Exploring Winter Lentil and Winter Pea Production in the Northeastern United States

Winter Lentil Seeding Rate Trial

Year 1

In the first year, the on-farm lentil trial was challenged by poor winter survival as well as annual weed competition, issues that also arose in a neighboring production winter lentil field. On-farm lentil plots were harvested with great effort, but weights were not recorded due to stand inconsistency and weed binding in harvest equipment. The first-year research station lentil seeding rate trial was also heavily infested with chickweed after planting, compromising results by affecting population establishment.

No significant difference in stand count, grain yield, lentil biomass, or weed biomass was detected in the research station trial (Table 2), likely due to extensive weed pressure.

Table 2. Results of one-way analysis of variance by seeding rate in Year 1 trial of 'Morton' winter lentil sown at five seeding rates. NS indicates no significant difference among treatments.

Trait means in the research trial varied numerically among seeding rate treatments but without explicable trend (Table 3). For example, stand count and lentil biomass were numerically (though not significantly) higher using the lowest seeding rate of 25 lbs/ac, while a much higher seeding rate of 47.5 lbs/ac showed much lower stand count and lentil biomass. It seems that weed pressure overwhelmed the effect of seeding rate treatments, so our approach for 2023-24 focuses on mitigating weed pressure through use of biculture with oat.

Table 3. Trait means by seeding rate in Year 1 trial of 'Morton' winter lentil sown at five seeding rates.

Winter Lentil Trait Means 2022-23

As a result of collaboration surrounding this project, Martens and researchers decided to use an oat nurse crop (target 60 lbs/ac) interplanted with lentil in 2023-24, in an effort to suppress weeds and aid winter survival by increasing surface residue. Nurse crops have been used successfully by researchers for improving weed suppression and winter survival in hairy vetch, and we hypothesize that this technique holds promise for winter lentil as well. Though not a formal research question of the proposed work, experimentation with a cereal nurse crop for winter lentil is a potentially valuable learning outcome in the case of this novel winter legume crop for our region. If successful, this technique will facilitate better trial data and improved commercial production in future seasons.

Year 2

Winter lentil plots sown in biculture with winter oats established well in both research station and on-farm settings. Importantly, significant differences were observed in spring stand (Table 4), with mean stand count increasing directly with seeding rate (Table 5). Stand count was not significantly different among the lower three seeding rates (25, 32.5, and  40 lbs/ac) or top two seeding rates (47.5 and 55 lbs/ac), but the lowest three seeding rates were significantly different in stand count from the highest 55 lbs/ac seeding rate.

However, these differences in lentil plant density among seeding rate treatments did not result in significant differences in weed biomass, lentil yield, or oat yield. The similarity of weed biomass and oat yield across seeding rate treatments reinforces the notion that a similar growing environment was present across plots, such that seeding rate treatments could have resulted in different lentil yield. While differences in lentil yield across treatments did not meet the threshold of statistical significance, some trends are evident. Lentil grain yield was very similar – between 75 and 81 g-- among the four higher seeding rate treatments (32, 40, 47.5, and 55 lbs/ac), while yield for the 25 lb/ac seeding rate was numerically lower by about one-third. 

Table 4. Results of one-way analysis of variance by seeding rate in Year 2 trial of 'Morton' winter lentil sown at five seeding rates. The *** symbol indicates significance at p<0.001, and NS indicates no significant difference among treatments.

Table 5. Trait means by seeding rate in Year 2 trial of 'Morton' winter lentil sown at five seeding rates.

Winter Lentil Trait Means 2023-24

On-farm lentil trials cannot be statistically analyzed, but with the exception of the 47.5 lb/ac seeding rate plot, stand counts increased with seeding rate (Table 6). However, stand counts were lower in the on-farm than research setting for each given seeding rate, and notably, stand count was quite similar for the on-farm 40 lb/ac and  55 lb/ac plots. Lentil yields largely followed the trend in seeding rates; the highest and lowest lentil yield (208 lbs and 119 lbs) resulted from the highest and lowest seeding rate, respectively. The 32.5 lb/ac and 40 lb/ac seeding rates yielded progressively more than the lowest seeding rate; the only seeding rate not showing a commensurate increase in lentil yield was the 47.5 lbs/ac seeding rate, likely due to its poorer stand establishment. 

Table 6: Trait values by lentil seeding rate in on-farm trial, sorted by grain yield.

