Breeding Alfalfa for Semiarid Regions in the Northern Great Plains: History and Additional Genetic Evaluations of Novel Germplasm

: Yellow-ﬂowered alfalfa ( Medicago sativa subsp. falcata ) (also known as sickle medic) has been the cornerstone for breeding alfalfa for dual grazing and hay production in the semiarid regions of the northern Great Plains in the US and Canada. Most, if not all, of the cultivars developed for the northern Great Plains during the 20th century, had parentage tracing back to introductions by Niels Ebbesen Hansen that were obtained from expeditions to Russia, primarily the province of Siberia, on behalf of the United States Department of Agriculture during the early 1900s. The M. falcata genome contains alleles for high levels of drought-tolerance, winter hardiness, and tolerance to grazing, but is generally deﬁcient for commercial seed production traits, such as non-shatter, compared with common alfalfa ( M. sativa ). A naturalized population, tracing to USDA plant introductions to Perkins County South Dakota by N.E. Hansen in early 1900, and subsequently, facilitated by the determined seed increase and interseeding of a population by a local rancher, Norman ‘Bud’ Smith, has shown highly desirable in situ characteristics for improving rangelands in the northern Great Plains. This includes adequate seed production to build a seed bank in the soil for natural seedling recruitment and population maintenance / expansion and support the production of a commercial seed source. This review documents the seminal events in the development of cultivars to date and describes novel germplasm with potential for new cultivars in the future.


History
Breeding alfalfa (Medicago sativa L.) for semiarid regions of the northern Great Plains, predominantly North Dakota (ND), South Dakota (SD), southern Saskatchewan and southern Alberta west of the 100th meridian and the eastern half of Montana, arguably began with the first intentional introduction of yellow-flowered alfalfa (Medicago sativa subsp. falcata (L.) Arcang.) (hereafter M. falcata) for forage purposes in cold and dry environments by Niels Ebbesen Hansen (hereafter N.E Hansen) to SD in leafhopper yellowing. Results revealed extensive among and within cultivar/population/strain/PI variation for morphological, physiological, and forage quality traits useful in breeding programs for grassland improvement.

Additional Genetic Evaluations
Variation for desirable traits for breeding new cultivars of alfalfa for hay production and grazing systems in the semiarid northern Great Plains [45] were evident in situ in the GRNG population [24] ( Figure 1). In addition, this population demonstrated a capacity for recruitment from reseeding under grazing, putatively facilitated by sporadic production of large crops of hard seed, which composed nearly 100% of the alfalfa soil seed bank [24]. To our knowledge, these critically important traits have not been reported for any other alfalfa germplasms [23]. However, Rumbaugh [32] observed reseeding enough to maintain plant density in pure stands of alfalfa over a 25-year period in UT, where plants were grazed early spring and seeds were produced from regrowth. However, seed production by the GRNG population in northwestern SD is more episodic, depending on when the annual grazing period occurs and the intensity of insect feeding that destroys floral meristems (alfalfa weevil (Hypera postica (Gyllenhall))) or developing seeds (alfalfa seed chalcid (Bruchophagus roddi Guss.)).
Agronomy 2020, 10, x FOR PEER REVIEW 5 of 18 variation for morphological, physiological, and forage quality traits useful in breeding programs for grassland improvement.

Additional Genetic Evaluations
Variation for desirable traits for breeding new cultivars of alfalfa for hay production and grazing systems in the semiarid northern Great Plains [45] were evident in situ in the GRNG population [24] ( Figure 1). In addition, this population demonstrated a capacity for recruitment from reseeding under grazing, putatively facilitated by sporadic production of large crops of hard seed, which composed nearly 100% of the alfalfa soil seed bank [24]. To our knowledge, these critically important traits have not been reported for any other alfalfa germplasms [23]. However, Rumbaugh [32] observed reseeding enough to maintain plant density in pure stands of alfalfa over a 25-year period in UT, where plants were grazed early spring and seeds were produced from regrowth. However, seed production by the GRNG population in northwestern SD is more episodic, depending on when the annual grazing period occurs and the intensity of insect feeding that destroys floral meristems (alfalfa weevil (Hypera postica)) or developing seeds (alfalfa seed chalcid (Bruchophagus roddi Guss.)). Another important attribute of the GRNG population is the capacity to establish naturally within stands of crested wheatgrass (Agropyron cristatum (L.) Gaertn.). This introduced cool-season grass was widely planted in the semiarid northern Great Plains to revegetate and permanently stabilize abandoned cropland as directed by federal programs, such as the Bankhead-Jones Farm Tenet Act of 1937, prior to the formation of the National Grassland System [24]. Both species are highly competitive and relatively impervious to the introduction and long-term establishment of other plant species in semiarid areas [45,46]. About eight million hectares of crested wheatgrass, much in monoculture, occur in the US. The benefits of incorporating alfalfa include economic and environmental parameters [46] such as increased forage production, improved forage quality, and enhanced wildlife habitat. Another important attribute of the GRNG population is the capacity to establish naturally within stands of crested wheatgrass (Agropyron cristatum (L.) Gaertn.). This introduced cool-season grass was widely planted in the semiarid northern Great Plains to revegetate and permanently stabilize abandoned cropland as directed by federal programs, such as the Bankhead-Jones Farm Tenet Act of 1937, prior to the formation of the National Grassland System [24]. This species is highly competitive and relatively impervious to the introduction and long-term establishment of other plant species in semiarid areas [45,46]. About eight million hectares of crested wheatgrass, much in monoculture, occur in the US. The benefits of incorporating alfalfa include economic and environmental parameters [46] such as increased forage production, improved forage quality, and enhanced wildlife habitat.
