Recent results have demonstrated that the genus Desmanthus is non-monophyletic because the genus Kanaloa is nested within it, with a single species, Desmanthus balsensis placed as sister to the clade comprising Kanaloa plus the remaining species of Desmanthus. Here we transfer D. balsensis to a new segregate genus Mezcala, discuss the morphological features supporting this new genus, present a key to distinguish Mezcala from closely related genera in the Leucaena subclade, and provide a distribution map of M. balsensis.

Desmanthus, Fabaceae, generic delimitation, Kanaloa, monophyly, taxonomy

In the 35 years since Desmanthus balsensis J.L. Contr. was first described (Contreras 1986) and the more than 20 years since the monograph of the genus Desmanthus Willd. was published (Luckow 1993), discovery and description of the monospecific Hawaiian endemic genus Kanaloa Lorence and K.R. Wood (Lorence and Wood 1994) and assembly of molecular phylogenetic evidence (Luckow et al. 2003; Hughes et al. 2003; Luckow et al. 2005; Ringelberg et al. 2022), have demonstrated that the genus Desmanthus is non-monophyletic. To remedy this non-monophyly, we here segregate D. balsensis as a new genus, Mezcala, thereby rendering Desmanthus s.s. monophyletic.

The molecular evidence for the non-monophyly of Desmanthus with Kanaloa nested within the genus and D. balsensis as sister to the clade comprising Kanaloa and the rest of Desmanthus, presented by Ringelberg et al. (2022), is compelling as it is based on DNA sequences of 997 nuclear genes obtained via targeted enrichment (hybrid capture, or Hybseq) using a slightly modified version of the Mimobaits bait set of Koenen et al. (2020). This large phylogenomic dataset yields a phylogeny that receives maximal bootstrap and posterior probability support in concatenated phylogenetic analyses and shows a high fraction of gene trees supporting this species tree topology (Fig. 1; Ringelberg et al. 2022). Furthermore, analysis of accompanying plastome DNA sequence data obtained from off-target DNA sequence reads from the Hybseq data, confirms this non-monophyly of Desmanthus (Ringelberg et al. 2022). This non-monophyly was already hinted at in previous phylogenetic analyses based on small numbers of traditional DNA sequence loci (ITS, trnL-trnF, trnK-matK), which showed that D. balsensis is an outlier in the genus, but that the relationships between D. balsensis, Kanaloa and the rest of Desmanthus were either weakly supported or formed a polytomy (Hughes et al. 2003; Luckow et al. 2003; Luckow et al. 2005).

Figure 1. 

Phylogeny showing sister group relationships of the genera of the informal Leucaena group based on analysis of 997 nuclear gene sequences. The root of the phylogeny is indicated with an arbitrary branch length. Pie charts show the fraction of gene trees supporting that bipartition in blue, the fraction of gene trees supporting the most likely alternative configuration in green, the fraction of gene trees supporting additional conflicting configurations in red, and the fraction of uninformative gene trees in grey. Numbers above pie charts are Extended Quadripartition Internode Certainty scores. Branch lengths are expressed in coalescent units, and terminal branches were assigned an arbitrary uniform length for visual clarity, see Ringelberg et al. (2022).

In her monograph of Desmanthus, Luckow (1993) pointed out that D. balsensis is morphologically unique within the genus. First, the fruits of D. balsensis are unusual within Desmanthus, being terete or sub-cylindrical in cross-section with thickened valves which are woody when ripe, held erect above the foliage (Fig. 2A–C), and tardily dehiscent along both sutures, the valves recurving from the apex as they open (Fig. 2B) and remaining attached at the base at least briefly after dehiscence. They are quite distinct from the dorsi-ventrally flattened fruits with chartaceous or coriaceous valves and passive dehiscence along one or both sutures which occur in the remaining species of Desmanthus. Second, the anthers of D. balsensis are capped by caducous terminal stipitate claviform glands (Fig. 2D; Luckow 1993: fig. 2F), which are lacking in the remainder of species in the genus and also absent in the genus Kanaloa (Lorence and Wood 1994). Third, the pollen of D. balsensis is also unique within the genus, being arranged in tetrahedral tetrads (Fig. 2E; Luckow 1993: fig. 3A), while the remaining species of Desmanthus and Kanaloa have eumonads. This suite of morphological character state differences resulted in placement of D. balsensis as sister to the rest of Desmanthus in cladistic analyses of morphological data (Luckow 1993: fig. 15), a set of analyses lacking the genus Kanaloa which had not been described at that time. These morphological differences, alongside the molecular evidence of non-monophyly, further support segregation of D. balsensis as a distinct genus.

