Life history and biocontrol potential of the first female-producing parthenogenetic species of Diglyphus (Hymenoptera: Eulophidae) against agromyzid leafminers

Diglyphus species are dominant biocontrol agents for suppressing outbreaks of agromyzid leafminers in fields. In July 2015, we collected a new thelytokous species of Diglyphus in Qinghai, China. The wasp is here named as D. wani Liu, Zhu & Yefremova sp. nov., based on morphological and molecular analyses. The life history and biocontrol potential of the wasp were studied in the laboratory and by providing Liriomyza sativae larvae. The intrinsic rate of increase, finite rate of increase, and mean generation time were 0.2373 d−1, 1.2678 d−1, and 15.9 d, respectively. The wasps showed three types of host-killing behaviour, namely parasitism, host feeding and host stinging, resulting in 47.6, 94.4 and 3.4 host larvae killed per wasp over a lifetime, respectively. The finite total host-killing rate was 1.0071 and the wasp showed strong synovigeny. The wasp may be a crucial biocontrol agent against agromyzid leafminers.

Thorax (Figs 1f,g and 2d,e,f). Pronotum shorter than mesoscutum, reticulate. Pronotum without transverse carina. Pronotum, mesontum and scutellum metallic green. Mesoscutum with notauli, with strong reticulation and areolae larger than that on scutellum, with 2 pairs of long setae on mid lobe. Mesoscutum 1.1× as long as scutellum. Scutellum as long as broad. Scutellum finely reticulate, with two straight sublateral grooves, with two anterior short setae and two posterior long setae. Dorsellum smooth. Propodeum 3.7× as broad as long, smooth, without median carina. Spiracle round, small with paraspiracular carina. Callus with 2 setae in first row and 2 setae in the second row.
Remarks. Diglyphus wani sp. nov. shows some characteristics that are similar to those of D. isaea, D. pulchripes, D. crassinervis, D. minoeus, D. sensilis, and D.chabrias 14,29 . The new species is similar to D.isaea, D. crassinervis, D. minoeus and D. chabrias owing to the scape cylindrical, scape predominantly dark, femora dark brown and metallic with apical white. However, in D. isaea the speculum is absent, POL 1.4× OOL, mouth 2.0× malar space, propodeum with median carina, clava 2.0× as long as F2, and tibia metallic with base and apical 1/5-1/6 yellow, wheras in D. wani sp. nov. the speculum is very small with sparse setations, POL 2.7× OOL, mouth 1.6× malar space, propodeum without median carina, clava 2.2× as long as F2, tibia with 2/3 brown distal part. In addition, D. crassinervis and D. minoeus have the same appearance of the antennae and cubital vein (not curved at base) as D. wani sp. nov., but in D. crassinervis clava 1.1-1.2× as long as funicle, mouth almost equal to malar space, POL 1.66× OOL, speculum extending along MV, and scutellum concolorous with mesoscutum, whereas in D. minoeus mouth 1.5× or more malar space, speculum extending along parastigma, and scutellum with a purplish shine in contrast to dark metallic blue-black mesoscutum. On account of the cubital vein not curved, fore and middle tibia with basal brown markings, D. wani sp. nov. and D. pulchripes are similar. Nonetheless, in D. pulchripes the clava is as long as funicle, coxae and femora green, hind tibia brown, and base of scape white. Diglyphus wani sp. nov. is closed to D. sensilis owing to terminal spine of clava, absent median carina on propodeum, but differ in POL 2.25× OOL, mouth 1.33× of malar space, clava 3.8× as long as F2, and tibia with two dark bands in D. sensilis. Diglyphus wani sp. nov. is closed to D. chabrias with body colour, subquadrate scutellum, small speculum with sparse setations on forewing and differs in the following characters: POL 2.25× as OOL, mesoscutum 1.5× as long as scutellum, gaster 1.8× as long as broad, and fore tibia with anterior surface white and posterior surface dark and metallic or dark brown.
Hosts and biology.  Longevity. Longevity under the host treatment was significantly longer than under the control starvation treatment (Table 2).
