Aculeate Hymenoptera: Phylogeny and Classification
- 190 Downloads
The Hymenoptera are one of the largest orders of insects, comprising almost 160,000 described extant species with a true total of possibly over one million species. Most species in the larval stage are parasitoids of other insects [1365体育网站]. The order is divided into two subgroups. The Symphyta, or sawflies and horntails, are a paraphyletic assemblage of mainly plant feeders with caterpillar-like larvae and in adults a broad connection between the thorax and abdomen. The much more diverse Apocrita are a monophyletic group, with legless larvae and mainly parasitoids or predators of arthropods, the adults with a strong constriction between the first and second abdominal segments. This “wasp waist” confers great maneuverability on the functional abdomen, or metasoma. Within the Apocrita, there are again two major groupings, sometimes considered infraorders, but more usually regarded as informal ones. The larger group is the Parasitica (another paraphyletic assemblage), mainly parasitoids of other insects and with well-developed ovipositors for laying eggs in or on the host. The smaller is the biologically more diverse Aculeata, a monophyletic group in which the ovipositor has been modified as a stinger and has lost its function in egg laying.
365体育网站Although sociality has arisen separately in many insect groups, the order containing the great majority of social species is the Hymenoptera. Within this, the Aculeata is the group with almost all the social species. The only exceptions are a few sawflies with brood care by the females, which does not amount to eusociality. Consequently, attempts to understand the origins of sociality in Hymenoptera must be informed by knowledge of the probable evolutionary history (phylogeny) of the aculeates.
The Basis of Phylogenetic Analysis
In order for the branching patterns representing the relationships of lineages to reflect the actual evolutionary histories of those lineages, they must be based on the evolutionary principle of descent with modification. This requires the identification of characteristics reflecting derived modifications (apomorphies), as distinct from the ancestral conditions (plesiomorphies). Where different lineages share those derived modifications (synapomorphies), we infer that this is due to descent from a common ancestor. Nested groupings of this sort produce the familiar tree structures, or cladograms, reflecting the hypothesized evolutionary history of the organisms involved. Logically, the preferred trees are those that require the fewest assumptions of independent change (homoplasy) and the fewest changes of conditions (states) across the whole tree.
Such cladistic methods were first developed for phenotypic (morphological or behavioral) characters and utilize “maximum parsimony”; they are distinguished from phenetic methods which merely calculate measures of overall similarity or difference between groups without taking shared ancestry into account and thus do not aim to reflect the evolutionary pattern itself. It must be recognized, however, that the most parsimonious cladogram is merely the initial best estimate of the true evolutionary tree, since evolution is heavily influenced by random events and is thus not necessarily parsimonious in itself. Our best estimates are thus almost certainly not completely accurate reflections of the true evolutionary history – which is inherently unknowable – but are nevertheless the most reasonable bases for further investigations.
Phenotypic characters are the products of interactions between the influences of many genes. In this context, information about the molecular composition (e.g., DNA sequences) of genes may be viewed as somehow more fundamental and therefore potentially more informative in tracing evolutionary history. This idea has spurred the recent explosion of molecular genetic analyses employing a great variety of approaches and techniques. Genetic data are, however, very difficult to interpret in this context and require the application of various tests and manipulations before being used. Furthermore, it is not feasible to develop evolutionary hypotheses for each character since it is usually not known how they function and interact or how likely it may be for functionally equivalent changes to have occurred independently.
Consequently, maximum parsimony is seldom preferred in purely molecular analyses, but Bayesian and/or maximum-likelihood approaches are generally used. These are modifications of methods to estimate overall similarity or difference, rather than grouping by shared derived characters (synapomorphies). As for phenotypic characters, such methods will closely estimate the actual evolutionary patterns only if the incidence of convergences (independent origins of the same character state) and reversals of states is extremely low. This is almost impossible to quantify for molecular data, but indications are that convergences and reversals are not uncommon. The accuracy with which molecular approaches estimate the true evolutionary sequences is thus questionable, especially since different types of molecular analyses often produce different results.
