximally and distally with an irregular narrow brown area, thus the yellow coloration is in the middle part.

Another dark mark detected in the yellow hind coxae is a dorsal elongated brown-black spot on the proximal part, this pattern was observed in three species (G. ninazitaniae, Fig. 175J; G. sureshnaiki, Fig. 210J; and G. taniaariasae, Fig. 213A), and in one species (G. mikesharkeyi, Fig. 168J) the size of the spot was smaller.

In three species, the dark coloration is located in the same place, covering the proximal third of the hind coxae, the remaining area is yellow. Glyptapanteles nataliaivanovae female exhibits unevenly dark brown blotches (Fig. 171J); however, in the male those marks are more extensive, covering until the proximal half (Fig. 172I); in G. stephaniekirkae (Fig. 206J) both sexes have the brown-black coloration uniform, but in male the area is more extensive covering until the proximal half (Fig. 207A); and in G. thibautdelsinnei male (Fig. 215K) the brown-black coloration also is homogeneous; however, its expansion is unknown in the female.

In four species the dark coloration on hind coxae is more extensive, covering the half or more than half the length of the hind coxae. Glyptapanteles carlossarmientoi female (Fig. 44A) has the proximal half light brown (distal half yellow), but in the male (Fig. 45A) the hind coxae are completely light brown; in both sexes of G. nealweberi (Figs 173H, 174J) the proximal half is brown-black and distal half is yellow; and in two species (G. pamitchellae, Fig. 178A and G. charlesmicheneri, Fig. 48A) the brown-black coloration covers the proximal 3/4 and only the distal 1/4 is yellow.

In three species (G. charlesporteri both sexes, Fig. 50A; G. scottshawi only female, Fig. 199A; and G. sydneycameronae only female, Fig. 212J) the distal third has a yellow area which is more extensive ventrally, reaching almost the distal half. Thus, brown-black is the dominant color on hind coxae.

In 25 species the hind coxae are brown or brown-black but distally have a yellow area which is distinctively tiny: G. annettewalkerae, G. betogarciai, G. billbrowni, G. bobkulai, G. bobwhartoni, G. boharti, G. brianestjaquesae, G. carlhuffakeri, G. claudiamartinezae, G. davesmithi (male with fore and middle coxae yellow-brown instead of light yellow), G. diegocamposi, G. edwinnarvaezi, G. felipesotoi, G. ferfernandezi, G. genorodriguezae, G. haroldgreeneyi, G. helmuthaguirrei, G. henryhespenheidei, G. jaquioconnorae (dorso-distally takes on a somewhat lighter color, yellow-brown), G. keithwillmotti, G. kevinjohnsoni, G. kyleparksi, G. linghsiuae G. marcepsteini, and G. wonyoungchoi. However, in four species (G. bobhanneri, G. erictepei, G. grantgentryi, and G. gunnarbrehmi) the distal tiny area is lighter (yellow-brown) than the remaining brown-black area.

In one species (G. sujeevanratnasinghami) the yellow-brown coloration covers the distal half of the hind coxae, the proximal half is brown-black; in two species (G. edgardpalacioi and G. johnstiremani) the distal third is yellow-brown (proximal 2/3 is brown-black), and in two species (G. andydeansi and G. jeremydewaardi) the yellow-brown coloration is restricted to ventro-distal part and covers a small area, the remaining area is brown-black.

The body and the hind coxae dark, and the fore and the middle coxae pale. In total, 29 species exhibits hind coxae completely dark, that coloration coincides with the dark body coloration; the fore and the middle coxae are completely yellow. Those species are: G. alexborisenkoi, G. alvarowillei, G. carinachicaizae, G. dorislagosae, G. jerrypowelli, G. jesusugaldei, G. jimmilleri, G. josesimbanai, G. juanvargasi, G. jumamuturii, G. luchosalagajei, G. malleyneae, G. mamiae, G. markshawi, G. marshawheelerae, G. michelleduennesae, G. mikegatesi, G. mikepoguei, G. montywoodi, G. pachopinasi, G. paulheberti, G. paulhurdi, G. petermarzi, G. phildevriesi, G. rafamanitioi, G. shelbystedenfeldae, G. sondrawardae, G. suniae, and G. suzannegreenae. However, in three species the hind coxae are lighter (yellow-brown/light brown) than body coloration (G. andybennetti, Figs 14A, 15A; G. barneyburksi, Figs 25A, 26A; and G. lubomasneri, Figs 142A, 143A).