Mortenson et al. [47] reported relatively uniform alfalfa plant densities of 5.5 plants m −2 and 4.4 plants m −2 in interspaces between interseeded rows (152-cm row spacing) in 2004 from interseedings in mixed-grass prairie made by Norman Smith [23] in 1965 and 1987, respectively. This suggested the alfalfa populations had reached equilibrium, through natural reseeding, with the native grass-dominated plant community under the grazing management system used on the Norman Smith Ranch for several decades. Forage production in interspace areas containing alfalfa from natural reseeding was 67% greater in the 1965 interseeding and 240% greater in the 1987 interseeding than in the control areas. In the 1998 interseeding, forage production within 1 m 2 spanning the interseeded row was 42% greater than the adjacent area, where alfalfa had not yet reseeded. The alfalfa plant population density in planted rows in the 1998 interseeding was 2.2 plants m −2 in 2004. They also reported increases in soil nitrogen and soil organic carbon, and suggested climate change mitigation aspects, on the Norman Smith Ranch in association with inter-seeding adapted yellow-flowered alfalfa [48].
Therefore, to establish a base population for a breeding program, the seeds were collected and kept separate from about 230 individual plants across about 5 km 2 on the GRNG in 2000. The amounts of seeds collected ranged from <0.25 g to >3.5 g plant −1 . From that seed collection, half-sib family nurseries were established in 2001 at Highmore, SD (SDSU), Mandan, ND (USDA-ARS, Northern Plains Research Center), and Ames, IA (Iowa State University) and at Millville, UT (USDA-ARS, Forage and Range Research Center) in 2003. Data collected in subsequent years included forage production, growth habit, flower color, pod shape, and potato leafhopper yellowing. This was a collaborative USDA NIFA Multi-State Research Objective (NE 1010 Breeding and Genetics of Forage Crops to Improve Productivity, Quality, and Industrial Uses).
Presently, the GRNG population is also proving useful for finding genetic sources of drought-resistance [49] and freezing tolerance [50] in alfalfa. Plants grown from seeds collected on the GRNG maintained root growth, relative water content, and chlorophyll content, with increasing water use efficiency, as deficit became severe, better than 10 other alfalfa accessions.
Oakley and Garver [1] determined that forage production of M. falcata was greater in hills and space-planted stands compared with broadcast or seeded stands, presumably due to plasticity in crown diameter and shoot density traits. They pointed out in its natural habitats in Europe and Asia, plants of M. falcata were generally widely spaced, relative to populations of M. sativa. Research is needed to elucidate the relationship between plant population (i.e., inter-plant spacing) and forage production. A recent study of interseeding alfalfa in semiarid native grassland in Texas revealed an advantage of wide row (71 cm) spacing compared with narrow row (36 cm) spacing for water use and forage production and quality [51].