Given the different emphases on modes of analysis, it is not surprising that the results of molecular analyses mostly differ from those based on phenotypic characters. The two kinds of analysis approach the problem in fundamentally different ways, and it may well be that an approach that combines their results to reach some sort of compromise may turn out to be the best, particularly where the same broad groupings are found by different analyses.
365体育网站Whatever approach is adopted, however, the number of taxa/individuals included and the number and variety of characters employed are critically important. This is particularly relevant where taxa are internally variable, which is almost always the case. The limited availability of suitable specimens (and funding) for molecular studies has often meant that sampling has been inadequate to produce convincing results, so that consecutive phylogenetic estimates of the same groups may differ considerably (see below). Although potential specimen representation for phenotypic studies is generally less problematic, the time and effort involved in surveying them and establishing accurate definitions of characters and states is another major limitation.
Phylogeny of the Aculeata
There is general agreement that the monophyletic Aculeata arose from within the “Parasitica,” but not on which parasitic group is most closely related to it. In other words, there is not yet consensus on which group of parasitic Hymenoptera is the sister group of the aculeates. The first rigorous analysis of the phylogeny of the components of the Aculeata, applying cladistic principles to numerous morphological characters for which their various states had been defined, was performed by Brothers . The prime aim was to clarify the relationships of the various groups then considered components of the family Mutillidae, both to each other and to other wasp groups. Accordingly, other groups had to be included in order to evaluate the placement of the mutillid components in the broader context of the Aculeata. The combined most-likely ancestral states (ground plan) for each of a wide variety of taxa, and a similar ground plan for the aculeates as a whole, were analyzed under a maximum-parsimony criterion. Several unexpected relationships appeared and were confirmed in later analyses incorporating new information .
Advances in molecular analyses of genetic material or its products, and phylogenetic analyses thereof, have led to several studies of the molecular phylogenetics of the Hymenoptera as a whole or various of its components. Early studies, with limited genetic scope, concentrated on non-aculeate groups since those were the ones for which the uncertainty about relationships was greatest. It is only relatively recently that there has been greater representation of aculeate taxa. Although these later studies have been based on extensive information derived from numerous genes or their products, all have still been unbalanced in their taxon sampling. This has meant that the genetic variability within the various subtaxa has mostly been inadequately estimated.
Early studies including genetic data also generally attempted to take account of the results based on morphological characters that were then available. The Aculeata was consistently found to be monophyletic, but the relationships of its components varied considerably. Sharkey [12365体育网站] presented a tree diagram derived from a compilation of the results of various studies, both morphological and molecular, to the level of superfamily. In this the three superfamilies of Aculeata were represented, but as a trichotomy, and with the vespoids indicated as possibly not monophyletic. All of these primarily molecular analyses involved the Hymenoptera as a whole and therefore did not explore the relationships of the components of the Aculeata in any critical way. They nevertheless suggested that the chrysidoids and apoids formed monophyletic groups, but that the vespoids might not do so.
Pilgrim et al.  explored this issue by analyzing DNA sequences from four nuclear genes for representatives of almost all subfamilies considered to belong to the Vespoidea and a few chrysidoid and apoid representatives. They did not include any non-aculeates and so could not root the Aculeata as a whole, but used the chrysidoids to root their trees. Their various analytical methods included some involving simultaneous analysis of morphological characters simplistically and inaccurately derived from Brothers . The results were varied, but generally showed the vespoids as a paraphyletic group and some components of three families to be misplaced; their “preferred” result disregarded the morphological data. This was the first molecular-based paper to concentrate on the Aculeata, although with a rather unbalanced representation of exemplars, since the emphasis was on the vespoids, and even with some vespoid subfamilies not represented or by only a single species.
Davis et al.  took a novel approach involving a matrix-based supertree analysis that combined the results of 79 trees showing relationships among families of Hymenoptera published from 1970 to 2008. Trees were chosen to minimize non-independence of their underlying data. The supertrees obtained were constrained to show the Aculeata as monophyletic, and all showed its components falling into three monophyletic clades, the placements of the terminals of which generally agreed with those in Ref. .