The body, the middle and the hind coxae dark, and the fore coxae yellow. In four species (G. carlrettenmeyeri, Fig. 46A; G. daveschindeli, Fig. 64A; G. johnheratyi, Fig. 127A; and G. toluagunbiadeae, Fig. 217A) the coloration of the coxae changes from light to dark. Thus, the fore coxae are yellow, the middle coxae yellow-brown, and the hind coxae brown-black. In all of them, the body coloration corresponds with the hind coxae coloration, brown-black.

The body and all the coxae dark. In 24 species the dark coloration (dark brown or brown-black) of all the coxae matches with the dark body coloration. Those species are: G. alejandrovalerioi, G. andysuarezi, G. andywarreni, G. celsoazevedoi, G. daveroubiki, G. davidwahli, G. garygibsoni, G. gavinbroadi, G. gerarddelvarei, G. henrytownesi, G. jjrodriguezae, G. johnburnsi, G. johnnoyesi, G. malloryvanwyngaardenae, G. marcpolleti, G. paulhansoni, G. ronaldzunigai, G. roysnellingi, G. scottmilleri, G. tanyadapkeyae, G. thomaspapei, G. tomwallai, G. wilmersimbanai, and G. yalizhangae. However, in five species all the coxae exhibit a coloration lighter than body: G. chrisgrinteri, G. eowilsoni, G. howelldalyi, G. hugokonsi, and G. marjorietownesae. However, in five species all the coxae are lighter than body coloration: G. chrisgrinteri, G. eowilsoni, G. howelldalyi, G. hugokonsi, and G. marjorietownesae.

When comparing the coxae coloration, some species exhibit the fore and middle coxae slightly light or lighter (yellow-brown or light brown) than hind coxae (brown-black). In all of them, the hind coxae coloration matches the body coloration. The 15 species with this pattern are G. chrisdarlingi, G. christerhanssoni, G. corriemoreauae, G. donquickei, G. iangauldi, G. ianyarrowi, G. ilarisaaksjarvi, G. jacklonginoi, G. jamesrobertsoni, G. johnlasallei, G. mehrdadhajibabaei, G. mikeschauffi, G. philwardi, G. robbinthorpi, and G. victoriapookae.

Femoral coloration. Five types of femoral coloration were recorded: all the femora pale; the fore femora with two colorations, and the middle and the hind ones dark; the fore and the middle femora with two colorations, and the hind ones dark; the fore and the middle femora pale, and the hind ones dark; and the fore and the middle femora pale, and the hind ones with two colorations.

All the femora pale. In nine species, the femora in all the three pairs of legs are completely pale: G. alexborisenkoi, G. ankitaguptae, G. boharti, G. charlesporteri, G. gavinbroadi, G. haroldgreeneyi, G. mayberenbaumae, G. paulheberti, and G. stephaniecluttsae. A variation was also registered in species with all the femora yellow. A narrow dorsal dark strip from top to bottom can be present in all of them (G. celsoazevedoi, G. johnstiremani, G. marcepsteini, and G. petermarzi) or only in the fore and the middle femora (G. phildevriesi) or only in the middle and the hind femora (G. luchosalagajei and G. mamiae) or only in the hind femora (G. andybennetti and G. johnnoyesi).

The fore femora with two colorations, and the middle and the hind ones dark. Only one species fits in this category, G. johnburnsi.

The fore and the middle femora with two colorations, and the hind ones dark. Two species registered this pattern: G. alejandrovalerioi and G. shelbystedenfeldae.

The fore and the middle femora pale, and the hind ones dark. Only G. mikegatesi is in this category.

The fore and the middle femora pale, and the hind ones with two colorations. The vast majority of species (114 spp.) are in this type of coloration. The variation in the dark coloration on hind femora mainly concerns quantity. The dark coloration can cover most of the apex, a tiny dot at the tip, the half of the length of the femora, or most to the entire structure.

Tibial coloration. Two types of tibial coloration were observed: all the tibiae pale, and the fore and the middle tibiae pale, and the hind tibiae with two colorations.

All the tibiae pale. Six species matches this pattern: G. mayberenbaumae, G. mikesharkeyi, G. nealweberi, G. ninazitaniae, G. stephaniekirkae, and G. sureshnaiki. Additionally, in two species, G. ankitaguptae and G. johnstiremani, all the pale tibiae have a narrow dorsal brown strip from top to bottom.