Our objectives here are to report results for: (1) The Multistate Research half-sib family evaluation trials for adaptation, forage production, pest resistance, and morphological traits at multiple locations (Highmore, SD, Mandan, ND, Millville, UT, and Ames, IA), and three ancillary experiments at Brookings, SD. The three ancillary experiments: (1) Compared morphological characteristics of plants derived from five bulk-seed samples collected from different topographical subdivisions of the GRNG, (2) compared eight cultivars/strains/experimental populations of predominantly M. falcata, including populations from divergent phenotypic selection for pod shape, and four standard hay-type and pasture-type cultivars for three years in a delayed single-harvest forage management system, and (3) evaluated forage production of 'Sholty' M. falcata with an M. media collection from rangeland in western SD, at low plant population densities. All three experiments were conducted in transplant nurseries at Brookings, SD. Each investigator/location used standard forage breeding spaced-plant nursery designs appropriate for their specific environments. For example, at Highmore, SD, the alfalfa nursery was broadcast-seeded with crested wheatgrass after transplanting in 2001 and again in April 2002 to provide strong interspecific competition typical of what would occur in inter-seeded situations in rangelands in the northern Great Plains (Figure 2). At the other locations, transplanting was into tilled soil, with weed control by tillage, hand, or herbicides. Each investigator/location used standard forage breeding spaced-plant nursery designs appropriate for their specific environments. For example, at Highmore, SD, the alfalfa nursery was broadcastseeded with crested wheatgrass after transplanting in 2001 and again in April 2002 to provide strong interspecific competition typical of what would occur in inter-seeded situations in rangelands in the northern Great Plains (Figure 2). At the other locations, transplanting was into tilled soil, with weed control by tillage, hand, or herbicides. At Highmore, the experimental design was two replications of 224 half-sib families in a randomized complete block design. Smith's falcata, '5454', and 'Vernal' were included as checks. Each family row was composed of seven plants with 0.3-m interplant spacing. Alley and inter-row spacings were 0.9 m. Soil type was a Glenham (fine-loamy, mixed, superactive, mesic Typic Argiustolls)-Java (fine-loamy, mixed, superactive, mesic Entic Haplustolls)-Prosper (fine-loamy, mixed, superactive, mesic Pachic Argiustolls) loam. Plots were harvested at 10-cm stubble height for dry matter forage production at late bloom to well-developed pod stage during late July for each of 2002 and 2005. During late July in each of 2003 and 2004, the standing crop was swathed and removed from the experimental area. This system provided estimates of forage production at early-and fullyestablished (alfalfa and over-sown desert wheatgrass) stages. Data for growth habits (scale was 1 = prostrate, 2 = semisprawling, 3 = bowl-shaped, 4 = fully erect) was devised to coincide with Sinskaya's [41] ecoelement designations based on heritable morphological variation for growth habits in M. falcata in Russia. Variation for flower color was described as the frequency of families that occurred for each of 11 color class combinations (i.e., monochrome purple, variegated, monochrome cream, and monochrome yellow and their two-way and three-way combinations). Data for growth habits and flower color were collected at full to late bloom on 25 July 2005. At Highmore, the experimental design was two replications of 224 half-sib families in a randomized complete block design. Smith's falcata, '5454', and 'Vernal' were included as checks. Each family row was composed of seven plants with 0.3-m interplant spacing. Alley and inter-row spacings were 0.9 m. Soil type was a Glenham (fine-loamy, mixed, superactive, mesic Typic Argiustolls)-Java (fine-loamy, mixed, superactive, mesic Entic Haplustolls)-Prosper (fine-loamy, mixed, superactive, mesic Pachic Argiustolls) loam. Plots were harvested at 10-cm stubble height for dry matter forage production at late bloom to well-developed pod stage during late July for each of 2002 and 2005. During late July in each of 2003 and 2004, the standing crop was swathed and removed from the experimental area. This system provided estimates of forage production at early-and fully-established (alfalfa and over-sown crested wheatgrass) stages. Data for growth habits (scale was 1 = prostrate, 2 = semisprawling, 3 = bowl-shaped, 4 = fully erect) was devised to coincide with Sinskaya's [41] ecoelement designations based on heritable morphological variation for growth habits in M. falcata in Russia. Variation for flower color was described as the frequency of families that occurred for each of 11 color class combinations (i.e., monochrome purple, variegated, monochrome cream, and monochrome yellow and their two-way and three-way combinations). Data for growth habits and flower color were collected at full to late bloom on 25 July 2005.
At Mandan, ND, two replications of single-row plots of 128 families were transplanted in a randomized complete block design on 12 August 2002. Each family row had eight plants with 0.6-m interplant spacing. Inter-row spacing was 1.8 m. Soil type was a Parshall (coarse-loamy, mixed, superlative, frigid, Pachic Haplustolls)-Manning (coarse-loamy, mixed, superactive, frigid Typic Haplustolls) sandy loam. Plots were harvested for dry matter forage production on 12 August 2003 and 2 August 2004, when plants were in late bloom to well-developed pod stages. Growth habits were determined on an individual plant basis using the same scale as for Highmore. Flower color was defined using a 1 = purple, 2 = variegated, 3 = cream, 4 = yellow, 5 = white scale developed by Barnes [52]. Data for growth habits and flower color were collected on 20 June 2003. Mean pod shape, determined using a graduated scale from 1 = sickle to 3 = two complete coils based on Sinskaya [41], was estimated for a sample of 3 random plants within each family row on 2 August 2004. The number of plants in row −1 was also determined at that time. Individual plant data for flower color and pod shape were subjected to analyses of variance to partition the total variance into within and among family components and generate intraclass correlation coefficients.
