Twelve new species of Dipara Walker, 1833 (Hymenoptera, Chalcidoidea, Pteromalidae, Diparinae) from Kenya, with a key to the Afrotropical species

Abstract Based on 261 female specimens of the genus Dipara Walker, 1833 from leaf litter samples of the Kakamega Forest in Kenya, we describe the following twelve new species: Diparaandreabalzeraesp. nov., Diparacoronasp. nov., Diparafastigatasp. nov., Diparakakamegensissp. nov., Diparaluxsp. nov., Diparanigroscutellatasp. nov., Diparanyanisp. nov., Diparareticulatasp. nov., Dipararodneymullenisp. nov., Diparasapphirussp. nov., Diparatenebrasp. nov., and Diparatigrinasp. nov. For Diparaalbomaculata (Hedqvist, 1963) and Diparanigrita Hedqvist, 1969, we give new distribution records. We examined the available type material of all described Dipara species from the Afrotropical mainland, i.e., Diparaalbomaculata (Hedqvist, 1963), Diparamachadoi (Hedqvist, 1971), Diparamaculata (Hedqvist, 1963), Diparanigrita Hedqvist, 1969, Diparapallida (Hedqvist, 1969), Diparapunctulata (Hedqvist, 1969), Diparasaetosa (Delucchi, 1962), Diparastraminea (Hedqvist, 1969), Diparastriata (Hedqvist, 1969), and Diparaturneri Hedqvist, 1969. We provide figures, descriptions, and diagnoses of the newly described species and figures and diagnoses of the ten known species as well as an identification key to all species of the Afrotropical mainland.


Introduction
In this study, we contribute to the taxonomic knowledge of the Afrotropical fauna of the genus Dipara Walker, 1833, with the first alpha-taxonomic treatment of this group and region in 50 years. We describe twelve new species and diagnose and key all new and previously described Afrotropical mainland species. Dipara belongs to the subfamily Diparinae within the chalcidoid family Pteromalidae (Heraty et al. 2012). The genus shows a cosmopolitan distribution (Desjardins 2007) with a total of 56 described species (Noyes 2019). The phylogenetic position of Diparinae is still unclear (Desjardins 2007;Heraty et al. 2012). Currently, it is still classified within Pteromalidae, which, however, is polyphyletic . Diparinae were shown to be monophyletic and can be identified by the following diagnostic characters: presence of a cercal brush and absence of a smooth convex dorsellum (Desjardins 2007). The genus Dipara is well characterized by a number of diagnostic characters (see below or Desjardins 2007 for a full diagnosis and a list of genera previously synonymized under Dipara).
The early taxonomic work on Dipara was confounded by the strong sexual dimorphism in this group. Males are usually macropterous and have filiform antennae. Females can range from macropterous to apterous and have clavate antennae. Additionally, males tend to be extremely similar even across different genera while females show a lot of interspecific morphological variation (Desjardins 2007). This led to the genus originally being described by Walker (1833) based on a male specimen of Dipara petiolata Walker, 1833 and Dipara females originally being described as Tricoryphus by Förster (1856) and as Hispanolelaps by Mercet (1927). The two genera were later synonymized with Dipara by Domenichini (1953). Because of the strong resemblance of males of different species and the morphological variation of females, most species level taxonomic work on Dipara (and other Diparinae) is based on female specimens (Delucchi 1962;Hedqvist 1963Hedqvist , 1969Hedqvist , 1971. For males, Desjardins (2007) provided a genus level key. Matching females and males of the same species based on morphological features is currently not possible. Mitroiu (2019) suggested to match conspecific females and males based on molecular sequence data (e.g., the DNA barcode) and this is certainly the way to go. Unfortunately, the material available for this study was not suitable for standard DNA sequencing and consistently failed in a pre-study trial (unpublished). Accordingly, our work is based solely on morphological characters of females.
A peculiar characteristic of Diparinae females is their intraspecific variation in the wing form with macropterous and brachypterous specimens being found in the same species (Bouček 1988;Desjardins 2007;Mitroiu 2019). To deal with this potentially confounding fact, we used a multivariate morphometric approach (Baur and Leuenberger 2011) in morphologically similar species with different wing forms, which has been applied successfully numerous times for taxonomic studies on parasitoid wasps (e.g., László et al. 2013;Baur et al. 2014;Baur 2015;Gebiola et al. 2017;Werner and Peters 2018). Additionally, we checked the state of the posterior notal wing process which Desjardins (2007) suggested to be a "measure of potential wing size", i.e., a possible hint on the intraspecific wing form variation.
