Parasitoid assemblage size and host ranges in a parasitoid (Hymenoptera)–agromyzid (Diptera) system from central Argentina

Parasitoid assemblage size and host ranges in a parasitoid (Hymenoptera)-Agromyzidae (Diptera) system from central Argentina A. Salvo* and G. Valladares Centro de Investigaciones Entomologicas de Cordoba, Universidad Nacional de Cordoba, Avda. Velez Sarsfield 299, CP 5000 Cordoba, Argentina Abstract A parasitoid community on Agromyzidae leafminers from Cordoba, Argentina was analyzed in terms of parasitoid assemblage size and host ranges of parasitoid species. Samples were taken during 1991-1995 at natural, urban and agricultural habitats. The system consisted of 69 parasitoid species and 51 leafminer species on 109 plant species. On average, 12 species parasitized each host, when only numerically well represented leafminer species were considered for analysis. Each parasitoid species exploited on average seven species and three genera of Agromyzidae leafminers. Both parasitoid assemblage size and parasitoid host range were notably high when compared with similar Systems from other regions. ———————————————————————————————————————

Introduction Leafminers are insects whose larvae feed between the epidermises of leaves, consuming foliar mesophyllum (Askew & Shaw, 1974). In the galleries thus excavated the feeding larvae find shelter against harsh environmental conditions but little protection against parasitoids, which are the most frequent natural enemies of the group. In fact, this guild of phytophagous insects supports the richest parasitoid communities, seemingly due to the physical characteristics of leaf mines (Hawkins, 1988). Parasitoid assemblages of phytophagous insects are usually studied in relation to a definite host species in a definite habitat. But knowledge of the entire community of parasitoids attacking all available host species within an ecological guild or a phylogenetically related group, on a regional scale over a variety of habitats, is essential to obtain reliable information about parasitoid host ranges, number of shared parasitoid species and host species, etc. (Askew & Shaw, 1986; Godfray, 1994). We aim to contribute to that knowledge by analyzing a parasitoid-leafminer-host plant system in Central Argentina. The fauna of leafminers in the study area is largely dominated by Agromyzidae (Diptera) (Valladares, 1982), a family which is well represented worldwide (Spencer, 1990). Knowledge about parasitoids associated with Agromyzidae in Argentina is limited (Diaz & Valladares, 1979; Neder de Roman & Arce de Hamity, 1985; Valladares et al., 1982; Salvo, 1996; Salvo & Valladares, 1993, 1996, 1997) despite the hosts being relatively well known (e.g. Valladares, 1982, 1984, 1991, 1996). This paper provides a description of the system, analyzing the size of parasitoid assemblages on Agromyzidae species and parasitoid host ranges. Materials and methods Mined leaves were sampled from 1991 to 1995 at 30 localities in Cordoba province, central Argentina, including natural, urban and agricultural habitats. Some localities were sampled monthly whereas others were visited occasionally. Mined leaves of each available host plant were cut and placed in plastic bags, in which they were carried to the laboratory and kept until flies and parasitoids emerged. Special ——————————————————— * Fax: 5451 332097