Winter Lentil On-Farm Trial Trait Means 2023-24

Winter Pea Variety Trial

Year 1
While some deer browsing was apparent in the research station winter pea variety trial in Year 1 , significant differences were found among varieties for all traits except weed biomass (Table 7). For fall vigor, grain yield, pea biomass, and protein, statistical analysis served to identify one or two lower-performing varieties (e.g. Windham for fall vigor, grain yield, and pea biomass and FP6101 for grain yield, pea biomass, and protein) but did not show significant distinction among the top-ranking 7 or 8 varieties. For spring vigor, one variety (KeystoneIcicle) emerged as significantly more vigorous than several others: FP6101, Icicle, Vail, and Windham. 

Table 7. Year 1 results of one-way analysis of variance by winter pea variety.

***, **, and * indicate one-way fixed effect ANOVA tests significant at p<0.001, p < 0.01, and p < 0.05, respectively. NS indicates no significant difference among varieties. Significance for grain yield, pea biomass, and weed biomass calculated using square root transformed data. Stand count data not shown due to deer browsing.

A means table sorted by grain yield shows that varieties sorted into three general groups (Table 8). A group comprising Blaze, Vail, and Goldenwood shows numerically higher yield and protein content from shorter, more heavily tillered plants. A second group -- KurtWood, KeystoneIcicle, and Keystone showed moderately high yield from taller, less tillered plants with numerically lower protein content. Finally, a third group composed of Icicle, FP6101, and Windham showed lower yield and biomass with varied plant stature.

Table 8. Trait means by winter pea variety in Year 1 research station trial, sorted by grain yield.
Winter Pea Trait Means 2022-23

On-farm variety trial plots established well and reflected trends observed in the research station trial (Table 9). As in the replicated trial, Blaze, Goldenwood, and Vail showed the highest grain yield, and Goldenwood yielded less biomass than Blaze or Vail. Similarly, both replicated and on-farm trials showed Keystone and Kurtwood to be intermediate for yield but strong for biomass production. Finally, both trials revealed Icicle and FP6101 as poorer-yielding varieties.

Table 9: Trait values by winter pea variety in Year 1 on-farm trial, sorted by grain yield.
On-Farm Winter Pea Variety Trial Trait Values 2022-23

Trait correlations from the research station trial suggest trends in the way pea plant architecture and vigor associate with yield traits (Figure 1). Spring vigor shows a modest but significant correlation with both grain yield and pea biomass (r=0.36, p < 0.05 for both traits), but fall vigor is not significantly correlated with either yield trait. This illustrates a known tradeoff between fall vigor and biomass in cold climates. A high, very significant correlation (r=0.87, p < 0.001) between grain yield and biomass suggests that both traits may be selected for concurrently and that similar varieties could suit both grain and forage / cover crop applications.

Our results show high, very significant correlations between stem count and two yield variables: grain yield and pea biomass (r=0.74 and 0.64, p<0.001, respectively). This indicates, unsurprisingly, that varieties with more stems are more productive for both grain and biomass. While plant height is significantly and inversely correlated with stem count (r=-.40, p<0.05), plant height itself is not significantly correlated with grain yield or pea biomass. Thus, stem count – regardless of plant height – is the trait most strongly associated with both grain and biomass yield. However, plant height does show a significant inverse correlation with protein content (r=-0.46, p<0.01), suggesting that short-stature pea varieties may offer higher protein content than longer-vining varieties. 

Figure 1: Trait correlations from Year 1 winter pea variety trial.
Winter Pea Variety Trial Trait Correlations 2022-23 

This winter pea variety trial was designed to compare multiple grain pea varieties against Keystone, which is the variety most commonly used by grain pea growers in the Northeast. We are interested and gratified that several varieties -- most notably shorter varieties Blaze, Vail, and Goldenwood -- showed numerically higher performance than Keystone for both grain yield and biomass. Finally, grain peas are generally bred for monoculture production, but our trial uses biculture with a small grain, creating an environment of resource competition. We note that varieties with abundant tillers performed well in this environment, and we are curious whether these results will be reflected in the trial's second year.

Year 2

The 2023-24 replicated winter pea variety trial showed significant differences among varieties for all traits except spring stand, total biomass, and triticale yield (Table 10). The consistency in pea plant population, whole plot biomass, and triticale yield shows that strong and consistent pea-triticale biculture stands were established, providing a reasonably uniform trial environment. While significant differences in fall vigor were evident (Table 11), relative vigor among varieties had changed by spring. By spring, two vining, normal-leaf varieties -- FP6101 and Icicle -- showed significantly more vigor than Blaze, Goldenwood, and Windham, three compact, semileafless varieties. 

Table 10. Year 2 results of mixed model analysis of variance by winter pea variety.

