The conventional system of classifying and naming chromosomes follows that of Levan et al. (1964) which is based on the ratio of length between the long and short arms. Thus, we have metacentric (m), submetacentric (sm), subtelocentric (st), and acrocentric (a) chromosomes, and this is widely used for most plant and animal taxa. In contrast a rather more detailed, and complex system of naming chromosomes has been used by Imai and coworkers and has been applied particularly to the chromosomes of the Hymenoptera (Hoshiba and Imai, 1993; Imai, 1978). It is worth going into this in some detail as it forms the basis of the minimum interaction theory to account for chromosomal evolution in the Hymenoptera (Imai et al., 1986; Hoshiba and Imai, 1993). This theory is interesting and important because it attempts an adaptive explanation for changes in chromosome number and of the amount and distribution of heterochromatin observed.

The system of chromosome morphology, called the TAM system by Imai (1978, 1991) classifies chromosomes into three basic categories, telocentric (T) acrocentric (A) and metacentric ( M ¯ ) chromosomes (Hoshiba and Imai, 1993). The A and M chromosomes are defined by two features; (i) chromosome arms are euchromatic or heterochromatic, and (ii) ratio of the width (i.e., length) of short (W_S) and long arms (W_L) as follows:

Modified types of the A chromosomes are A^e, A^M (pseudoacrocentric), and A^h which are defined as

Heterochromatic regions are located by the C-banding technique (see below), and C-band(s) located at the pericentromeric, interstitial or, terminal regions of chromosome arms are represented as A^C, Aⁱ, A^e, M^C, M^j, M^t, etc. The terms “A group” and “M group” chromosomes are used by Hoshiba and Imai (1993), and include respectively A, A^M, A^Mc, A^Mc, etc., and M, M^t, M^c, Mⁱ, etc. Some examples of the A and M group chromosomes with C-banding found in bees and wasps show in Figure 2 , which is modified from Hoshiba and Imai (1993). It is the cyclic transition between these types of chromosomes that is the essence of the minimum interaction theory (Imai, 1978). Imai (1978) proposed that centric fission of an M chromosome → T + T (two telocentric chromosomes), this is followed by tandem growth of heterochromatin, thus T → A, and finally pericentric inversion reverting A → M (Imai, 1978). Thus, the overall trend is the evolution of higher chromosome numbers and numbers of chromosome arms within lineages (Imai, 1978). Telocentric chromosomes although rare in nature, do occur (Marks, 1957; White, 1977); for example, are common in some birds (Takagi and Sasaki, 1974). Telocentric chromosomes (or telosomes) are thought to be unstable as Koo et al. (2015) state “they arise through misdivision of centromeres in normal chromosomes, and their cytological stability depends on the structure of their kinetochores. The instability of telosomes may be attributed to the relative centromere size and the degree of completeness of their kinetochore”. Thus, in their analysis of chromosomes evolution in the Hymenoptera Hoshiba and Imai (1993) treat the T chromosome as only a theoretical construct and it is used only for theoretical discussions of chromosome evolution, and is regarded as a member of A chromosome group (Hoshiba and Imai, 1993). In this system the T chromosomes transform into acrocentrics through increase in heterochromatin (T → A, see above).

From the study of the diversity of mammalian chromosome numbers three different mechanisms or hypotheses emerged to account for this karyotype evolution (Imai and Crozier, 1980; Menezes et al., 2014). The hypotheses are general enough to be applicable to any group of animals:

As an alternative to all of the above, as already mentioned, Imai et al. (1986) have proposed a modified form of the fission hypothesis, which is:

4. The minimum interaction theory (MIT) to account for chromosomal evolution in ants, bees and wasps (Imai et al., 1986; Hoshiba and Imai, 1993). As succinctly put by Hoshiba and Imai (1993) under the minimum interaction hypothesis “…chromosome evolution proceeds toward minimizing chromosomal interactions which induce deleterious chromosomal mutations such as reciprocal translocation, and predicts that increasing chromosome number by centric fission is one of the adaptive solutions”.

