PhD Project - Discussion Chapter

This page is a quick conversion to text format, for figures, tables, and references, see the pdf file.

Co-dispersal and co-evolution of native legumes and rhizobia


This thesis has shown that the native legumes of New Zealand are effectively nodulated by genotypically diverse Mesorhizobium strains, that contain three nodA genotypes. Native legumes are also ineffectively nodulated by other rhizobia, predominantly R.leguminosarum, that contain different nodA genotypes than those present in native Mesorhizobium spp.

To understand how this relationship between native legumes, and nodulating strains developed, it is necessary to examine the history of legumes in New Zealand.

The historical presence of native legumes in New Zealand, and their phylogeny is discussed in Section legume-history. In summary Carmichaelia, Clianthus, Montigena, and the Australian genus Swainsona form a coherent clade (Carmichaelinae); the ancestors of which arrived in New Zealand about 5--8 mya. The other native legume genus, Sophora, is phylogenetically distinct and arrived in New Zealand about 2--5 mya, from a different geographical origin.

It appears that the evolutionary history of the native legumes is reflected in their symbiotic associations. Field isolates indicate that the two native legume lineages (Carmichaelinae and Sophora) are nodulated by Mesorhizobium strains with different nodA genes. However, host-range testing showed that native legumes of both lineages can cross-nodulate to a limited extent. Apart from this cross-nodulation, there was no effective nodulation with isolates from distantly related legumes.

Nodulation of the Carmichaelinae

In the field Carmichaelia, Clianthus, and Montigena were nodulated by various genomic groups, but only by two nodA gene types (1 and 2). Type 1 is a novel genotype (little similarity to any other nodA sequence), whereas type 2 groups with known Mesorhizobium nodA genes. Despite the differences between type 1 and 2 nodA genes and surrounding chromosomal areas (inferred from the location of primer binding---see Section s-nodA-chromo) the nodulation patterns are indistinguishable. Both can nodulate various native legumes and also exotic related legumes, whilst most differences are strain-to-strain within those groups.

In host-range studies of this thesis (Table t-Meso-native), Greenwood69, and Crow81, it was determined that Meso-NZL strains isolated from Carmichaelinae species could effectively nodulate the original host legume and other Carmichaelinae species. Although nodulation of Montigena was not determined in this thesis, Montigena and Australian Swainsona were previously shown to nodulate with New Zealand Carmichaelinae isolates Greenwood69.

In an investigation of legumes likely to be nodulated by Mesorhizobium spp., it was found that Meso-NZL strains were only able to nodulate legumes in related tribes. Work of this thesis (Table t-Meso-exotic) and that of Greenwood77,Greenwood78a and Crow81 showed that Meso-NZL strains (isolated from Carmichaelinae members) nodulated legume species from the tribes Carmichaelinae, Galegeae, and Hedysareae, while being unable to nodulate tested members of the Mimoseae, Loteae, Trifolieae, Phaseoleae, and Vicieae tribes (see Section s-Exotic-meso). An exception to the inability to nodulate unrelated tribes was the ability of some Meso-NZL strains to induce effective and ineffective nodules on New Zealand Sophora spp. (Table t-Meso-native).

These data together indicate that the Carmichaelinae are not only genotypically related to Galegeae and Hedysareae members, but their rhizobia are also able to cross-nodulate, and generally are incapable of nodulating other legumes.

Nodulation of New Zealand Sophora

The taxonomic and origin differences of Sophora spp.compared to the Carmichaelinae species seem to be reflected in the experiments of this thesis.

All Mesorhizobium strains isolated from Sophora in the field were found in a single clade in the multi-locus gene sequencing data---genomic Group A (see Section s-coherence). Sequencing of the nodA gene revealed that the three sequences derived from Sophora isolates, here called `type 3' or `Sophora' formed a distinct clade. Since no other nodA genes have ever been sequenced from Sophora isolates, it is difficult to tell if this is a New Zealand adaption and unique, or if it is specific to isolates from the Edwardsia sector, or related to more diverse Sophora isolates.

The Sophora isolates were also unique in host-range testing. Although all Sophora isolates were able to re-nodulate Sophora spp., ineffective and absence of nodulation occurred with Carmichaelinae species. Two isolates from Sophora formed ineffective nodules on Carmichaelia species (effective on Sophora and Clianthus). Likewise, two Sophora species were the only species not nodulated by three Carmichaelinae strains (Table t-Meso-native). There did not seem to be a clear relationship to nodA type or genomic group in these cases.