At Millville, UT, two replications of 84 half-sib families were transplanted in 10-plant single-row plots with 0.5-m intra-row and 1.0-m inter-row spacing on 13 May 2003. The soil at the Evans Farm is a Nibley silty clay loam (fine, mixed, mesic, Aquic Argiustolls). Area annual precipitation averages 472 mm (1980-2011), with July and August being the driest and hottest months (http: //www.usclimatedata.com, 8 June 2020). Plots were harvested for dry matter forage production on 17 June and 5 August 2004 and 1 July and 19 September 2005. Data for growth habits were as described for Highmore [41]. Flower color was determined using the same scale used at Mandan, ND, but was based on the most frequently occurring color in a 10-plant family plot.
At Ames, IA, two replications in a randomized complete block design of seedlings of 212 half-sib families, Smith's falcata, 'Sholty', Vernal, and 5454 were transplanted during mid-May 2001 in single-row plots with 1 to 8 plants plot −1 (mean was 6.5 plants plot −1 ) with 0.3-m intra-and 0.9-m inter-plot spacing. The soil type was a Nicollet loam (fine-loamy, mixed, superactive, mesic Aquic Hapludolls). Data were collected for growth habits and potato leafhopper yellowing (PLY) tolerance on 14 July 2003 and for regrowth (1 = low to 5 = high) on 4 August 2004 for forage accumulated since harvest at mid-bloom during June. The primary trait of interest was tolerance to potato leafhopper yellowing. The expectation was that stockpiling forage until mid-July in south-central Iowa would provide an ideal situation for the development of heavy leafhopper infestation.
Analyses of variance were used to partition phenotypic variation into sources associated with fixed (year, entry, and year × entry) effects for Highmore and Mandan for dry matter forage production, growth habit, and pod shape. For Millville, the same analyses were conducted for dry matter forage production and growth habit.
The number of families and plants within families evaluated at each location varied by amounts of seeds available and local limitations for field plot area and support staff. Traits measured at each location were based on the preferences of individual investigators and time limitations. Ultimately, the dataset was intended to provide the value of diverse environments for measuring the plasticity of phenotypic expression and appropriateness to target environments in the semiarid northern Great Plains and perhaps beyond.  Table 1). The cessation of top growth during severe drought by M. falcata is an adaptation whereby plants devote energy to the production of rhizomes for resumption of growth when conditions become favorable [1].
Check cultivars at each location were 'Vernal', '5454', and 'Smith's falcata'. At Highmore, Smith's falcata ranked 52, whereas Vernal and 5454 ranked higher than 150 out of 224 entries. At Mandan, Smith's falcata ranked 33, Vernal ranked 75, and 5454 ranked 112 out of 128 entries. At Millville, UT in the Intermountain West, dry matter forage production from a two-harvest system was about four times that of a delayed single harvest in the northern Great Plains. The first harvest was taken at early bloom near peak standing crop. The checks at Millville were hay-type . These data pointed out the superiority of numerous GRNG families, compared with conventional hay-type cultivars, for forage production from initial spring growth. However, the data also showed that several families exhibited regrowth potential comparable to the hay-types.
Since only two replicates of single rows of spaced plants were established at each location, caution should certainly be taken when declaring differences among families as a basis for selecting for forage production, in particular. However, at this initial stage of germplasm evaluation, our goal was to estimate the breadth of variation in the GRNG population for morphological and forage production traits. By collecting seeds from individual plants on the GRNG, rather than bulking, we had an identifiable source (i.e., family) useful in partitioning the variation and providing valuable insights for future more detailed genetic studies.

Growth Habit
Mean growth habits for 1875 individual plants, from 128 families, at Mandan was 2.9 ± 0.01. In comparison, when scored on a family-row basis, mean growth habits were 3.1 ± 0.06 for 84 families at Millville and 2.5 ± 0.02 for 224 families at Highmore. At Ames, mean growth habits score was 3.0 ± 0.03 for 215 families. These results quantified the visual descriptive assessment that the preponderance of plants in this population had a distinctive bowl-shaped (i.e., ascending) morphology [1,41] that was consistent across a wide geographical area. Sholty was selected for this morphotype [20] for stockpiling until mid-July in the northern Great Plains for combined forage and wildlife habitat purposes.

