, Volume 167, Issue 1, pp 149–155

Does a facultative mutualism limit species range expansion?


    • Department of Ecology, Evolution and BehaviorUniversity of Minnesota, Twin Cities
  • Carolyn G. Anderson
    • Department of Land, Air and Water ResourcesUniversity of California
Community ecology - original paper

DOI: 10.1007/s00442-011-1958-4

Cite this article as:
Stanton-Geddes, J. & Anderson, C.G. Oecologia (2011) 167: 149. doi:10.1007/s00442-011-1958-4


The availability and quality of mutualists beyond a species’ range edge may limit range expansion. With the legume Chamaecrista fasciculata, we asked to what extent the availability and quality of rhizobia beyond the range edge limits host range expansion. We tested the effect of rhizobia availability on plant growth by transplanting seed from three locations into five sites spanning C. fasciculata’s range (interior, at the northern and western range edges, and beyond the range edges), and inoculating half the seeds with rhizobia. We recorded growth of all surviving plants, and, for the uninoculated plants, whether they had formed nodules or not. We isolated rhizobia from nodules collected on the uninoculated plants, and cross-inoculated seed from four populations (both range edge and interior populations) in the greenhouse to determine whether the quality of rhizobia differed between regions. We found that seeds transplanted beyond the range edge were less likely to be nodulated when they were not experimentally inoculated, and there was benefit to inoculation at all sites. In the greenhouse, the three inocula that formed nodules on plants, from the range interior, northern edge and beyond the northern edge, did not detectably differ in their effect on plant growth. These results suggest that low densities of suitable rhizobia beyond the range edge may limit range expansion of legume species.


Range limitsMutualismChamaecrista fasciculataRhizobiaTransplant study


Many factors including climate (Angert and Schemske 2005; Geber and Eckhart 2005; Purves 2009; Rehfeldt et al. 2008), competitors (Case and Taper 2000; Price and Kirkpatrick 2009), natural enemies (Holt and Barfield 2009) and pathogens (Antonovics 2009) have been implicated in limiting species’ geographical distributions. However, the role of a mutualism has received little attention as a potential factor limiting species’ distributions. Mutualisms can constrain range expansion of a host species if the symbiont is absent beyond the range edge (Parker 2001), reducing fitness below that necessary to maintain positive population growth rate. Furthermore, even if the symbiont is present, low abundance or poor quality of the symbiont for the host (Burdon et al. 1999; Heath and Tiffin 2007) may constrain range expansion. As the effect of mutualists on plant performance has been shown to be equal in magnitude to the effect of enemies (Morris et al. 2007), the role of mutualisms in limiting range expansion needs to be further considered. For example, in the case of plant invasion, it has been shown that the absence or low densities of mutualistic rhizobia (Parker et al. 2006) and mycorrhizae (Nuñez et al. 2009; Pringle et al. 2009) may limit invasion success. However, the role that mutualisms may have in limiting native ranges is unclear.

The symbiosis between legumes (Fabaceae) and rhizobia (i.e., nitrogen-fixing soil-dwelling bacteria) is well suited for examining the role of a facultative mutualism in limiting plant range expansion for both empirical and theoretical reasons. First, legumes experience significant reductions of fitness if compatible rhizobia are not present, demonstrating the dependence of legume population growth on both the availability and identity of symbionts (Burdon et al. 1999; Bushnell and Sarles 1937; Heath 2010). Second, legumes are under-represented in island flora relative to nearby mainland areas (Parker 2001), suggesting that their establishment is constrained by factors other than climate. Third, Medicago species that exhibit less specificity for mutualists have larger geographic distributions (Béna et al. 2005), suggesting that the evolution of reduced symbiont specificity allows range expansion. Further, there are theoretical reasons to expect the availability and quality of rhizobia to differ beyond the range edge. Models show that both partners may reach an equilibrium at a low population size (Parker 2001), which may happen in marginal conditions at the edge of a species range (see examples and counter-examples in Sagarin and Gaines 2002), and thus are more likely to face local extinction. If present beyond the range edge, rhizobia may be of lower quality (i.e., provide less benefit compared to other strains) to a plant host because selection is not acting to maintain the symbiosis between rhizobia and the plant host. Any rhizobia present either persist saprophytically in the soil, with selection favoring survival in the soil potentially at a trade-off to symbiotic nitrogen fixation (Ratcliff et al. 2008), or the rhizobia persist by forming symbioses with other legume species.