Sophora also responded differently to inoculation by R.leguminosarum. Most R.leguminosarum strains nodulated Carmichaelia and Clianthus, whilst Sophora was only nodulated by strain 14642 (ex Sophora).

In summary these data indicate that the nodulation response of Sophora is distinct from Carmichaelinae members, although there is cross-nodulation between isolates from the two groups of New Zealand legumes. This difference may be explained by the different origin of bacterial strains nodulating the two groups.

Origin of native legume symbioses

Ancient bacterial dispersal

The ancestral Rhizobiales were probably free-living and widely dispersed throughout the supercontinent of Pangaea. Subsequent diversification of strains (around 500 mya for the divergence of fast and slow growing species) would have led to the current diverse rhizobial species Turner00. It is therefore likely that New Zealand's bacterial populations may have been well established before the breakup of Gondwana (80 mya), and would have included ancestors of Rhizobium, Mesorhizobium and Bradyrhizobium. The diverse Mesorhizobium and Bradyrhizobium strains found in this study (Fig.p-16S-ML) are evidence for their ubiquitous presence and diversity. The close relationship of 16S rRNA genes of some native rhizobial strains to overseas strains may indicate that long distance dispersal occurred after New Zealand became isolated, and is perhaps recurrent through mechanisms such as wind and human activity. Alternatively, the 16S genes may be highly conserved and have not changed significantly since dispersal.

New Zealand would thus have had a pre-existing population of bacterial species in typical `rhizobial genera' prior to the arrival of legumes. It is likely that these bacteria did not harbour any symbiotic genes, as the evolution of rhizobia--legume symbiosis occurred subsequent to rhizobial species differentiation. Indeed the presence of non-symbiotic Mesorhizobium species in New Zealand soils has been demonstrated by Greenwood78a,Sullivan96, and perhaps by this thesis where few effective mesorhizobia were detected in pristine soils.

Co-dispersal of effective rhizobia and legumes

The ancestors of the current legume species probably arrived as seed, with the Carmichaelinae arriving from Australia, and Sophora from the north western Pacific (see Section legume-history). The seed of the Carmichaelinae is small (Table t-legume-seed), as presumably was its ancestors, and because of its size it may have arrived in mud attached to the feet of migratory birds. Alternatively a proportion of the seed does float and could have been dispersed by ocean currents.

Sophora seed is large in comparison, and is buoyant, and probably arrived via ocean currents Hurr99,Mitchell02,Heenan04. Although rhizobia have been shown to transfer with seed Perez98, the transoceanic dispersal of Sophora seed makes survival of rhizobia adhered to floating seeds unlikely. However, it may be possible as the salt tolerance of some rhizobia is very high, at least 65 days at 92 seawater equivalent Singleton82.

The results of cross-nodulation host-range of this study and others indicate that the host-range of native nodulating rhizobia extends to encompass members of related legume tribes, but little further. The most reasonable hypothesis to explain this is that the ancestor of these plants developed a specific symbiosis, and that rhizobial species (carrying transmissible symbiosis islands or plasmids) were dispersed along with their hosts, to retain this specific relationship. These strains would have been co-distributed, along with their hosts to New Zealand. Upon arrival in New Zealand these symbiosis regions may have been transferred to the locally adapted non-symbiotic population, or retained in the original hosts. The presence of three different nodA genotypes nodulating native legumes may result from separate introduction events.

The ability of Mesorhizobium isolates to cross-nodulate between the unrelated Carmichaelinae and Sophora lineages cannot be explained by this mechanism. It is possible that after arrival of these species into New Zealand, subsequent co-evolution or horizontal transfer of genes other than nodA between rhizobia, broadened the host range further. This local adaptation hypothesis could be tested by conducting nodulation studies with rhizobia nodulating Sophora sect.Edwardsia in countries with no Carmichaelinae members, e.g. Hawaii, and Chile.

Ineffective nodulation of native legumes

An unexpected result of this research was that R.leguminosarum nodulated the native legumes of New Zealand, despite literature searches revealing few exceptions to the established host-range of R.leguminosarum nodulating its typical hosts (with the exception of deliberately genetically modified strains).

