Perennial forages are widely cultivated across North America and globally. For example, alfalfa is one of the oldest and most important perennial forages adapted for temperate and continental climates and is grown in over 30 million hectares around the world, either in pure stands or mixed with perennial grasses (Moot et al., 2012). Orchardgrass is one of the top four forage grasses grown globally, because of its ability to adapt to a wide range of environments (Wu et al., 2023). In the northeastern United States, cool-season perennial forages, such as alfalfa and orchardgrass, are grown in >404,685 ha (>1 million acres) and support the region's dairy and livestock industries (Castonguay et al., 2006; NASS, 2014; Winsten et al., 2010).
Perennial forage crops are vulnerable to weather variability resulting from climate change because, unlike annual crops, they are exposed to weather extremes year-round (i.e., both over the growing season and the winter period). Perennial forage legumes are especially vulnerable to weather extremes during the establishment year when smaller plants have not yet accumulated energy reserves in root crowns equivalent to larger, established plants. Yet the high establishment costs of perennial forages make the establishment year crucial to recoup returns on investment. Therefore, the productivity and stability of cool-season perennial forages, like alfalfa (Medicago sativa L.) and orchardgrass (Dactylis glomerata L.), are likely to be impacted by increases in temperature, as well as greater temperature variability anticipated to occur with climate change.
Temperate regions, including the Northeast region of North America, are already experiencing changes in seasonal weather patterns resulting from climate change (Kunkel et al., 2013). Temperatures in the Northeast region have increased by 1.1°C from 1895 to 2011 (Tobin et al., 2015). Regional annual and seasonal temperature anomalies have been occurring much more frequently in the last 30 years (Kunkel et al., 2013), and are expected to continue increasing through 2100. In temperate regions, winter is warming faster than any other season (Karmalkar & Bradley, 2017; Kunkel et al., 2013), and climatological models predict an increase in the minimum winter temperature of 2–8°C by 2050 (Burakowski et al., 2022). Rising winter temperatures are lengthening the frost-free season, with later first-frost dates in the fall and earlier last-frost dates in the spring (Kunkel et al., 2013; Lane et al., 2019). These changes in winter season temperatures are likely to impact winter survival and persistence of alfalfa and other cool-season perennial forages, as well as their subsequent regrowth and competitive interactions (Castonguay et al., 2006; Wells et al., 2014).
Greater winter weather variability is likely to decrease alfalfa vigor and result in stand loss and overall yield reductions (Bélanger et al., 2002; Jing et al., 2020). Elevated temperatures in the fall may delay plant hardening, thereby decreasing alfalfa's winter hardiness (Bélanger et al., 2002; Sharratt, 1993). Milder winter temperatures punctuated by prolonged and extreme cold (Casson et al., 2019) may then lead to alfalfa frost injury and reduced winter survival (Bélanger et al., 2002; Wells et al., 2014). This elevated injury arises when warmer winter temperatures and reduced snow cover decrease the insulation that the snowpack provides against fluctuating winter air temperatures for overwintering perennial forages (Sharratt, 1993). A reduction in snow cover may also increase soil freeze–thaw cycles (Henry, 2008), which can lead to frost heaving of the alfalfa crowns, frost injury, ice encasement, and plant mortality (Bélanger et al., 2002; Gu et al., 2008). Hence, the ongoing trend of increasing winter weather variability may increase the risks of winter kill and subsequent forage yield reductions in the northeastern United States and elsewhere (Belanger et al., 2006).
Perennial forage stand density often declines as alfalfa stems die off in the establishment year due to increased winter weather variability. This can increase the potential for winter-hardy weeds to establish, further decreasing forage yield and quality (Dillehay et al., 2011). For fall-seeded alfalfa, winter annual weeds such as chickweed (Stellaria media) can emerge with the forage and directly compete for resources. Throughout the winter, when sudden freezes cause alfalfa frost injury or even winter kill, many winter annual weeds are more frost tolerant compared with alfalfa, enabling them to resume vegetative growth earlier and expand into open spaces created by alfalfa stand loss, thereby consuming nutrients to the detriment of forage yield and quality (Aldrich, 1957; Cici & Van Acker, 2009; Undersander, 2011). If an alfalfa stand is not covering more than 80% of the ground or does not have more than 25 plants per square foot in the seeding year, the potential for weeds to establish increases and the stand must be replaced (Wang & Brackenrich, 2023).