There is a severe lack of information about the biology of Dipara species. One of their main habitats is supposed to be leaf litter (Desjardins 2007). The only published information about their hosts is that of an unidentified Indian Dipara species which was reared from a curculionid beetle feeding on the roots of a Cyperus species (Bouček 1988). Additional host records from curculionids in Lelaps Walker, 1843led Desjardins (2007 to suggest that the more common and typical Diparinae (like Lelaps and Dipara species) may parasitize soil-inhabiting beetles and maybe curculionids more specifically.
We based our work on an extraordinary series of 261 female Dipara specimens from the Kakamega Forest in Kenya. Collection of the specimens was done in the framework of the BIOTA (BIOdiversity Monitoring Transect Analysis in Africa) East Africa project (Ross et al. 2018). The Kakamega Forest is a montane rainforest fragment in western Kenya and the easternmost remnant of the Guineo-Congolian rainforest belt (Kokwaro 1988;Clausnitzer 2005;Holstein 2015). Due to high rural population density around the Kakamega Forest it is under high threat from deforestation and habitat destruction (KIFCON 1994). Parts of its plant and animal fauna have already been studied in detail (e.g., Althof 2005;Clausnitzer 2005;Kühne 2008;Hita-Garcia et al. 2013). To preserve biodiversity, it is a most urgent and necessary task to contribute to the knowledge of highly diverse, threatened habitats, including knowledge on parasitoid wasps of these areas, by increasing visibility of species from this region and making specimens from it available.
With the description of twelve new Dipara species from the leaf litter of Kakamega Forest in Kenya we can show that the species diversity of the genus has not been sufficiently studied and the true diversity of Afrotropical Dipara, and presumably other Diparinae, has been underestimated. Since our very much geographically limited study already more than doubles the number of known species, we expect that numerous additional species of Afrotropical Dipara still await discovery and description. This study may serve as a starting point for future in-depth investigations, including thorough taxonomic revisions of the Afrotropical Diparinae, Chalcidoidea or, more generally, parasitoid wasp fauna.

Materials and methods
In the following, abbreviations are given of the museums where the material used in this study is stored. The abbreviations will be used throughout the text.

MDLA
Laboratório de Biologia, Dundo, Lunda, Angola NHMUK Natural History Museum, London, UK NMK National Museums of Kenya, Nairobi RMCA Royal Museum for Central Africa, Tervuren, Belgium ZFMK Zoologisches Forschungsmuseum Alexander Koenig, Bonn, Germany The terminology is based on Gibson (1997) and the Hymenoptera Anatomy Ontology portal (Yoder et al. 2010). For the terminology of the surface sculpture, we used Harris (1979).
A total of 261 female Dipara specimens from the Kakamega Forest in Kenya were examined. They were collected in 2007 and 2008 using Winkler extraction of a 1 m² leaf litter sample in multiple transects throughout the Kakamega Forest (Ross et al. 2018) and stored in 70% ethanol at room temperature at the ZFMK. All female Dipara specimens were isolated from the collective leaf litter samples and transferred to 99.8% ethanol. After presorting and examination, the specimens were critical point dried using a Leica EM CPD 300 AUTO and mounted on small, pointed cardboard plates with shellac-based glue. Morphological examinations were done with a Zeiss Discovery V8 stereomicroscope with a Plan S 1.0× FWD 81mm objective and PI 10×/23 eyepieces.
Digital imaging was done with a Keyence VHX-2000 digital microscope. For images of the dorsal and lateral habitus and the head the VHX-J250 objective (250-2500×) was used. The images were stacked and edited in brightness, coloration and contrast using the Keyence internal software. Further editing of figures was done with Microsoft Power Point 2010. For the images for the morphometric measurements the Keyence VH-Z20R objective (20-200×) with a magnification of 200× was used. For the body length and the gaster length magnifications of 100× or 150× were used if the character did not fit into an image with 200× magnification. After calibration, measurements were done using ImageJ 1.53a. Characters used for morphometric measurements are given and explained in Table 1.
