Host plant associations in Western Palaearctic Longitarsus flea beetles (Chrysomelidae, Galerucinae, Alticini): a preliminary phylogenetic assessment

Abstract Longitarsus Latreille (Chrysomelidae, Galerucinae, Alticini) is a very large genus of phytophagous insects, with more than 700 species distributed in all zoogeographical regions. Patterns of host use have been a central topic in phytophagous insect research. In this study a first assessment is provided to test the hypothesis that host-plant association is phylogenetically conserved in Western Palaearctic Longitarsus species. Maximum Likelihood and Bayesian Inference methods were used to infer a phylogeny based on DNA sequence data from two mitochondrial genes from 52 Longitarsus species from the Western Palaearctic. In agreement with the host phylogenetic conservatism hypothesis, a strict association between most of the recovered clades and specific plant families was found, except for species associated with Boraginaceae. Low phylogenetic resolution at deep nodes limited the evaluation of whether closely related Longitarsus clades are associated with the same plant family or to closely related plant families.


Introduction
Longitarsus Latreille, 1829 is a mega-diverse genus of phytophagous insects and the most speciose among flea beetles (Chrysomelidae, Galerucinae, Alticini) with more than 700 known species. It is widespread through all zoogeographical regions (Furth 2007, Biondi and D'Alessandro 2010, Döberl 2010, Prathapan and Viraktamath 2011, Reid 2017. Longitarsus is also ecologically diversified with specialized feeders, monophagous or oligophagous (Schoonhoven et al. 2005), on different angiosperm families. Larvae feed mostly on roots, and adults target leaves of their host plants (Dobler et al. 2000, Furth 1980). The monophyly of Longitarsus is accepted based on molecular evidence (Gómez- Rodríguez et al. 2015, Nie et al. 2018. Members of the genus are recognized mainly by the length of first metatarsomere, exceeding half-length of hind tibia, along with confuse elytral punctuation and absence of dorsal pubescence (Biondi and D'Alessandro 2012).
Relationships among Chrysomelidae and their host plants has been investigated from a biochemical, behavioural, and phylogenetic point of view, and at various taxonomic levels, often with the aim of understanding the biology of actual or potential pests (Jolivet and Hawkeswood 1995, Becerra and Vernable 1999, Dobler et al. 2000, Clark et al. 2004, Fernandez and Hilker 2007, Kergoat et al. 2007, Reid 2017. Understanding the mechanisms which drive observed host-use patterns has been a central topic in phytophagous insect research. Among the non-mutually exclusive hypotheses proposed (Gripenberg et al. 2010, Balagawi et al. 2013, Charlery de la Masselière et al. 2017, Kergoat et al. 2017, Lima Bergamini et al. 2017, Jones et al. 2019, the phylogenetic conservatism states that the phylogeny of host plants strongly constrains host affiliations. The phylogenetic conservatism hypothesis is widely demonstrated, even though it can be masked to varying degrees of convergent evolution in both plant and herbivore traits, and/or by phenotypic plasticity of both plants and herbivores; in addition, evolutionary processes that have generated phylogenetic conservatism patterns often remain unclear (Fernandez and Hilker 2007, Kergoat et al. 2017, Lima Bergamini et al. 2017. A comprehensive phylogenetic assessment of the insect-host plant relationship in the genus Longitarsus is still lacking, as well as comprehensive systematic studies on the genus. In this paper, we conduct a molecular phylogenetic analysis on 52 Western Palaearctic Longitarsus species to map host plant data in order to assess whether the pattern observed is consistent with the phylogenetic conservatism hypothesis. If it is consistent, we would expect to find a) closely related Longitarsus species (single clades) to be associated with a specific plant family; b) Longitarsus clades associated with a specific plant family to form a single lineage.