Figure 2. 

Morphology of Mezcala balsensis A unripe fruits held erect above branchlets B, C ripe fruits, tardily dehiscent from the apex, terete or sub-cylindrical in cross-section with thickened valves, and rhombic, four-angled seeds D claviform anther gland E tetrahedral tetrad of tricolporate pollen grains F exine of pollen showing striate ornamentation. Photos A–C José Luis Contreras Jiménez D Alejandro Martínez Mena, Facultad de Ciencias, Universidad Nacional Autónoma de México E, F Egon Köhler, Humboldt University, Berlin.

An alternative generic delimitation to ensure generic monophyly would be to transfer Kanaloa to Desmanthus. However, Kanaloa itself has unique morphological features including tergeminately bipinnate leaves, a leaf formula not seen elsewhere in any genera of the informal Leucaena group (although not uncommon elsewhere in mimosoids); absence of an involucel of floral bracts subtending the capitula; a very broad funnelform anvil-shaped, flanged stigma on a style held below the anthers (Anna Palomino, pers. comm.), small, coriaceous, ovate or elliptic, monospermous fruits and unusual large cordiform seeds (Lorence and Wood 1994), a set of characters that clearly distinguish it from both Desmanthus and Mezcala, although the fruits of the unusual Baja California endemic D. oligospermus Brandegee are also monospermous and somewhat reminiscent of Kanaloa pods (Luckow 1993). Furthermore, this alternative generic delimitation would not solve, but rather would accentuate the morphological heterogeneity within the genus Desmanthus, much of which is attributable to inclusion of D. balsensis. In addition, subsuming Kanaloa into Desmanthus would have the undesirable impact of detracting attention from the plight of Kanaloa and from the ongoing conservation battle to save this monospecific Hawaiian genus. When it was first described, Kanaloa was known from just a handful of individuals on a sea stack off the coast of the island of Kaho’olawe, the smallest of the main islands of the Hawaiian archipelago (Lorence and Wood 1994), and it is now thought to be extinct in the wild and is the focus of ongoing ex-situ conservation efforts at the Olinda Rare Plant Propagation Facility, Maui.

Mezcala is placed in the informal Leucaena group, a subclade now composed of five genera: Leucaena Benth., Schleinitzia Warb. ex Guinet, Mezcala, Kanaloa and Desmanthus (Fig. 1, Table 1; Ringelberg et al. 2022), placed within the wider Dichrostachys clade sensu Koenen et al. (2020). This wider Dichrostachys clade includes many taxa with heteromorphic inflorescences comprised of variable proportions of neuter (non-functional staminate), male and hermaphrodite flowers, sometimes with showy staminodia at the base of the inflorescences which are characteristic of the clade even though not universal within it (Koenen et al. 2020). This lability in flower types within an inflorescence is mirrored in Mezcala and Kanaloa. While inflorescences of both these genera lack showy basal staminodes, inflorescences of Mezcala frequently include a proportion of sterile basal flowers and have highly variable ratios of sterile, male and hermaphrodite flowers, including some inflorescences with entirely sterile or male flowers (Contreras 1986). The same apparently applies to Kanaloa, given that, when the genus was first described, only male flowers were found (Lorence and Wood 1994). This variation is also apparent in Desmanthus s.s., but here a number of species have long-exserted, flattened, fused and showy staminodes at the base of the inflorescence (Luckow 1993), while Leucaena and Schleinitzia apparently lack sterile flowers. Similar evolutionary lability is apparent in other morphological characters across the genera of the Leucaena subclade, and notably pollen and anther glands. As indicated above, the occurrence of pollen in tetrahedral tetrads (Fig. 2E) and the presence of anther glands (Fig. 2D) separate Mezcala from Kanaloa and Desmanthus s.s. which have eumonad pollen and lack anther glands. In fact, the pollen arranged in tetrahedral tetrads with striate exine ornamentation (Fig. 2E, F), and the stipitate claviform anther glands of Mezcala and Schleinitzia species are nearly identical (Nevling and Niezgoda 1978; Luckow 1993), reflecting the sister group relationship of Schleinitzia to the Mezcala + Kanaloa + Desmanthus s.s. clade (Fig. 1; Ringelberg et al. 2022). This close similarity of Mezcala and Schleinitzia may also be related to the likely allopolyploid origin of the genus Schleinitzia potentially involving the ancestor of Mezcala as one of the parents – see below (Ringelberg et al. unpubl. data), which is also reflected in the sister group relationship between these two genera in the plastome phylogeny of Ringelberg et al. (2022). Pollen also varies within the genus Leucaena which includes species with tricolporate eumonads as well as others with pollen in polyads (Hughes 1997), including distinctive acalymmate polyads made up of porate monad units that are quite different from the tetrahedral tetrads of Schleinitzia and Mezcala. While most species of Leucaena lack anther glands, a few have rounded or ‘hooded’ apiculae that have been equated as homologous with stipitate anther glands (Hughes 1997). Thus, these five genera display a mosaic of character state combinations that reflects extensive morphological homoplasy across this clade, as well as apparently complex and poorly understood variation in the reproductive biology of these species involving presence or absence of anther glands, presence or absence of pollen aggregated into polyads, and highly variable ratios of sterile, male and hermaphrodite flowers within an inflorescence (Table 1).

