Obtaining a better taxonomic understanding of native bees: where do we start?

There are more than 20 000 bee species worldwide (Michener, 2007), many of them vital pollinators of both native and cultivated plants. In the United States alone, native bees (i.e except Apis spp.) are estimated to be essential to the production of about 3 billion dollars worth of fruit each year (Losey & Vaughn, 2006). Bees are also deeply ingrained in our cultural and biological histories. Archeological and anthropological records indicate that bees were, and remain, an integral part of the cultural knowledge of many indigenous peoples around the world, including that of ancient civilizations such as the Egyptians and the Mayans (e.g. Crane, 1983; Ransome, 1986; Sheppard et al., 2001; Cohn, 2005a,b; Ayala et al., 2013). Bees may even have played a significant role in the hominid Paleolithic diet and perhaps in fueling human brain evolution, as honey and bee larvae are excellent sources of calories and proteins (Crittenden, 2011). Doubtless, recent studies of bee population and species declines in Europe, North America and Brazil have raised global environmental and economic concerns (Biesmeijer et al., 2006; Martins & Melo, 2010; Potts et al., 2010; Cameron et al., 2011; Martins et al., 2013). Some data are particularly disturbing. For instance, in the last 20 years the relative abundance of four North American bumble bees has declined by up to 96% (Cameron et al., 2011) and original geographical ranges contracted by up to 87%. Thus, it is not surprising that such reports increasingly highlight the need for a better taxonomic understanding and more accessible information on native bees to assess the status of pollinators and pollination services (Berenbaum et al., 2007). Obtaining a better taxonomic understanding of bees and making that information more accessible to the scientific community and society at large appear to be simple, straightforward tasks, but they are not. A better taxonomic understanding implies rigorously tested species hypotheses, sound phylogenetic hypotheses and stable classifications. Likewise, making that taxonomic information more accessible to science and society implies significant additional investments in financial and human capital. Herein we argue that despite more than 250 years of taxonomic research, a considerable amount of work remains to be done even in areas of the world where faunas are relatively well known, as well as in groups that have received attention because of their cultural and economic importance. To achieve a better taxonomic understanding of bees, we suggest that efforts should be focused on revisionary species-level studies for the following reasons: First, the higher level classification of bees has remained remarkably stable for the last 65 years and modern phylogenetic studies have confirmed cases that have been discussed or suspected for a long time. This is not to imply that these types of studies are insignificant to science or bee systematics. Second, bee taxonomy is based on morphological characters and most species were proposed by early scientists who had limited morphological knowledge due to the scientific equipment available at the time; additionally, these scientists operated under presently outmoded, inconsistent or even nonexistent species concepts. This means that detailed morphological characters, potentially useful in species recognition and diagnosis, remain to be explored, documented and analysed in a phylogenetic context. It also means that we have entered the 21st century with outdated, untested species hypotheses for the vast majority of bees, which translates into a high likelihood of taxonomic and nomenclatural instability. And third, species-level studies are aligned with current societal values and needs. We cannot assess the magnitude of a decline of pollinators and pollinator services if we cannot correctly identify them. Additionally, we discuss briefly the wealth of exciting research and collaboration opportunities that can be developed today while making that taxonomic information more accessible to science and society. In this context, guidelines to assess the current taxonomic status of a particular bee group are also provided and illustrated with bumble bees. Bees are unquestionably a monophyletic group that arose from the apoid wasps at least 125 Ma ago and are likely the sister group of the digger wasp family Crabronidae (e.g. Engel, 2001; Michener, 2007; Ohl & Engel, 2007; Debevec et al., 2012; Michez et al., 2012). The most widely accepted classification for bees recognizes 7 extant families, 21 subfamilies, 50 tribes and about 450 genera worldwide (Michener, 2007). Although recent higher level phylogenetic studies have clarified the historical relationships among bees as well as redefined concepts of some supraspecific taxa, the higher level classification has been relatively stable for the last 65 years in comparison to that of other insects, and even among aculeate (stinging) Hymenoptera. For example, many families, subfamilies and even tribes within Coleoptera are paraphyletic or polyphyletic, resulting in numerous taxa being shifted from one higher taxon to another (e.g. Verdú et al., 2004; Davis, 2010; Franz & Engel, 2010; Oberprieler, 2010). Similar cases occur within the Aculeata, including potentially Vespoidea, Chrysidoidea, Mutillidae and Tiphiidae (Pilgrim et al., 2008; Debevec