rearing methods were developed for leaves with high water content or those that rapidly dried once excised. In the former case, a box covered with gauze was used instead of a bag, or the mined portion of the leaf was cut out from the rest, to reduce the risk of the mine becoming infected with mould. To avoid excessive desiccation, stems with several leaves were collected or humid cotton wool pieces were added to the bags. Adults were stored in small glass tubes plugged with cotton wool and after emergence had ceased they were counted and identified. A reference collection of both Agromyzidae and parasitoids has been deposited in Catedra de Entomologia, Universidad Nacional de Cordoba. Average, maximum, minimum and standard error (X, max., mm., and SE) were calculated for the parasitoid assemblage size of agromyzid species and for host ranges of the parasitoid species. These figures were estimated separately for the total number of species and for those species with a sample size of over 100 agromyzid or ten parasitoid adults, in order to reduce errors attributable to the use of small samples. Parasitoid assemblage size was estimated for both species and genera of leafminers, whereas host ranges of parasitic species were assessed at both the specific and generic level of host identification. The influence of sample size on both parasitoid assemblage size and host range of parasitoids was described by using linear regressions. Although the overall relationships must eventually be asymptotic, some nonlinear models tested did not show a better adjustment of the data. Parasitoid assemblage size, host range and sample size were log transformed prior to analysis. The effect of host taxonomic isolation on parasitoid assemblage size was assessed by regressing the latter on the number of agromyzid species ii’ each genus occurring in the study area. The residuals from the regression of parasitoid assemblage size on agromyzid genus sample size were then regressed against agromyzid generic species richness to remove sample size effects. Possible differences in host ranges among taxonomic families of parasitoids were analyzed by ANCOVA, with specific and generic host ranges of all parasitoids or parasitoids with more than ten individuals as dependent variables and sample size as covariate, in order to incorporate its effect (Steel & Torrie, 1981). Results and Discussion The leafminer-host plant-parasitoid community Leaf miners Fifty-one leaf mining agromyzid species in ten genera were found in central Argentina during the study period. The most speciose genera were Liriomyza Mik 15 species), Calycomyza Hendel (ten species) and Haplopeodes Steyskal (nine species). Ophiomyia Braschnikov and Phytamyza Fallen were represented by three species each, while just one species was reared for Agromyza Fallen, Amauromyza Hendel, Chromatamyia Hardy, Japanagromyza Sasakawa and Phytoliriomyza Hendel. Host plants Agromyzid species mined leaves from 31 families, 55 genera and 109 species of plants in the study area. A preliminary analysis of the relative importance of leafminer and plant affiliation on the host ranges of some of the parasitoid species has previously been reported (Salvo & Valladares, 1997). Parasitoids A rich parasitoid community exploited agromyzid leafminers in this region. All the parasitoid species known from the Agromyzidae in Argentina were represented in the area, and many more were recorded for the first time from this study (Salvo & Valladares, in press; Van Achterberg & Salvo, 1997).

Out of 41,534 adult insects reared (flies and parasitoids), hymenopterous parasitoids comprised 40% percent, in three super families, five families and 69 species. Seventeen of the latter were identified to species level, and the rest were considered morphospecies, 37 at the genus level, two at the subfamily level and 14 at the family level. Eulophidae, Pteromalidae and Braconidae each accounted for almost a third of the parasitoid species, while Eucoilidae and Mymaridae had little representation. However, in terms of the abundance of parasitoid individuals, the Eulophidae was distinctly the best-represented family, followed by the Braconidae (table 1). Parasitoid assemblage size on agromyzid leafminers Parasitoid assemblage size was largely determined by sample size (P < 0.0001, R2 = 0.74) (fig.1), an effect noticed almost whenever this factor has been considered (Askew & Shaw, 1974; Hawkins & Lawton, 1987; Hawkins, 1988, 1990; Towner, 1992; God fray, 1994). To a lesser degree, sample size also influenced parasitoid assemblage size when genera instead of species of leafminers were considered (P=0.02, Model: R2= (0.50, log (y + 1) =0.672+0.189* lot,, (x + 1)). On average, each agromyzid species hosted 9.2 (SE = 0.78, max. = 23, mm. = 0, n = 51) parasitoid species in the study site. When only those host species with more than 100 reared adults were considered (fig. 2) the average rose to 12.3 (SE = 0.89, max. = 23, mm. = 1, n = 30); and reached to 19.3 (SE = 1.92, max. = 23, mm. = 6, n = 6) parasitoid species per leafminer species with 1000 or more adults. These figures seem remarkably high when compared to the average of two species in the parasitoid assemblages of leafminers from three insect orders in Costa Rica (Memmot et al., 1994). Similarly, each host species in a dipteran leafmining community in Great Britain supported just over three parasitoid species (Towner, 1992). Based on a literature survey, Hawkins 1990) recorded an average of six parasitoid species for Neotropical leafminer species. At a global scale, 11 parasitoid species were found on average for leafminer species, with sample size above 1000 individuals (Hawkins, 1994). Therefore, parasitoid communities of agromyzid flies in Central Argentina seem to be notably rich. Although the abovementioned studies included leafminers other than Agromyzidae, there is no obvious reason for this group to sustain a higher parasitoid assemblage size than leafminers in other insect orders or families. Differences in sampling methods and objectives of the studies (specifically designed to obtain information on this topic in the present paper) could explain the higher species richness of the parasitoid assemblages in central Argentina, in relation to those studied by Hawkins, but not in relation to the Costa Rican and the British systems. In the latter cases, sampling intensity might be a more satisfactory explanation since only 13 and 12 leafminer species, respectively, were represented by 100 or more individuals in those studies, compared to 28 leafminer species in the present study. Table 1. Representation of each parasitoid family as a percentage of total parasitoid species and total parasitoid individuals reared. ——————————————————————————————————————————————— Taxonomic family Individuals (%) Species (%) (n=16,543) (n = 69) ——————————————————————————————————————————————— Braconidae 26.8 27.5 Eucoilidae 5.7 7.3 Eulophidae 57.8 33.3 Mymaridae 1.2 1.5 Pteromalidae 8.5 30.4 ———————————————————————————————————————————————