Here I am following Imai et al. (2001) by describing “chromosome evolution’’ as a general term covering three distinct concepts: (i) morphological alteration of individual chromosomes, (ii) evolution of individual karyotypes, and (iii) evolution of mass-karyotypes, which refers to evolution of karyotype at the generic, familial or ordinal level Imai et al. (2001).

Keeping these hypotheses in mind, I will briefly summarize some trends in chromosomal evolution as seen in some of the other major groups of Hymenoptera, and then turn to the bees in detail.

Chromosome counts for 1846 species of Hymenoptera are given in Table 1 . Chromosome numbers vary widely in the Hymenoptera, from an n = 1 to a high of n = 60 (both ants)! and their numbers, karyotypes and evolution have been studied intensively in many groups and have been reviewed elsewhere (e.g. Crozier, 1975; Gokhman, 2023; Lorite and Palomeque, 2010; Ross et al., 2015). However, in spite of this variation, haploid numbers of the vast majority of Hymenoptera lie within the range typical for animals as a whole which range from 6-20, and which presumably represents the limits of the spindle apparatus to function with either large or small numbers of chromosomes (White, 1977). Interestingly Hymenoptera do exhibit some of the lowest chromosome numbers in the animal kingdom. The lowest possible chromosome number, n = 1, in the Australian ant Myrmeca croslandi (Crossland and Crozier, 1986) is rivaled only by one species of nematode (White, 1977), nonetheless the low haploid number of three (n=3) is found in some bees, parasitoids and ants ( Table 1 ). On the other hand, the highest n of 60 in the South American ant Dinoponera lucida (Mariano et al., 2008) is much less than the highest chromosome numbers recorded in animals, the record being an n = 217–223 in the butterfly Lysandra atlantica (White, 1977).

Polyploidy (of the germ line, as distinct from somatic polyploidy) has been suggested as a cause of major trends in hymenopteran chromosomal evolution (reviewed by Crozier, 1975) and in certain groups (for example bees, Kerr and Silveira, 1972). However, is little evidence for this except in some specific cases and other types of chromosomal rearrangements can account for the observed karyotypic evolution in the Hymenoptera (Crozier, 1975). An interesting exception is the gall wasp Diplolepis eglanteriae (Sanderson, 1988). Three other species in this genus have an n=18, whereas as D. eglanteriae has n=27, and triploidy is strikingly seen by the presence of three distinctively large chromosomes ( Figure 3 , in Sanderson, 1988). Similarly, Naito and Inomata (2006) identified a triploid thelytokous sawfly, Pachyprotasis youngiae. Thirty chromosomes were seen at mitotic metaphase and the karyotype clearly consists of three sets of n = 10 chromosomes ( Figure 4 , Naito and Inomata, 2006). Given that most other Japanese species of Pachyprotasis have a haploid number of 10, triploidy of P. youngiae is strongly implied (Naito and Inomata, 2006).

In terms of total DNA content, it is worth noting that there is no obvious relationship between C-value and chromosome number in the Hymenoptera (Ardila-Garcia et al., 2010). This is the so-called C-value paradox (Thomas, 1971), for which there appears to be no one satisfactory explanation (Lakhotia, 2023). The haploid genome sizes (picograms, pg) for 131 species of Hymenoptera range from 0.1 pg in a braconid wasp to 1.14 pg in the yellow mud dauber Sceliphron caementarium (Ardila-Garcia et al., 2010). Ardila-Garcia et al. (2010) give additional genome size data for 89 species of Hymenoptera, which have a mean genome size of 0.38 pg ± 0.20 SD. For the 18 species of bees (Anthophila) they examined, taking the data from their Table 1 , gives a mean of 0.57 pg ± 0.18 SD. The values ranged from 0.24 pg in Apis mellifera to 0.90 pg in the halicted Augochloropsis metallica. Thus, there appears to be no clear connection between chromosome number and genome size in the Hymenoptera as a whole, or the bees in particular.

