Identifying Climate-resilient Warm Season Food and Forage Species in Western Oregon

Species effects on dry farmed forage biomass

The main determinants of dry farmed forage crop biomass in this study were Treatment and Site, which were fixed and random effects of the general linear mixed model (GLMM) model, respectively. The analysis of biomass yields using the GLMM revealed significant treatment effects (F(9, 192) = 7.66, p < 0.001) among the ten forage species evaluated (Figure 1, Table 3). Mean biomass production ranged from a low of 632.6 ± 676.8 kg/ha for teff to a high of 5129.8 ± 676.8 kg/ha for sorghum, with the latter significantly outperforming all other treatments. Among the legumes, cowpea yielded the highest biomass (1987.9 ± 676.8 kg/ha), followed by sunn hemp (1743.8 ± 676.8 kg/ha) and lablab (1392.6 ± 676.8 kg/ha). The model explained approximately 41.6% of the variation in biomass production (R² = 0.416), with site-to-site variability accounting for 34.2% of the total variance component. Wide confidence intervals observed across treatments suggest considerable variability in performance within each species. Tukey's HSD test (α = 0.05) was employed for pairwise comparisons to distinguish significant differences between treatment means, with only sorghum emerging as a significantly more productive species compared to the other treatments (p < 0.001).

Figure 1: Estimated Least Squares Means (LSM) contributed by each forage species, based on generalized linear mixed model (GLMM).

Table 3: Estimated Least Squares Means (LSM) contributed by each forage species, based on generalized linear mixed model (GLMM)

Site effects on dry farmed forage biomass

Sites differed in soil type, climate, management, and pest and weed pressure, all of which may have had an impact on dry farmed forage productivity. Soil texture ranged from loam to clay across sites; some sites significantly different in soil texture between 1 and 3 ft depths. Estimated available water holding capacity of soils (AWHC; estimate of plant-available water, reported by USDA NRCS) ranged between 8 to 11.7 inches of water between 0 - 5 ft. depth. Soil fertility varied significantly between sites, and between sampling depths within site (Table 5). Despite this variation in soil factors across sites, no singular soil factors were significantly related to forage biomass production.

Table 4: Overview of sites of our studies. Lewis Brown and Oak Creek were OSU research farms; other sites were privately managed by producers.

Table 5: Soil characteristics of each site compared to average dry biomass per site.

Sites were generally similar in terms of climate, as they were located in the central and south Willamette Valley. The exception was working farm Lane, which was approximately 5 F cooler than the other sites, and located on the eastern side of the Coast Range Mountains; all other sites were located in the Willamette Valley.

On some sites (Lewis Brown, Yamhill1), a poor stand was the main cause of low biomass. On other sites experienced notable weed or pest pressure on some or all species planted. Later planting dates appeared to favor greater grass production, but not legumes or mixed swards (Figure 2).

Figure 2: Relationship between seeding date and the planting date of, grass, legume, and mixed species in our study.

Seed yield of diverse forage species

Seed yield was recorded at 3 sites: Lewis Brown, Oak Creek, and Benton, from crops in which mature seed was available by October 1st. Sorghum, pearl millet, mung bean, and tepary varieties all produced mature seed. These yields varied substantially between sites (Table 6). Sites with lower seed yield also had lower dry biomass; in particularly drought-stressed locations (Lewis Brown and Benton) significant amounts of sorghum and pearl millet heads were unfilled with grains. Some mung bean and tepary bean yield was lost due to shattering pods in the field.

Table 6: Average seed yield (kg/ha) at each site in which seed yield was assessed

Discussion

Grasses: The earliest warm-season forage trials in Oregon were carried out in the 1940s and identified sorghum as the most productive warm-season annual available for producers   our results, nearly nine decades later, confirm this is still the case. Sorghum was a reliable forage, producing at least 1300kg/ha DM at all of the ten sites, with a maximum yield of 18250 kg/ha. Pearl millet produced less than 1000 kg/ha at half the sites, yet on the other half produced between 2160 and 11240 kg/ha. Teff produced little enough forage at nearly every site that it ended up being the least productive species of any tested; to ensure these results are representative of the species rather than the source of seed used in the 2024 trials, we will need to do further testing of teff. These results demonstrate that sorghum is still the most reliably productive species tested overall, as well as being the most productive grass species. Nevertheless, sorghum also may present issues with prussic acid toxicity, a potentially deadly situation in which stresses to the plant—such as drought, frost, herbicide exposure, or recent grazing/mowing--can precipitate elevated prussic acid levels. Another name for prussic acid is hydrogen cyanide, indicating the severe danger posed by feeding plants with prussic acid to livestock. Simple, affordable, fast field testing kits are available from several sources, so this risk is not like gambling with the lives of your herd or flock. By testing immediately before grazing, producers can prevent livestock losses to prussic acid. We used a rapid prussic acid testing kit on sorghum leaves during biomass sampling, and found only intermediate levels of the toxin on one site; other sites had no or very low levels of prussic acid.

While pearl millet is less reliably productive than sorghum in western Oregon, it also doesn’t pose some of the same dangers as sorghum because it has no prussic acid at any life stage. However, both pearl millet and sorghum can also accumulate dangerous levels of nitrates, the consumption of which can cause livestock to essentially asphyxiate while breathing due its interference with respiratory processes. Pearl millet can actually accumulate nitrates more readily than sorghum, though both do so when stressed. Like prussic acid issues, however, proactive testing of soils and forages can be done beforehand to mitigate risks. Ultimately, for warm-season annual grasses, our results confirmed sorghum and pearl millet to be western Oregon’s best options.