Flower Color
The occurrence of flower color types data at Highmore, Mandan, and Millville are presented in Table 2. These data were generated from different approaches at each location; however, the interpretations are similar. At Mandan, the intraclass correlation, These data poin superiority of numerous GRNG families, compared with conventional hay-type cultivars production from initial spring growth. However, the data also showed that several familie regrowth potential comparable to the hay-types.
Since only two replicates of single rows of spaced plants were established at eac caution should certainly be taken when declaring differences among families as a basis fo for forage production, in particular. However, at this initial stage of germplasm evaluatio was to estimate the breadth of variation in the GRNG population for morphological production traits. By collecting seeds from individual plants on the GRNG, rather than b had an identifiable source (i.e., family) useful in partitioning the variation and providin insights for future more detailed genetic studies.

Growth Habit
Mean growth habits for 1875 individual plants, from 128 families, at Mandan was 2. comparison, when scored on a family-row basis, mean growth habits were 3.1 ± 0.06 for at Millville and 2.5 ± 0.02 for 224 families at Highmore. At Ames, mean growth habits sco ± 0.03 for 215 families. These results quantified the visual descriptive assessmen preponderance of plants in this population had a distinctive bowl-shaped (i.e., morphology [1,41] that was consistent across a wide geographical area. Sholty was selec morphotype [20] for stockpiling until mid-July in the northern Great Plains for combined wildlife habitat purposes.

Flower Color
The occurrence of flower color types data at Highmore, Mandan, and Millville are p Table 2. These data were generated from different approaches at each location; ho interpretations are similar. At Mandan, the intraclass correlation, ƿ = σ 2 F/(σ 2 F + σ 2 E), wher variance component associated with among family variance and σ 2 E is the variance c associated with within-family variance, was ƿ = 0.3733/(0.3733 + 0.6807) = 0.35. This is the between two randomly chosen individuals from the same family. This indicated a significa of variation among and within families.
Although >90% of the plants on the GRNG are yellow-flowered (Figure 1), variegate plants occur most noticeably with desert wheatgrass in swales [24]. The Norman Sm population contains a low frequency of plants with variegated flowers [46], presum = σ 2 F /(σ 2 F + σ 2 E ), where σ 2 F is the variance component associated with among family variance and σ 2 E is the variance component associated with within-family variance, was wildlife habitat purposes.

Flower Color
The occurrence of flower color types data at Highmore, Mandan, and Millville are presented in Table 2. These data were generated from different approaches at each location; however, the interpretations are similar. At Mandan, the intraclass correlation, ƿ = σ 2 F/(σ 2 F + σ 2 E), where σ 2 F is the variance component associated with among family variance and σ 2 E is the variance component associated with within-family variance, was ƿ = 0.3733/(0.3733 + 0.6807) = 0.35. This is the correlation between two randomly chosen individuals from the same family. This indicated a significant amount of variation among and within families.
Although >90% of the plants on the GRNG are yellow-flowered (Figure 1), variegated-flowered plants occur most noticeably with desert wheatgrass in swales [24]. The Norman Smith Ranch population contains a low frequency of plants with variegated flowers [46], presumably from intercrossing with past inter-seedings of Teton and Travois on the ranch, and seedings of hay-type cultivars in the immediate region [23].
= 0.3733/(0.3733 + 0.6807) = 0.35. This is the correlation between two randomly chosen individuals from the same family. This indicated a significant amount of variation among and within families. Although >90% of the plants on the GRNG are yellow-flowered (Figure 1), variegated-flowered plants occur most noticeably with crested wheatgrass in swales [24]. The Norman Smith Ranch population contains a low frequency of plants with variegated flowers [46], presumably from intercrossing with past inter-seedings of Teton and Travois on the ranch, and seedings of hay-type cultivars in the immediate region [23].

Pod Shape
Significant (p < 0.001) differences were found among families for pod shape, estimated on a graduated scale (0.1 units) from 1.0 to 3.0, at Mandan. Ninety-five percent of the plants had pods that were in the range of 1.0-1.7 (i.e., <1 complete coil). Only five percent had pods with from one to two complete coils. These results provided strong morphological evidence of a major contribution of M. falcata to the parentage of the population. The intraclass correlation was 0.51, which is the correlation between two individuals within the same family. As expected from the magnitude of the within-family variation for flower color, a significant amount of variation for pod shape occurred among and within families.