Only a small fraction (about 360/7170 ≈ 5%, Table 1 ) of the described species in this suborder have been studied cytogenetically and for many families chromosome numbers are unknown (Westendorff, 2006). Likewise modal chromosome numbers cannot be confidently given for all families which have been studied (Westendorff, 2006). However, some families, such as the large family Tenthreidae are relatively well studied (Westendorff et al., 1999; Kuznetsova et al., 2001), but few species have been karyotyped in the Argidae, Cimbicidae and Cephidae (Westendorff and Taeger, 2002) Overall, haploid (n) chromosome numbers in the Symphyta vary from 5-35 ( Table 1 ) (Westendorff, 2006), with some genera showing great variation both in number and morphology of chromosomes. For example, Cephalcia (Pamphiliidae) n varies from 23-35, and chromosome morphology is also highly diverse with metacentrics predominating in some species and acrocentrics in others (Westendorff, 2006). Nevertheless, given the phylogeny of the suborder (Vilhelmsen, 2006) and its position as a basal group of the Hymenoptera, it is reasonable to conclude that low n values of 7–8 is ancestral to the group, and as Hoshiba and Imai (1993) and Gokhman and Quicke (1995) suggest the ancestral chromosome number of the Hymenoptera and also the Apocrita was low; such as n = 8 or less. Thus, in the Symphyta and Apocrita chromosome evolution has, to a large extent, proceeded by fission, a major tenet of the minimum interaction theory of Imai et al. (1986). Interestingly there is direct evidence of chromosomal fission in sawflies as Rousselet et al. (2000) suggested that the origin of Neodiprion abietis with n = 8 was due to fission from a karyotype with n = 7 (typical for the Diprionidae), and that the break point of the chromosome fission was located close to an rRNA gene cluster.

Gokhman (2006) reviewed the trends in chromosome evolution of parasitic Hymenoptera. Haploid chromosome numbers of parasitic wasps (Parasitica + Chrysoidea) for the approximately 400 species examined range from 3-23 (Gokhman, 2006, 2009, 2023) ( Table 1 ). He considers chromosome numbers (n) of 14–17 with a symmetrical karyotype to be ancestral. A symmetrical karyotype is one containing metacentric chromosomes of a similar size, in contrast to an asymmetrical karyotype which has predominantly acrocentric and telocentric chromosomes (intrachromosomal asymmetry) and highly variable chromosome sizes (interchromosomal asymmetry) (White, 1977; Peruzzi and Eroğlu, 2013).

The two major trends have been independent reduction in chromosome number in the major lineages and increasing asymmetry of the karyotype. Reductions to n ≤ 10–11 occurred in some groups of the Ichneumonoidea, and independently in the common ancestor of the Proctotrupoidea, Ceraphronoida, Cynipodea and Chalcidoidea (Gokhman, 2006).

Ants (Formicidae) are the most variable chromosomally of all Hymenoptera ( Table 1 ), and a great deal is known about their karyotypic evolution, as reviewed by Lorite and Palomeque (2010). Robersonian centric fusions and fissions, and also inversions and translocations appear to be the most important mechanisms for karyotypic evolution in ants whereas polyploidy and aneuploidy probably have a minor role (Lorite and Palomeque, 2010). For example, Santos et al. (2012) found large numbers of acrocentic chromosomes in four closely related Dinoponera ants, D. australis (n = 57), D. gigantea, (n = 41), D. lucida, (n = 59/60), D. quadriceps (n = 46) suggesting that these arose from centric fissions of metacentric chromosomes. Imai et al. (1986) have proposed the minimum-interaction theory (MIT) to account for ant chromosomal evolution, and under which chromosome numbers tend to increase, but obviously to an upper limit (Lorite and Palomeque, 2010; Cardoso and Cristiano, 2021).