The fore and the middle tibiae pale, and the hind tibiae with two colorations. The largest variation in the tibiae coloration is focused mainly on the hind tibiae. Most of the species here described (128 spp.) are included in this category. The variation in the dark coloration on hind tibiae includes the location (distally or at both ends) and quantity (mostly to entirely). In two species, G. marcelotavaresi and G. marcpolleti, the fore and middle tibiae are pale, but additionally, they have a narrow dorsal dark strip from top to bottom.

More details in the body parts which coloration varies are specified in the species description under the coloration section.

One problem that many taxonomists face during the identification of a specific taxon is the absence of drawings, images, or visual aids. Here, a detailed compilation of body part images was undertaken instead of a minimalist descriptive prose approach. It is expected that this high-resolution material will be of great help for future species identification and also serve as a source that facilitates the search of new morphological characters.

Approximately 2,300 original figures populate the morphological image library, which was used to create 222 plates. All of the 136 species described in this work were photographed, of which 84 species have plates for both sexes; for 39 species, only females were considered for the plates even though males were unknown for only ten species. In contrast, females of 13 species were unknown so digital images were taken from males in those cases. Most of the species (90%) were described based on females. A small fraction (10%, 13 spp.) of holotypes correspond to males, due to females not having been caught: G. alexwildi, G. ankitaguptae, G. celsoazevedoi, G. dorislagosae, G. josesimbanai, G. juanvargasi, G. malleyneae, G. marcpolleti, G. montywoodi, G. pachopinasi, G. shelbystedenfeldae, G. tanyadapkeyae, and G. thomaspapei.

In this revision, the specimens collected by Malaise traps came only from Costa Rica (material from Ecuador was also available but was not included in this revision) and constitute 27% (21 spp.) of the ACG species described here from that country (out of 78 spp.). Twelve species, of those 21 caught by Malaise traps, were also obtained from reared material. Thus, it was possible to assign two species as solitary (G. nealweberi and G. sujeevanratnasinghami) and the other ten species as gregarious (G. andybennetti, G. charlesporteri, G. daveschindeli, G. eowilsoni, G. henrytownesi, G. hugokonsi, G. ilarisaaksjarvi [also collected with YPT], G. jamesrobertsoni, G. jesusugaldei, and G. philwardi). Lifestyle, herbivore hosts, and host food plants for nine species continue to be unknown: G. mikesharkeyi, G. nataliaivanovae, G. ninazitaniae, G. pamitchellae, G. scottshawi, G. shelbystedenfeldae, G. sondrawardae, G. stephaniekirkae, and G. sureshnaiki. It is worth emphasizing that the project to date has reared more than 650,000 wild-caught caterpillars and 1,182 Glyptapanteles-attacked caterpillars, yet those nine species have not yet been reared. This gap between sampling methods emphasizes the importance of using diverse methods to assess biodiversity in a specific place. In addition, abundance of those species caught by Malaise traps and yet not recovered from wild caterpillars suggests that probably they are locally common: G. mikesharkeyi (133 specimens), G. pamitchellae (95), G. scottshawi (126), G. sondrawardae (148), G. stephaniekirkae (84), G. sureshnaiki (59), and to a lesser extent G. nataliaivanovae (11), G. ninazitaniae (5), and G. shelbystedenfeldae (13). It is unknown as to whether those species are solitary or gregarious, but that is irrelevant to their abundance in a Malaise trap. Since tens of thousands of exposed caterpillars living on herbs, small trees and bushes, and a very large number of leaf-rolling microlepidoptera have been well-sampled in both projects, the absence of these species in reared material suggests that these species may selectively parasitize either very cryptic caterpillars (leaf rolls, grass moths) or caterpillars that feed higher up in the canopy.

Considering the whole body of examined specimens, females were more abundant (12,609) than males (4,054). Female-biased sex ratios are common under the conditions of local mate competition (LMC, Hamilton 1967). LMC predicts that in environments consisting of several isolated patches where the number of mated females is small, it is advantageous for inseminated females to lay female-biased clutches. This is achievable because in Hymenoptera the act of fertilization is under the voluntary control of the egg-laying female. Adult females can alter progeny sex ratios by producing unfertilized male eggs and fertilized female eggs. Other factors such as host size, host quality, and superparasitism (Tagawa 2000) also play an important role when controlling the progeny of sex ratios, but their significance needs to be assessed in Neotropical Glyptapanteles.