Biocontrol potential. Parasitism, and host-feeding, host-stinging and host-killing events. Parasitism. In the host treatment (n = 32), the parasitism number was 47.6 per parasitoid wasp (Table 2) over the whole life span, whereas average daily parasitism was 6.3 hosts. The curve for the daily parasitism showed a parabolic tendency, which fluctuated slightly on the 2nd and 9th days (Fig. 4). Host feeding. The parasitoids fed on 94.4 hosts during a lifetime and killed 12.2 hosts per day ( Table 2). The host-feeding events accounted for 66.5% of the host-killing events. The trend in daily host-feeding events gradually increased and there was a small decrease between 4th day and 5th day. The host-feeding event was highest on the 4th day at 14.9 hosts (Fig. 4).
Host stinging. The number of host-stinging events was 3.4 hosts in a lifetime (Table 2), whereas the average daily host-stinging number was 0.5 hosts.
Total host mortality. Average of 142.1 host killings over a lifetime was observed (Table 2), with individuals killing 18.5 hosts daily. On the 1st day, the host killing was 7.7 hosts, and peaked on the 4th day, with up to 23.7 hosts were killed, and thereafter declined slowly (Fig. 4). Consumption rate of three host-killing behaviour. Host-feeding rate. There was no consumption during the egg, larval, and pre-pupal to pupal stages, hence we calculated the host-feeding rate using the host feeding of adults. Thus, there was a blank stage before the females emerged. The maximum daily host-feeding rate of D. wani sp. nov. was 14.3 hosts on the 17th day (Fig. 5a). The net daily host feeding was 12.6 hosts on the 14th day (Fig. 5a). The C 0 , ψ, ω and Q q values were 86.3429 ± 6.8367, 0.5143 ± 0.033, 0.6520 ± 0.0442, and 1.9974 ± 3.0821, respectively.
Host-stinging rate. The maximum daily age-specific host-stinging rate was 0.8 hosts at 17th day of age (Fig. 5c). Both the daily age-specific and net age-specific host-stinging rate curves gradually increased with fluctuations prior to 17th day and then declined dramatically (Fig. 5c). The C 0 , ψ, ω and Q q values were 3.1429 ± 0.5154, 0.0193 ± 0.0029, 0.0244 ± 0.0037 and 0.0727 ± 0.0109, respectively.

Discussion
Based on distinct morphological characteristics and mitochondrial COI sequence analyses, the species was named Diglyphus wani Liu, Zhu & Yefremova sp. nov. To our knowledge, this is the first thelytokous species described in the genus Diglyphus. This wasp is a numerically dominant member of the parasitoid assembly of agromyzid leafminers in fields in Qinghai, Hebei, Beijing and Shanxi Provinces, China (unpublished data). The thelytokous parthenogenesis of this wasp, in which females produce daughters without mating, was observed. So far, the males of D. wani sp. nov. are absent in the field or in the laboratory colonies. According to the review by Ma and Schwander 26 , female-producing parthenogenesis arises by three main evolutionary mechanisms. First, thelytokous parthenogenesis could be induced by endosymbionts that increased their transmission by manipulating host reproduction, which was considerated to be a main route. Wolbachia, Cardinium and Rickettsia genera of parthenogenesis-inducing bacteria occured in Hymenoptrea. Endosymbiont-induced parthenogenesis that was known or suspected in 124 host species from seven arthropod taxa, of which Wolbachia was the most frequent endosymbiont (in 56-75% of host species) 26 . Stouthamer 30 indicated that at least 270 hymenopteran species showed female-producing parthenogenesis. Among Hymenoptera, Cardinium bacteria were known to occur only in chalcidoid wasps. At present, thelytokous reproduction caused by a parthenogenesis-inducing bacterium Rickettsia is known only in two hymenopteran parasitoids, Neochrysocharis formosa and Pnigalio soemius, and one hymenopteran pest, Leptocybe invasa [31][32][33] . Endosymbiont-induced parthenogenesis appears to facilitate the maintenance of reproductive polymorphism, with at least 50% of species consisting of both thelytokous (infected) and arrhenotokous (uninfected) strains 26 . Second, thelytokous colony can arise as spontaneous mutations, as has occurred in, Daphnia water fleas 34 and monogonont rotifers 35 . However, it is rare in the Figure 5. (a) Age-specific host-feeding rate (k x ), net age-specific host-feeding rate (q x ), and cumulative hostfeeding rate (C x ) of D. wani sp. nov. (b) Age-specific parasitism rate (k x ), net age-specific parasitism rate (q x ), and cumulative parasitism rate (C x ) of D. wani sp. nov. (c) Age-specific host-stinging rate (k x ), net age-specific host-stinging rate (q x ), and cumulative host -stinging rate (C x ) of D. wani sp. nov. (d) Age-specific total hostkilling rate (u x ), net age-specific total host-killing rate (w x ), and cumulative total host-killing rate (Z x ) of D. wani sp. nov. Hymenoptera. Finally, hybridization between two sexually reproducing species can induce transformation of the sexual reproduction into thelytokous parthenogenesis, however, this phenomenon appears to happen frequently among vertebrates. In addition, the mechanisms of thelytokous parthenogenesis in some species are not clear (like thelytokous strain of Venturia canescens 36 and Aptinothrips rufu 26 ). Given that the laboratory-reared colony of D. wani sp. nov. about 55 generations did not produce males under laboratory conditions and no arrhenotokous strain of D. wani sp. nov. was found in the field. These results showed that D. wani sp. nov. was a thelytokous or female-producing parthenogenetic species rather than a cyclical parthenogenetic organism. Although Rickettsia was detected in four individuals among ten individuals selected from the original colony (210 individuals), females only produced daughters from F1 to F5 generations under antibiotic treatment (20 mg/ml tetracycline) (unpublished data). The cause of female-producing parthenogenesis of D.wani sp.nov. need to study further in the future.
Diglyphus wani sp. nov. is indicated to be a parasitoid with three types of host-killing behaviour. All host-killing parasitoids show host-feeding behaviour. More than 140 species from 17 families of Hymenoptera feed on their hosts after adult eclosion 37 . The total number of host-feeding parasitoid species has been estimated at almost 100,000, or about one-third of all parasitoid species 38 . Recently, additional host-feeding parasitoid species, especially eulophids, that attack leafminers have observed to exhibit three types of host-killing behaviour 13,[19][20][21][22][23][24] . However, the host-killing rates vary in different parasitoids, such as D. isaea, N. formosa, Hemiptarsenus varicornis, and P. soemius 13,18,23,24 . These species have been widely studied and all are common parasitoid wasps against leafminers. The proportions of parasitism, host feeding and host stinging among the host-killing events of D. wani sp. nov. were 30.7%, 66.5% and 2.8%, respectively. For the ectoparasitoid H. varicornis, the proportions of the three behaviour were 25.7%, 58.2% and 16.1%, respectively, whereas for the endoparasitoid N. formosa the proportions were 44.4%, 31.1% and 24.6%, respectively 13,18,23,24 . These results may reflect a trade-off strategy, which is determined by the decision whether to invest in current or future reproduction of insect parasitoids 39 .
In addition, all host-feeding parasitoids studied to date are synovigenic 18,19,21,24,40 . The female wasps must feed on host tissues and host haemolymph to promote continuous oogenesis and egg maturation because most adult parasitoids cannot generate lipids de novo 41,42 . As a host feeder, D. wani sp. nov. is also strongly synovigenic. Furthermore, such parasitoids feed on hosts to increase their longevity 13,19,24,43 , as does D. wani sp. nov. The present results and previous results show that the host-feeding behaviour of these parasitoids have significant strongly positive linear relationships with longevity and fecundity 13,18,19,24 . However, the values of single host-feeding behaviour in different parasitoids were diverse. For example, when a female D. wani sp. nov. generated a host-feeding behaviour, the fecundity and longevity increased by 0.5 eggs and 0.1 d, respectively. However, Cheng et al. 13 found that when a female H. varicornis fed on a host L. trifolii, it increased the fecundity by 0.2 eggs and the longevity by 0.1 d. Both D. isaea and N. formosa that fed on a host L. sativae increased their fecundity levels by 2.6 and 1.4 eggs, respectively, and their longevities by 0.4 d and 0.2 d, respectively 18 . All of these species are widely used to control leafminers. Variation in the contributions of host-feeding behaviour among these species may be dependent on differences in the nutrient accumulation and utilisation capacities of diverse parasitic wasps 19,44,45 . Compared with the results of Zhang et al. 19 , in which D. isaea reared on L. sativae at 25 °C, the values for parasitism, host stinging, host mortality, and longevity of over the whole