To assess the association between phylogeny and patterns of host use, we used information on host plants for each species of Longitarsus from field observations over many years, and from a critical analysis of information reported in the literature (Furth 1980, Doguet 1994, Biondi 1995a, b, 1996, Bienkovski 2004, Konstantinov 2005, Aslan and Gok 2006, Gruev and Tomov 2007. We ignored isolated observations or reports of one or a few individual herbivores seen only one time on a host, as well as cases of postseason host refugium (Furth 1980). According to Biondi (1996), species feeding on one or two phylogenetically very closely related plant genera were considered as monophagous (MON); species feeding on one or two phylogenetically very closely related plant families were considered as oligophagous (OLI); species feeding on more plant species not phylogenetically closely related were considered as polyphagous (POL).
Phylogenetic relationships among plant families were discussed according to The Angiosperm Phylogeny Group 2016 (hereafter The APG 2016). Molecular analyses included DNA sequences from the mitochondrial genes cytochrome oxidase subunit I (cox1) and 16S rDNA (16S). These sequences were obtained from 20 alcohol-preserved specimens, each representing a distinct species, and retrieved from GenBank for 32 species (see Table 1 for information on specimens). We selected these two genes because they are the most represented in GenBank for Longitarsus species and because their phylogenetic utility at the genus level have been demonstrated by Nie et al. (2018). Details on sample data, along with GenBank accession numbers are provided in Table 1.

Phylogenetic analyses
Multiple sequence alignment was performed with MAFFT v.7 (Katoh and Standley 2013) using the E-INS-i iterative refinement algorithm. Phylogenetic analyses were conducted using Maximum Likelihood (ML) and Bayesian Inference (BI) approaches on the concatenated alignment of cox1 and 16S sequences, using as outgroup Batophila aerata (Marsham, 1802). This outgroup belongs to Altica group, and has been demonstrated as a closely related lineage to Longitarsus group based on molecular evidence by Nie et al. (2018). Maximum Likelihood analyses were performed in raxmlGUI 1.5b2 (Silvestro and Michalak 2012), a graphical front-end for RAxML 8. . Two independent Markov chain Monte Carlo (MCMC) analyses with 6 chains each were run in parallel for 50 million generations, sampling every 5000 generations. The first 25% were discarded as burn-in. MCMC chains convergence was verified by average standard deviation of split frequencies values below 0.0035 and confirmed in Tracer 1.7 (Rambaut et al. 2018). A majority rule consensus tree with posterior probability of each node was calculated with the sumt command in MrBayes.
Phylogenetic results were summarised using the ML tree and reporting for each node both BS and BPP from the Bayesian analysis. Nodes with bootstrap values (BS) between 70 and 90% and Bayesian posterior probability (BPP) between 0.95 and 0.98 were considered as supported, and those with BS greater than 90% and BPP greater than 0.98 as highly supported.
Among the 52 Longitarsus species used in our molecular phylogenetic analyses (Table 1), 49 are monophagous or oligophagous; L. atricillus and L. aeneicollis are polyphagous, while no information is available about host plants of L. bedelii. Such a high number of specialized species allows a straightforward assessment of phylogenetic conservatism in host plant use, given a phylogenetic tree of the Longitarsus species and the relationships among host plant families.
Phylogenetic analyses based on ML and BI methods gave consistent results and identified the same supported clades (Fig. 2). Most of these clades have been previously recognized as distinct species-groups based on morphology (external morphology, aedeagus and/or spermatheca): (1) clade H includes species of the tabidus group sensu Leonardi (1972); (2) species of the pratensis group sensu Leonardi and Doguet (1990) are clustered in the clade B2; (3) L. pulmonariae, L. exsoletus, and L. cerinthes (clade O) were already known to be closely related (Leonardi 1972); (4) L. anchusae and L. saulicus (within clade D) belong to the anchusae group sensu Biondi (1995a).