A hypothesis was formed that these strains had acquired a transmissible Mesorhizobium plasmid or symbiosis island from native rhizobia, and this permitted nodulation of native legumes. However, sequences of the nodA gene revealed that the strains were typical for R.leguminosarum. A comprehensive test was then devised to determine the host-range symbiotic capacity of these strains, and of three biovar strains. All strains (bar one) were able to nodulate native legumes---but the symbiosis was ineffective. These strains were still able to nodulate their typical legume hosts---implying that the full set of standard nod genes and accessory symbiosis components were present. This showed that R.leguminosarum strains nodulating native legumes were apparently entirely typical.

In other experiments, strains of Rhizobium spp., Bradyrhizobium spp., and Paenibacillus were found to ineffectively nodulate native legumes baited in pristine rhizosphere soils (in additional work to this thesis, see section s-Johns-work). These strains formed nodules---gaining the benefits of a constant food supply and living in a protected environment. Yet they do not fix nitrogen for the host plant and in this manner they could be considered parasitic.

These isolates would have a competitive advantage over other rhizobia, by diverting more energy for reproduction than nitrogen fixing strains. Nevertheless, nodule isolates from native legumes were predominantly effective Mesorhizobium spp. It is possible that native plants defend against parasitic behaviour by restriction of oxygen supply to an ineffective nodule, reducing their reproduction as demonstrated in soybean Denison04b. Alternatively it may be that Mesorhizobium spp.are more competitive for nodulation, or better adapted for local soil conditions.

Thus whilst effective nodulation of the native legumes is restricted to specific Mesorhizobium strains, other strains can form ineffective symbioses. The mechanisms that allow strains distinct from the native Mesorhizobium to nodulate are unknown, but could be investigated by characterising Nod factors and their receptors.


The host-range of Mesorhizobium strains isolated from native legumes extends to include related legume species from the northern hemisphere and Australia. The ability to nodulate only related legumes probably indicates that effective strains co-dispersed along with their hosts, during radiation from the northern hemisphere, through Australia, to New Zealand.

The presence of novel nodA genotypes and the ability of isolates from New Zealand's two native legume lineages to cross-nodulate may be due to co-evolution of legume and rhizobia after arrival in New Zealand.

Other strains (predominantly R.leguminosarum) form ineffective parasitic nodules on native legumes, the extent and effect of this parasitism is unknown.

Future work

Some of the hypotheses of this discussion chapter are based on a single nod gene. Characterisation of other nod and nif genes, and the Nod factor receptor may allow more robust interpretation.

Further characterisation of nod genes would allow a more complex description of the Nod factor molecule. Through sequencing nodZ (fucosylation), nolL (acetylation), and noeI (methylation), or equivalents, the modifications of the molecule could be determined. This would allow more complete description of the novel Nod factors associated with New Zealand legumes. In addition, comparison of these sequences (perhaps along with nodB and nodC) may help to determine if the patterns in nodA seen in this thesis, are repeated in other nod genes.

Sequencing of nod genes from exotic Sophora rhizobia is required to gain insight into the origin of rhizobia nodulating New Zealand Sophora species. There are currently no nodA sequences available in the literature from rhizobia nodulating any exotic Sophora or close relations of the Carmichaelinae such as Swainsona. The comparison of nodA genes from exotic and related species may help to determine their origin and if the nod genotypes found in New Zealand are unique or are more widely distributed.

Sequencing the entire symbiosis region of rhizobia nodulating native legumes would assist in the understanding of the genetic elements regulating Mesorhizobium symbioses. Such elements have only been described for Lotus nodulating species Kaneko00,Sullivan02. The nature of the symbiosis region (plasmid or chromosomal) could also be determined by nod gene probes, extending the work of McCallum96. Alternatively, the presence of a symbiosis island could be inferred by sequencing the Phe-tRNA gene to intS region that borders a symbiosis island inserted in the Phe-tRNA gene Nandasena05.

Another avenue of research is characterisation of the Nod factor receptor (NFR) from the host plant. This correlation between the phylogenies of legume determinant of Nod factor perception and that of bacterial nod genes would provide extra data to support or reject a co-evolution hypothesis. The sequence of the NFR of native legumes is to be determined in future work in collaboration with Tomasz Stepkowski, from the Polish Academy of Sciences.