The haploid chromosome numbers of the wasps (Vespidae) vary widely from 5-34 ( Table 1 ). Hoshiba and Imai (1993) concluded that the pattern of chromosome evolution in both of these groups conformed with the MIT starting with an ancestral low n of around three, and then proceeding in a characteristic zig-zag pattern through the karyograph and evolving high numbers. However, a more recent study by Menezes et al. (2014) on the Epiponini (swarm-founding Polistine wasps) concluded that in this group a high chromosome number of n = 33 was ancestral and that a gradual reduction in chromosome number had occurred. For example, chromosome numbers had decreased to an n = 16 in Polybia, and even within this genus a similar reduction in n had occurred within each subgenus (Menezes et al., 2014). These trends contradict the MIT of Imai et al. (1986) (Menezes et al., 2021). Menezes et al. (2014) used the computer program chromEvol v1.3 of Mayrose et al. (2010) to evaluate the direction and type of chromosome change. The program uses a probabilistic approach to estimate the most likely chromosome changes (polyploidy, gain or loss of chromosomes by fission or fusion) that have occurred when as set of chromosomes is mapped onto a known phylogeny (Mayrose et al., 2010). Menezes et al. (2014) found that a process of constant gain and loss, and no duplication (polyploidy) gave the best fit to the data and indicated the ancestral n of 33.

Twenty one species from 10 genera of the subfamily Eumeninae have been karyotyped with n values ranging from 5-18 (Tavares and Teixeira, 2021). Tavares and Teixeira (2021, 2022, 2023) have examined in detail chromosomal evolution in various solitary wasps. Notable among these is the variation they observed in Ancistrocerus flavomarginatus (Tavares and Teixeira, 2023). In the karyotype they found two larger chromosome pairs, which were almost entirely heterochromatic, and many subtelocentric chromosomes with heterochromatic short arms. They concluded that the latter resulted from chromosomal fissions (Tavares and Teixeira, 2023).

As with humans (O’Connor, 2008), sectioning of tissue embedded in paraffin was previously used for hymenopteran chromosomes counts; for example, an incorrect (Owen et al., 1995) n = 12 for Bombus fervidus was obtained by Whelden (1954). Although reliable chromosome counts have been obtained with sectioning (e.g. Deodikar et al., 1959), generally this is an unreliable technique (Crozier, 1975). Squash methods using fresh brain, testis, or ovary tissue yield much more better chromosome preparations ( Figure 5 ) and have been successfully used to count bee chromosomes (e.g. Owen, 1983; Owen et al., 1995). Following dissection and fixation, the tissue can be stained using either the Feulgen technique or aceto-carmine method (Darlington and LaCour, 1976; Owen, 1983; Owen et al., 1995). Slides are prepared by teasing out a small piece of tissue into a drop of 50% acetic acid and then squashing under a cover slip ( Figures 1 , 5 ). Although the results are acceptable, a simpler air-drying technique that yields superior preparations was developed by Imai et al. (1977) and is widely used ( Figure 6 ). In brief, the tissue to be examined is, after treatment in a colchicine-hypotonic and Fixative (I) solution, macerated on a slide under a dissecting microscope to separate out single cells and clumps of cells, then Fixative II solution is added which is then drained off the slide (Imai et al., 1977). The slide is left to dry for one day (Barcia, 2007) and then stained with Giemsa solution, and after brief washing the slide is again left to dry; it is now permanent and needs no coverslip (Imai et al., 1977). Furthermore, it also gives C-bands spontaneously (Imai et al., 1977). C-bands occur where Giemsa stains intensely, and identify regions of constitutive heterochromatin on the chromosome; however, it is not clear whether this results from DNA denaturation-renaturation (Gokhman, 2009; Kumar et al., 2021) or staining of heterochromatin-specific proteins (Gokhman, 2009). C-banding has been widely used on hymenopteran chromosomes, particularly by Imai and coworkers on bees and wasps (Hoshiba, 1984a, b; Hoshiba and Imai, 1993; Imai, 1991). The classic Giemsa stain, perfected by Gustav Giemsa in 1904, and developed originally to identify malarial plasmodia in the blood, was successful due to its stability and because of its ability to stain chromatin deeply (Fleischer, 2004; Barcia, 2007).