Several analyses of the aculeates or components thereof, whether as the main focus or as part of an analysis of the Hymenoptera as a whole, and based on various types of molecular/genetic data (e.g., targeted DNA sequences, molecular information extracted en masse from Genbank, transcriptomic and genomic data, DNA-based ultraconserved elements (UCEs), or mitochondrial genomes) were published from 2011 onward. Although the Aculeata mostly appeared as monophyletic, as did Apoidea, the relationships of the various components varied considerably from study to study. The authors of several of the studies acknowledged their unsatisfactory nature, especially the highly unbalanced representation of exemplars, a problem common to all of these studies. Because of their popularity, ants and some bees dominated many of the samples, to the detriment of other groups. This is cause for critical skepticism.
Two recent simultaneously published studies [2, 7], independent of each other, presented analyses of extensive genetic data across the Hymenoptera, although with some emphasis on the Aculeata. One of these  analyzed 3256 protein-coding genes derived from sequencing whole-body transcriptomes of 168 hymenopterans, 112 of them aculeates (6 chrysidoids, 82 apoids of which 41 were bees, but only 24 “vespoids” of which 3 were ants and 6 vespids). They again found the Aculeata to be monophyletic. The chrysidoids were also monophyletic and sister to the remaining aculeates. The vespoids were paraphyletic, comprising four monophyletic groups diverging sequentially: (i) the vespids sister to the remaining aculeates; (ii) a group comprising most of the other vespoids sister to the remaining aculeates; (iii) the scoliids sister to the remainder; and (iv) the ants sister to the monophyletic apoids. Within the apoids the crabronids were paraphyletic both with respect to the sphecids and the monophyletic bees.
The two noted papers of 2017 produced rather similar results using somewhat different analytical techniques. Even so, a comparison of them shows that many uncertainties remain, especially regarding the actual branching patterns within the broad groupings. Are the chrysidoids monophyletic or paraphyletic? Are the vespids sister to the remaining vespoids + apoids or part of a monophyletic vespoid group? What are the relationships of the many unsampled tiphiid, bradynobaenid, and mutillid taxa? It is clear that additional sampling and incorporation of morphological characters derived from the sampled taxa themselves are needed for more robust results. Nevertheless, many similarities are seen between these results and those derived from phenotypic characters (Fig. 2b).
Relevance of Phylogeny to the Evolution of Eusociality in Aculeata
Considering the various attempts to elucidate the phylogeny of the Aculeata and its various components, it is evident that major discrepancies persist between the results obtained using different types of data, whether phenotypic or genotypic/molecular, different taxon samplings, and different methods of analysis. Attempts to explain the patterns of evolution of sociality in the different taxa need to take these considerable uncertainties into account.
How Phylogenetic Analysis informs Classification
365体育网站The functions of biological classifications are very broad, underpinning all biological investigations, and acting as organizing systems to summarize information about taxa, heuristic systems enabling predictions about characters of poorly known taxa, and fundamentally acting as indexes to such information. Classifications are based on phylogenies since they represent our best estimates of actual evolutionary events, are essentially hierarchical in structure, and reflect nested systems of characters, enabling efficient summaries of them to be compiled. The accuracy of these phylogenetic estimates is, however, confounded by the pervasive effects of homoplasy (convergent or parallel origins of apparently identical character states from different ancestors), whether evidenced by phenotypic (e.g., morphological or behavioral characters) or genotypic/molecular (e.g., DNA sequences or ECUs) factors, not to mention alternative modes of analysis. As a consequence, whatever estimate of phylogeny is produced, it is almost guaranteed not to be an accurate reflection of the actual evolutionary history (which is really never definitely knowable), but merely a “best” estimate of it. Nevertheless, mapping of character state changes on putative phylogenies enables character-by-character estimation of the evolutionary “reasonableness” of those changes (e.g., through consideration of Dollo’s law) and so provides information which can be used in deciding which tree to prefer as the “best” estimate of the true phylogeny, even though such a tree may be slightly longer than the most parsimonious one. Unfortunately, such analyses are only possible for phenotypic (especially morphological) characters since the functionality of individual molecular characters is unknown; phylogenies derived only from molecular information thus cannot be evaluated for reasonableness and need to be treated with caution.