lifespan were lower. However, the daily total host mortality (18.5 hosts d −1 ) and the average daily fecundity (6.3 eggs d −1 ) of D. wani sp. nov. were greater than those of D. isaea (12.6 hosts d −1 and 2.6 eggs d −1 , respectively) 18 . In addition, D. wani sp. nov. showed a higher daily reproductive ability than D. begini or D. isaea, whereas the number of hosts parasitised per day for D. wani sp. nov. was lower than that of D. intermedius 18,46,47 . When the overall number of host-feeding events of D. wani sp. nov. (94.4 hosts female −1 ) was higher compared with those of other parasitoids that attack leafminers, such as D. isaea and N. formosa cultured from L. sativae with means of 56.6 and 43.7 hosts female −1 , respectively 18 . In contrast, the host-stinging to total mortality ratio of D. wani sp. nov. was less than those of other parasitoids. For instance, the host stinging-to-total mortality of D. isaea and P. soemius attained 39.9% and 52.0%, respectively 19,23 . Patel et al. 47 suggested that the proportions of different host-killing events depended on the density of leafminer larvae on individual leaflets.
Host-killing parasitoids may show greater biocontrol potential than parasitoids capable of only reproductive parasitisation owing to their strong destructive non-reproductive host-killing capacities 13,18,20,24,37,48 . For instance, in the present study, the non-reproductive rate of D. wani sp. nov. was 69.3%. Given the non-reproductive behaviour of host-killing parasitoids, the assessments of their biocontrol abilities differ from other parasitoid wasps that only kill hosts by parasitism. Thus, numerous attempts have been made to determine the biocontrol potential of host-feeding parasitoids that attacked the same hosts. In addition, to assess the population growth potential and the efficacies of the biocontrol potential, we determined an age-specific survival rate and finite host-killing rate. For the parasitoid wasp D. wani sp. nov., the intrinsic rate of increase, net reproductive rate, and finite host-killing rate were high, whereas the mean generation time was short. The present study illustrated that the short developmental duration, thelytokous reproduction and three host-killing behaviour render D. wani sp. nov. to a good natural enemy for the biological control of leafminers.
Although D. isaea is considered to show good biocontrol potential and broad application prospects, the species is easily prone to sex ratio imbalances in artificially generated rearing processes that affect its application 7 . Thelytokous wasps show a higher rate of population increase because individuals do not mate and waste eggs 26,27 . Therefore, D. wani sp. nov. might be a favourable biocontrol agent against agromyzid leafminers.
Temperature is a factor that commonly influences parasitoid life-history traits and biocontrol potential. Despite the distribution of Diglyphus in the Holarctic region and its adaptation to a cool climate, some parasitoid species in the genus may be used in warm environments 7 . In the field, parasitoids can simultaneously rely on hosts and non-host nutrient sources, such as floral tissues, hemipteran honeydew and pollen, and various artificial diets have been used in mass rearing 43 . To provide insights into the development of integrated control strategies for leafminer pests in field-and greenhouse-grown crops, the temperature-related adaptability of thelytokous parasitoids and the impact of adult diet on parasitoid reproduction require further study.

Methods
Insect cultures. The parasitoids (210 individuals) used in this study were originally collected from leafminer C. horticola infesting leaves of P. sativum in Xining, Qinghai, China (36°43′23.52″N, 101°45′1.51″E) in July 2015. This colony from one line was reared on Phaseolus vulgaris leaves infested with L. sativae larvae and supplemented with 20% honey water in gauze cages over 15 generations. The colony of L. sativae was obtained from P. vulgaris in Langfang, Hebei, China (39°35′14.77″N, 116°47′37.27″E) and cultured for 20 generations on P. vulgaris supplemented with 20% honey water. The leafminers and parasitoid wasps were maintained in climate chambers set to 25 ± 1 °C, relative humidity of 30 ± 5%, and a photoperiod of 14:10 h (light:dark). All tests were performed in climate chambers at the State Key Laboratory for Biology of Plant Diseases and Insect Pests, Institute of Plant Protection, Chinese Academy of Agricultural Sciences (Beijing, China).