In agreement with the host phylogenetic conservatism hypothesis, we recovered a strict association between most of the recovered Longitarsus clades and specific plant families, except for species associated with Boraginaceae (Fig. 2). Longitarsus species associated with Plantaginaceae form two closely related and supported clades (clades B1 and B2) and those associated with Scrophulariaceae form a distinct, well supported  clade (clade H). Species associated with Lamiaceae are included in four clades (C, E, F, G) within a major lineage of Western Palaearctic Longitarsus (clade A). Relationships between clades within this lineage are poorly resolved; additional molecular data will be required to clarify relationships within clade A and to assess whether clades associated with Lamiaceae are truly polyphyletic or instead if increased phylogenetic resolution will allow recovering them as a monophyletic assemblage. All nine species associated with Asteraceae are grouped in the clade I; this clade also includes L. brunneus feeding on Ranunculaceae, and might represent an instance of host-shift towards an unrelated plant family (The APG 2016).
The two species associated with Convolvulaceae, L. nigrocillus, and L. pellucidus, cluster together with high support in clade N. The polyphagous species, L. atricillus and L. aeneicollis, plus L. bedelii, for which no host plants are known, form the highly supported clade M. On the other hand, clades grouping species associated with Boraginaceae are distant in the phylogenetic tree: clade D with six species and the isolated branch of L. curtus are included in clade A, whereas clade O with four species occupies a basal position of the phylogenetic tree together with other four species with poorly resolved phylogenetic position (L. fallax, L. quadriguttatus, L. lateripunctatus, L. linnaei).
Overall the phylogenetic tree of Western Palaearctic Longitarsus shows a decrease of statistical support from the tips to the root, with highly supported terminal clades and weakly supported basal relationships (Fig. 2). Therefore, the inference of phylogenetic conservatism in host-plant association is only robust at the lower hierarchical level. While the strong association between closely related Longitarsus species to the same plant family is clear, it is difficult to identify an association between closely related Longitarsus clades and closely related plant families. Most of the clades of Longitarsus (clades B-H) belonging to the same main lineage (clade A) are associated with plant families (Plantaginaceae, Scrophulariaceae, and Lamiaceae) that belong to the same order Lamiales (The APG 2016).
Ideally, for a conclusive assessment of phylogenetic conservatisms in host-plant association between closely related Longitarsus clades and closely related plant families we would require well-resolved phylogenies for both insects and plants at all taxonomic levels. While limited uncertainty exists for interrelationships between plant families (e.g., over the exact placement of Boraginaceae family (The APG 2016)), our molecular phylogenetic analysis only provides a first appraisal of relationships between Western Palaearctic Longitarsus. To assess whether the pattern of basal polytomy we observed (Fig. 2) is a solid polytomy or is due to a lack of data (see Mendes et al. 2016) further studies based on increased taxon and marker sampling are required. Improving field research is also crucial because the ecology and feeding biology of several species are still unknown. True host affiliation can be difficult to detect, due to the different interaction that phytophagous insects can have with plant species (Furth 1980, Schoonhoven et al. 2005, even though the use of molecular techniques, such as DNA barcoding of gut contents, can help to detect real trophic interactions (Jurado-Rivera et al. 2009).

Conclusions
In this study, we provided first evidence that host-use patterns are phylogenetically constrained in Western Palaearctic Longitarsus. Despite the limited set of species analysed, we found a clear association between closely related Longitarsus species and specific plant families (Plantaginaceae, Asteraceae, Scrophulariaceae, and Convolvulaceae). However, relationships between clades of species were poorly resolved thus preventing the assessment of whether all Longitarsus clades associated with a specific plant family, or to related plant families, represent a single lineage. Such a relationship is unlikely for those Longitarsus species feeding on Boraginaceae which were resolved in unrelated clades. A better understanding of the phylogenetic relationships between Longitarsus species associated with Boraginaceae is of great interest also from a biogeographical point of view. In fact, two groups of species feeding on Boraginaceae and sharing a number of striking morphological features show a disjunct Mediterranean-South African distribution (Biondi 1995a, Biondi and D'Alessandro 2008, 2017. Molecular studies with additional markers are in progress on an extended set of species to further our understanding of hostplant relationships in Longitarsus.