The analysis of mitochondrial and nuclear genetic markers indicate that Asecodes lucens is not one generalist species but at least three species with a more narrow host use. Two parasitoid species seem to be specialist on their respective host, Galerucella sagittariae and G. lineola. The three remaining Galerucella species are seemingly attacked by the same parasitoid species, even though these parasitoids also show some population differentiation. There were tendencies that the parasitoids attacking G. tenella, which feeds on F. ulmaria, had diverged from the parasitoids attacking G. calmariensis and G. pusilla, which both feed on L. salicaria. The mitochondrial dating furthermore show that speciation in parasitoids are sequential events following speciation in their host, and a recent but perhaps not completed split between parasitoids attacking G. tenella and G. calmariensis/pusilla, would follow this pattern.
The derived phylogeny of Galerucella matches the previous analysis by Borghuis et al.  where the split between what is considered two subgenera, Galerucella and Neogalerucella, is most ancient. In our data set, the subgenus Galerucella only includes G. sagittariae whereas the other species belong to the subgenus Neogalerucella. Borghuis et al. based their analysis only on mitochondrial gene fragments (CO1 and NADH-2), and found strong support for the monophyly of G. tenella and G. lineola but could not resolve G. pusilla and G. calmariensis as reciprocally monophyletic, indicating either a recent divergence or mitochondrial introgression. Our study also included two nuclear DNA fragments (the D2 region in 28S and ITS2), which provides a test of the alternative explanations. Although our nuclear DNA datasets were small, divergence between G. pusilla and G. calmariensis was strongly supported. These two taxa both use Lythrum salicaria (Lythraceae) as host plant, and both ecological and morphological information suggest that the two species are reproductively isolated. The male copulatory organs are distinctly different, the body size differs and both larval and adult colour differs [58, 59]. In these characters, there is quite small overlap. In addition, Nokkala and Nokkala  found karyotypic differences which further support species status. In contrast to the conclusion drawn by Borghuis et al., based on the results from mitochondrial gene fragments, the lack of monophyly in the mitochondrial genes for G. pusilla and G. calmarensis might rather indicate some recent “phenomenon” such as introgression, Wolbachia infestation etc. However, a larger sample from each population is necessary to establish if lineage sorting is indeed complete for the nuclear markers, which theoretically should sort slower than mitochondrial counterparts. Due to the maternal inheritage of the mitochondrion, effective population size is lower (1/4th that of nuclear genes) and lineage sorting is therefore faster than for nuclear genes.
The Bayesian species delimitation analysis for the A. lucens group was performed on the same set of nuclear and mitochondrial genes as the analysis for Galerucella, with the addition of the nuclear gene PGD, and provided strong evidence for population differentiation. It is possible to conclude that A. lucens should be split at least into three species but additional data may strengthen the indication of further splits; the species delimitation analysis suggested 3–5 species. The molecular data have been confirmed by morphological studies that found differences in wing patterns among at least 3 taxa . Comparisons with type specimens of available (synonymized) names suggested the identity of two taxa, A. lucens parasitizing G. sagittariae and A. parviclava (Thomson) parasitizing G. tenella, G. calmariensis and G. pusilla. A third species, parasitizing G. lineola, represents a new species named A. lineophagum Hansson & Hambäck . There were no morphological characters supporting a further subdivision of A. parviclava.
The BPP method distinguishes populations as different species if the per generation migration rate Nm << 1 , and the interpretation is therefore unambiguous in sympatry but distance-decay patterns in allopatry may alone contribute to partial genetic isolation. Allopatry could explain why the analysis with all data included identified population differentiation between parasitoids collected on G. tenella versus G. pusilla/calmariensis. In the southern area, populations of G. tenella and G. pusilla/calmariensis are often found in slightly different habitats and were typically collected in different localities. In the northern area, where population differentiation was seemingly weaker, G. tenella and G. calmariensis co-occur on Baltic shore lines and were collected in the same localities. It is possible that the southern population of A. parviclava is in an early stage of speciation, but the current data are insufficient to confirm this suspicion.