The morphological terminology followed that used by Gibson 49 . Abbreviations used are: F1-F2, first through second flagellomeres; SMV and MV, submarginal and marginal, veins, respectively; OOL, minimum distance between an eye margin and the adjacent posterior ocellus; and POL, minimum distance between the posterior ocelli.
Measurements in millimetres (mm) of body and forewing lengths were recorded using a stereomicroscope (Olympus Corporation, VHX-2000, Tokyo, Japan). For all other dimensions, relative measurements were used. Photographs of the species were taken using an Olympus CX31 microscope with a Helicon Focus system. Pictures were taken of critical-point-dried specimens (30 min after emergence) using a scanning electron microscope (Hitachi Corporation, SU8010, Tokyo, Japan) at an acceleration voltage of 10.0 kV.

Molecular diagnosis.
Individual wasps from the laboratory colony were randomly chosen and killed by freezing at −80 °C for 10 min. The each body of an individual was used for DNA extraction following the method described by De Barro and Driver 50 .
After a positive PCR amplification, PCR products were purified and sequenced by Sangon, Shanghai, China. Forward and reverse strands were first aligned using MEGA 7.0 53 with default parameters. The sequences were aligned in the NCBI BLAST database. The other sequences for other Diglyphus species (such as D. isaea, D. pachyneurus, D. pulchripes, D. minoeus and D. crassinervus), which show similar morphological features, were downloaded from NCBI for phylogenetic, sequence divergence and haplotype analyses using MEGA 7.0 and DNA sp 5.10.01 54 . Phylogenetic tree reconstruction was implemented using the neighbour-joining approach based on the uncorrected "p" distances model in MEGA 7.0. A bootstrap analysis with 1,000 replicates was performed to assess branch support.
Life history and biocontrol potential. Immature development. To obtain data on immature development, kidney bean leaves infested with late 2nd and early 3rd instar L. sativae larvae (30-35 leaf −1 ) were cut from the plants. To maintain the leaves fresh, 10 ml water-agar (1%) was trickled into a petri dish. After refrigeration in the lab, the leaves were individually and immediately placed on the agar gel surfaces in the petri dishes and covered with cling film pricked with a dissecting needle to allow air circulation. Then, 5-10 parasitoid wasps (3 d after emergence) obtained from the laboratory colony were placed in each petri dish with 20% honey water to collect parasitoid eggs. After 3 h, the parasitoid wasps were removed and petri dishes with both parasitised and unparasitised leafminers were maintained in a climate chamber set at 25 ± 1 °C, relative humidity of 50 ± 5%, and a photoperiod of 14:10 h (light:dark). In total, 35 eggs were used at the beginning of the immature development study. Eggs were individually recorded and checked at different stages using a stereomicroscope every 12 h until adult emergence. The prepupae were placed individually in vials, which were tightly sealed with cotton to maintain moisture, for rearing in the laboratory and the immature development of D. wani sp. nov. was recorded every 6 h.
Adults life history: longevity and three host-killing behaviour. Newly emerged parasitoid wasps (08:00-10:00) of uniform body length were randomly assigned to the control (starvation treatment, n = 30) and host (n = 32) treatments under the above mentioned conditions in a climate chamber. One adult was added to each petri dish. In the control, the wasps were only allowed to feed on water, and the longevity was recorded and checked every 12 h (08:00 and 20:00). In the host treatment, one female was introduced to a petri dish containing one leaf infested with 30-35 late 2nd and early 3rd instar larvae of L. sativae, and allowed to implement host-killing behaviour for 24 h. Each petri dish was renewed daily until the wasp died. Meanwhile, the treated petri dish was placed in the same climate chamber as mentioned above. After 48 h, eggs or larvae of parasitoid wasps were observed under a stereomicroscope, and the three types of host-killing behaviour (host feeding, parasitism, and host stinging) were identified using the method of Cheng et al. 13 . The longevity, host feeding, parasitism, host stinging, and host mortality were recorded every 24 h.