Fig. 1. Relationship parasitoid assemblage size of agromyzid leafminer species and sample size. Model log (y + 1) = 0.33 +0.28 * log (x + 1)

Fig. 2. Frequency distribution of leafminer species (sample size ≥ 100 adults) in relation to the size of their parasitoid assemblage.

Fig. 3. Relationship between parasitoid assemblage size of agromyzid leafminer genera and number of species in the genus. Model P = 0.003, R2 = 0.67 log (y + 1) = 0.96 + 0.50 log (x + 1).

Fig. 4. Mean host range (±SE), at species (❏) and genus (■) level of identification, of parasitoid species depending on the sample size of the species considered in the analysis.Number within the bars indicate number of species included in each category.

Fig. 5. Relationship between species host ranges of parasitoid species and sample size. Model log (y + 1) = 0.087 * log (x + 1)

Fig. 6. Relationship between parasitoid taxonomic family and mean host range (species (❏) and genera (■)) adjusted for parasitoid sample size by ANCOVA. Data from parasitoids with sample size ≥ 10 individuals.

Host taxonomic isolation is usually included among the many potential explanatory factors for parasitoid assemblage size patterns (Hawkins & Lawton, 1987; Hawkins, 1988). In the present system, the richest agromyzid genera support the richest parasitoid assemblages (fig. 3), reflecting their higher apparency as a colonizable resource for parasitoids, and the consequent accumulation of parasitoid species over evolutionary and ecological time. This effect was maintained even after the effect of sample size was removed (P = 0.05 R2 = 0.39). Parasitoid host ranges Thirty out of 69 parasitoid species recorded in this study were apparently monophagous, being reared from only one agromyzid species, while 33 were restricted to one agromyzid genus. But only 19% (eight species) or 26% (ten species) of the parasitoids represented by ten or more individuals restricted their attack to one species or genus of host, respectively. On average, each parasitoid species attacked nearly seven species of agromyzid leafminers in the study region, and these values rose considerably when the analysis was restricted to parasitoids with larger sample sizes (fig. 4). It has been proposed that hosts in concealed situations, such as leafminers, should sustain a generalistdominated parasitoid fauna (Hawkins et al., 1992), which is fully supported by the present data. But even then, the figures reported above suggest a remarkable degree of polyphagy for the parasitoids of agromyzid leafminers in central Argentina. Monophagous species were much better represented (63% of species) in the system studied by Memmot et al. (1994) in Costa Rica, where an average host range of two leafminer species was recorded. This difference is even more noticeable given the much higher number of available host species in that study (90) than in the present one (51). Dipteran leafminer parasitoids in Great Britain also presented a much narrower average host range (three species) than the one discussed here (Towner, 1992). Again, comparisons must be made carefully given the differences in the sampling intensity of the different studies. Sample size greatly influenced the number of host species (P c 0.00~, R2 = 0.80) (fig. 5) and to a lesser degree host genera (P < 0.0001 R2 =0.26 Model: log (y + 1) = 0.318 + 0.139 * log (x + 1)) exploited by each parasitoid species. This renders the polyphagy of parasitoids in central Argentina even more noticeable, since a greater sampling effort would be expected to reveal even broader host ranges. Also, it must be noted that very few individuals mainly represent ‘monophagous’ species. This contradicts the expectancy of a supposedly higher efficiency of resource utilization by specialists. Their scarcity suggests they could actually be generalists, possibly feeding on various groups of leafminers (at suprafamily level), among which Agromyzidae could be non-preferred hosts. Specialization or generalism could be phylogenetically conserved (Godfray, 1994), and thus host ranges could be linked to parasitoid taxonomy. Host ranges (measured at both the specific and generic level) varied among parasitoid families. When an ANCOVA analysis was performed, the preliminary test for homogeneity of slopes was not significant for either the species (F = 1.56, P = 0.20) or the generic (F = 1.09, P = 0.36) host ranges. In the final model, sample size highly affected host ranges, both at generic and specific level (P