Invasive introduced legumes

An underlying question of this thesis was whether the invasiveness of introduced woody legumes was influenced by the nature of their rhizobial symbioses. Their invasive ability is certainly enhanced by their nitrogen-fixing symbiosis with rhizobia. However, the source of the rhizobia that nodulate introduced legumes is an enigma---since exotic legumes were recently introduced (in contrast with native legumes), and New Zealand is so geographically distant from the natural habitat of gorse and broom (Europe).

Three hypotheses were initially proposed to explain the nodulation of introduced woody weeds: 1) Introduced legumes are promiscuous and use the same rhizobia as native legumes. 2) Introduced legumes use specific rhizobia that were recently introduced---perhaps in conjunction with exotic legumes. 3) Introduced legumes use specific rhizobia that were already present in New Zealand.

The data presented in this thesis quite clearly eliminates the first possibility. Multilocus gene sequences placed all strains from introduced legumes into the Bradyrhizobium genus---distinct from the native legumes that are nodulated by Mesorhizobium species. Additionally, an investigation of the nodA gene revealed significant differences between rhizobia nodulating introduced and native legumes. Further to this, host-range cross inoculation tests showed that Mesorhizobium strains were unable to nodulate introduced legumes (and vice versa), these data show that invasive weeds are not nodulated by the same rhizobia as native legumes.

The choice between the remaining two hypotheses is somewhat harder to elucidate. The nodA gene of introduced legumes was very similar to sequences found overseas---unlike some sequences from the native legumes which were unique. This supports the notion of recent introduction from an external source.

On the other hand, an investigation into pristine New Zealand soils showed the presence of genotypically diverse Bradyrhizobium spp., that were geographically widely dispersed and were found in areas that had little human influence. This may suggest a ubiquitous free-living distribution of strains, and a long history in New Zealand.

In order to resolve this apparently confounding evidence it may be necessary to treat New Zealand Bradyrhizobium two groups---the `Acacia' and `Genisteae' as defined by nodA gene type and effective nodulation ability.

`Acacia' strains have a similar nodA gene to Acacia isolates from Australia, and are similar in 16S gene sequences Lafay01. This suggests that they may have dispersed here from Australia---the source of the current Acacia population. Many insects, birds, and plants arrived in New Zealand from Australia via the `west wind drift' Cook05, which was initiated about 31 mya along with the Antarctic circumpolar current Florindo03,Lawver03. It is conceivable that aerosols of soil could disperse bacteria to New Zealand. Indeed, even in contemporary times, snow on the Southern Alps was stained red with aeolian Australian chromosols Knight95,Kiefert96,McGowan05.

An alternative to this dispersal hypothesis, is that extant Bradyrhizobium spp.derive from symbionts associated with New Zealand's once native Acacia population, which was present during the Neogene, but became extinct in the last ice age. For this to be correct, the Bradyrhizobium population would have had to remain viable and effective in the soil, for more than 10000 years. This is perhaps unlikely as symbiosis regions may be lost, or accumulate deleterious mutations, after an extended time and passage though multiple generations with no immediate need for expression of host-specific functions.

The `recent importation' hypothesis is consistent with other studies. An investigation of Western Australian lupins determined that the nodulating strains were of European origin Stepkowski05. The `Genisteae' nodA type of this thesis groups closely with these lupin strains---implying that the New Zealand strains could also be of European origin. The `Genisteae' strains may have been introduced from Europe with the settlers, or even blown over from Australia, since the establishment of lupins there. This conclusion is further supported by the observation that the nodA gene is too similar to have diverged for many millions of years. Strains carrying these specific nod genes must therefore have arrived at some point, as Genisteae is a Northern Hemisphere legume clade.

As an avenue for future work, recent or ancestral transfer of Bradyrhizobium strains to New Zealand could be investigated by examining mutation rates in nod genes of Bradyrhizobium pristine New Zealand soils, that have never grown gorse or broom. Since nod genes in this situation would be under no selective pressure, random mutations would be expected to have accumulated if the strains were ancient Zhao97. If they were recent introductions, however, the genes would be relatively unchanged from European strains. Obviously this work could not be done by baiting---as this would miss nod genes that have mutated to the point of non-functionality or been lost altogether.

Implications for conservation and biosecurity

Conservation of native legumes

Some native legumes are considered critically endangered, and restoration projects are underway to conserve and protect these species Clemens01,Walker03,Stanley05. The work of this thesis may help to better understand the rhizobial aspects of conservation of native legumes.