The range of the morphometric measurements is given in the species description with the value for the holotype in parentheses. If more than five specimens were present, five specimens were used for the morphometric measurements, and their respective collection numbers are given in parentheses at the beginning of the taxonomic treatment (see below). If five or less specimens were available, all specimens were used. In a few cases the number of specimens used for a certain measurement varies from the total number of specimens used, either because the measured character was not visible in some specimens or because more specimens were used for the in-depth morphomet-ric analysis, using a subset of the characters (see below). In these cases, the collection numbers of specimens used are given in parentheses directly after the respective measurement (see Suppl. material 1: Table S1 and Suppl. material 2: Table S2). Some morphometric characters were used to calculate ratios. For these ratios different categories were defined to simplify the description of shape (Table 2). The shape ratios are given in Suppl. material 3: Table S3.  Graham (1969), Gibson (1997) and Baur (2015)). Characters highlighted in bold were used for the morphometric analysis of D. kakamegensis sp. nov. and D. nyani sp. nov. (see Tables S1 and S2 for

Morphometric analysis
Two putative species were found that clearly differed in the wing form, with one being macropterous and the other being brachypterous, but they were otherwise very similar with no obvious qualitative characters found to separate them. Since wing form might vary within species (Bouček 1988;Desjardins 2007;Mitroiu 2019), we chose to apply a quantitative approach based on multivariate morphometric analysis (Baur andLeuenberger 2011, 2020;Baur et al. 2015). For this purpose, the characters highlighted in bold in Table 1 were imaged and measured, as explained before, for 30 specimens of the first species (that later became D. kakamegensis sp. nov., see below) and all five specimens of the second species (that later became D. nyani sp. nov., see below). The measurements (given in Suppl. material 1: Table S1) were subsequently analyzed using R 4.0.2 and the R script package by Baur and Leuenberger (2020). Missing data was added using the imputation function of the mice R package.

Posterior notal wing process
As suggested by Desjardins (2007), the posterior notal wing process (pnwp) can be used as a "measure of potential wing size". The pnwp can be absent in brachypterous or apterous species, leading to the assumption that a fully developed pnwp in a brachypterous species could mean that macropterous individuals of this species exist. Desjardins (2007) lists four different character states: present and pointed, present but squarely truncate, present but truncate and rounded, and absent. We examined the state of the pnwp in each new species and imaged specimens of four different species with varying wing forms, from brachypterous to macropterous (Fig. 4), using the Keyence with the VHXJ-250 objective as described above.

Morphometric analysis
The morphometric analysis of specimens of the two morphologically similar putative species showed that they can be reliably separated ( Fig. 1 and 3). In the following, they will be treated as D. kakamegensis sp. nov. and D. nyani sp. nov. Based on the results of the scree graph (not shown), only the first and second principal component (PC) were relevant for the further analysis of shape. The results of the shape PCA (Fig. 1A) of the two species show that they are separated by shape. The ratio spectrum (Fig. 1B) shows which ratios had the highest impact on the first shape PC. To confirm that these differences are based on true shape differences and not allometric size effects, the isometric size was plotted against the first shape PC ( Fig. 2A). The species overlap in size but lie on different allometry axis, confirming that the separation is based on shape and not on allometry effects.
The LDA ratio extractor (Baur and Leuenberger 2020) found the best ratios to separate the two species: mss.l/sctl.l was the best ratio, clv.l/prn.l was the second best ratio (Fig. 3).
The allometry ratio spectrum (Fig. 2B) reveals the allometric variation of ratios. The characters of the best ratio (mss.l/sctl.l) lie close to each other, indicating no strong allometric effects. The characters of the second-best ratio (clv.l/prn.l) show a higher allometric effect than the first one but still not a considerable one. This confirms that the differences in these characters are based on shape and not on allometric effects.
The separating ratios were used for the diagnoses of the two species in the descriptions below.  Table 1 A scatterplot of first against second shape PC B ratio spectrum of the first shape PC; horizontal bars represent 68% confidence intervals based on 1000 bootstrap replicates.

Posterior notal wing process
Examination of the posterior notal wing process (pnwp) in the newly described species showed that it was present and pointed in all cases. Figure 4 shows a selection of pnwps from species with different wing forms. The uniformity of this character can be interpreted as it either being unsuitable as a measure of potential wing size (see Desjardins 2007) or as indicating that all species, including the brachypterous ones, harbor also macropterous specimens. Accordingly, it proved little help in delimiting species, especially in the case of the very similar, but morphometrically discriminated macropterous D. nyani sp. nov. and brachypterous D. kakamegensis sp. nov. (see morphometric analysis above and taxonomic treatment below).   Desjardins (2007) states that the petiole of females of the genus Dipara is usually less than 1.5× as long as wide. In the present study, seven out of twelve newly described species have a petiole length exceeding this, going up to being 2.8× as long as wide. Accordingly, the genus level key of Desjardins (2007) might be misleading. However, the diagnosis by Desjardins (2007) uses other characters than the petiole length and can be kept unmodified (see above).
Remarks. Dipara andreabalzerae is similar to D. albomaculata, D. fastigata, D. nigroscutellata, and D. saetosa in having a black mesoscutellum while the general body coloration is not black. Dipara andreabalzerae differs from D. albomaculata, D. nigroscutellata and D. saetosa in different propodeum sculpture. It differs from D. fastigata in body coloration, which is much darker in D. fastigata and the more obtuse angle formed by the anterior part of the mesoscutellum and the frenum in lateral view.