In this study, we examine how the legume–rhizobia mutualism affects the potential for range expansion of the native annual legume Chamaecrista fasciculata. Specifically, in a transplant experiment, we ask how rhizobia availability influences fitness at different geographic range locations by inoculating some seeds with rhizobia and leaving other seeds uninoculated. We emphasize that, because rhizobia are ubiquitous, we are testing for the presence of strains that have the ability to infect and benefit C. fasciculata, and not simply the presence of rhizobia. Finding that uninoculated plants are less likely to nodulate than inoculated plants, and that the inoculated plants have greater fitness than uninoculated plants beyond the range edge would suggest that rhizobia availability does limit legume range expansion. Using rhizobia strains isolated from the field experiment, we conducted a second experiment in the greenhouse asking to what extent rhizobia quality differs between geographic range locations. Finding that rhizobia from the range interior provide a greater benefit to plant growth than rhizobia from the range edge or beyond would suggest that rhizobia quality may also limit plant range expansion.

Materials and methods

Study system

Chamaecrista fasciculata (Fabaceae), an insect-pollinated native annual legume, is one of the most northerly species in the genus Chamaecrista, with a distribution from Mexico or Central America in the south to north-central and eastern United States in the north (Irwin and Barneby 1982). Symbiosis with rhizobia is common in the genus Chamaecrista, which is one of few genera in the subfamily Caesalpinioideae with species known to nodulate (Doyle and Luckow 2003). In North America, C. fasciculata is nodulated by rhizobia in the genus Bradyrhizobium (Parker and Kennedy 2006; Tlusty et al. 2004), specifically the lineage B. elkanii (Parker and Kennedy 2006). C. fasciculata is also known to nodulate with rhizobia isolated from co-occuring species (e.g., Dalea purpurea) in the laboratory, indicating that it has the potential to utilize rhizobia from alternate hosts in the soil (Tlusty et al. 2004), though the relatedness of these strains and the prevalence of sharing of rhizobia strains in the soil is unknown. Throughout the text, we use rhizobia to refer to the functional type, as opposed to the generic identity, of the bacteria.

Field experiment assessing rhizobia availability

Plant growth at different geographic range locations may be limited by the availability of compatible rhizobia in the soil. To test this, we examined the growth of C. fasciculata plants inoculated with rhizobia (+rhiz) or left uninoculated (−rhiz) at five sites: the range interior [CERA: Conard Environmental Research Area (Grinnell College), Kellogg, Iowa], western edge (RNHA: Reller Natural History Area, Spague, Nebraska), beyond western edge (CPBS: Cedar Point Biological Station, Ogallala, Nebraska), northern edge (SCWRS: St. Croix Watershed Research Station, Marine-on-St. Croix, Minnesota) and beyond northern edge (ACNW: Audubon Center of the Northwoods, Sandstone, Minnesota) (map in Online Resource 1). We selected these sites to span two distinct range edges (northern and western) of C. fasciculata that may be limited by different combinations of environmental factors. At the interior and edge sites, C. fasciculata was not growing at the site where the seeds were planted, but was found within 1 km. We planted 20 seeds collected in 2008 from each of three source populations (interior, northern edge, and western edge; Online Resource 1) randomly assigned to locations 1 m apart at each site between late April and early May 2009. Seeds were randomly assigned to either inoculation with 1 mL inoculum or 1 mL sterile H2O at the time of planting. The rhizobia inoculum was a mixture of strains UMR6404 and UMR6437, which are known to be beneficial to C. fasciculata, from the University of Minnesota Rhizobium collection provided by the late Dr. Peter Graham. Though the identity of these strains has not been confirmed, slow growth suggests that they are Bradyrhzobium spp., likely B. elkanii given the only previous study of rhizobia taxonomy from C. fasciculata (Parker and Kennedy 2006). We used multiple plant populations and a mixture of rhizobia strains to reduce the chance of complete incompatibility between plant genotype and rhizobia inocula (Heath 2010). To inoculate the plants, the rhizobia strains were grown separately in tryptone yeast (TY) media (Somasegaran and Hoben 1994) until turbid (approx. 106 cells/mL, about 5 days), mixed and stored at 4°C until the inoculum was applied to the seeds (not longer than 4 days).