Reticulation may also have contributed to the morphological homoplasy across the genera of the Leucaena subclade, with independent whole genome duplications subtending two of the genera, Schleinitzia and Leucaena, which, in the case of Schleinitzia, is suggested to have involved an allopolyploid event most likely involving parental lineages from the Mezcala + Kanaloa + Desmanthus clade, one of which was likely the ancestor of Mezcala (Ringelberg et al. unpubl. data).

Mexico has been an important centre of legume diversity potentially throughout the Cenozoic (Centeno-González et al. 2021), and apparently had an especially rich Oligocene fossil legume flora that included many elements assigned to subfamily Caesalpinioideae and the mimosoid clade (Calvillo-Canadell and Cevallos-Ferriz 2005; Magallón-Puebla and Cevallos-Ferriz 1994), and Mexico remains an extremely important global centre of legume diversity today (Sousa and Delgado 1993). The segregation of Mezcala as a distinct genus adds to the tally of legume genera endemic to Mexico. This includes three other genera in subfamily Caesalpinioideae – Heteroflorum M. Sousa, Conzattia Rose and Calliandropsis H.M. Hern. and P. Guinet – which are also monospecific, and which also grow in similar seasonally dry tropical forest and scrubland habitats to Mezcala in south-central Mexico. Age estimates for the divergence times of these three monospecific genera and for Mezcala from their sister groups are strikingly congruent, all of them falling in the mid- to late-Miocene, 11–16 Myr (Ringelberg et al. in prep). This shows that these depauperon Mexican endemic dry habitat Caesalpinioid legume lineages are palaeoendemics and suggests they may be best viewed as potential relics of a formerly richer Oligocene / early Miocene Mexican seasonally dry tropical legume flora. All this further emphasizes the conservation importance of Mezcala and of the rich diversity of distinctive and deeply divergent legume lineages endemic to Mexico more generally (Sousa and Delgado 1993).

We thank Alejandro Martínez Mena, Facultad de Ciencias, Universidad Nacional Autónoma de México for photographing the anther gland, Egon Köhler (1932–2015), Humboldt University, Berlin for the SEM images of pollen, Anna Palomino of the Olinda Rare Plant Propagation Facility, Maui for information about and images of flowers of Kanaloa, Erik Koenen for contributions to constructing the Caesalpinioideae phylogeny, Gwilym Lewis for comments on an earlier draft of the manuscript and editorial inputs, Marianne Le Roux and Héctor Hernández for constructive reviews, and the Swiss National Science Foundation (grants 310003A_156140 and 31003A_182453/1 to C.E.H.) for funding.