Male. Unknown. Etymology. As the first author, I dedicate this species to my mother, Andrea Balzer, who sadly passed away in 2017.
Remarks. Dipara corona is similar to D. turneri in having a distinct transverse broad black stripe on the median and lateral areas of the mesoscutum. In other not completely black species the black spots on the mesoscutum are restricted to the lateral area.
Dipara corona differs from D. turneri in the wing form and in the different petiole shape. The petiole is distinctly longer than wide in D. corona and slightly wider than long in D. turneri. Other differences include the body coloration, the shape of the mesoscutellum and the shape of the metacoxa.
Male. Unknown. Etymology. Named after the Latin word corona for crown because of the raised and shiny part between the ocelli in frontal view, and additionally as a reference to the pandemic in 2020 and the following years caused by SARS-CoV-2, also known as the Corona virus.
Remarks. Dipara fastigata is similar to D. andreabalzerae, D. albomaculata, D. nigroscutellata and D. saetosa in having a black mesoscutellum while the general body coloration is not black. Dipara fastigata differs from D. albomaculata, D. nigroscutellata and D. saetosa in different propodeum sculpture. It differs from D. andreabalzerae in general body coloration, which is much lighter in D. andreabalzerae, and the 90° angle formed by the anterior part of the mesoscutellum and the frenum in lateral view.
Male. Unknown. Etymology. Named after the Latin adjective fastigatus for pointed or sharp. The name refers to the raised mesoscutellum.
Biology  Diagnosis. Body bright yellowish brown (Fig. 8); face with two transverse stripes of very dark brown coloration just at the level of toruli and at the level of the ventral margin of the eye, interrupted in interantennal area and supraclypeal area, enclosing a stripe of pale white coloration (Fig. 8B); legs yellowish brown except for metacoxa white (Fig. 8A); brachypterous, fore wing reaching middle of gt1 (Fig. 8A); mesoscutellum small, mesosoma length 3.90-4.86× (4.86) (Fig. 8C) (specimens used for measurement: ZFMK-HYM-00037140 to ZFMK-HYM-00037170) mesoscutellum length; petiole short to medium, 1.15-1.72× (1.15) as long as wide in dorsal view. Description. Size: small to medium sized, body length 1483-2227 (2027) µm. Coloration: body bright yellowish brown (Fig. 8); ventral part of scape and clava pale yellowish white, dorsal part of scape and last three funicle segments brown, rest of funicle segments and pedicel yellowish brown (Fig. 8A); face with two transverse dark brown stripes just at the level of toruli and at the level of the ventral margin of the eye, interrupted in interantennal area and supraclypeal area, enclosing a stripe of pale white coloration (Fig. 8B); two black spots medially on lateral areas of mesoscutum (Fig. 8C); middle part and tip of the fore wing infuscate (Fig. 8A); legs yellowish brown except for metacoxa white (Fig. 8A); nucha and posterior 2/3 of petiole pale yellowish white, rest of petiole bright yellowish brown (Fig. 8C); some darker brown stripes dorsally on gaster (Fig. 8A); brown spots on gt6 and gt7 around cerci (Fig. 8A); tip of ovipositor sheath brown (Fig. 8A).
Variations. The bristles on the forewing can vary from five to 15. This variation is found in the bristles at the tip of the wing while along the edges there are constantly five bristles. In some specimens there are just a few larger bristles at the tip and in others there can be up to ten small bristles at the tip. The number of bristles can vary between left and right wing in one specimen. The surface sculpture of the median part of the propodeum can vary from rugose to smooth.
Remarks. Dipara kakamegensis is very similar to D. nyani. It differs from D. nyani in the following characters: D. kakamegensis is brachypterous and the mesoscutellum is smaller relative to the mesosoma length, based on the morphometric analysis (Fig. 3). The stripes across the face are similar in D. maculata, D. reticulata and D. rodneymulleni. Dipara kakamegensis differs from D. maculata in having a yellowish brown mesocoxa and petiole. Dipara kakamegensis differs from D. rodneymulleni in many characters: D. kakamegensis is brachypterous, the body coloration, the length of the petiole and general body shape. Dipara kakamegensis differs from D. reticulata in having smooth gastral tergites while they are reticulated in D. reticulata.
Male  (Fig. 9); face with dark brown to black stripe from one eye to the other at the level of the ventral margin of the eye (Fig. 9B); vertex reticulate (Fig. 9B); petiole very long, 2.50-2.61× (2.50) longer than wide in dorsal view (Fig. 9C).