Early-season survival was recorded about 4 weeks after planting to determine whether the treatments influenced germination and survival. We recorded plant height, which is correlated with fitness (r2 = 0.49 between height and seed pod production; J.S.-G., unpublished data), when plants began to flower in July so that we could also record nodulation before the nodules began to senesce, which happens prior to seed maturation (J.S.-G., personal observation). Sample size at this stage was low because of drought and herbivory [ACNW: 11 (−rhiz), 9 (+rhiz); CPBS: 4 (−rhiz), 5 (+rhiz), SCWRS: 6 (−rhiz), 7 (+rhiz); RNHA: 10 (−rhiz), 3 (+ rhiz); CERA: 18 (−rhiz), 15 (+rhiz)]. To record nodulation, we excavated the roots of all surviving uninoculated plants and collected nodules. We recorded only observations of at least one nodule, and not nodule number, because of the difficulty of extracting all roots. Nodules were stored at 4°C with a dessicant (Drierite; W.A. Hammond).

While geographic range location may influence the effect of rhizobia on plant growth, abiotic soil nitrogen can also influence plant growth. At high levels of soil nitrogen, C. fasciculata does not form symbiotic nodules with rhizobia (Naisbitt and Sprent 1993), and, thus, all plants should grow equally well regardless of inoculation treatment. To examine how soil nitrogen influenced nodulation and plant growth, we determined total nitrogen in soil samples collected from each site using a Costech ECS 4010 elemental analyzer (Costech Analytical Technologies, Valencia, CA, USA).

Greenhouse experiment to assess rhizobia quality

To examine whether rhizobia from the edge or beyond a host plant’s range differed in quality from rhizobia from within the range, we cross-inoculated four plant populations from different geographic regions with rhizobia strains sampled from within and beyond the range in a greenhouse experiment. C. fasciculata seed was collected in September 2008 from four sites; two within the species range, one at the northern edge, and one at the western edge (Online Resource 1). We obtained rhizobia from the nodules collected on uninoculated plants at each site in the field experiment, plus an additional interior site (TYS) so as to have two rhizobia strains for each region (interior, edge, beyond). Nodules were rehydrated in sterile H2O, surface sterilized with sodium hypochlorite (NaOCl) and 95% ethanol, crushed in 1.5-mL tubes in liquid TY media and the supernatant was spread on TY plates. Where possible (TYS, CERA, SCWRS), we selected three nodules from three different plants, but at other sites (ACNW, RNHA, CPBS), most plants did not form many nodules so we used three nodules from one plant. We grew the bacteria at 30°C for 5–7 days, and then randomly selected one colony from each nodule (three per site), except for the CPBS site where no rhizobia grew on the plates. The strains were re-streaked to check for contamination. We grew each strain individually in 50 mL TY media for 7 days at 30°C, and then we standardized the concentration of each strain at approximately 106 cells per mL with OD600 readings, likely far greater than the concentration of rhizobia in natural settings. We mixed the three strains isolated from each site in equal proportions to generate a single inoculum representative of that site (designated rCERA, rRNHA, rSCWRS, rACNW, respectively) and applied the inoculum to the plants within 24 h. A mixture was used to avoid genotype-specific effects, as we were interested in the effects of the rhizobia population at each geographic location, and not any specific rhizobia strain by plant genotype interactions.

