Introduction

Prairie reconstruction is one of the few remaining options to increase connectivity among existing fragments of remnant prairie, buffer remnants from disturbance and pesticide drift, and increase native prairie abundance (Gerla and others, 2012), which is critically important to conservation of pollinators and other native prairie biota. Reconstructed prairies also are increasingly viewed as highly valuable to mainstream production agriculture in the Midwest because they offer habitat for pollinators and other beneficial insects that provide valuable services in agroecosystems, such as control of weeds and pest arthropods (Schulte and others, 2017), and highly land-efficient conservation of soil and water. Thus, from the perspectives of nature preservation and agroecology, reconstructions can be of critical importance; however, the biodiversity conservation value of prairie reconstructions depends strongly on their species richness (Nicolson and Wright, 2017). Unfortunately, reconstruction practitioners struggle to reproduce the species richness and ecosystem services provided by large remnant prairies (Larson and others, 2018). An experimental study of eight reconstructed prairies in Minnesota and Iowa determined that one-half of the planted species declined or failed to establish (Larson and others, 2017). More reliable and cost-effective prairie reconstruction methods are needed. Calls to increase habitat for struggling Danaus plexippus (Linnaeus, 1758; monarch butterflies) and declining pollinators have further emphasized the need for improved reconstruction methods.

Arbuscular mycorrhizal fungi (AMF) may play an important role in improving the efficiency of reconstruction methods. AMF are root-associated symbionts of many prairie plant species and have been documented repeatedly to have strongly mutualistic relationships with such species (Middleton and Bever, 2012). In particular, AMF can assist in the development of species-rich native prairie reconstructions, especially if late-successional species are desired (Middleton and Bever, 2012), some of which seem to have obligate relationships with AMF taxa (Koziol and Bever, 2017). In addition, AMF have been determined to increase pollinator visitation to a flowering plant by increasing general robustness of growth and flower production (Wolfe and others, 2005). Unfortunately, intensive land uses such as agriculture that precede reconstruction commonly leave soils in a degraded state (McLauchlan, 2006). AMF in particular have been determined to have different, and depauperate, community composition and lower abundance in cropland than in grassland (Oehl and others, 2003). Degraded soil conditions are commonly highly stable (Suding and others, 2004), and thus, substantial effort may be required to restore key soil attributes for achieving diversity-related reconstruction goals, such as plant-soil mutualisms with native plant species (Asmelash and others, 2016).

Active inoculation of AMF can be useful for establishment of native prairie plants, but outcomes of such inoculation are complex and contingent on many biotic and abiotic variables. Commercial AMF inoculant does not consistently improve plant establishment and growth (Maltz and Treseder, 2015). The primary role of AMF in mutualistic relationships with prairie plants is in enhancing water and nutrient uptake in exchange for sugars; however, the relationship is not always mutually beneficial. Under some conditions, AMF may act parasitically to obtain carbon resources without providing substantial benefit to the plant. This parasitism may, in part, explain the unreliability of commercial AMF inoculants as tools for restoration (Maltz and Treseder, 2015). Moreover, Koziol and Bever (2016) determined that specificity of the AMF-plant mutualism was especially pronounced for late-successional plant species, species that are typically the most difficult to establish and maintain in prairie reconstructions.

The method by which AMF are introduced into a reconstruction also is an important consideration. A metaanalysis that included 28 manipulative field-based trials, two of which were on temperate grasslands, indicated that inoculum from a reference site was as effective as inoculum cultured from individual species, and both were superior to commercial inoculum (Maltz and Treseder, 2015). Locally sourced inoculum also may have the advantage of including AMF taxa that are adapted to local conditions; AMF ecotypes are not necessarily interchangeable in their functioning (Sanders and Rodriguez, 2016). Maltz and Treseder (2015) also cautioned that sourcing inoculum from remnants should be carefully planned so that the remnant is not damaged in the process.