365体育网站Although a classification is based on the hypothesized phylogeny of the group, it cannot (and need not) replicate it in its entirety. Given their above functions, classifications form the fundamental bases for communication across different disciplines, whether pure or applied. Such systems must be relatively stable over time to fulfil such functions effectively, and changes to classifications should thus be conservative and only made when there is a measure of agreement on the likely accuracy of a new phylogeny, with strong support from adequate data. However, since classifications entail summaries of data, attempts to trace the evolutionary histories of particular characteristics, such as sociality, must rest on the best available phylogenies, rather than classifications. The fundamental differences between phylogenies and classifications, and their functions, must thus not be overlooked.
Classification of the Aculeata
Prior to the earliest cladistic analysis , seven or eight superfamilies of Aculeata were generally recognized (e.g., Ref. , as reflected in a highly regarded textbook of the time): Chrysidoidea (Chrysididae only); Bethyloidea (Bethylidae, Sclerogibbidae, Sierolomorphidae, Dryinidae, Embolemidae, Cleptidae); Pompiloidea (Rhopalosomatidae, Pompilidae); Scolioidea (Scoliidae, Mutillidae, Tiphiidae, Sapygidae, Plumariidae); Vespoidea (Masaridae, Vespidae, Eumenidae); Sphecoidea (Ampulicidae, Sphecidae); Apoidea (Colletidae, Halictidae, Andrenidae, Melittidae, Fideliidae, Megachilidae, Anthophoridae, Apidae); and Formicoidea (Formicidae only). It took several years, but Brothers’s three-superfamily scheme (Fig. 1) became generally adopted.
Based on a tree derived only from molecular data, Pilgrim et al.  proposed a dramatic reclassification of the vespoids, recognizing six superfamilies for them. Such drastic proposals should, however, be based on analyses which include much better representation across all taxa (to be able to estimate variability within taxa adequately). Combination of molecular data with accurate scoring of morphological character states for each individual terminal (instead of using scores based on putative ground plans, an approach which eliminates intra-taxon variability) would also have given more confidence in the results. Because of these inadequacies, it is premature to adopt their proposed reclassification (despite several more recent studies having done so). The supertrees produced by Davis et al.  did not include the Pilgrim et al.  tree; unsurprisingly, since they produced trees reflecting a consensus between previous trees, they confirmed recognition of the three superfamilies, Chrysidoidea, Vespoidea, and Apoidea. Subsequent molecular analyses, culminating in Peters et al.  and Branstetter et al. , confirmed the probable paraphyly of the Vespoidea, however, and suggested that Chrysidoidea may also be paraphyletic, but again confirmed the monophyly of the Apoidea and of the bees (the remaining apoids, the spheciform wasps, included a multiply paraphyletic Crabronidae). Since taxon sampling of many components remained inadequate, and only molecular data were used, neither paper proposed changes to the higher classification, but both adopted the dubious Pilgrim et al.  reclassification of the vespoids.
To repeat, the major functions of a biological classification are to provide a hierarchical system of names that enables efficient communication, summaries of relevant information, and the heuristic ability to make predictions about unknown characteristics with a certain degree of confidence. This means that classifications form the backbone of the data storage and retrieval systems used universally by all involved in biological research, whether systematic, ecological, or physiological, and should therefore be relatively stable. Most users of classifications are not systematists, a circumstance that must be borne in mind when proposing revisions to existing classifications. The temptation to change classifications drastically whenever new phylogenetic patterns are derived should therefore be resisted until there is overwhelming evidence that the new patterns are consistently found using different techniques and diverse data sets which include sufficient representatives of each subtaxon in a balanced fashion across the entire group as a whole and that the patterns make evolutionary sense.