There are numerous other staining techniques in addition to C-banding that give chromosome bands (e.g. Q, G, R, T) or identify certain regions on the chromosome (Kumar et al., 2021), one of these is NOR banding. Nucleolus organizer regions (NOR) of metaphase chromosomes can be visualized by silver staining techniques, either Ag-As or Ag (Kumar et al., 2021). Maffei et al. (2001) localized NORs in the bees Euglossa sp., Melipona marginate, Plebia sp., and the parasitic wasp Mellitobia australica, with AgNOR. Similarly, Rocha et al. (2002) used NOR banding on ten Brazilian species of Melipona. Similarly, Beye and Moritz (1993) used DNA probes specific to Drosophila melanogaster, coding for 28S and 18S rRNA, and found that these hybridized in situ to distinct regions of two chromosomes of the honeybee, identifying NORs.

More recently the development of single-stranded DNA probes that anneal to complementary DNA (Bishop, 2010) has allowed very specific regions of chromosomes to be identified. Of these Fluorescence In Situ Hybridization (FISH) has been used extensively for the Hymenoptera (Gokhman, 2009), with ribosomal DNA and microsatellite-satellite sequences commonly mapped (Cunha et al., 2023). For example, Cunha et al. (2023) used probes for four microsatellites to help identify possible Robertson fusion events in Neotropical meliponid bees (see below).

There is great interspecific variation in the number and the size of chromosomes within a taxonomic group, (White, 1977) which are generally inversely correlated because if we view this as a resulting from “…merely in packaging, if we regard chromosomes as packages of genetic material.” (Hsu and Mead, 1969). Gokhman (2009) points out this applies equally well to the Hymenoptera. Hymenopteran chromosomes at metaphase have an average length of 3-5 μm, but range in size from 0.5-17 μm (Gokhman, 2009). A dramatic example of size variation is seen in the ant subfamily Myrmeciinae where the Dinosaur Ant, Nothomyrmecia macrops, with n=47 has chromosomes of sizes 1-4 μm (Imai et al., 1988, 1990) whereas Myrmecia croslandi (Crossland and Crozier, 1986) a most unusual species in the same subfamily, has a single pair of chromosomes (the minimum possible haploid number, n=1) of length 17 μm (Gokhman, 2009). However, although the total genome size may be about the same in species with very different chromosome numbers and sizes there is no doubt that karyotypes are of adaptive significance and not just differences in “packaging” (White, 1977). Having an n of 1 may be risky since any chromosomal change such as a deletion however small could be lethal, whereas individuals with higher chromosome numbers might be able to tolerate this. Indeed M. crosslandi is known from only a few locations and is one of a set of sibling species with various chromosome numbers (Crossland and Crozier, 1986; Taylor, 1991). Ants themselves exhibit the total range of chromosome number as found in the Hymenoptera as a whole (Imai et al., 1990; Cardoso and Cristiano, 2021) from the Australian M. croslandi with n=1 to the Brazilian giant ant Dinoponera lucida with n=60 (Mariano et al., 2008). Bees show a smaller range of variation with the lowest found in the andrenid Andrena togashii (n=3) and the highest (n=28) of colletid Hylaeus sp. 2, both from Japan (Hoshiba and Imai, 1993).

The range of bee haploid chromosome numbers is shown in Figure 7 with data taken from the Bee Chromosome Database (www.bees.ufop.br) established by Cunha et al. (2021) which

“… is an online resource to gather information on chromosome number and nuclear genome size on bee species from all over the world. Considering the importance of cytogenetic studies for taxonomy, phylogeny, genetics, systematics, conservation, and evolution, the main goal of this database is to outline what has been done in the field of bee cytogenetics over the last century.”

This is a very valuable initiative and it is to be hoped that as new studies are published that the authors will upload their results to the database.