Ovigeny index. To describe the relationship between ovigeny and life history, Jervis et al. 22 devised a quantitative "Ovigeny index" that was calculated as the proportion of the maximum potential lifetime complement that is mature upon female emergence. In synovigenic species the index value is less than 1, whereas in pro-ovigenic species the value is 1 22 .
From among the newly emerged parasitoid wasps (08:00-8:30) that were reared on late 2nd and early 3rd instar larvae of L. sativae in an artificial climate chamber, 32 individuals were randomly selected and frozen for 2 min at −80 °C. Subsequently, the wasps were dissected on a microscope slide with PBS buffer (pH = 7.2), and the initial eggs were counted under a stereomicroscope. The real lifetime fecundity data were obtained from the above section, and the initial egg loads were used to calculate the ovigeny index value.
Gains in longevity and fecundity per host-feeding event. Based on previous methods, we estimated the number of mature eggs and the number of days added to the lifespan per host-feeding event for a given female 13,18,19,37 . To obtain the greatest number of mature eggs, we compared different starvation treatments (6, 12, 24, and 48 h; n = 5 for each) for females only allowed distilled water. After the 12 h starvation treatment, 32 females were dissected and the number of mature eggs recorded because this group had the greatest number of mature eggs in a preliminary experiment. The longevity of females in the control was determined as described in the preceding section.
Life table. According to methods of the age-stage, two-sex life table [54][55][56] , the life history raw data were analysed using the TWOSEX-MSChart software 57 . Life table parameters, such as intrinsic rate of increase (r), finite rate of increase (λ), net reproductive rate (R 0 ) and mean generation time (T), were calculated in accordance with the previous studies.
Finite host-killing rate. To analyse the host-feeding, parasitism, host-stinging, and total host-killing rates, we followed the methods of Wang 58 using the CONSUME-MSChart software 59 with slight modifications.
The age-specific consumption (host feeding/parasitism/host stinging) rate (k x ) represents the number of hosts killed by the parasitoids of individuals at age x and was calculated according to the following equation: where c xj is the age-stage-specific host-killing (host feeding/parasitism/host stinging) rate of individuals at age x and life stage j, whereas s xj is the age-stage-specific survival rate of individuals at age x and stage j. When the age-specific survival rate (l x ), which is the number of individuals that survive to age x, was taken into consideration, the net age-specific host-killing (host feeding/parasitism/host stinging) rate (q x ) was calculated as follows: SCIENtIfIC REPORts | (2018) 8:3222 | DOI:10.1038/s41598-018-20972-3 The cumulative host-killing (host feeding/parasitism/host stinging) rate (C x ) is defined as the number of hosts killed by a parasitoid from birth to age x and it was calculated as follows: The stable host-killing (host feeding/parasitism/host stinging) rate (ψ) represents the proportion of individuals belonging to age x and stage j in a stable age-stage distribution (ɑ xj ): To assess the host-killing (host feeding/parasitism/host stinging) potential, the finite host-killing (host feeding/parasitism/host stinging) rate (ω) was calculated as follows: The total host-killing rate (u x ) represents the total number of host feeding, parasitism and host stinging events, as follows: The age-stage-specific total host-killing rate of individuals at age x and stage j (p xj ) was the summation of c xj of host feeding, c xj of parasitism and c xj of host stinging, respectively: xj xj host feeding x j parasitism x j host stinging ( ) ( ) ( ) The net age-stage-specific total host-killing rate (w x ) was calculated as follows: The net total host-killing rate (Z 0 ) represents the total number of three host-killing events during one generation of the wasps. It was calculated as: The cumulative total host-killing rate (Z x ) represents the total number of hosts killed by a parasitoid wasp from birth to age x. It was calculated as: x i x i i 0 To compare the overall host-killing potential, the finite total host-killing rate (θ) and the stable total host-killing rate (ϑ) were as calculated as follows: The transformation rate from host population to parasitoid offspring (Q P ) is the mean number of hosts a parasitoid needs to kill to produce one offspring. It is calculated as: p 0 0 Data analyses. Data on the three types of host-killing behaviour were analysed with a one-way ANOVA using Fisher's model under the 0.05 level of significance. Linear regression was used to assess the relationships between host feeding and longevity, parasitism, host stinging and host mortality.