Description. Size: medium sized, body length 2243-2772 (2772) µm. Coloration: body yellowish brown (Fig. 9); scape, pedicel and f1-3 yellowish brown, f4 yellowish brown to dark brown, f5-7 dark brown, clava yellowish brown (Fig. 9A); face with dark brown to black stripe from one eye to the other at the level of the ventral margin of the eye (Fig. 9B); mesoscutum with two black spots medially on lateral area (Fig. 9C); fore leg with distal tip of coxa brown and rest of coxa white, trochanter brown, rest yellowish brown (Fig. 9A); mid leg with coxa and trochanter white, rest yellowish brown (Fig. 9A); hind leg with anterior part of coxa white and posterior part dark brown, anterior part of femur white, rest yellowish brown (Fig. 9A); gt6 and gt7 with dark brown spots around cerci (Fig. 9A); posterior tip of gt7 dark brown (Fig.  9A); tip of ovipositor sheath dark brown, rest white (Fig. 9A).
Remarks. Dipara lux is similar to D. corona, D. machadoi, D. striata, D. tenebra, D. tigrina and D. turneri in having one dark brown to black stripe across the face. Dipara lux is different from D. machadoi in having distinct notauli, which are lacking in D. machadoi. It differs from D. corona, D. striata, D. turneri and D. tigrina in having a very long petiole. Dipara lux and D. tenebra are very similar in body shape and differ in their body coloration which is much brighter in D. lux and in the surface sculpture of the head. They share the otherwise unique character of having a dense brush of setae close to the proximal end of the marginal vein on the fore wing.
Variation. The bristles on the gt1 and the tip of the forewing can sometimes be missing. In this case the pit where the bristles are supposed to be is still visible.
Remarks. Dipara nigroscutellata is similar to D. andreabalzerae, D. albomaculata, D. fastigata, and D. saetosa in having a black mesoscutellum while the general body coloration is not black. Dipara nigroscutellata differs from D. andreabalzerae and D. fastigata in having a pair of large bristles dorso-anteriorly on the gt1. It differs from D. albomaculata and D. saetosa in the general body coloration, which is much brighter in D. nigroscutellata and in the coloration of the lateral area of the mesoscutum. In D. nigroscutellata the lateral area is laterally yellowish brown and in D. albomaculata and D. saetosa the lateral area is completely black.
Dipara nigroscutellata is similar to D. straminea in sharing the bristles on the gt1 and in propodeum sculpture. It differs from D. straminea in having a black mesoscutellum.
Coloration: body bright yellowish brown (Fig. 11A); scape ventrally yellowish white, dorsally brown, pedicel and f1-4 yellowish brown, f5-7 brown, clava pale yellowish white (Fig. 11A); face with two transverse stripes of dark brown coloration just at the level of toruli and at the level of the ventral margin of the eye, interrupted in interantennal area and supraclypeal area enclosing a stripe of pale white coloration (Fig. 11B); mesoscutum with pair of black spots medially on lateral area (Fig. 11C); two infuscate spots at the upper edge of the fore wing, one at 1/3 of the length and the other one in the middle (Fig. 11A); legs yellowish brown except for metacoxa white (Fig. 11A); gt6 and gt7 with brown spots around cerci (Fig. 11A); tip of ovipositor sheath brown (Fig. 11A).
Remarks. Dipara nyani is very similar to D. kakamegensis. It differs from it in the following characters: D. nyani is macropterous and the mesoscutellum is larger relative to the mesosoma length, based on the results of the morphometric analysis (Fig. 3). The stripes across the face are similar in D. maculata, D. reticulata and D. rodneymulleni. Dipara nyani differs from D. maculata in having a yellowish brown mesocoxa and petiole. Dipara nyani differs from D. rodneymulleni in many characters: the body coloration, the length of the petiole and the body shape. Dipara nyani differs from D. reticulata in having smooth gastral tergites while they are reticulated in D. reticulata.
Male. Unknown. Etymology. Named after the word for monkey in the national language of Kenya, Swahili, because of the dorsal black dots and the mesoscutellum which resemble the face of a monkey.
Remarks. Dipara reticulata is similar to D. kakamegensis, D. maculata, D. nyani, and D. rodneymulleni  Diagnosis. Female. Face with two dark brown stripes at the level of the ventral margin of the eye, interrupted in supraclypeal area, and at the level of the toruli (Fig. 13B); absence of black spots on median area of mesoscutum (Fig. 13C); petiole very long, 2.53-2.80× (2.79) longer than wide (Fig. 13C).