In January 2010, we surface-sterilized 60 C. fasciculata seeds from each population in 10% sodium hypochlorite. We scarified the seeds, germinated them on filter paper, and planted each emerging seed in a container filled with a 50:50 soil mixture of Turface and SunGro SB-500 to facilitate nodule counting at the end of the experiment. The soil mix was steam-sterilized before planting to reduce potential rhizobia contamination. We used a split-plot experimental design with each of three blocks divided into ten whole plots (container racks) that were randomly assigned a rhizobia treatment (one of five rhizobia inoculum or a control) to reduce the chance of cross-contamination between treatments. Each treatment was represented twice in two blocks and once in the third block because of space constraints on the benches (i.e., blocks). Each plot was divided into eight sub-plots, with two individuals from each plant population randomly assigned. Because of differences in emergence rate, seeds were not all planted on the same day.

Approximately 2 weeks after planting, we inoculated all C. fasciculata plants on the same day with 1.5 mL of the prepared inoculum or sterile H2O for the control. Plants were watered daily with a fine-nozzle spray hose to reduce splashing between racks. We recorded initial plant height on the day of inoculation, and when plants began flowering (after about 8 weeks), we measured height again, harvested all plants and counted all the nodules on each plant. Two inocula (rCERA and rRNHA) did not form any nodules on most plants, and it was unclear if this was due to biological reasons or experimental error. Thus, we analyzed the data without these two rhizobia inocula, though the results were similar if they were included (Online Resource 1).

Data analysis

To test the influence of site, population, treatment and the site × treatment and population × treatment interactions on early-season survival, we used a generalized linear model with a binomial link function using glm in R (R Development Core Team 2009). For the plants that survived, we tested the effects of these terms on height by ANOVA (aov in R). For both responses, we fit a full model and then compared nested models using the likelihood ratio test, beginning with the interactions, and removing terms that did not significantly (P < 0.05) improve fit of the model to the data. We calculated estimated mean values and standard errors from the final model for plotting.

To assess whether the percentage of uninoculated plants that formed nodules differed beyond the range than within the range, we grouped data from the two sites beyond the range, and the three sites within the range or at the edge. We compared frequencies of nodulated plants between these two groups (beyond vs interior) using Fisher’s exact test because of small sample sizes. To determine if soil nitrogen influenced nodulation, we tested if there was a significant correlation between site nitrogen and the percent increase in growth between uninoculated and inoculated plants. Finding a significant negative correlation would suggest that, at sites with high soil nitrogen, plants forgo forming symbiosis with rhizobia and thus receive less benefit from inoculation.

To assess differences in rhizobia quality on plant growth in the greenhouse, we examined the effects of rhizobia treatment, population, rhizobia × population and the covariates of block and date planted on the difference in plant growth from the date of inoculation to the date of harvest. One plant died and was removed from the analysis. We performed a split-plot ANOVA with the effect of rhizobia treatment and rhizobia × plant population tested against the variance among plots, here the rhizobia × block interaction while the main effects of population and date planted were tested with the residual variance as the appropriate error term (Milliken and Johnson 2009). We repeated these analyses for nodule number, with a ¼ power transformation of nodule number which came closest to the assumption of normality for the residuals.