A close look at Malaise-trapped specimens shows that the vast majority of samples exhibit the opposite pattern, with males exceeding females: G. nataliaivanovae (1♀, 10♂), G. pamitchellae (29♀, 66♂), G. scottshawi (17♀, 109♂), G. shelbystedenfeldae (1♀, 11♂), G. sondrawardae (8♀, 140♂), G. stephaniekirkae (10♀, 74♂), and G. sureshnaiki (13♀, 46♂). Males seem to be common near the forest floor where Malaise traps usually are set up while females are comparatively rare at the same place. It is also perhaps males are the dispersing-sex, moving around to look for females. As a result, males are easier to capture.

A total of 127 species is described here from the reared material, and nine species were collected only by Malaise traps. From the reared specimens, 92 Glyptapanteles species seem completely gregarious while 26 seem exclusively solitary (Table 2). It is worth noting that nine species reared showed both solitary and gregarious lifestyles: G. alexborisenkoi (with one sample gregarious of three specimens, and only one solitary), G. bobkulai (seven samples solitary, only one gregarious of two specimens), G. boharti (ten samples gregarious with two, three, five or ten specimens, and only one solitary), G. bobwhartoni (ten samples gregarious with no more than five specimens per sample, only one solitary), G. carlhuffakeri (16 samples gregarious with maximum of six or eleven specimens in few vials, and only one solitary), G. haroldgreeneyi (eleven samples gregarious with maximum three specimens, only one solitary), G. jimmilleri (five samples gregarious with two to four, seven specimens, only one solitary), G. mamiae (eight samples gregarious some with only two specimens, and only one solitary) and G. meganmiltonae ten samples gregarious, some of them with two, five, nine, maximum 13 specimens, and only one solitary). The presence of two specimens in a sample automatically qualifies it as “gregarious” (as obviously a misnomer, but as used traditionally in Hymenoptera natural history). The occurrence of a single solitary sample in a species which a majority of samples are gregarious can be explained by human error from transferring specimens from rearing containers to vials with ethanol or during handling of the material in the laboratory, or that simply one wasp larva survived the arduous trip from egg to emerging adult. Loss of specimens during these procedures is feasible due to the small size of the adult parasitoids. Almost all of the species mentioned above have gregarious samples with the minimum number of siblings (one) that categorizes a sample as gregarious, a fact that supports the idea of a technical artifact rather than a real biological phenomenon. However, it has been reported in other Microgastrinae genera that females of some species laid one to three eggs per host, but usually, only one offspring survived to adulthood. One example comes from Microplitis demolitor Wilkinson that parasitizes the caterpillar Heliothis virescens (F.) (Noctuidae) (Strand et al. 1988).

In the majority of gregarious species, the number of wasps emerging from one host caterpillar does not exceed one hundred individuals. Only seven species overstep this amount. Such is the case of G. howelldalyi (from 100 to 161 adults), G. donquickei (106), G. andydeansi (from 108 to 190), G. iangauldi (from 114 to 196), G. andybennetti (138), G. billbrowni (185), and G. sydneycameronae (212). It is assumed that offspring is the result of a single ovipositing female, but need not be. The hosts of these species belong to the Lepidoptera families Sphingidae, Noctuidae, and Apatelodidae (Table 4) with body sizes relatively large.

Fifteen lepidopteran families were reported as hosts of Glyptapanteles of which five families are reported for the first time as hosts from Neotropical Glyptapanteles: Crambidae, Depressariidae, Euteliidae, Hesperiidae, and Sphingidae (Table 3). However, in the older literature, almost any record of Elachistidae would be reported as Depressariidae today. A total of 88 species within 84 genera was identified as hosts. Glyptapanteles attacks mainly members of the family Noctuidae, followed by Erebidae, and then distantly by Geometridae (Table 3). In contrast, Apanteles, another of the “super genera” within Microgastrinae, parasitizes (in the same region) principally species of Hesperiidae, Elachistidae, and Crambidae (Fernández-Triana et al. 2014a). As for the very species-rich microgastrine genus, Cotesia, the most frequent hosts are Nymphalidae, Saturniidae, and Hesperiidae (O’Connor 2011). None of these host families used by Apanteles and Cotesia is the core target of Glyptapanteles. Putative Glyptapanteles species waiting for description and which belong to both projects have emerged from different Lepidoptera families from those reported here. This is the case for hosts from Bombycidae, Dalceridae, Gelechiidae, Nolidae, Riodinidae, and Tortricidae (Janzen and Dyer pers. obs.), demonstrating a wider breadth of host range within Glyptapanteles.