In addition to warmer winters, summer temperatures are also increasing and likely affecting perennial forage growth. Cool-season perennial forages are vulnerable to the “summer slump,” also known as a “feed gap,” which is a period of declined growth rates when temperatures reach above the optimal range for perennial forages (Duiker et al., 2023). For example, the optimal temperature for alfalfa growth is between 15 and 25°C (Understander et al., 2011), while the optimal temperature for orchardgrass is ~20°C (NRCS, 2000). Alfalfa and orchardgrass can become stressed under high daytime summer temperatures that surpass their optimal temperature for regrowth, resulting in lower yields, especially if there is no sufficient moisture (Sharratt et al., 1987).
The goal of our study was to examine how in situ warming affects alfalfa frost injury, alfalfa and orchardgrass yield, and the abundance of a winter annual weed, chickweed (S. media), during the establishment year in an alfalfa–orchardgrass mixture. We hypothesized that warmer temperatures would decrease frost injury in alfalfa and chickweed, but that temperature fluctuations during abrupt cold spells would eliminate these benefits. We also hypothesized that warmer temperatures would increase forage yields early in the growing season but would exacerbate yield reductions during the “summer slump” period (late July–August).
To do this, we used open-top chambers (OTCs) to impose the warming treatments in the field (Figure 1). OTCs are commonly used to passively increase temperature in a range of different ecosystems because they are relatively inexpensive and easy to construct, they do not require a power source (i.e., they rely on passive heating), and they increase both air and soil temperature compared with other warming structures that often do only one or the other (Hollister et al., 2022). While OTCs have been used across a range of ecosystems, there have been only a handful of studies that have evaluated OTCs in agroecosystems. Therefore, an additional objective of our study was to evaluate the performance of OTCs in an agroecosystem common in the northeastern United States.
Enlarge this image.FIGURE 1. Open-top chambers used to increase temperatures in the alfalfa–orchardgrass experiment at the Russell E. Larson Agricultural Research Center in Rock Springs, PA. Photo credit: Ashley Isaacson.
Field studies occurred at The Pennsylvania State University Russell E. Larson Research and Education Center in Rock Springs, PA (40.72° N, 77.92° W), in both 2021 (Year 1) and 2022 (Year 2). The climate (1991–2020) at the site is characterized by a mean annual air temperature of 10°C, and mean annual rainfall and snowfall of 1006 and 1140 mm, respectively (Palecki et al., 2021). Soils at the site are comprised of Hagerstown silt loam (fine, mixed, mesic Typic Hapludalfs).
Experimental designWe conducted two complementary experiments to examine the effects of warming on frost injury and growing season yield of cool-season perennial forages and weeds. The first experiment spanned the entire study period and evaluated the effects of warmer and more variable temperatures on frost injury and productivity of an alfalfa–orchardgrass mixture, as well as a winter annual weed, in the establishment year, and hereafter referred to as “warming.” In the second year, we added an additional factor to the warming experiment to more explicitly investigate how warming affects winter annual weed suppression of forage yields. This second experiment, hereafter warming × weediness, included both the original treatments in the warming experiment plus two weediness treatments (weeded and weedy).
Experiment 1: WarmingThis experiment examined how warmer and more variable temperatures affected frost injury and productivity of an alfalfa–orchardgrass mixture and a winter annual weed in the establishment year. Treatments consisted of (1) constant warming, (2) fluctuating warming, and (3) control with no imposed warming (ambient air temperature). In the constant warming treatment, the OTCs were present throughout the duration of the study and only removed for harvests. The goal of the fluctuating warming treatment was to exacerbate temperature fluctuations that can occur over short periods when mild winters are punctuated by extreme cold. In addition to being removed for harvests, the OTCs in the fluctuating warming treatment were also removed in winter when the ambient temperature was predicted to decrease from above ≥1.5 to below ≤−2°C in a 24-h period (see Appendix S1: Table S1 for OTC removal and reinstallation dates in fluctuating warming plots). Finally, the control treatments were established plots that were the same size and managed in the same way as the constant and warming treatments; however, they did not contain an OTC.
The OTCs constructed for our study were adapted from Marion et al. (1997) and consisted of hexagonal, open-top structures made up of six fiberglass panels 91 cm tall and 150 cm in diameter. OTCs were constructed from Sun-Lite HP fiberglass (purchased from the Kal-Lite Division of the Solar Components Corporation, Bow, Manchester, NH). Panels were connected via hex nuts and bolts, but with an eyelet bolt at the top attached with paracord, which tethered the OTCs to tent stakes in the ground.
