The environmental impacts of dams on forest-river processes have been well documented (Hall et al., 2011; Reidy Liermann et al., 2012) and include alterations to aquatic animal populations, riparian vegetation, hydrology, and river deltas (Li et al., 1987; Jansson et al., 2000; Sharma, 2001; Perry et al., 2016). Damming rivers in the Pacific Northwest (PNW) obstructs many historic anadromous fish migration routes, leading to major species declines (Duda et al., 2008). Riparian and upland forests are inundated through the creation of reservoirs that impede downstream flow and sediment exchange and cause deeply channelized riverbeds that become disconnected from floodplains (Shafroth, 1999; Rood and Mahoney, 2000). Further, dams decrease the delivery of large woody debris from forests to riverine systems, which negatively affects floodplain, delta, and instream habitat complexity, and the nearshore environment (Andersson et al., 2000; Gregory et al., 2002). Inhibiting the passage of anadromous fish affects the overall health of riparian ecosystems by limiting the transfer of marine-derived nutrients to adjacent forests (Helfield and Naiman, 2001; Drake et al., 2002).

Until recently, much of the Elwha River Basin was inaccessible to sea-run fish due to the Elwha and Glines Canyon dams. These dams blocked access to 90% of the spawning habitat, impeded sediment transfer to the nearshore, and degraded stream and riparian habitat (Pess et al., 2008; Shaffer et al., 2008; Duda et al., 2011). Despite the environmental impact, the two dams operated with little regulation for over 50 years until the Federal Power Act (late 1960s–1970s) required the owner to license the dams with the Federal Energy Regulatory Commission (Sadin and Vogel, 2011). This required maintenance for structural integrity and construction of fish passageways (Service, 2011). The Elwha River Ecosystem and Fisheries Restoration Act was enacted in 1992, requiring dam removal to restore riverine habitats, native salmon runs, and other ecosystem processes (DOI (U.S. Department of Interior) et al., 1996; Winter and Crain, 2008). The costs/benefit analysis, combined with the potential to restore the Lower Elwha Klallam Tribe’s salmon runs, began to sway public and political opinion in favor of dam removal (Gregory et al., 2002; Duda et al., 2008).

The deconstruction of both dams and lake drawdown was completed by 2014, restoring access to the upper river and marking the largest dam removal project to date in the United States. Given the unprecedented nature of the large-scale dam removal, there was considerable uncertainty about how the biota would respond to the dewatering of the two reservoirs, which were estimated to store ~21,000,000 m³ of sediment prior to removal (Warrick et al., 2015; Major et al., 2017). A major concern was the impact on terrestrial and aquatic habitats of this increased sediment load (Pizzuto, 2002; Stanley and Doyle, 2003). During drawdown, coarse-grained terraces approximately 3.3 to 7 m thick formed along the Lake Mills bed above the Glines Canyon dam (East et al., 2015). These terraces, composed of unconsolidated sand, gravel, and cobble, resulted in novel landforms perched above the hyporheic zone and situated between the riparian and forest corridors along the Elwha River (Bauman and Kardouni, 2018).

Revegetating these landforms, a fundamental goal of the Elwha River and Ecosystem Restoration Project, required the restoration of river-forest processes and native communities while limiting the establishment of non-native species (DOI National Park Service, 2015). Revegetation plans were designed to promote rapid succession to native forests, thus limiting erosion, regulating water temperature, and providing critical habitats (Chenoweth et al., 2011). In the Lake Mills Basin of the Elwha River, 44.0 ha of newly exposed sediments were actively revegetated from 2013 to 2015. This included broadcasting 2,800 kg of native seed and planting 205,000 herbaceous and woody plants representing 64 species (Chenoweth et al., 2022). Conifers such as Douglas fir (Pseudotsuga menziesii Mirb Franco), grand fir (Abies grandis [Douglas ex D. Don] Lindl.), and western white pine (Pinus monticola Douglas ex D. Don) were incorporated into the planting plan and are considered important indicators of forest recovery, as well as iconic species to the PNW forests (Franklin, 1988; Shafroth et al., 2002).