Legumes: Cowpeas and sunn hemp stood out as the most consistently productive legume species, with a slight edge to cowpeas in terms of mean yield, a slight edge to sunn hemp in terms of maximum yield, and a tie when it came to median yield. Cowpeas showed good adaptation to both heavy and well-drained soils, while sunn hemp was best suited to well-drained soils including hillsides.  Both had a mean yield nearing 2000kg/ha. In terms of animal performance, cowpea has broader leaves which are easier to preferentially browse by livestock; sunn hemp, in comparison, has myriad smaller leaves and a more fibrous stem, reducing intake rate and digestibility. However, sunn hemp also has an upright growth form which can compete well in dense swards of tall bunchgrasses such as the warm-season grass species tested in this research. On the other hand, cowpea’s vining lateral growth fills in gaps between bunchgrasses and acts as an effective ground cover and weed suppression smother crop.

Other legumes were less reliable and/or less productive, with lablab leading the remainder in terms of mean yield, followed by mung beans and tepary beans. Tepary’s yields were impacted by pests on at least three sites, with some plants exhibiting complete defoliation due to rabbit and invertebrate herbivory. Lablab’s performance was marked by inconsistency, with three sites yielding between approximately 2700 and 4700 kg/ha means it should not form the primary foundation of a warm-season planting’s legume component, but on sites where it did establish, it generally had good yields. It also has a vining habit which paired well with adjacent sorghum or even tall weedy species like thistles wherein the lablab would twine its way up the taller neighboring plant, much like beans in a Three Sisters planting. For these reasons, lablab may be a useful companion species to more reliable and consistent legumes like cowpeas or sunn hemp, but should not be the sole legume in a warm-season annual pasture mix. Mung bean had inconsistent yields similar to lablab, but with less productivity even on its higher-yielding sites; the only potential upside to its use seemed to be its early germination relative to other species.

It should be noted that these results are not assumed to be generalizable across entire species of these crops; this is because for each species selected in this study, we only used a particular variety or cultivar. These varieties were selected mostly based upon commercial availability of their seed, and some were not bred for forage production per se. For example, the tepary and mung bean varieties chosen were known more for their edible seed production, rather than being particularly high-biomass varieties. We recommend future research to trial diverse warm season grass and legume varieties that are more suitable for livestock forage.

The heavy soil and well-drained soil mixes were selected for this study based upon known traits of the grass and legume species comprising these mixes (Table 2). In most cases, these mixes were outperformed by monocultures (sorghum, millet, cowpea, sunn hemp and lablab).  It is likely that these mixes underperformed as they contained lesser-performing species (e.g., teff and tepary bean), and did not have as high of a seeding density of better-performing species to produce adequate biomass. There may also have been unknown species incompatibilities in these mixes. Though yielding dry biomass in the middle of the range compared to the other treatments, these mixes are likely to be more nutritionally well-rounded than monocultures. More research should go into intercropping of different combinations of warm season forage species, and determining proper seeding rates for multi-species mixes.

Aside from species-specific outcomes, some generalizable site-specific observations for dry-farming warm-season forages emerged from this study. Sites with lower available water holding capacity (e.g., 8” compared to 11” of available water) did not correlate with lower yields, indicating effective scavenging of available moisture by the dry-farmed forage species. At least two sites (Lane and Benton) may have had root-restrictive qualities between 1-3ft in depth, as their soil moisture at 3 ft did not appreciably change throughout the season, indicating sorghum roots were not reaching those depths. Rooting progression might have been slowed by heavy soil texture at deeper layers, inadequate soil nutrients, and/or a low pH. In dry farming, the effective rooting depth will be a key factor that limits water availability to the plants, so managing soils for deeper rooting is a key practice to consider.

 Seeding method also did not have a significant impact on yield, with both the highest and lowest-yielding sites having been broadcast. Two sites with broadcast seeding were in the same valley and within walking distance of one another—however, one of these sites had yields in the middle of the pack, so to speak, and the other was the lowest-producing site. We hypothesize pest pressure played a role in the latter case, as the site had a defined area of low productivity that did not align with soil types and had changed in extent and shape from one year to the next on the farm. We found that grasses seeded in late May or early June performed better than earlier-seeded sites; cooler weather in mid-to-late May could have slowed the establishment of warm-season grasses, making them more susceptible to competition with cool-season grasses and forbs or other growth-limiting factors.

Some of the warm season forage species were chosen in this study due to the possibility of dual-use as both animal forage and for human consumption of the grains and pulses. Of the three sites in which seed yield was assessed, only two grasses (sorghum and pearl millet) and two legumes (mung bean and tepary) had mature, harvestable seed by the end of the season (Table 6). The cowpea and lablab varieties chosen for this study did not mature in time in the climate of Western Oregon, but other cowpea varieties do yield seed in this climate, and it is possible that certain lablab varieties do as well. Of those species that produced seed, sorghum and pearl millet appear to be the best candidates for dual use as human food, as their seed heads are readily harvestable, above the other competing vegetation. However, more drought-stressed sorghum and millet plants had poor seed fill of their seed heads; pearl millet seed production appeared to be more sensitive to drought stress than sorghum in this regard. Also, these grasses are susceptible to bird predation, which may have affected yield in this study. Sorghum and millet seed heads are relatively easy to thresh, and a promising for production of dry-farmed, edible grains for humans and livestock like poultry; moreover, and their small seed size and large amount of seed per head suggests on-farm seed saving in a relatively small designated area could be a viable strategy to maintain seed stock for future planting of these varieties. Tepary and mung beans are low-growing legumes, and their harvest would be not feasible on a larger scale unless grown as a row crop, or otherwise by using costly weed control methods. Pods of these legume species tended to shatter easily in the field, and could feasibly self-seed in pastures in future years.