Potato Leafhopper Yellowing (PLY) and Regrowth
These two traits were measured only at Ames, IA, PLY in 2003, and regrowth in 2004. The grand mean for the half-sib families was 2.7 for PLY, indicating general moderate resistance/tolerance. Twelve out of 215 families had a PLY < 2.0. On the other hand, the check standard hay-type cultivars, Vernal and 5454, were highly susceptible, with ratings of 4. Garver [8] observed in early 20th century cultivar trials at Redfield, SD that Ladak was more resistant to PLY than cultivars, such as Grimm and Ranger, which had smaller contributions of M. falcata, relative to Ladak, in their parentage. Of course, consideration needs to be given to differences in regrowth between the families and the checks contributing to attractiveness for leafhoppers, as the simple linear correlation between PLY and regrowth was highly significant (r = 0.52, p < 0.01).

Standard Cultivars and Selected M. falcata Populations
Results in the 2006 transplant nursery at Brookings mirrored, in some respects, what Misar et al. [22] reported for the same germplasms in northwestern South Dakota. That is, Smith's falcata, a selection for sickle-shaped pods from the GRNG population, and Sholty produced more forage than conventional hay-and pasture-type cultivars (i.e., Alfagraze, 5454, and 6200 HT) in a delayed 1-harvest system in eastern South Dakota during each of 2007 and 2008 when early summer precipitation was severely lacking (Table 3). However, forage production of Vernal was similar to Smith's falcata, Sickle pod, and Sholty in both years (Table 3). Morphologically, the yield advantage of the M. falcata populations could be attributed to wider crowns, compared to conventional types, at a uniform population density. During May through July of the transplant year (2006), total precipitation was 40% of the 30-yr average, with <6% of the normal amount for July. Similarly, total precipitation during May through July 2007 was only 47% of normal, with <3% of the normal amount for July. In addition, during May through July 2008, precipitation was 95% of normal, but was still <50% for July (Table 3). However, May through July precipitation was 88% of normal in 2009, with 100% of normal in July. When a 2-harvest system was applied in 2009, Smith's falcata and selections for pod shape from it produced just as much forage as the highest yielding hay types. In addition, these three germplasms exhibited regrowth similar to the top-yielding hay-and pasture-types. As expected, the two diploid M. falcata cultivars, Don and Sholty, produced less forage from regrowth than standard hay-and pasture types under favorable early summer precipitation totals (Table 4). Significant differences were found among the five bulks for flower color and growth habit, but not for g dry matter plant −1 . Plants derived from seeds collected adjacent to the Smith Ranch and plants high on the landscape (i.e., shoulder position) produced more yellow-flowered plants than bulks collected across swales that had Kentucky bluegrass or crested wheatgrass as co-dominant species (Figure 2). This indicated genetic variation for morphological traits of interest and perhaps for physiological traits associated with spatial environmental variation in soil and vegetation (native vs. predominantly introduced plant communities) related to landscape position on the GRNG [24].

Effect of Population Density on Forage Yield of M. falcata
Significant (p < 0.01) differences occurred between populations and between plant population densities for forage dry matter yield at Brookings in 2011. Sholty produced 25% more forage (5.6 Mg ha −1 ) than the TBC population. The 3.0 plants m −2 population density produced 30% more forage (5.7 Mg ha −1 ) than the 1.8 plants m −2 population density. Similarly, in 2013, mean dry matter yields were significantly (p < 0.01) greater for Sholty (7.6 Mg ha −1 ) than TBC (5.6 Mg ha −1 ); however, no difference (p > 0.05) was found between the two population densities (6.5 Mg ha −1 ). The population × plant density was non-significant in both years.
Significant (p < 0.01) differences occurred between Sholty and TBC at a plant population density of 2.4 plants m −2 in both 2011 and 2013 at Aurora. Mean dry matter yield averaged across the two years was 6.3 Mg ha −1 for Sholty and 3.4 Mg ha −1 for TBC.
The TBC population had variegated flowers and narrower crowns than Sholty; however, both populations showed no effect of differences in population density on forage yield during the fourth growing season. The yield advantage of Sholty over TBC increased at both locations between the second (1.8 times greater) and fourth growing season (2.1 times greater), indicating, as described by Oakley and Garver [1] the capability of M. falcata for abundant forage production at low population densities in mature stands.

Perspectives/Implications
As pointed out by Adams [53], N.E. Hansen envisioned yellow-flowered alfalfa (M. falcata), introduced from the steppe of Siberia to rangelands in South Dakota, "would hold its own with the native plants found there, and is certainly a work worthwhile".
Adams expanded, "we have fallen heir not only to this vision and challenge, but also to plant materials by which they soon may be realized". It is tempting to conclude that the pattern of naturalization of yellow-flowered alfalfa on the GRNG is an example of exactly what Hansen imagined. Bolton [9] stated, "it was impossible to overestimate the value of Hansen's introductions of alfalfa" to alfalfa breeding in North America.