Description. Size: medium to large, body length 2718-3397 (3397) µm. Coloration: body brown (Fig. 13); distal quarter of scape and pedicel, all funicle segments and small proximal part of the first claval segment (c1) dark brown, rest of scape white, rest of pedicel and clava yellowish brown (Fig. 13A); face with two dark brown stripes at the level of the ventral margin of the eye, interrupted in supraclypeal area, and at the level of the toruli (Fig. 13B); fore leg yellowish brown (Fig. 13A); mid leg with coxa and trochanter white, rest brown (Fig. 13A); hind leg with coxa white with dark brown coloration on posterior part, tibia dark brown, rest brown (Fig. 13A); gt7 with dark brown coloration around cerci and on posterior 1/2, rest of gt7 yellowish brown (Fig. 13A); ovipositor sheath yellowish brown on anterior 1/2 and posterior tip dark brown (Fig. 13A).
Metasoma: petiole very long, 2.53-2.80× (2.79) longer than wide, with anterior 2/3 rugose and rest carinate, with lateral pair of large white setae visible in dorsal view (Fig. 13C); gaster medium, 1.46-1.53× (1.53) longer than mesosoma in dorsal view (Fig. 13C); gt1 covering ~1/3 of gaster, gt2-6 ca. equal in size (Fig. 13C); gt7 and ovipositor sheath slender and elongated, together ca. as long as rest of gaster, sparsely setose (Fig. 13C). Remarks. Dipara rodneymulleni shares the stripes across the face with D. maculata, D. nyani, D. kakamegensis, and D. reticulata and but other than that has a completely different morphology and coloration. The most obvious characters to distinguish D. rodneymulleni are the very long petiole and the absence of black spots on the lateral areas of the mesoscutum.
Male. Unknown. Etymology. Named after professional skateboarder Rodney Mullen who revolutionized street skating like no other, reflecting the first author's lifelong passion for skateboarding.
Diagnosis. Strong blue metallic tint on the following areas: vertex between ocelli, pronotum laterally, median area of mesoscutum posteriorly between notauli, lateral area of mesoscutum and mesoscutellum (Fig. 14C).
Remarks. In body shape, D. sapphirus is similar to D. lux and D. tenebra but can be distinguished from all other Dipara species by having a very distinct blue metallic tint on the following body parts: vertex between ocelli, pronotum laterally, median area of mesoscutum posteriorly between notauli, lateral area of mesoscutum and mesoscutellum (Fig. 14C) (Fig. 15); face with dark brown to black stripe from one eye to the other at the level of the ventral margin of the eye, interrupted in supraclypeal area (Fig. 15B); vertex smooth (Fig. 15B); petiole very long, 2.51-2.77× (2.51) longer than wide in dorsal view (Fig. 15C).
Description. Size: medium sized, body length 2293-2474 (2293) µm. Coloration: body brown to dark brown (Fig. 15); scape and f1 yellowish brown, pedicel, f23, and f7 yellowish brown to brown, f4-6 brown, clava white (Fig. 15A); face with dark brown to black stripe from one eye to the other at the level of the ventral margin of the eye, interrupted in supraclypeal area (Fig. 15B); mesoscutum with two black spots medially on lateral area (Fig. 15C); fore leg with distal tip of coxa brown, rest of coxa white, trochanter and femur brown, tibia and tarsus yellowish brown (Fig. 15A); mid leg with coxa and trochanter white and rest yellowish brown (Fig. 15A); hind leg with anterior part of coxa, trochanter and anterior part of femur white, posterior part of coxa dark brown to black, rest of hind leg yellowish brown (Fig. 15A); gt1 brown, anterior 2/3 of gt7 yellowish brown, rest of gaster dark brown (Fig. 15A).
Remarks. Dipara tenebra is similar to D. corona, D. lux, D. machadoi, D. striata, D. tigrina, and D. turneri and in having one dark brown to black stripe across the face. Dipara tenebra is different from D. machadoi in having distinct notauli, which are lacking in D. machadoi. It differs from D. corona, D. striata, D. tigrina, and D. turneri in having a very long petiole. Dipara tenebra and D. lux are very similar in body shape and differ in their body coloration which is much darker in D. tenebra and in the surface sculpture of the head. They share the otherwise unique character of having a dense brush of setae close to the proximal end of the marginal vein on the fore wing.
Male. Unknown. Etymology. Named after the Latin word tenebra for darkness, in contrast to D. lux which looks very similar but is much lighter in coloration.
Remarks. Dipara tigrina is similar to D. corona, D. lux, D. machadoi, D. striata, D. tenebra, and D. turneri in having one dark brown to black stripe across the face. It differs from D. corona, D. lux, D. machadoi, D. tenebra, and D. turneri in the propodeum sculpture. The propodeum sculpture is similar in D. punctulata and D. striata. They show a very distinct surface sculpture with a striated subcarinate pattern extending to the nucha. Dipara tigrina differs from D. punctulata and D. striata in having more setae laterally on the propodeum and in having a reticulated pattern medially between the carinae on the propodeum.
The only available specimen of this species has an irregular black spot on the propodeum. This spot is considered an aberration and thus is not part of the species description.