In this study, we evaluated the difference in AMF taxa colonizing roots of plant species collected in high-quality remnant versus reconstructed prairies. Objectives of the study were to complete the following:

Methods

Eight reconstruction sites on National Wildlife Refuge (NWR) system lands in Iowa and Minnesota were selected (fig. 1). The sites were part of a long-term study of effects of seed mix richness and planting method on establishment success of prairie reconstructions (Larson and others, 2011, 2017). Reconstructions were established in 2005, and cover of native and invasive plant species was evaluated in 2005–7, 2010, and 2015 to determine which species seeded into the reconstructions had increased or declined in abundance over time. No AMF amendments were used in these reconstructions.

Results

AMF communities in sampled plant roots differed between prairie site type (that is, reconstruction or remnant; table 2, fig. 2A), and with a few exceptions, this pattern was consistent. Nonetheless, interactions also were detected between site type and plant species (fig. 2B) and between site type and site cluster (fig. 2C). AMF communities in reconstructions were more similar to each other than AMF communities in remnants (fig. 2A for all sampled species, fig. 1.1 for only those species that were in both remnant and reconstructed sites; results were qualitatively similar). This finding was confirmed with a beta dispersion test, which indicated dispersion was significantly different between remnants and reconstructions (F=7.82, DF=1, and P=0.008 with 999 permutations, where F is the test statistic, DF is degrees of freedom, and P is significance) with remnants indicating a higher mean distance to their median (that is, were more dispersed) than the reconstructions (0.65 versus 0.63). For most plant species, AMF communities differed between remnant and reconstruction (table 2, fig. 2B); however, it was not uncommon to find one site in which remnant AMF communities were more similar to reconstruction communities for an individual plant species (for example, Andropogon gerardii [andger], Heliopsis helianthoides [helhel]).

The separation between AMF communities associated with remnants and reconstructions was consistent for most plant species at most site clusters (figs. 2C, 3). Overall, AMF communities differed along axis 1 of the NMDS, with remnant AMF communities on the left-hand side and reconstruction AMF communities on the right-hand side of the graph. The exceptions to this pattern of separation were site cluster A in Minnesota and B in Iowa, where the AMF communities of the remnant site were on the right-hand side of axis 1, indicating that they were similar to those of reconstruction sites (both in that cluster and the other reconstruction sites at the other clusters). In cluster A, only a few of the plant species were similar to species of the reconstruction sites, whereas in cluster B, almost all the plant species of the remnant (Coneflower) harbored AMF communities that resembled those of reconstructions. In none of the site clusters did the reverse happen; no reconstruction sites had AMF communities that resembled those of most of the remnants (that is, were placed on the left-hand side of axis 1).

Contrary to our expectation, AMF community composition was not significantly related to declining or increasing species (determined in our monitoring study; table 3). The clear differences in AMF communities between remnants and reconstructions could be explained to a degree by edaphic soil characteristics (fig. 3B, table 3, appendix 1). The AMF communities of prairie remnants correlated in particular with higher organic matter, total carbon, and total nitrogen, and to a lesser degree with higher levels of ammonium, nitrate, magnesium, and calcium in the soil. AMF communities in reconstructions were particularly correlated with higher phosphorus and manganese levels. The separation between the sites within reconstructions and remnants was strongly related to iron and pH, and to a lesser degree with soil potassium and silt levels, and mean plant height at each site. Interestingly, the Coneflower remnant site had levels of manganese, iron, and phosphorus that were more in line with those of reconstruction sites (Larson and others, 2021 [“2019 prairie remnant and reconstruction soil properties data v1.1” file]).

Discussion

Overall, we determined that AMF communities differed between remnant and reconstructed prairies and that AMF associated with individual plant species also tended to differ, depending on whether the plant species’ roots were collected from remnant or reconstructed prairie. On the other hand, we did not determine that, as a group, plant species in decline in the reconstructions we monitored (Larson and others, 2017) were more likely to harbor different AMF communities compared to other species, perhaps owing to the generally different AMF communities in the two site types: nearly all the species demonstrated these differences. Significant interactions between site type and clusters indicate geographic variation in AMF communities, making the consistent differences between remnants and reconstructions along axis 1 of the NMDS even more striking.