Keller [45] in his "Wanted: A Paragon for the Range" lauded the value of crested wheatgrass as largely responsible for restoring millions of acres of denuded farmland in the northern Great Plains and Intermountain West [46,54]. However, he pointed out that tracts of rangeland are highly variable aggregations of large numbers of irregular sites, making breeding plants for their improvement difficult. Regarding approaches for plant improvement, Keller noted rangelands in need of improvement are not represented by environments on experiment farms. Therefore, a range-plant breeding program should be conducted on rangelands. He also acknowledged that, in pastures, a grass-legume association is more productive than grass alone and deserved considerable attention.
To accomplish what Keller [45] outlined requires a complex selection screen which subjects several to perhaps many generations to fluctuating abiotic (e.g., drought) and biotic (e.g., cattle grazing, insect feeding) environmental stresses that occur at different times during a growing season, depending on the year. Clearly, populations of alfalfa that have adapted to fluctuating climatic and other environmental stresses, of which Smith's falcata and the GRNG population are examples, have considerable promise for immediate interseeeding of rangelands and as breeding stocks for the development of new cultivars for future rangeland and monotypic grassland improvement in the northern Great Plains. They have also demonstrated higher seed production potential in situ (GRNG) and in commercial seed production practices (Smith's falcata) than pure M. falcata populations. Therefore, they will not likely require additional direct selection for seed production to be commercially viable.
Alfalfa populations developed for improving mature stands of crested wheatgrass and/or depleted rangeland in the northern Great Plains for forage, conservation, and ecological purposes need to be: (1) Well-adapted to the climate of the region, as a result of natural selection; (2) tolerant of grazing, yet capable of producing sufficient amounts of high-quality forage to improve performance of grazing animals; (3) drought-tolerant and winterhardy; (4) able to establish and persist in monocultures of highly competitive grasses, such as crested wheatgrass, from mechanical interseeding and natural recruitment, without curtailing grazing. The commercial seed increase of Smith's falcata demonstrated these traits in northwestern SD [23], southwestern ND, and eastern WY [46].
This study found the GRNG population was highly variable, with a genetic variation for morphological traits. However, it also showed that environmental variation may have created genetically different subpopulations (e.g., topographical variation for physical soil characteristics and moisture-holding capacity and the associated native or introduced plant communities) based on a few morphological traits. Therefore, seed increase of the highly heterogeneous GRNG population, as described here, would likely result in rapid genetic shifts depending on where the increase occurred. Therefore, we propose conservation of this highly heterogeneous naturalized [24] cross-pollinating population should employ both in situ and ex situ strategies, similar to what might be prudent for landraces [55]. Since seed production episodes are sporadic, due to unpredictable abiotic and biotic factors, including variation in grazing time and intensity, ex situ methods for standing seed crops may be possible only once or twice during a decade. However, ex situ conservation of the population through a sampling of the soil seed bank could be done at will, as the hard seed characteristic of the population presumably provides long-term storage/viability of seeds in the soil [24]. Clearly, in situ conservation, under the past and current grazing management strategies employed by the U.S. Forest Service, appears to be a viable long-term strategy for the preservation of the genetically heterogeneous germplasm. In addition, it provides for genetic recombination for natural selection and a continuous germplasm source for timely ex situ seed collections. Rincker [56] showed that seed production in widely-spaced plants of alfalfa exceeded that of rows in an environment that was conducive to seed production.
The work of Misar et al. [46], in particular, provides instructive information regarding the value of Smith's falcata, and presumably the GRNG population, as well, for interseeding into crested wheatgrass, based on results from >20 year-old stands of this species at four locations in ND, SD, and WY. The degree of interseeding success with Smith's falcata was site-specific (e.g., amount of wheatgrass cover, soil texture) and related to cultural practices (e.g., seeding date, seeding rate, and type of herbicide). Schellenberg et al. [57] determined that killing crested wheatgrass in a strip alongside the furrow where seeds were planted was crucial for alfalfa establishment, the wider the strip the better. They also reported similar establishment success and forage yield for Rangelander (M. media), a creeping rooted cultivar, and Yellowhead (M. falcata), a taprooted cultivar.