Remarks. The holotype of D. albomaculata is supposed to be stored at the MDLA but we were unable to get in contact with the museum and thus the holotype could not be located and examined. Two paratypes are stored at the BMNH and one of them was examined.
Dipara albomaculata is similar to D. andreabalzerae, D. fastigata, D. nigroscutellata and D. saetosa in having a black mesoscutellum while the general body coloration is not black. It differs from D. andreabalzerae and D. fastigata in having a pair of bristles dorso-anteriorly on the gt1. It differs from D. nigroscutellata in the general body coloration, which is much darker and in the coloration of the lateral area of the mesoscutum. In D. albomaculata the lateral area is completely black and D. nigroscutellata has a small yellowish brown area on its most lateral part. The differences to D. saetosa can be found in the smooth vertex and the white pro-and metacoxa.
Dipara albomaculata is similar to D. straminea in sharing the bristles on the gt1 and the propodeum sculpture. It differs from D. straminea in having a black mesoscutellum.
Additional specimens from this species were found in the Kakamega Forest in Kenya and the distribution is updated accordingly.

Diagnosis. Female. Notauli absent.
Remarks. The holotype of D. machadoi is supposed to be stored at the MDLA but we were unable to get in contact with the museum and thus the holotype could not be located and examined. Based on the original description by Hedqvist (1971) D. machadoi differs from all other Afrotropical Dipara species in having no notauli. (Hedqvist, 1963) Afrolelaps maculata Hedqvist 1963: 47-49. Diagnosis. Female. Face with two transverse stripes of dark brown coloration just at the level of toruli and at the level of the ventral margin of the eye, enclosing a stripe of pale yellowish white coloration; mesocoxa and petiole white.

Dipara maculata
Remarks. The holotype of D. maculata is supposed to be stored at the MDLA but we were unable to get in contact with the museum and thus the holotype could not be located and examined. Based on the original description by Hedqvist (1963) it is similar to D. kakamegensis, D. nyani, and D. rodneymulleni in having two transverse stripes on the face. It differs from D. rodneymulleni in having a much shorter petiole. In contrast to D. kakamegensis and D. nyani, D. maculata has a white petiole and mesocoxa. Fig. 18A-D, 19A-C Dipara nigrita Hedqvist, 1969: 195.  (Hedqvist, 1963) (Figs 18, 19). Variation. Dipara nigrita was originally described as brachypterous (Fig. 18) by Hedqvist (1969). In the examined material from Kenya, we found specimens that we consider to be the macropterous form of this species (Fig. 19). Differences in the wing form within Diparinae are reported from several other species (Bouček 1988;Mitroiu 2019) and the slight differences found between the macropterous forms and the brachypterous holotype were not enough to justify describing the macropterous form as a new species. Those differences were found in the color of the first claval segment. It can vary from light brown to white.  Macropterous individuals have fully developed wings with the fore wings reaching the gt7 (Fig. 19). Brachypterous individuals show much shorter wings with the fore wings reaching approximately the posterior margin of the petiole (Fig. 18).

Dipara nigrita Hedqvist, 1969
Remarks. Dipara nigrita is the only species which shows a completely black coloration of the head and mesosoma. Darker specimens of D. albomaculata sometimes have a partly very dark brown to black head and mesosoma but never completely black. Additionally, the coxa of D. albomaculata are white in contrast to the dark brown coxa of D. nigrita.
Additional specimens from the species were found in the Kakamega Forest in Kenya and the distribution is updated accordingly.
Remarks. Dipara saetosa is similar to D. albomaculata, D. nigroscutellata, and D. straminea in having a pair of large bristles dorso-anteriorly on the gt1. It differs from D. straminea in having a black mesoscutellum. In contrast to D. nigroscutellata the lateral area of the mesoscutum is completely black. Differences to D. albomaculata Figure 21. Holotype of Dipara punctulata (Hedqvist, 1969) A habitus in lateral view B face in frontal view C head and mesosoma in dorsal view D metasoma in dorsal view; red arrow: long lateral bristles on the petiole.
can be found in the reticulation between the ocelli and the coloration of the clava and the pro-and metacoxa.
Dipara straminea (Hedqvist, 1969) (Delucchi, 1962) A habitus in lateral view B face in frontal view C body in dorsal view D labels; red arrows: dorsal bristles on gt1. Scale bar: 100 µm. Figure 23. Holotype of Dipara straminea (Hedqvist, 1969) A habitus in lateral view B face in frontal view C body in dorsal view D labels; red arrows: dorsal bristles on gt1. Scale bar: 100 µm.