Asmelash and others (2016) reviewed the ways in which AMF can be used to improve ecological restoration of degraded lands. They emphasized the lack of host specificity among AMF because of the relatively small number of AMF taxa compared to the number of plant species that form interactions with them (but see Hausmann and Hawkes [2010] and Werner and Kiers [2015]). Nonetheless, benefits of the mutualism vary among partners (Bever, 2002), over seasons (Bittebiere and others, 2020), and in differing environmental contexts (Revillini and others, 2016). Fungal taxa have been determined to vary in their “carbon sink strength,” such that some taxa are more beneficial for early development of individual plants (requiring little carbon from their hosts) and others for increasing biomass of larger plants that are able to supply more carbon (Martignoni and others, 2021). In the context of prairie reconstruction, an understanding of carbon sink strength, how it varies depending on AMF and host identity, and how to manipulate it to the advantage of the reconstruction may be an important aspect of characterizing functionally and ecologically significant differences in AMF taxa. Taxa beneficial to mature plants may not benefit seedlings. If taxa that benefit mature plants do not benefit seedlings, a single inoculation may not be sufficient to provide benefits over the life of a reconstruction, especially if AMF are unable to disperse naturally to a reconstruction site. All of our samples were collected from roots of mature plants; future work could evaluate effects of AMF collected in this way on seedling establishment and growth.

Only three of the soil characteristics we measured were clearly related to site type, with mean total nitrogen and carbon (and therefore also organic matter) about twice as high in remnants and mean phosphorus nearly six times higher in reconstructions, possibly as a legacy of farming. Excessive phosphorus, in particular, reduces the value of AMF to plants (Revillini and others, 2016). These measures (and several others) also were significantly related to AMF communities in these site types; however, it should be noted that levels of most environmental characteristics varied considerably at the site level, indicating a more complicated relationship with AMF taxa. This relationship is demonstrated by the remnant site in cluster B (Coneflower) whose AMF communities strongly resembled those of reconstructions. At this site, phosphorus, manganese, and iron were elevated relative to the mean levels for reconstructions; however, several reconstruction sites had low levels of the same elements, yet their AMF communities did not resemble those of remnants. The implications of such variation, especially with respect to prairie reconstruction, may not only relate to soil characteristics, but also relate to local adaptation of plants and AMF (Jack and others, 2021), as revealed by a “home and away” study that demonstrated coadaptation of plant genotypes and AMF communities in nutrient limited sites (Johnson and others, 2010). Likewise, the wider soil microbial community can influence the effect of AMF colonization on plant growth (Hetrick and others, 1988, 1990).

Studies in Estonian forest ecosystems have determined that generalist plant species associate with generalist AMF taxa, whereas habitat specialist plants may associate with generalist and specialist AMF (Davison and others, 2011). We determined greater dispersion of AMF communities in remnant prairies than in reconstructed prairies, perhaps reflecting greater abundance of specialist plants in remnants, each interacting with a few specialist AMF taxa. A global-scale metaanalysis of AMF communities associated with plant functional groups (cool-season and warm-season grasses, forbs, and legumes) determined that grasses tended to associate with a greater breadth of AMF taxa than did forbs (Davison and others, 2020); Tipton and others (2019) likewise determined the warm-season grass, Schizachyrium scoparium, to be a generalist with respect to AMF taxa. On average, 10 years after planting, 35–40 percent of total live cover consisted of warm-season grasses and 5–27 percent of total live cover consisted of forbs at the Iowa reconstructions (Drobney and others, 2020) we sampled for this study, but we do not have corresponding plant community data for the remnants. Although inspection of the graphs of AMF communities for individual species does not indicate widespread variation between warm-season grasses and forbs at our study sites, the relative abundance of the grasses compared to forbs may account for some of the overall variation in AMF communities.