et al., 2012; Sharkey et al., 2012). This is not the case in bees. With few exceptions, the monophyly of the families, subfamilies and even tribes is unquestioned. Modern phylogenetic analyses, using either morphology or molecular characters, or both, have confirmed cases that have been discussed or suspected for a long time, such as the demotion of the family Oxaeidae, removal of the paraphyletic Anthophoridae, and the proper association of Ctenoplectridae and Fideliidae (Michener, 2007). None of these were ground-breaking changes, instead often representing classificatory modifications reflecting relationships already suspected or known (e.g. see Almeida & Danforth, 2009 for Colletidae; Michez et al., 2009 for Melittidae; Gonzalez et al., 2012 for Megachilidae). Thus, the higher level classification of bees has remained basically the same since it was developed by Michener (1944). Other classificatory proposals exist, but they are regional in scope or deal with semantic issues about the rank given to the supraspecific groups rather than restructuring them to reflect phylogeny. We are not implying that higher level phylogenetic studies are insignificant for bee systematics or that they are merely useful to shift names around. Certainly, higher level investigations have not only improved our understanding of the historical relationships among bees, but have also served as predictive frameworks to explore biogeographical patterns or the evolution of a particular biological trait. For instance, the conclusions derived from some of those studies have challenged previous evolutionary ideas, such as that social behaviour can be reversed (e.g. Wcislo & Danforth, 1997) and that oligolecty is an ancestral rather than derived condition among bees (e.g. Danforth et al., 2006; Praz et al., 2008). Regrettably, a few groups of bees have received disproportionate attention while the relationships among most bees as well as their species diversity remain to be fully explored. For example, while a great amount of time, effort and resources have been invested in exploring the relationships among the four tribes of corbiculate apines (e.g. Cameron, 1993; Koulianos et al., 1999; Cameron & Mardulyn, 2001; Engel, 2001; Schultz et al., 2001; Cardinal & Packer, 2007; Kawakita et al., 2008; Cardinal & Danforth, 2011), which include the economically important bumble bees (genus Bombus Latreille), stingless bees (Meliponini), honey bees (genus Apis Linnaeus) and orchid bees (Euglossini), the species status of many of them remain questionable, many species are new to science but remain undescribed, and traditional identification keys are outdated or nonexistent. Such an interest in exploring the relationships among these four tribes to determine the possible origins of eusocial behaviour contrasts highly with the urgent need by society as well as by other scientists worldwide to rapidly understand and access the information on the great diversity of species found within these groups, excluding honey bees. Species concepts are hypotheses subject to test, refutation and change when other species, specimens, or characters (behavioural, morphological, or molecular) are discovered and analysed (Fig. 1) (Wheeler, 2004, 2008). The species status of the vast majority of bees has never been tested since first proposed by early scientists who operated under presently outmoded, inconsistent or even nonexistent concepts of species and possessed limited material for study, restricted knowledge of their biology and distribution, and, at most, a vague idea of their intricate morphology given the equipment available at the time. For example, roughly 28% of the known bee species worldwide were described by four individuals: Frederick Smith (1805–1879), Joseph Vachal (1838–1911), Heinrich Friese (1860–1948) and Theodore D.A. Cockerell (1866–1948) (Rasmussen, 2012). These scientists proposed species using hand lenses or stereomicroscopes with limited optics (e.g. Michener, 1948: 173), often recognized species primarily on differences in body colour or size (Fig. 2) and usually described them from a single specimen or small number of specimens. Given that bee systematics is based primarily on morphological characters, high taxonomic and nomenclatural instability, as well as an unexpected hidden diversity, is therefore anticipated. For example, Cockerell proposed 54 names in wool carder bees of the genus Anthidium Fabricius (Megachilidae, Anthidiini) for the Americas, 43 of which turned out to be synonyms (Gonzalez & Griswold, 2013). Detailed morphological characters, potentially useful in species recognition and diagnosis, remain to be explored, documented and analysed across most bees. Such characters might also prove to be useful when building sound phylogenies that provide stable classifications with predictive values for diverse biological and ecological traits. Until recently, species recognition in Anthidium was primarily based on colour differences because of the conspicuous yellow or cream maculations of these bees. Except for distinctive colour patterns that were easily seen with hand lenses, most morphological characters now proven to be taxonomically important, such as the diameter and density of punctures of the integument (Fig. 3) or shapes of genitalic structures, were overlooked. Consequently, in a recent revision of Anthidium (Gonzalez & Griswold, 2013), A. emarginatum (Say) –considered a single species for the last 188 years—turned out to be composed of multiple species when further material was examined, finer morphological characters were detected, and geographical patterns were analysed using today's methodological tools. The issues outlined above explain why even today the nomenclature of some groups continues to change, why modern species-level revisions are imperative, and why redescriptions of taxa are not a waste of effort and data. These issues apply not only to bees from remote areas of the world or to those minute, rarely collected species. A significant amount of taxonomic work remains to be done even in areas of the world where faunas are relatively well known. For example, even in the United States many species are known from a single sex making species identification often difficult or even impossible; many new species remain to be studied and described; and traditional identification keys are either nonexistent or, when available, are often outdated, poorly illustrated or not illustrated at all, and often not easily accessible to other researchers or the public. Similar problems are found throughout groups that have received attention because of their cultural and economic importance. These problems can be illustrated by a brief review of the mason, leaf cutter, bumble and stingless bees, but are reflective of conditions found across all bee taxa. Among megachilid bees, the tribes Megachilini and Osmiini are of particular importance because they include the majority of non-Apis managed pollinators [e.g. Megachile rotundata (Fabricius), Osmia lignaria Say, Osmia cornifrons (Radoszkowski)], some now introduced to many parts of the world. A preliminary assessment of the number of undescribed species in North American Osmiini based on the appraisal of museum specimens deposited in the US National Pollinating Insect Collection (Logan, Utah) resulted in the recognition of new taxa in six of the eight genera (T.L. Griswold, personal observation). This assessment also suggests a conservative estimate of 28% of the fauna undocumented (Table 1). For some genera, known proportions of undescribed species are higher (e.g. Atoposmia, 33%; Heriades, 57%). These estimates are likely conservative because narrow endemism is common in osmiines and additional undetected species are expected given that many areas in western and southern North America are still poorly sampled. With the exception of a few genera or subgenera, existing taxonomic revisions in both tribes are mostly over 50 years old and are largely outdated (e.g. revision of Osmia; Sandhouse, 1939). Also, the status of several species is doubtful (even the common, widely distributed, economically important O. lignaria may be composed of three cryptic species), no reliable taxonomic keys are available, and many species remain known only from one sex (e.g. 37% of North American Megachile Latreille; Sheffield & Westby, 2007). Bumble bees are a diverse group containing 240–250 species worldwide (Williams, 1998). They are particularly diverse in temperate areas, although relatively few species are abundant in high tropical environments (Michener, 2007). Bumble bees are among the most important wild pollinators of crops; some, such as Bombus terrestris (Linnaeus), are already widely used in managed pollination inside greenhouses, while others are being scrutinized as potential candidates. Bumble bees are taxonomically among the best known bees, with most species already known to science; however, even in North America and Europe many doubts remain about the species status of some. For example, molecular and chemical studies (Bertsch et al., 2005; Murray et al., 2008; Bertsch, 2009; Waters et al., 2010) suggest that B. lucorum (Linnaeus), one of the most common species in Europe and commercially reared as a pollinator of greenhouse crops, is composed of three distinct species: B. cryptarum (Fabricius), B. lucorum and B. magnus (Vogt). In North America, the species status of some threatened species are also controversial, namely between B. occidentalis and B. terricola Kirby, and between B. californicus Smith and B. fervidus (Fabricius) (Milliron, 1973a,b; Thorp et al., 1983). Similar problems are also found in other regions with even fewer bumble bee species but where they are also sought for pollination and conservation purposes. For example, in some South American countries, such as Colombia, it is still not clear how many and which species are present, with lists ranging from 9 to 12 species (Lievano et al., 1994; Ascher & Pickering, 2013). Like bumble bees, stingless bees are also highly appreciated, mainly because of their honey and other products that were, and remain an integral part of the culture and economy of many indigenous and non-indigenous Some species are being reared as bees are among the most bees in the of the world. There are about species most of them in the & 2007; Michener, 2007; Ascher & Pickering, 2013). and are often found in these bees, such as the of some species of & and the of stingless bees of the genera and at et al., bees are by an abundance of cryptic species species only by detailed morphological characters as well as by differences in biology and (Michener, 2007). Additionally, large areas in tropical America have never or are and their stingless bee fauna remains For example, the diversity of stingless bees in such as Colombia, and is at