Kowhai (Sophora) species are distributed throughout New Zealand, and are not considered threatened, although S.fulvida is in gradual decline, and S.longicarinata and S.molloyi are range restricted deLange04.

Most Carmichaelia species are abundant, but C.curta, C.juncea, C.kirkii, and C.williamsii are endangered, and C.hollowayi and C.muritai are critically endangered Heenan99,deLange04.

Clianthus is a common garden plant in New Zealand and abroad. In the wild there were only about 200 adult plants left in 1997 Shaw97, and despite an active restoration project, as of December 2005, there were only 153 mature wild plants recorded at 20 sites Stanley05. Part of the problem lies in the fact that although Clianthus germinates readily, competition and other environmental factors in the wild results in few seedlings developing into mature plants. Adults also only have an approximately seven year functional life and therefore recruitment is slow [David][]King-PC. Both species (C.puniceus, C.maximus) are considered critically endangered.

Native species have become endangered through destruction of their natural habitat, and invasion by competing species. Although rhizobial associations play a part in the ecology of these species, until now they have been largely ignored.

The results presented in this thesis are generally positive for native legume conservation. It appears that all tested members of the Carmichaelinae subtribe can cross inoculate effectively with Mesorhizobium spp.Sophora can also nodulate effectively with Mesorhizobium spp.but to a lesser extent. This effective rhizobial association indicates that nitrogen deficiency should not be a growth limiting factor in most situations. The very low nodulation of natives in pristine soil however, means that restoration into areas that are currently devoid of native legumes may require inoculation to achieve the best growth.

This research also showed that native legumes can be nodulated ineffectively by other strains. These strains are effectively parasitic, in that they do not produce nitrogen for the plant but consume resources. This may be a problem when attempting to establish native legumes near or downstream of pasture, or where exotic legumes are present. It is unknown if R.leguminosarum or other ineffective strains are more or less competitive for nodulation than native Mesorhizobium strains (although Mesorhizobium strains were found more often in nodules). The relative competitive ability of effective and ineffective strains could be investigated in future studies.

Biosecurity implication of introduced legumes

Gorse, broom and wattles are all serious weeds in New Zealand. Although much work has been done on other aspects of their ecology Hamilton90,Richardson98,Fogarty99,Hill01,Buckley03, their rhizobial symbionts have largely been ignored Richardson00,Parker01. It appears that effective strains have been introduced into New Zealand through natural dispersal or human activity. Bradyrhizobium spp.are present in pristine soils (although in low numbers) implying that the invasion of weed legumes is unlikely to be hindered by the absence of an effective symbiont.

In Hawaii, gorse is also a major problem, but there, the predominant native legume is Acacia koa which can cross-nodulate with gorse rhizobia (and vice versa), assisting invasion Leary06. This may also be the case in Australia with its native Acacia legume population. In New Zealand there is no natural reservoir of legumes nodulated by Bradyrhizobium spp., other than the invasive weeds.

The rhizobia--legume association is unlikely to become a target for bio-control, although some research has been done on broom-Bradyrhizobium specific bacteriophages Malek05. This is unlikely to be a realistic method of control due to the unknown danger of releasing bacterial viruses into the complex microbial ecology of the soil.

The work presented in this thesis helps to fill in the gaps of knowledge of the nitrogen fixing ability of invasive legume weeds, and partially explains their rapid colonisation of large areas of New Zealand.

Final Conclusion

This thesis set out to identify the nature of the nitrogen-fixing symbioses of New Zealand's native and exotic woody legumes. Through sequencing of housekeeping and symbiosis genes, it has been established that native and exotic legumes form effective symbioses with distinctly different species of bacteria.

The origins of these bacteria can not be categorically determined. However evidence is presented to suggest that symbionts of native legumes, the Mesorhizobium, derived from bacteria that were distributed along with their hosts on arrival in New Zealand. This would have introduced effective symbiosis genes. Whether these genes were subsequently transferred to the existing locally adapted Mesorhizobium population is unknown. The source of effective Bradyrhizobium, which nodulate exotic legumes, is suggested to be more recently introduced, possibly from Australia. Further work is required to confirm these hypotheses.

Collectively, the work presented in this thesis provides new insights into the nature of rhizobial symbionts of native and exotic legumes. An understanding of the specificity of nodulation and nitrogen fixing capability may help in the conservation management of endangered native legumes, whilst knowledge of the nitrogen fixing ability of woody weeds goes some way to explain their success as invasive weeds.