Differences in plant habit and leaf type were again reflected in plant height measurements, in which FP6101, Icicle, and KeystoneIcicle -- a mixture of normal-leaf, vining Icicle and compact, semileafless Keystone -- were all significantly taller than the other (all semileafless) varieties. Conditions were moisture-rich, due to the soil's high water-holding capacity, partial shade from a neighboring woods, and abundant precipitation. These conditions promoted disease -- particularly fusarium root rot -- and prolific plant growth, such that substantial lodging occurred. Significant differences in lodging exactly followed significant differences in plant height, and disease severity was significantly higher in Keystone and Vail (two semileafless types) than vining varieties FP6101 and Icicle. It seems that in these wet, disease-prone conditions, normal-leaf, indeterminate varieties were able to generate new tissue to replace diseased tissue, while determinate semileafless varieties tended to succumb more easily to disease.

KeystoneIcicle, the mixed-determinacy variety, showed significantly higher yield than four semileafless cultivars and one-third more than the next-highest-yielding variety, Icicle. It's possible that while determinate Keystone struggled with disease when planted individually, it benefitted from the protection of vining Icicle when planted in mixture. Finally, significant differences in percent crude protein were identified, with FP6101, Vail, and Windham showing significantly higher protein content than Kurtwood and Payback.

Table 11: Trait means by winter pea variety in Year 2 research station trial, sorted by grain yield.

Winter Pea Variety Trial Trait Means 2023-24

Trait correlations offer clues as to the interaction of plant habit and disease in the Year 2 replicated winter pea trial. Disease severity was positively correlated with standability and negatively correlated with spring vigor, both of which characterize determinate, semileafless varieties. Disease severity also shows significant negative correlation with total biomass and pea grain yield, suggesting that disease differentially impacted semileafless varieties and decreased their yield. Spring vigor -- which was significantly higher for vining varieties -- is significantly correlated with both total biomass and pe yield. Thus, the replicated trial's wet environment seemed to favor indeterminate varieties while constraining the growth potential of determinate varieties.

Figure 2: Trait correlations from Year 2 winter pea variety trial.

Winter Grain Pea Trial Trait Correlations 2023-24

The Year 2 on-farm unreplicated trial did not experience the excessively wet conditions that the replicated trial did, resulting in less disease pressure, less lodging, and earlier grain maturity. Blaze, a compact semileafless variety, was the top yielding variety for the second year running, and Vail was in the top 3 varieties both years (Table 12). The only vining variety in the on-farm trial -- Icicle, which yielded second-best in the replicated trial -- ranked last for yield on-farm. Kurtwood and Keystone showed intermediate yields on-farm in both years. 

Table 12: Trait values by winter pea variety in Year 2 on-farm trial, sorted by grain yield.

Winter Pea On-Farm Trial Trait Means 2023-24

Two-year variety trial analysis

Combined two-year analysis of key traits  shows both entry and entry x year interaction for all traits analyzed: standardized grain pea yield, standardized total plot biomass, and percent crude protein (Table 13). 

Table 13. Years 1-2 results of mixed model analysis of variance by winter pea variety.

An interaction plots shows substantial crossover interaction for standardized grain yield (Figure 3). Blaze and Vail yielded considerably worse than average in 2024 than 2023, while Windham, FP6101, Icicle, and KeystoneIcicle yielded much better than average in 2024. Interestingly, three varieties showed consistent performance (relative to same-year average yield) between years: KurtWood and Goldenwood both showed an average of 117% same-year yield in both years, while Keystone showed an average of 84% same-year mean yield in both years.

Figure 3: Year 1-2 interactions for grain pea yield

Winter Grain Pea Standardized Yield Interactions 2023-24

Standardized biomass showed much more variance in 2023 than 2024 (Figure 4). FP6101, Windham, and Keystone produced much less biomass than average in 2023 but converged toward average biomass in 2024. Blaze, Vail , and Goldenwood yielded more biomass than average in 2023 but converged toward average biomass in 2024. Kurtwood, Icicle, Keystone, and the KeystoneIcicle mixture showed little change in relative biomass year to year. 

Figure 4: Year 1-2 interactions for total biomass

Winter Grain Pea Standardized Biomass Interactions 2023-24

Finally, interactions for percent crude protein showed much less crossover than those for grain yield and biomass (Figure 5). Blaze decreased in percent crude protein slightly, but all other varieties had higher protein content in 2024 than 2023. The degree of variation between years varied dramatically by variety; for instance, FP6101 had much higher protein in 2024, while Goldenwood increased only slightly year-to-year. No significant correlations with percent crude protein are present in Year 2 (Figure 2), so while increased moisture seems to be associated with higher protein content, we cannot hypothesize disease or plant stature as a mechanism of increased protein accumulation in Year 2.