One seeded forb species, native riverbank lupine (Lupinus rivularis Douglas ex Lindl.), was included in the seed mix and thrived in the coarse sediments along the Lake Mills terraces (Chenoweth et al., 2022). As a legume, it assimilates nitrogen (N) through the endosymbiotic association with Rhizobium, a N-fixing bacterium (Staniewski, 1970). Myrold and Huss-Danell (2003) demonstrated that N-fixing species such as lupine can enhance N availability and organic matter in nutrient-limited soils (Halvorson et al., 1991). Riverbank lupine is an annual, biennial, or short-lived perennial that can grow up to 150 cm in height and acts as a pioneer species adapted to sandy or gravelly soils (Darris and Young-Mathews, 2012). The 2-to-3-year life cycle of riverbank lupine starts with a vegetative phase followed by seeding and senescence with subsequent grow-back periods by seeded progeny.

One of the most well-studied examples of lupine colonization was in plant successional studies conducted after the eruption of Mount Saint Helens in the Pacific Northwest region of the United States. Lupinus lepidus Douglas ex Lindl. was the first colonizing herbaceous plant and formed dense to patchy colonies within a decade after the eruption (del Moral et al., 1995). Its presence was shown to have both facilitative and inhibitory effects on invading plant species, where living colonies inhibited ruderal species, while dead lupine patches facilitated recruitment (Morris and Wood, 1989). Elsewhere at Mount Saint Helens, del Moral and Rozzell (2005) demonstrated that L. lepidus inhibited colonization of forbs in mature lupine colonies. However, older colonies demonstrated a subsequent increase in plant richness and diversity. Further, varying levels of lupine allowed different sets of species to be established, suggesting that variation in lupine abundance may lead to different successional trajectories (del Moral and Rozzell, 2005; del Moral, 2007). As lupine densities gradually decrease, it is hypothesized that their legacy will contribute to soil development and ultimately accelerate plant succession (Vitousek et al., 1987; del Moral, 2007). Therefore, developing our understanding of lupine during primary succession may be an important component of forest restoration projects given its adaptability, ability to facilitate the establishment of neighboring plants, and N-fixing capability (Bishop, 2002).

The planting of native tree seedlings with and into herbaceous cover is common in restoration; however, not all N-fixing herbaceous species facilitate the establishment of neighboring trees (Weidlich et al., 2020). Previous studies utilizing various Lupinus species have illustrated positive tree growth response and increased N acquisition (Prietzel et al., 2008; Mauer et al., 2013; Oldřich et al., 2013). Other projects have also shown that neighboring vegetation acts as “nurse plants” that moderate soil temperatures, increase water and nutrient availability, and promote microbial interactions (Bertness and Callaway, 1994; Raffaele and Veblen, 1998; Flores and Jurado, 2003). L. lepidus has also been reported to stimulate microbial activity in primary successional soils at Mount Saint Helens (Halvorson et al., 1991). Ectomycorrhizal fungi are key components of PNW forest soils, forming symbioses with conifers (Kranabetter et al., 2015). ECM fungi perform a variety of functions that contribute to the establishment, survivability, and resilience of planted seedlings (Massicotte et al., 1999; Smith and Read, 2008). Mature soils typically support diverse and abundant fungal communities essential to forest health (Balestrini et al., 2016). However, dam removal and deposition of coarser sediments may not be conducive to the formation of these symbioses, thus limiting tree establishment after dam removal (Cortese and Bunn, 2017). Therefore, lupine’s presence may facilitate soil development to promote ECM colonization of conifer roots, thereby supporting plant establishment and native plant community recovery.

In this study, we explored how varying densities of riverbank lupine affected the performance of planted conifers (Douglas fir, grand fir, and western white pine), ECM root colonization, soil N and organic matter (OM), and neighboring plant communities in coarse substrates on the exposed terraces of Lake Mills along the Elwha River. We hypothesized that greater lupine densities would result in greater conifer growth, foliar N concentrations, and greater ECM taxonomic richness and colonization. We also hypothesized that greater lupine densities would yield greater C:N, OM, and N content in coarse sediment. Finally, we hypothesized that plant species richness would be lower and community composition would differ in areas of greater vs. lesser lupine density. Documenting species interactions that accelerate succession by aiding the growth of conifers and the development of native plant communities and soils may help future projects that will have to manage the revegetation of coarse sediments left behind after dam removal.