Rhizobia availability in field

Inoculation with rhizobia had no effect on early-season survival, but significantly increased mid-season plant height (Table 1) with inoculated plants being taller at all sites (Fig. 1). Though the site by treatment interaction was not significant for plant height (Table 1), there was a trend for a greater benefit of inoculation at sites beyond the range edge (179% increase beyond the northern edge, 48% increase beyond the western edge, compared to an average of 40% increase at the edge and interior sites), which is suggestive given the small sample size. Site had a significant effect on early-season survival (Table 1) with high survival at the interior (63%) and beyond northern edge site (61%), intermediate survival at the western edge (42%) and low survival at the northern edge (15%) and beyond western edge site (15%). Plant height in July was also significantly influenced by site (Table 1), with plants being tallest in the interior site and shortest in the beyond range sites (Fig. 1). Plant population source had a significant effect on early-season survival but not plant height (Table 1). The interaction between population and treatment was not significant for either early-season survival or height (Table 1).
Table 1

Summary of results from likelihood ratio tests of the effects of site, plant population, rhizobia treatment (±rhizobia) and interactions on early-season survival (GLM with binomial link) and mid-season height (ANOVA) in the field

Early-season survival





P > χ2



F value

P > F


4, 293



4, 82




Plant pop

2, 293



2, 80





1, 292



1, 80




Site × Rhiz

4, 286



4, 74




Pop × Rhiz

2, 290



2, 78




Residual df differ for height and early-season survival because fewer plants survived to mid-season

Fig. 1

Estimated mean (±SE, n = 88) of mid-season height for uninoculated (dark bars) and rhizobia inoculated (light bars) plants grown in the field at five sites: the interior of Chamaecrista fasciculata’s range, at the western (W) and northern (N) range edges, and beyond the western and northern range edges. Specific locations are given in Supplementary Figure S1 (Online Resource 1)

The percentage of uninoculated plants that had nodules differed between sites beyond the range and sites within the range (Fisher’s exact test, P = 0.002; result consistent using GLM with binomial link, dev1,44 = 10.1, P = 0.0001). Specifically, 13% of uninoculated plants beyond the range were nodulated, compared to 64% within the range, though this result should be regarded as tentative because of the small sample size due to low survival. Soil nitrogen varied from 0.13 to 0.25% and there was no significant correlation between soil nitrogen and the effect of inoculation (r = −0.67, P = 0.2).

Greenhouse experiment to assess rhizobia quality

Plant height was significantly influenced by rhizobia treatment (Table 2). Rhizobia mixtures from the inocula that formed nodules (one interior site, the northern range edge site and the beyond northern range edge site) all significantly increased plant growth relative to the control treatment, though they did not differ from each other (Fig. 2). Plants inoculated with the two rhizobia mixtures that did not form nodules (one interior site, one western range edge site) did not grow significantly taller than the control plants (Online Resource 1). Plant population also influenced plant height, but there was no significant interaction between plant population and rhizobia treatment (Table 2).
Table 2

Summary of results of split-plot ANOVA testing for effects of rhizobia treatment and plant population on change in plant height and nodule number

Change in plant height

Nodule number




F value

P > F



F value

P > F



















Rhiz × Block





Date planted





Plant × Rhiz









Plant pop














The effects of block and rhizobia treatment were tested against the variance among plots (rhizobia × block interaction), while the remaining terms were tested against the overall variance. The error terms for each level of the split-plot design are designated in italics below their respective terms

Fig. 2

Estimated mean (±SE, n = 239) of change in height since inoculation for plants grown in the greenhouse with or without one of three rhizobia inocula (interior, northern range edge or beyond northern range edge)

The number of nodules per plant differed significantly with rhizobia treatment but not by plant population or the interaction between the two (Table 2). Control plants had zero nodules in 28 of 40 cases and never had more than two nodules, allowing us to compare ecologically relevant differences in infection and benefit, though limiting our ability to detect potential low levels of infection. Plants inoculated with the interior inoculum produced the most nodules (~98 nodules/plant), the inoculum from beyond the northern range edge next (~57 nodules/plant) and the northern edge inoculum fewest of those that produced nodules (~37 nodules/plant) (Online Resource 1).


A mutualism may limit species range expansion if either the availability or quality of mutualistic partners is decreased beyond the range edge. In the C. fasciculata–rhizobia facultative mutualism, we found uninoculated seeds were less likely to form nodules when transplanted beyond the range edge than within the range or at the edge, and that there was a benefit to plant growth from inoculation, indicating that rhizobia availability does limit range expansion. We emphasize that we observed this pattern beyond two distinct range edges, reinforcing the generality of this conclusion. However, we did not find evidence that rhizobia from beyond the range edge were of lower quality in their effect on growth of the legume.