Even though it has been shown to be highly probable that Vespoidea, at least, is a paraphyletic group, the relationships of its components, and those of taxa elsewhere in the Aculeata, are still insufficiently analyzed for them to be generally agreed. Therefore, it seems sensible not yet to tinker with the classification of the Aculeata, but still recognize the three superfamilies, Chrysidoidea, Vespoidea, and Apoidea, realizing that one or more of those may actually be paraphyletic. Within the superfamilies, continued recognition of their component families as reflected in Fig. 2b is also wise until there is convincing evidence that different arrangements of subtaxa are fully warranted. As far as the classification within the Vespidae is concerned, uncertainty about the phylogenetic positions of some components has little effect on the recognition of the taxa involved, so the established classification again (reflected in Fig. 3a) should prevail, at least until further analyses including both phenotypic and molecular data have been carried out.
- 1.Aguiar, A. P., Deans, A. R., Engel, M. S., Forshage, M., Huber, J. T., Jennings, J. T., Johnson, N. F., Lelej, A. S., Longino, J. T., Lohrmann, V., Mikó, I., Ohl, M., Rasmussen, C., Taeger, A., & Yu, D. S. K. (2013). Order Hymenoptera. In Z.-Q. Zhang (Ed.), Animal biodiversity: An outline of higher-level classification and survey of taxonomic richness (Addenda 2013). Zootaxa, 3703: 51–62. .
- 2.Branstetter, M. G., Danforth, B. N., Pitts, J. P., Faircloth, B. C., Ward, P. S., Buffington, M. L., Gates, M. W., Kula, R. R., & Brady, S. G. (2017). Phylogenomic insights into the evolution of stinging wasps and the origins of ants and bees. Current Biology, 27, 1019–1025. .
- 3.Brothers, D. J. (1975). Phylogeny and classification of the aculeate Hymenoptera, with special reference to Mutillidae (Vol. 50, pp. 483–648). Lawrence: University of Kansas Science Bulletin.
- 4.Brothers, D. J. (1999). Phylogeny and evolution of wasps, ants and bees (Hymenoptera, Chrysidoidea, Vespoidea and Apoidea). Zoologica Scripta, 28, 233–249. .
- 5.Carpenter, J. M. 1981 (“1982”). The phylogenetic relationships and natural classification of the Vespoidea (Hymenoptera). Systematic Entomology, 7, 11–38. .
- 6.Davis, R. B., Baldauf, S. L., & Mayhew, P. J. (2010). The origins of species richness in the Hymenoptera: Insights from a family-level supertree. BMC Evolutionary Biology, 10, 109. .
- 7.Peters, R. S., Krogmann, L., Mayer, C., Donath, A., Gunkel, S., Meusemann, K., Kozlov, A., Podsiadlowski, L., Petersen, M., Lanfear, R., Diez, P. A., Heraty, J., Kjer, K. M., Klopfstein, S., Meier, R., Polidori, C., Schmitt, T., Liu, S., Zhou, X., Wappler, T., Rust, J., Misof, B., & Niehuis, O. (2017). Evolutionary history of the Hymenoptera. Current Biology, 27, 1013–1018. .
- 8.Pickett, K. M., & Carpenter, J. M. (2010). Simultaneous analysis and the origin of eusociality in the Vespidae (Insecta: Hymenoptera). Arthropod Systematics and Phylogeny, 68, 3–33.
- 9.Piekarski, P. K., Carpenter, J. M., Lemmon, A. R., Lemmon, E. M., & Sharanowski, B. J. (2018). Molecular Biology and Evolution, 35, 2097–2109. .
- 10.Pilgrim, E. M., von Dohlen, C. D., & Pitts, J. P. (2008). Molecular phylogenetics of Vespoidea indicate paraphyly of the superfamily and novel relationships of its component families and subfamilies. Zoologica Scripta, 37, 537–560. .
- 11.Riek, E. F. (Michener, C. D., Brown, W. L., Jr., & Taylor, R. W.) (1970). Hymenoptera (wasps, bees, ants). In I. M. Mackerras (Ed.), The Insects of Australia: A textbook for students and research workers (pp. 867–959). Carlton: Melbourne University Press.
- 12.Sharkey, M. J. (2007). Phylogeny and classification of Hymenoptera. In Z.-Q. Zhang & W. A. Shear (Eds.), Linnaeus tercentenary: Progress in invertebrate taxonomy (pp. 541–528). Auckland: Magnolia Press. Zootaxa, 1668: 1–766.