We documented lupine’s ability to thrive in highly disturbed landscapes without reducing the growth of planted conifers. A direct benefit included an increase in foliar N concentrations among seedlings growing in the medium and dense lupine cover classes. An unexpected finding was the reduction of ECM on conifer roots in medium and dense lupine plots. In addition, areas of dense lupine cover had fewer, and a different composition of, species than did areas of sparse lupine cover.

There was no difference in tree height or diameter among the three lupine cover classes. This differed from Mauer et al. (2013), who demonstrated Norway spruce (Picea abies), European beech (Fagus sylvatica), and Scots pine (Pinus sylvestris) to have an increased height and collar diameter when sowed among three species of lupine (Lupinus angustifolius, Lupinus albus, and Lupinus polyphyllus) after a 5-year growing period. In our study, greater foliar N was correlated with greater cover of riverbank lupine—at concentrations well above those deemed adequate for conifers in the PNW (N > 1.25%; Radwan and Brix, 1986; Walker and Gessel, 1990). Scots pines also showed an increase in foliar N levels during the first decade of growth when lupine was present (Prietzel et al., 2008). N-fixation by both Alnus incana and Lupinus nootkatensis was incorporated into the foliage of birch trees after 20 years on degraded forest soils (Myrold and Huss-Danell, 2003).

We also demonstrated a significant reduction of ECM colonization on conifer roots in the medium- and dense-cover plots. One possible explanation for the decrease in ECM colonization is that conifers may reduce carbon allocation to their ECM symbiont when N is not limiting (Corrêa et al., 2008). Our results support other research illustrating that carbon transfer from the host plant to the fungal symbiont depends on the N status of the tree (Nilsson et al., 2005). Field studies have documented that nitrogen fertilization can directly reduce ECM growth and colonization rates on longleaf pine (Pinus palustris; Sims et al., 2007). Ostonen et al. (2011) demonstrated that as N deposition loads increased, ECM biomass decreased in stands of Norwegian spruce (P. abies). This phenomenon was also observed by Nilsson and Wallander (2003) where Norwegian spruce under N additions resulted in a 50% ECM reduction when compared to trees within non-fertilized plots. This finding supports hypotheses that speculate mycorrhizal relationships range from mutualistic to parasitic, depending on soil nutrient status (Johnson, 1993; Johnson et al., 1997). Therefore, we propose a new hypothesis that investing carbon in non-photosynthetic symbionts is more costly than acquiring N from neighboring N-fixers.

We found less ECM species richness on novel terraces. Of the taxa identified, T. terrestris and W. mikolae were the most abundant. Both of these species are noted as among the most ubiquitous greenhouse and field ectomycorrhizas symbiotic with grand fir, Douglas fir, and western white pine (Massicotte et al., 1999; Hilszczańska and Sierota, 2013). Rare species included S. luteus, T. pacificum, and a Rhizopogon sp., all considered prevalent in PNW forests (Dahlberg, 2001). In addition, the average root colonization was rather low (averaged 30% ECM root tips). Soil organic matter is important for mycorrhizal symbioses, and studies have shown that mycorrhizal infection increases with organic matter amendments (Baar and DeVries, 1995; Lunt and Hedger, 2003). This low rate of colonization and small species list likely reflect a soil environment insufficient to support functional mycorrhizas (Cortese and Bunn, 2017).

As nitrogen fixers, Lupinus species develop N pools via atmospheric N fixation, which become available to plants via the mineralization of organic N into ammonium (Vitousek et al., 2002; Robertson and Groffman, 2007). While this study illustrated lupine’s ability to influence bioavailable N to neighboring conifers, we did not see differences in total soil N. In addition, we observed no differences when C:N and OM were compared among the lupine cover classes. The inability to detect differences in soil N may have been due to immediate bio-assimilation by conifers and neighboring vegetation coupled with N leaching in sediments with low water/nutrient holding capacity. We acknowledge that the measure of soil N was a single point-in-time measurement of total N, which could explain the lack of difference in soil N among lupine cover classes. Future work would benefit from the use of ion exchange resins, which integrate N availability over time (Qian and Schoenau, 2002). We noted that lupine litter-fall was abundant on the soil surface 4 years after seeding/planting; additional time may be needed to observe changes in the soil profile (Vitousek et al., 2002). In comparison, Halvorson et al. (1992) and Halvorson and Smith (1995) measured significant organic matter production and total increased N near the lupine’s (L. lepidus and Lupinus latifolius) rhizosphere, 7 years after the Mount Saint Helens eruption.