Theoretical work has shown that low densities of compatible rhizobia, and not just the absence thereof, can limit the potential for plants to invade (Parker 2001). Our results support this prediction, as we found a reduction of nodulation beyond the range edge of C. fasciculata. While rhizobia availability was limited, some plants did nodulate and we were able to isolate rhizobia from the nodules collected at the northern site.

The potential to form a symbiosis, and the benefit that each partner gains in symbiosis, depends on both plant and rhizobia genotype (Burdon et al. 1999; Heath 2010; Spoerke et al. 1996). We expected to find differences in rhizobia quality between strains isolated from within C. fasciculata’s range compared to strains from beyond the range because selection between symbionts is no longer maintaining the mutualism. Specifically, given the trade-off between allocation of resources to carbon storage and symbiotic nitrogen fixation that has been shown for rhizobia (Ratcliff et al. 2008), we expected rhizobia from beyond the range edge to be of lower quality to the host. However, we did not detect any differences in the effect on plant growth of rhizobia inocula isolated from sites within C. fasciculata’s range, at the northern range edge and beyond the northern range edge (Fig. 2). One potential explanation is that, by using a mixture of strains isolated from each site, we averaged across genotypes that may differ in quality. Alternatively, these could be strains of rhizobia that are generalists and are maintained in the soil at these sites by alternate hosts, at no cost to symbiosis with C. fasciculata. Finally, these strains may be broadly distributed (e.g., dispersal by wind) and are the same as those within C. fasciculata’s range, though at lower densities where C. fasciculata is not present and thus has not had the opportunity to augment populations of compatible rhizobia via the symbiosis. Regardless of any differences among individual strains, this result suggests that at least at one site beyond C. fasciculata’s range, the average effect of rhizobia (even if not abundant) on host fitness is not strikingly different from that of rhizobia within the species’ range.

In conclusion, we find evidence that the availability of rhizobia in the soil beyond the range edge of the legume host limits plant growth, and thus fitness, but that rhizobia from beyond the range edge are not different in quality to the plant from rhizobia within the range. We emphasize that rhizobia alone do not prevent range expansion, as both survival and plant growth are reduced beyond the range edge even with added rhizobia (Fig. 1; J.S.-G., unpublished data), but that low densities of rhizobia beyond the range edge may interact with other environmental factors to constrain range expansion (e.g., Case et al. 2005). This result suggests that the rate at which legume species can shift their distributions to track climate change (Ackerly 2003; Davis and Shaw 2001) may be limited not only by the rate at which they can disperse, but the rate at which they encounter compatible rhizobia in the soil. More generally, it implies that the role of mutualisms should be further considered in studies of the factors that limit species’ ranges.


We thank Peter Tiffin, William Ratcliff and Ruth Shaw for helpful discussion and comments, Larissa Mottl at Conard Environmental Research Area, Jean Knops at University of Nebraska, Shawn Schottler at St. Croix Watershed Research Area, Bryan Wood at the Audubon Center of the Northwoods for the use of field sites, and Konza Prairie, Tyson Research Station, Conard Environmental Research Area and the Minnesota Department of Natural Resources for seed collection, and two anonymous reviewers for helpful comments. This work was funded by a University of Minnesota Undergraduate Research Opportunity Fellowship to C.A., the Center of Community Genetics Fellowship to J.S.-G. and National Science Foundation DEB 0515466 to Peter Tiffin. All work performed in this study complies with laws of the United States of America. Data accessibility: DRYAD entry: doi:10.5061/dryad.8566.

Supplementary material

442_2011_1958_MOESM1_ESM.doc (66 kb)
Supplementary material 1 (DOC 66 kb)

Copyright information

© Springer-Verlag 2011