Although relatively few chromosome counts are available for bees (Apoidea) - only 215 species out of a total of 1846 hymenopterans ( Table 1 ) and over 20,000 species of bees - it is possible that most of the entire range of chromosome numbers has been discovered ( Figure 7 ). However, the frequency distribution may not be completely representative, since over half of the counts (119) are of meliponid species which have a distinct clade with an n=9 (Cunha et al., 2023) as seen in Figure 7 . It would appear that the modal haploid number for the Apoidea as a whole is 17 and the second mode of 9 is exaggerated given the number of meliponid species that have been studied; but this is not to say that this is not of significance nor of importance. Higher chromosome numbers may be found, since ants have a maximum n value of 60 ( Table 1 ) numbers this high could occur in bees also.

In addition to the standard chromosome complement of a species there may be supernumerary or B chromosomes in some (or many) individuals in a population, which as defined by White (1977) are “ones additional to the normal karyotype and not homologous, or only partly homologous to members of the regular set”. They are usually heterochromatic and there can be geographical variation among populations (White, 1977). B chromosomes are found in some Hymenoptera; various species of ants, a sphecid wasp (Gokhman, 2009) and the meliponid bees Melipona quinquefasciata, Tetragonisca fiebrigi (Cunha et al., 2023), Partamona cupira and P. helleri (Costa et al., 1992; Brito et al., 1997).

Returning to the question of variation in chromosome size, it is clear that the cellular mechanisms, such as spindle formation, etc, and the processes of mitosis and meiosis, are intimately related to the number and morphology of the chromosomes (White, 1977). There are also genetical implications, for example it is well known that recombination rates increase with increasing chromosome number and not just total map length (White, 1977; Sherman, 1979; Templeton, 1979). There will be genetic consequences of chromosomal fusions as this inevitable results in some loss of genes (White, 1977). Within taxonomic groups there is variation of total chromosome complement length as well as variation in size of chromosomes in some species. For example, the bumble bee Bombus pennsylvanicus shows little variation in individual chromosome length while B. fervidus shows somewhat more ( Figure 8 ). The mean haploid total complement lengths (TCL) being 24.00 μm and 23.3 μm respectively (Owen et al., 1995). Owen et al. (1995) did find significant, but not great, differences in TCL among different bumble bee species (see also Table 2 ).

An early attempt to describe and understand chromosome evolution in bees was made by Kerr and Silveira (1972) and they advanced the idea that this had proceeded by various rounds of polyploidy ( Figure 4 ). In their own words:

“The chromosome numbers published for all bees … suggest that polyploidy originated independently at least 5 times: (1) From an ancestor of Augochloropsis sparsilis n=8, to Pseudoaugoclloropsis graminea n=16; (2) from an ancestor of Leurotrigona muelleri n=8 to Frieseomelitta (3 species) n=15; (3) from an ancestral Trigonini n=9 to Plebeia (6 species) n=18; (4) from an ancestor of Melipona quadrifasciata, Melipona arginate and other species n=9 to Melipona quinquefasciata n =18; (5) from an ancestor of Apis florea n = 8, to Apis cerana and Apis mellifera n = 16.”

At the time they were writing this was plausible given the data available at that time, however even soon after it was published it was regarded skeptically (Crozier, 1975) and is no longer accepted. There is no evidence of widespread polyploidy, and where it has rarely been found it is quite obviously unusual, as already discussed earlier with the triploid gall wasp (Sanderson, 1988) and sawfly (Naito and Inomata, 2006). Also, some of the chromosome numbers were incorrect; as we have seen it was not until 1977 that it was established that all Apis species had an n=16 (Fahrenhorst, 1977), also Kerr’s (1972) count of 16 for Melipona quinquefasciata was incorrect; it is in fact just 8 (Cunha et al., 2023).

Treating the bees as the superfamily Apoidea and the major groups as families ( Table 1 ) rather than as subfamilies as done by Cunha et al. (2021), we can map chromosome numbers on a phylogeny (Bossert et al., 2019), as shown in Figure 9 . This gives the modal haploid chromosome numbers for the major families of bees with the Apidae split into the five recognized tribes (Bossert et al., 2019).

