Putative evolution of symbiotic patterns within Lotononis s l

Edaphic heterogeneity driven by geomorphic change in the early and late Miocene (23.8–5.3 mya), and simultaneous climatic deterioration, are hypothesised to have triggered the radiation of the Cape floral lineages (including the Lotononis s.

l. clade) and their subsequent speciation (Cowling et al., 2009; Edwards & Hawkins,

2007). The edaphic heterogeneity of this landscape would presumably also promote the genotypic diversity of rhizobia available as potential symbiotic partners of these legume species, due to the varying saprophytic competencies of different rhizobial genera and strains (Graham, 2008).

It is proposed that in response to this, two specificity groupings have arisen in

Lotononis s. l. species. In the first group, Leobordea and Lotononis s. str. species are

more or less promiscuous and able to nodulate, with varying degrees of effectiveness, with soil rhizobia that have a diversity of both chromosomal backgrounds and symbiotic genes. In the second group, the adaptation of Listia spp.

to waterlogged habitats has consequently required the selection of microsymbionts that are more specialised inhabitants of these environments. The comparative rarity

of rhizobial species of Microvirga and Methylobacterium suggests that their capacity

for nodulation has been acquired through horizontal transfer of symbiotic genes from other rhizobial genera.

The mechanisms of HGT are favoured in closely related microorganisms (Ochman et al., 2000). This concept accords with the results of Wernegreen & Riley

(1999), who have proposed that HGT in rhizobia is restricted across major chromosomal subdivisions, but supported among congeneric strains. A recent review of rhizobial symbiovars also supports this concept, as the biovars were nearly always found in species belonging to the same rhizobial genus (Rogel et al., 2011).

Nandasena et al. (2007) and Sullivan & Ronson (1998) have demonstrated that in

mesorhizobia, HGT between closely related strains can result in the rapid evolution of symbiotic bacteria. In the Lotononis s. l. rhizobia, the 100% sequence identity of

the bradyrhizobial nodA genes and likewise the 100% sequence identity of nodA in

the Ensifer strains lend weight to the theory of HGT occurring among closely related

strains.

It is presumed therefore that HGT between distantly related bacteria is a rarer event than that between closely related strains. That HGT between unrelated strains does occur in rhizobia has been documented in a study of Brazilian soybean isolates, where symbiotic genes from Bradyrhizobium japonicum inoculant strains were transferred to a presumably more saprophytically competent indigenous Ensifer fredii strain (Barcellos et al., 2007). Newly acquired genes require integration into the existing cellular regulatory circuits (Masson-Boivin et al., 2009). Moulin et al.

Bradyrhizobium and distant rhizobial genera implies that symbiotic genes from the

former require the correct chromosomal background to function. Interestingly, a study of 165 sequenced microbial genomes shows that B. japonicum USDA110 is an important “hub” for HGT, as it has one of the highest numbers of HGT partners (Kunin et al., 2005).

Thus, in the proposed model of the evolution of symbiotic patterns in

Lotononis s. l., the radiation and subsequent speciation of ancestral plants has led to

transference of symbiotic genes from the originally associated rhizobia to diverse, more saprophytically competent strains. A selective advantage is conferred upon rhizobial strains that successfully integrate the symbiotic genes into their chromosomal background and are able to nodulate the host plants. The symbiotic genes may then be rapidly disseminated via HGT to closely related rhizobial populations. Over time, the symbiotic genes evolve within each background, leading to allelic variation.

Species within Leobordea and Lotononis s. str. appear to have maintained the

symbiotic promiscuity that Perret et al. (2000) consider to be the ancestral state. In

contrast, the adaptation of Listia species to waterlogged habitats appears to involve

the selection of microsymbionts that are more saprophytically competent in this environment. The waterlogged habitat may restrict the number of other rhizobial species that are available for HGT, in addition to the cellular mechanisms that restrict gene flow in genetically divergent organisms (Papke & Ward, 2004), thus contributing to symbiotic isolation for both the microsymbiont and the legume host. In the Methylobacterium-Listia symbiosis, the extreme symbiotic specificity that has

developed may be due to coevolution of the plant hosts and the rhizobia. Aguilar et

al. (2004) have previously suggested that coevolution occurs in the centres of host

genetic diversity for the symbiosis between Phaseolus vulgaris and Rhizobium etli.

The symbiosis of the more tropically adapted L. angolensis with rhizobial strains of

Microvirga appears to be a case of symbiont replacement. Such replacement has also

been documented in the Phaseoleae tribe, in which species of Rhizobium have

replaced Bradyrhizobium strains as the preferred microsymbionts in two Phaseolus

species (Martínez-Romero, 2009).

Symbiotic specificity, and the mechanisms that govern rhizobial infection and nodule initiation have been studied extensively in a small number of legume hosts (Goormachtig et al., 2004; Madsen et al., 2010; Oldroyd & Downie, 2008), but not

in Listia species. A first step might be to investigate the methods by which the

Methylobacterium and Microvirga rhizobia infect and nodulate their hosts. To this

end, the processes of infection and nodule initiation were studied in L. angolensis