Remarks. Dipara striata is similar to D. corona, D. lux, D. machadoi, D. tenebra, D. tigrina, and D. turneri in having one dark brown to black stripe across the face. It differs from D. corona, D. lux, D. machadoi, D. tenebra, and D. turneri in the propodeum sculpture. The propodeum sculpture is similar in D. punctulata and D. tigrina. They show a very distinct striated subcarinate pattern extending to the nucha. Dipara striata differs from D. punctulata in lacking a large bristle anterio-laterally on the petiole. Dipara striata differs from D. tigrina in having less setae laterally on the petiole and in lacking reticulation on the propodeum. Diagnosis. Female. Broad dark brown stripe across head from one eye to the other below toruli (Fig. 25B); median and lateral area of mesoscutum with distinct transverse broad black stripe (Fig. 25C); brachypterous, fore wing reaching slightly beyond petiole (Fig. 25A); petiole slightly wider than long (Fig. 25C).
Remarks. Dipara turneri is similar to D. corona in having a distinct transverse broad black stripe on the median and lateral area of the mesoscutum. In other not completely black species the black spots on the mesoscutum are restricted to the lateral area.
Dipara turneri differs from D. corona in the wing form and in the petiole shape. The petiole is longer than wide in D. corona and wider than long in D. turneri.

Discussion
Our results confirm that there is still a lot of undiscovered diversity within Microhymenoptera and the genus Dipara in particular (Desjardins 2007;Sharkey 2007;Aguiar et al. 2013;Forbes et al. 2018). Desjardins (2007) stated that there are "possibly hundreds of undescribed species" of Dipara left. We can support this statement based on the number of new Dipara species found only in the leaf litter in the small forest fragment Kakamega Forest, which more than doubled the number of known species from the Afrotropical mainland.
Some of the species descriptions in this study are based on so far unparalleled series of Dipara specimens. While most of the previously described Dipara species are known only from the holotype or just a few specimens, D. kakamegensis, for example, is described from 108 specimens and D. nigroscutellata from 86 specimens. These large series allowed for an advanced insight into intraspecific variation of Dipara species. The characters used for the species descriptions and diagnoses were found to be consistent among the large series, which gave us some confidence in delimiting species using the same characters in species with less specimens available. Our insights reveal that in most cases Afrotropical Dipara species seem to be reliably distinguishable by comparatively simple morphological characters of females like color patterns, surface sculpture or the number and position of setae or bristles. However, the intraspecific variation of the wing form of Diparinae females (Bouček 1988;Desjardins 2007;Mitroiu 2019) can pose a challenge for species delimitations, including those in this study. We decided to list wing related characters in the diagnoses and key but to always add additional non-wing characters.
While shedding more light on the species diversity of Dipara their biology remains largely unknown. All specimens were found in the leaf litter confirming that this might be their preferred habitat (Desjardins 2007). Reduced wings in females, which is found in 17 out of 22 Afrotropical species, can most likely be regarded as an adaptation to their grounddwelling lifestyle and their search for hosts in the leaf litter or the soil. To gain more information about the hosts of Dipara more studies focusing on their biology would be needed.
Adding information on the biology, taxonomy, and distribution of species, is a crucial task. We still have only very limited knowledge on the biodiversity on this planet. We are aware, though, that we are facing presumably unprecedented biodiversity loss, especially through habitat destruction, and that this is one of the most pressing problems of our time (Steffen et al. 2015). The tropics including the Afrotropics are especially under threat because of ongoing deforestation and changes in land use, while also being biodiversity hotspots (Brooks et al. 2002). For example, the Kakamega Forest is the last large continuous forest in Kenya (Holstein 2015) and an officially protected area, but it is still under threat of habitat destruction (KIFCON 1994;Bleher et al. 2006;Lung and Schaab 2006;Müller and Mburu 2009). Studying the diversity of parasitoid wasps or other species-rich, abundant but understudied taxa can be a decisive tool for highlighting their importance for ecosystems, for conservation efforts and for understanding the evolution of the insects' megadiversification. This contribution to our knowledge on the genus Dipara might serve as a small but valuable addition to the overwhelming picture of the biodiversity of the Afrotropics.

Table S3
Authors: Christoph Braun, Ralph S. Peters Data type: Xslx file. Explanation note: Shape ratios for the descriptions of the newly described Dipara species. See Table 1 for definitions of the characters and Table 1 for definitions of the characters and Table 2 for definitions of the shape categories. N: measurement not available. Copyright notice: This dataset is made available under the Open Database License (http://opendatacommons.org/licenses/odbl/1.0/). The Open Database License (ODbL) is a license agreement intended to allow users to freely share, modify, and use this Dataset while maintaining this same freedom for others, provided that the original source and author(s) are credited. Link: https://doi.org/10.3897/zookeys.1067.72395.suppl3