Because we were interested in warming effects during the establishment year, the experiment ran from planting the forages in autumn until the final harvest in the subsequent autumn (approximately 1-year duration). The first year of the experiment occurred from September 2020 to September 2021 (hereafter referred to as Year 1). We then repeated the experiment in a different nearby field in the second year (September 2021–October 2022, hereafter referred to as Year 2). In both years of the study, the experiment was arranged as a randomized complete block design with four replicates. The plot size for all treatments was 1 m2. To determine the location of OTC and control plots within the field, we first established the blocks and then the plots (three in Year 1 and five in Year 2) were evenly spaced within each block, which ensured that each plot (either with an OTC or a control plot) was at least 3 meters apart.
Prior to planting in both years, the field was sprayed with glyphosate (Roundup PowerMax, Bayer Crop Science, St. Louis, MO; rate of 0.85 kg acid equivalent ha−1), chisel plowed, cultimulched, and fertilized according to soil test results (see Appendix S1: Table S2 for dates of key field activities). Prior to the initiation of the study in 2020, the field was in winter wheat that was harvested in mid-July. We reduced volunteer wheat in the alfalfa–orchardgrass mixture prior to planting our study by chisel-plowing and then allowing two flushes of wheat emergence. After each flush, we sprayed with glyphosate and then cultimulched the soil to prepare for planting. The field used for the second year of the study was previously kept fallow.
Alfalfa (M. sativa L. var “SW5213,” Fall Dormancy rating of 5) was direct-seeded at 6.8 kg ha−1 and orchardgrass (D. glomerata L. var “Extend”) was direct-seeded at 2.3 kg ha−1 (seeding rates recommended for our region in Hall, 2021) in order to have 30%–40% grass in the forage stand. Alfalfa stands in fall were approximately 200 and 300 plants m−2 in Year 1 and Year 2, respectively. Orchardgrass stands were approximately 200 plants m−2 in Year 2 (we did not conduct orchardgrass stand counts in Year 1). Prior to planting, alfalfa seeds were inoculated with N-DURE Alfalfa/True Clover inoculant (Verdesian Life Sciences, Cary, NC). Lambda cyhalothrin (Warrior II, Syngenta Crop Protection, Greensboro, NC) was applied directly after the second harvest timepoint in both years at 0.034 kg ai ha−1 to control potato leafhoppers (Empoasca fabae).
To evaluate how warming would affect the productivity of a common winter annual weed that is often problematic in fall-planted forages, we overseeded chickweed (S. media), across all plots in both years. In the first year of the study, we used a split-plot design with warming treatment as the main plot and presence of chickweed as the split-plot, so that chickweed was only in half the plot. Chickweed, which is a sprawling winter annual, spread throughout the entire 1-m2 main plot to the point that many of the split-plots without sown chickweed had chickweed biomass equivalent to those sown with chickweed. Consequently, we dropped the weedy split-plot and instead analyzed all the data at the main-plot level. In both years, we overseeded with locally collected chickweed, and thinned down to 15 plants after emergence (in Year 1, all 15 plants were sown in 0.5 m2 but spread throughout the whole 1-m2 plot, while in Year 2, the 15 chickweed plants were sown over an entire 1 m2).
Experiment 2: Warming × weedinessTo examine how warming affected winter annual weed suppression of forage yields more thoroughly, an additional experiment was established in Year 2. Two additional treatments were added to the warming experiment to make a full factorial of warming and weediness, with each factor having two levels. The warming treatments consisted of control (no OTC) and constant warming (with OTC). The weediness treatments were weedy (no weed management, included ambient weeds plus overseeded chickweed) and weeded (all weeds were removed twice throughout the year, once in November, to remove the early flushes of winter annual weeds, and again in April, to remove any weeds that emerged in early spring). Together, the four treatments in the warming × weediness experiment were: (1) control–weedy (same plots as control in the warming-only experiment), (2) control–weeded, (3) constant warming–weedy (same plots as constant warming in the warming-only experiment), and (4) constant warming–weeded. Therefore, the warming × weediness experiment occurred in the same field as the warming experiment during Year 2 of the study, but we added two additional treatments to the experimental design. Weedy treatments were sown with locally collected chickweed seeds when alfalfa and orchardgrass began emerging. Field preparation for the warming × weediness experiment was the same as for the warming-only experiment, and alfalfa and orchardgrass were planted at the same seeding rate as in the warming-only experiment.