The GRNG alfalfa population includes a subpopulation of reproductive individuals from seedlings that are established in swales that support a plant community dominated by alfalfa, crested wheatgrass, and Kentucky bluegrass. The swales were more than likely privately farmed up to the 1930s, but then federally reclaimed and reseeded to wheatgrass for permanent management, predominantly for grazing starting in the 1950s [24]. The swales also contain large soil seed banks of hard alfalfa seed, up to 30,000 viable seeds m −2 , compared with a virtual paucity of seeds on shoulder and summit landscape positions. Existing plants and the soil seed bank represent a gene pool, likely a result of multiple generations of genetic recombination, from which to select for establishment and persistence of alfalfa in crested wheatgrass dominated grasslands in the semiarid northern Great Plains. Correspondingly, existing plants and the associated soil seed bank at other landscape positions should provide similar sources, through natural selection, of potentially adapted germplasm.
Since improved establishment, persistence, and recruitment from natural reseeding of alfalfa interseeded in crested wheatgrass or depleted native rangeland are primary plant breeding goals in the northern Great Plains, collecting seeds from surviving alfalfa plants in natural reseeded or successful mechanically interseeded stands, preferably after exposure to several cycles of the appropriate management system, would be a logical approach for initiating breeding programs for the development of new cultivars with specific attributes for solving those particular rangeland problems/needs. Smith's falcata [23] and the GRNG population [24] would seem to be good base populations for those situations in the semiarid regions of the northern Great Plains. However, the magnitude of genetic variation among and within families in the GRNG population for dry matter forage production, from initial and regrowth harvests, and morphological traits revealed by this research suggests their potential for the accomplishment of other breeding goals for alfalfa in the semiarid northern Great Plains, as well. A good example was reported by Berdahl et al. [21]. A germplasm derived from open-pollinated bulk seed collected from the Smith Ranch, and two germplasms derived by bulking open-pollinated seeds from stands planted in the early 1920s at Mandan, ND had the highest survival (78%) of 10 entries, including hay-and pasture-types, after grazing by sheep for three growing seasons at Mandan [21]. These three germplasms traced to introductions of M. falcata by N.E. Hansen to the Dakotas during the early 1900s.
Interseeding native rangeland with alfalfa in the semiarid northern Great Plains has met with mixed success, based on general assessments made at two [58], five [59], or seven years after planting [17]. Fewer studies have been conducted with alfalfa interseeded into mature (i.e., 20-to 30-year-old) stands of crested wheatgrass in the northern Great Plains [22,57]. The only study that included post establishment stand density data at six years after planting was that by Misar et al. [22]. In addition, a recent reconnaissance (i.e., July 2020) was conducted of the four experimental sites where crested wheatgrass stands greater than 20 years of age were interseeded with the first generation increase of seeds (Wind River Seed Company) of Smith's falcata in 2008 [22]. This recent reconnaissance revealed that two of the experimental plantings (Newcastle, WY (43 • 51 11" N, 104 • 12 34" W) and Hettinger, ND (46 • 0 3" N, 102 • 38 0" W)) of Smith's falcata made in 2008 had expanded beyond the post-establishment analysis in 2014 [22], primarily as a result of notable vegetative spread and crown expansion. These successful stands were achieved under two very different forage management strategies. The Newcastle interseeding has been hayed annually during early summer; whereas, the Hettinger interseeding has been grazed with sheep annually during spring [60]. Recently, interseeding alfalfa into native grassland in the southern High Plains showed initial seedling density and short-term post establishment trends for crown density similar to those reported by Rumbaugh and Thorn [58] in the northern Great Plains. The Texas study indicated that wide row spacing (71 cm) was more desirable than narrow spacing (36 cm) for seed cost, since the two different row spacing converged for crown density within three years after planting [61].
Currently, Smith's falcata is providing a valuable commercial seed source for improving native rangeland and mature competitive monotypic perennial cool-season grasses pastures, such as crested wheatgrass, in the northern Great Plains. The GRNG population, derived from Smith's falcata, represents further natural selection for the establishment, natural recruitment, and persistence under grazing. To preserve these novel germplasms, we recommend a combined in situ and ex situ conservation strategy similar to that used for the conservation of landraces.
The expansive natural distribution of M. falcata in Eurasia traverses 42 to 60 • N. latitude and 10 to 80 • E. longitude [62,63]. In many regions, indigenous strains provide essential forage for livestock and other ungulates [41]. Recently, more than 200 PIs of M. falcata were evaluated at multiple widely spaced locations in diverse environments in the United States and Canada [36]. These evaluations identified promising new genetically diverse M. falcata germplasms for potential cultivar development for multiple purposes in North America. It is reasonable to think that the two novel germplasms that became naturalized under grazing in northwestern SD may be of interest for evaluation to improve grasslands dominated by perennial cool-season grasses in steppes and prairies in temperate climates globally.