most a with available records in the number and of the species in each & 2007; Ascher & Pickering, 2013). the fauna of in comparison to the tropical Americas, stingless bees are poorly and documented across this and more of areas such as in with stingless bee fauna such as the species status of some and important species is still For example, the common which was highly by the is highly and it is still not clear it several cryptic species. and may be the same species 1999; Ayala et al., 2013). The of in such taxonomic issues is and in to the exciting opportunities for research and the efforts and being invested in and of these bees for pollination and as an of for For the first in more than 250 years of research, today have access to a number of that them to and more rapidly and more information than and as well as new (e.g. have the of not only morphological structures, but have also them with characters such as or remain to be in many groups, including bees 2011). Such morphological can also be easily illustrated (Fig. 3) as well as in such as molecular have another of characters in species hypotheses and phylogenetic Modern have the of both extant species and taxa. For example, of or to in in the (e.g. et al., 2011; et al., in are important to understand the evolutionary and of modern taxa, and although is a of of bees in and & Engel, they have been by most bee scientists when phylogenetic but not the has provided a new to the of by and the scientific knowledge developed over more than It has also an to knowledge to the of by making available information on species and specimens collected since et al., in other taxa, are specimen for bees, which are served data (e.g. and now are often used in species and studies to about some and with such There are also resources that provide species with and as well as include and served the of the the bumble bees of the world and the for bees of the is from these existing are restricted either in their geographical and taxonomic or in the and amount of information they Also, it is difficult to assess the relative of specimens or records based on outdated species concepts that are in the specimen records in available, specimens were in the of a recent revision (e.g. & are in need of an on bees, such as that developed for which the not only to large and on particular taxa, but also to the community of scientists Such a a great for scientists as well as the public. Given the taxonomic problems outlined above as well as the products that are already developed for some taxa, or can potentially be an assessment of the current taxonomic status of a particular group of bees is This be the first to where we and where we can from is to identify the level of we may have with a group of bees, and other Some of these products are also at the of or of as are the of products resulting from taxonomic The is not The level of with the number of products The of products to both the level of given the diversity of and the level of of the to scientists and society in of and of information to Although the of the as well as the into each level of is it is to the historical change in taxonomic It has been estimated that the of a single species by a is about US 000 & 2011). Thus, a higher is expected when these products are In to the and current status of taxonomic work of a particular taxon using such of we used and of with the to for on the taxonomy of bumble bees of the between and We found a of records but only 43 were to taxonomy and were available to small containing new records or were The number of within each was and in most taxonomic on bumble bees the first of the century in and while those of the in to research was largely restricted to or because of and it is perhaps how taxonomic research has been These are the same types of products that are the only available today for most bees. This is not to that a the of one or a few species not have a significant on society or Such a is common today for many taxa and it is particularly when it a significant important is that these of are now often with high and that Additionally, specimen data are available for Bees are highly to both science and society because of their products and including In of the recent species and population declines worldwide that may our a reliable of information and to that knowledge are species-level studies on bees are imperative, highly and aligned to current societal values and needs. focused on these studies the current taxonomic in pollinator status and pollination They also to global to the knowledge and conservation of bee have a than They also have available a of modern and methodological to such a societal There is no that such of a of into work efforts to the at which this work can be but it also of opportunities for collaboration and is several of outdated, untested species A. who had described more than of all known bee species worldwide at the of in and who rarely provided keys and never to a Michener have the now it is up to to a had described many species and now it was up to Michener to a Michener, with many over this and developed a higher level classification for bees that remains largely the is and now it is the of this of to and our For and to the discussed we are to and to B. A. Michener, Gonzalez and This was in part by the of and by National This is a of the of of The is not for the or of information by the than should be to the for the