The possibility of geographic variation in population differentiation of parasitoids is very interesting considering the previously documented parasitoid mediated indirect interactions between G. tenella and G. calmariensis in northern localities. Hambäck et al.  found that parasitism rates on G. tenella were higher at sites where this species co-occur with G. calmariensis compared with sites without G. calmariensis. The differences in parasitism rates translated into differences in the strength of interactions between the herbivore and its host plant . In sites with L. salicaria and G. calmariensis, the attack rates on F. ulmaria by G. tenella were lower and the seed set were higher compared to sites without L. salicaria. Other studies show that this apparent competition between the Galerucella species may have evolutionary consequences for F. ulmaria. Galerucella tenella has a very strong impact on plant fitness and studies on differently aged populations suggest that G. tenella causes a shift in the population towards plants with a lower height and with higher concentrations of potential antiherbivore compounds . In the field, the lower quality of F. ulmaria as food for the larvae of G. tenella caused the beetle to expand its diet towards other Rosaceae plants . Preliminary data suggest that these evolutionary changes in F. ulmaria only occur on islands without L. salicaria (Ericson & Stenberg, unpublished data). The hypothesis for these effects was that the parasitoids use both G. tenella and G. calmariensis, even though the behavioural mechanisms are not fully understood . It was therefore important to know whether parasitoids collected from the two host were indeed the same population and this seems to be the case. The potential for a larger population differentiation in southern localities suggest that a similar apparent competition is not likely in these localities, and the parasitism rates in these localities are also typically much lower (<10% vs >70%).
Speciation in the parasitoids for this system seem to follow the identity of the beetle host rather than the identity of the beetle’s host plant. This finding suggests that host finding or recognition cues originate from the larval host rather than from the host plant, such as beetle produced pheromones or beetle specific plant cues. Two contrasts in particular supports this view. First, G. sagittariae feed on multiple host plant species that are not closely related, but we nevertheless found no population differentiation among parasitoids hatching from larvae collected on different plants. Second, G. sagittariae and G. tenella are known to feed on the same host plant, but we find that their respective parasitoid belong to different taxa. It seems less likely that plant produced volatiles provide sufficient information for parasitoids to both differentiate between G. sagittariae and G. tenella, and at the same time to locate G. sagittariae on its different host plants. Beetle produced compounds therefore seem more likely. Previous studies show that at least two Galerucella species (G. pusilla and G. calmariensis) produce aggregation pheromones , but these compounds are produced by adult males and seem less likely to provide any information on the whereabouts of Galerucella larvae. To further understand the speciation process in Asecodes parasitoids, we are currently working to identify the compounds involved during the search process.
Speciation in parasitoids may follow different pathways according to published data, and the causes underlying this variability is not well understood. There are cases with host-association differentiation and seemingly tight cospeciation [66, 67], similar to the one suggested in this study for the Galerucella-Asecodes system. In other cases, speciation in parasitoids is less well connected to the phylogeny of their host [68, 69]. The current data both in this and most of the published cospeciation examples cannot differentiate whether this is ecological speciation where hosts or parasitoid species are not geographically isolated or cladogenesis where host and parasitoids are isolated in pairs . Differences in the speciation pattern may arise because the parasitism process involves several steps that create barriers to parasitoids when their host switches diet. Diet switching and speciation in herbivorous insects is often connected to enemy free space where the host is more difficult to either find or exploit on an alternate plant [71–73]. Different plants may produce quite different volatile profiles upon damage and this may cause problems for parasitoid females to either find their host or even to identify the larvae as such. There are cases where a parasitoid species has geographic variability in the type of cues used during host search , but this area of research is poorly exploited. Besides affecting host search cues, plant quality is also known to affect herbivore immunocompetence and switching to an alternate plant may affect resistance to parasitoid attack. The Galerucella-Asecodes system seems ideal to study these processes in progress, both because host race formation is common and well documented among Galerucella beetles [75–77] and because interaction strengths between host and parasitoid show quite large geographic differences. Further studies on species interactions in this system however necessitates a better understanding also on the small scale population differentiation, to identify strengths of direct and indirect species interactions and their ecological and evolutionary consequences.