As mentioned earlier, Hoshiba and Imai (1993) and Gokhman and Quicke (1995) suggest that the ancestral chromosome number of the Apocrita was low, n = 8 or less. Thus, according to this hypothesis, chromosome evolution in the Apoidea has proceeded by fission to generate the higher chromosome numbers in the families Halictidae, Megachilidae and the Apidae ( Figure 9 ) corresponding to the general predictions of the minimum interaction theory of Imai et al. (1986). Reduction in haploid number would have occurred in the Andrenidae. Although the overall trends of the MIT seem to be met it does not necessarily mean that within groups chromosome evolution always follow the MIT, and this is particularly true with the meliponid bees (Cunha et al., 2023).

Relatively few species (28 total) have been karyotyped in the families Andrenidae (1), Colletidae (5), Halictidae (14), and Megachilidae (8). The only andrenid bee kayyotyped, Andrena togashii with n=3, just happens to have a particularly low chromosome number. It clearly is of importance to obtain more chromosome counts for the Andrenidae. However, it must be noted that haploid numbers of two other Andrena species are given by Goodpasture, (1974 ¹ , unpublished doctoral dissertation) but which are used by Ross et al. (2015) and given in their Supplementary table of data. These are A. duboisi with n=3, and an Andrena sp. with n=10 (Goodpasture, 1974). These are not included in the Bee Chromosome Database (Cunha et al., 2021) and Cunha et al. (2021) do point out that only one andrenid bee (A. togashii) has been karyotyped. Since these observations were not published in the primary literature that is good reason to exclude them, nevertheless this does suggest a low n in this group but still with some variation. Also, it is important to recognize that Goodpasture (1974) published high quality chromosome preparations of five species of eumenid wasps so there is no reason to doubt his results for the Andrena species. The two of the three chromosomes of A. togashii are relatively long with lengths (when measured from Figure 6J of Hoshiba and Imai (1993) using their 5 μm as reference) of about 8.9 μm and 6.7 μm, the other being 3.3 μm (Hoshiba and Imai, 1993). In contrast all the 28 chromosomes of Hylaeus sp. 2 are of shorter length being about 3.3 μm ( Figure 6I of Hoshiba and Imai, 1993). This does illustrate the inverse relationship between size and number of chromosomes (Gokhman, 2009).

The five species of the Colletidae show the range of haploid numbers from 8 to the highest in the bees of 28, with a mode of 16. Again, is unlikely to be a coincidence that one of the five species examined happened to have the highest n of the bees, so one might expect other species to also have high haploid numbers. Since Hylaeus is a large genus with 47 subgenera and over 650 described species globally (Michener, 2007; Almeida and Danforth, 2009), much chromosome variation is likely. It would be particularly interesting to look at the 60 Hawaiian species since they form a single clade, although the arrival of the common ancestor and the subsequent rapid adaptive radiation only occurred about 0.4–0.7 MYBP (Magnacca and Danforth, 2006). It would also be very informative to karyotype bees in the closely allied genus of Chilean bees, Xeromelissa (Almeida and Danforth, 2009), some of which, such as X. rozeni, have extremely long tongues as a result of extreme ecological adaptation (Miklasevskaja and Packer, 2015).

The 14 Halictidae karyotyped have a mode of n =16, but range of 6-20 ( Figures 6 , 9 ). A count of 21 for the former Nomia nevadensis angelesia (Cockerell, 1910), now a subspecies of Dieunomia nevadensis, was reported by Goodpasture (1974a) and is not included by Cunha et al. (2021) in the bee chromosome database and is not included “officially” here.

Chromosome counts have been made for eight leafcutter bees (Megachilidae) all of which have n=16. However, another 12 species were also karyotyped by Goodpasture (1974a) and of these all had n=16 except for one with n=15 and another with n=17.

Turning now to the five tribes of the Apidae. The modal haploid chromosome number of each tribe where known, and number of species on which this is based is shown in Figure 9 .