With regard to the vegetation community, we found that species richness decreased in plots with abundant lupine. This was also documented following the Mount Saint Helens eruption; Morris and Wood (1989) demonstrated that Pacific lupine (L. lepidus) inhibited immediate colonization of forbs in the early years, while later studies demonstrated a subsequent increase in plant community species richness that was accelerated by the presence of lupine (del Moral, 2007). We also found compositional differences in plots with dense vs. sparse lupine cover, suggesting that lupine may influence the trajectory of community development during primary succession (del Moral and Rozzell, 2005). The increase observed in plant species in the sparse lupine plots included a few rare native and non-native plants; however, the non-native species count was not different among the cover classes, and neither species was in high abundance or a significant indicator of sparse cover class. Chenoweth et al. (2022) found that seeding herbaceous species (including lupine) on these coarse terraces deterred non-native species establishment. This is of particular importance in coarse, nutrient-poor substrates with the lack of canopy closure that creates an opportunity for the invasion of exotic ruderal (Shafroth et al., 2002; Tabacchi et al., 2005; Funk and Vitousek, 2007; Michel et al., 2011). Important to the overarching goal of accelerating forest succession, lupine’s potential role as a beneficial cover crop may aid in the establishment of planted conifers while deterring non-native plant communities (Fierke and Kauffman, 2006; Peltzer et al., 2009; Urgenson et al., 2009).

The damming and the subsequent dewatering of Lake Mills created a phenomenal disturbance, leaving novel terraces void of ecological legacies. Similar to the post-eruption landforms colonized by lupine at Mount Saint Helens, seeded lupine in the Elwha River Valley was able to establish on the barren substrates. Important to restoration projects, lupine’s cover did not impede conifer growth in the early years of planting. We acknowledge that we used natural variation in lupine density as a predictor of conifer performance and soil development; therefore, lupine density itself could be a response to variation in soils, vegetation, or other environmental features. Spatial variation in lupine abundance could be caused by or otherwise confounded with other environmental conditions that were not measured. Regardless, when comparing conifer growth within dense lupine cover, lupine canopies were not found to be detrimental.

Further, lupine’s contribution to bio-available N promoted adequate foliar nitrogen concentrations that may contribute to greater growth in the subsequent years. It is not clear whether this increase in available N was the mechanism behind the inverse relationship with lupine and ECM root colonization; however, current research is further investigating this. This may be an explanation of tradeoffs: ECM fungi aid in nutrient uptake/growth in low lupine density cover, but N-fixation by lupine serves this function at higher densities.

This study also illustrates the challenges for revegetation in coarse sediments and the important function of a cover crop on sediment surfaces. Lupine OM contributions are an important attribute for the successional trajectory, given its biennial life cycle and ability to tolerate resource-limited soils. Although lupine did not contribute to an increase in soil carbon or OM, we hypothesize that lupine will contribute important carbon pools to the forest soil-building process after a decade of recovery, which may also lead to an increase in ECM fungal colonization and native plant recruitment. Importantly, this study also documented the benefits of including a locally native N-fixing species (riverbank lupine) in the Elwha River Restoration project, which can inform future practitioners when considering strategies of revegetation after dam removal. Based on our findings and corroborative literature, we recommend the inclusion of native lupine in highly disturbed, N-limiting soils to contribute to the N pools for native tree establishment. These attributes may accelerate forest succession leading to closed canopies much faster than passive recovery on coarse soils.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

JK: literature review, study design, data collection and analysis, and synthesis. MDL: DNA extraction and PCR. AL: DNA extraction and PCR, and literature review. JB: literature review, study design, data collection and analysis, synthesis, and project oversite. All authors contributed to the article and approved the submitted version.