We did not find a clear signal in AMF community composition to explain declining frequency or poor establishment in some of the plant species we had monitored in the reconstructions (Larson and others, 2017). Given the myriad circumstances in which plant species and AMF may encounter each other, and the importance of environmental conditions in shaping the interaction, an observational study such as ours may lack sufficient power to detect such effects. Greater benefit is commonly observed for late-successional plants (Middleton and Bever, 2012; Koziol and others, 2021; Cheeke and others, 2022), but the plants we observed to decline do not fall neatly into this category. Other members of the microbial community, such as rhizobia (Bauer and others, 2012), also may play a role in successful establishment of some species. As noted previously, the dominant difference demonstrated in the NMDS was between remnants and reconstructions, so differences between declining and increasing species may have been masked by the much larger magnitude of the effects of site type. In addition, we did not evaluate the contribution of individual AMF taxa, which may have had important associations for which we did not account.

Soil disturbance, such as by plowing, has been determined to change the composition and function of AMF in grasslands (Schnoor and others, 2011). More intensive land management was observed to result in a similar number of AMF species but different species composition (Börstler and others, 2006). Our results support these observations in that roots from plants in reconstructions, where the soil ecosystem had been subjected to disturbance for row-crop agriculture, had a different AMF community composition than did those in remnants, where soil had never been plowed. Dispersal of AMF is poorly understood (Paz and others, 2021), so the likelihood of their unassisted arrival in new reconstructions is unclear. Oehl and others (2003) determined declining numbers and diversity of AMF in a disturbance gradient ranging from lightly managed grassland to permanent row-crop agriculture. It must be kept in mind, however, that grasslands are disturbance-dependent ecosystems that cannot persist without fire, grazing, or their anthropogenic analogs (Perkins and others, 2019). We have no data on the AMF taxa present before reconstruction, but if the sites were as depauperate as the row-crop sites in the study completed by Oehl and others (2003), our sites experienced substantial colonization by AMF in the decade and a half since reconstruction began. Moreover, these reconstructions were all established by seed, so the planting itself was unlikely to introduce AMF.

The geographic variation we saw in AMF community composition, combined with the differences between reconstructed and remnant prairies, indicates the possibility of short-distance dispersal of AMF, perhaps in a random fashion within the area of an agricultural field, with selection over time by plant species acting as filters that winnow out AMF that provide fewer benefits (Davison and others, 2011). In this regard, the similarity between reconstruction and remnant sites in cluster B in Iowa may be instructive. Although the site labeled a remnant in cluster B may not have been plowed, one of us (P. Drobney) was present when the site was first acquired by the refuge. Glacial till was evident where topsoil was lacking, and the plant community was depauperate. Over the years, the site was overseeded with a few conservative native plant species not originally present and Viola pedatifida was planted as plugs (K. Viste-Sparkman, U.S. Fish and Wildlife Service, unpub. data, 2022). The site’s originally depauperate state may have never supported the diversity of AMF we detected at other sites, and surrounding agriculture may have reduced opportunities for local dispersal.

Conclusion

Results of this study have provided insight into variation in AMF communities in reconstructed versus remnant prairies in Minnesota and Iowa. Consistently across this geographic range, remnant- and reconstruction-based AMF communities remained largely distinct from each other. A recent metaanalysis (Neuenkamp and others, 2019) revealed that inoculation with AMF promoted native and specialist plant species in restorations, which thereby became more similar to reference sites; the effect was especially pronounced in the most disturbed sites. In addition, Koziol and others (2021), in a restoration experiment that spanned grasslands in three States, detected greater resistance to exotic invasion in sites where AMF-inoculated late-successional nurse plants had been introduced. High-quality pollinator habitat requires the establishment of species-rich plant communities, which include late-successional and hard to establish forbs and resistance to exotic infestation. We have determined that even more than 10 years after planting, prairie reconstructions have AMF communities distinct from AMF communities in remnants. These reconstructions have been invaded by exotic cool-season grasses, while losing the native cool-season grasses that might compete with them (Larson and others, 2017). Careful application of AMF inocula may not only reduce the differences in AMF communities between reconstructions and remnants but also may improve multifunctionality and sustainability of increasingly rare prairies and the fauna they support. The differences we have detected indicate that further research to discern the effects of AMF derived from reconstructions versus remnants on plant performance is warranted.