As mentioned before, 127 Glyptapanteles species out of 136 were obtained from reared material of which 74 species (58%) were recovered multiple times. Thus, the remaining 53 species had a unique rearing occurrence. In total, 45 Glyptapanteles species have registered only family or subfamily hosts determinations. Two species, G. josesimbanai and G. marshawheelerae, lack of any level of information about lepidopteran host affiliations (Table 4).

Approximately 96% of the Glyptapanteles species with known host records parasitize a defined group of Lepidoptera, just a single host family or a narrower group, while a very small number (five species) use a slightly broader taxonomic range, parasitizing more than one Lepidoptera family [e.g., G. andrewdebeveci (Noctuidae and Pyralidae), G. bobwhartoni (Erebiidae and Saturniidae), G. edwinnarvaezi (Apatelodidae, Noctuidae, and Nymphalidae), G. luchosalagaje (Nymphalidae and Saturniidae), and G. tomwallai (Apatelodidae and Erebidae)]. All of these supposedly broader host ranges require more study before concluding that they are accurate, owing to potential errors in host caterpillar identification. However, misidentifications at the family level seem to have a low probability.

In total, 16 Glyptapanteles species were reared from more than one Lepidoptera species that belong to the same caterpillar family as well as the same subfamily: G. alexborisenkoi (2), G. andydeansi (3), G. annettewalkerae (2), G. brianestjaquesae (2), G. charlesporteri (2), G. daveschindeli (3), G. donquickei (2), G. garygibsoni (2), G. ianyarrowi (4), G. jimmilleri (3), G. johnburnsi (3), G. johnnoyesi (2), G. lubomasneri (2), G. mikeschauffi (2), G. nealweberi (2), and G. sydneycameronae (3). Only two parasitoid species emerged from hosts from different subfamilies within the same family: G. bobkulai attacks members of Sterrhinae and Larentiinae (Geometridae) and G. ilarisaaksjarvi specialized in Amphipyrinae and Plusiinae (Noctuidae) (Table 4). With six Noctuidae species hosts, G. ilarisaaksjarvi is the species with the greatest number of hosts recorded in this study.

Four duos and two trios of Glyptapanteles species share the same Lepidoptera host(s). Thus, G. erictepei and G. haroldgreeneyi have been reared from Actinote stratonice Latreille (Nymphalidae), G. donquickei and G. ilarisaaksjarvi have been reared from Condica cupienta (Cramer) (Noctuidae), G. linghsiuae and G. luchosalagajei have been reared from Hypanartia sp. Hübner (Nymphalidae), G. sydneycameronae and G. andydeansi have been reared from Aleuron carinata (Walker), Enyo ocypete (Linnaeus), and Pachygonidia drucei (Rothschild & Jordan) (Sphingidae). Glyptapanteles felipesotoi, G. ferfernandezi, and G. genorodriguezae have been reared from Memphis nr. lorna (Druce) (Nymphalidae), and G. davesmithi, G. jamesrobertsoni, and G. jesusugaldei have been reared from Antiblemma sp. Hübner (Erebidae). All these host records (and for every caterpillar) require more replications and additional scrutiny of host caterpillar identifications to be certain that they represent actual host affiliations.

It is worth mentioning that none of the species previously reported as Lepidoptera hosts in the scientific literature from the Neotropics were obtained by the rearing projects. Some plausible explanations for the lack of those records include: some Lepidoptera species occur naturally at low densities at study sites, time of foraging (early or late in the season) does not coincide with the collecting time, and larvae are well-camouflaged or semi-concealed (leafrollers, leaf tiers, shelter-building, grass moth, twig-like pose of some Geometridae) are even more difficult to spot in the field and, to a certain extent, never collected during censuses. Moreover, previous studies seem to be mostly restricted to field crops, not natural ecosystems. Additionally, some species are not reared successfully because natural conditions are difficult to replicate in the laboratory; caterpillars need special conditions or succumb to pathogens or other sources of mortality.