Data collectionTo evaluate the effect of OTCs on forage canopy temperature, HOBO Pendant data loggers (Onset Computer, Bourne, MA) were secured inside solar radiation shields and attached to stakes that were placed in the center of each plot, positioning the temperature sensors 10 cm above the ground. Air temperatures were logged hourly from study initiation to termination. Temperature data were removed from the dataset during mowing and baling of forages, as there were periods when the OTCs and temperature sensors were removed from the plots. Growing degree-days were calculated using Equation (1), where Tmax and Tmin are the daily maximum and minimum temperatures and Tbase is the base temperature recommended for alfalfa (3.5°C, Sharratt et al., 1989).[Image Omitted. See PDF]
In October of each year, following planting, 15 randomly selected alfalfa plants and 15 chickweed plants in each plot were marked with a colored skewer, so that we could identify the same plant in spring, which we then evaluated for frost injury the following April. Frost injury was evaluated using a visual injury rating scale from 0 to 1 in increments of 0.1 and was based on the proportion of the plant showing signs of browning or purpling from frost injury, with 0 being completely green with no signs of injury and 1 being completely brown and dead.
In both years, four harvests were conducted at the early flowering stage of alfalfa, which is the common number of harvests for our region. In Year 1, biomass was collected from each 1-m2 plot using hand shears and was sorted into alfalfa, orchardgrass, chickweed, wheat, and other weeds. In Year 2 of the study, biomass was collected from each plot using a 0.5-m2 quadrat (to reduce time spent sorting the mixtures to species) and hand shears. The biomass was cut at 8 cm, sorted, oven-dried at 68°C for a week, and then weighed. Immediately after collection of forage biomass in our focal plots (1 m2 in Year 1, 0.5 m2 in Year 2), the alfalfa–orchardgrass mixture was mowed, raked, and baled in the rest of the field to simulate a standard forage harvest. Once the final harvest was collected in Year 1, the OTCs were removed and transferred to the field site utilized for Year 2 of the study.
Data analysisTo examine whether the OTCs affected air temperature within the forage canopy (average daily temperature, daily minimum temperature, and daily maximum temperature as separate response variables), we used a linear mixed-effects model using the lme function in the nlme R package (Pinheiro et al., 2022), with treatment as a fixed factor and treatment nested within block nested within date as a random intercept. Years were analyzed separately. Data were checked for normality, and to account for repeated measurements, we used Akaike information criterion values to evaluate whether including a corAR1 autocorrelation structure (autocorrelation order 1) improved model parsimony, which it did in both years and was therefore included in the model.
Frost injury ratingsTo analyze the effects of warming treatments on alfalfa and chickweed frost injury, we used generalized linear mixed models using the glmmTMB function in the glmmTMB package in R fit with a beta family distribution, which is recommended for proportional data (Brooks et al., 2017). The fixed effects in the model included warming treatment and year, and because we included frost injury ratings on multiple (15) plants within the plot in the analysis, we used treatment nested within block as the random intercept effect. To account for zeroes and ones in our frost injury data, we applied the following transformation recommended for proportional data modeled with a beta distribution (Verkuilen & Smithson, 2012):[Image Omitted. See PDF]where N is the sample size (15 plants per plot) and inj is the injury rating on a 0–1 scale.
Warming experiment effects on weed biomass and dry matter yieldChickweed and total weed biomass were analyzed only for the first harvest because in both years weed biomass was negligible beyond this timepoint. A linear mixed-effects model was used to quantify the effect of the warming treatment on chickweed or total weed biomass. Treatment, year, and the interaction of treatment and year were fixed effects, and block was included as a random intercept. Both chickweed and total weed biomass were log-transformed to improve the normality of the residuals, and variances were grouped by year using the varIdent function because of heteroscedasticity. A post hoc Tukey's honestly significant difference test was used for pairwise comparisons. All data were analyzed using R (R Core Team, 2022).
To determine the effect of warming treatments on forage dry matter yield, we used linear mixed-effects models using the lme function in the nlme package in R (Pinheiro et al., 2022), with warming treatment and harvest date, as well as their interaction, as fixed effects and treatment nested within block as a random intercept. Because of variation in harvest timepoints per year, years were analyzed separately. Each model was checked for normality and homogeneity of variance, and when necessary, the data were either log-transformed or variances weighted using the varIdent function to meet model assumptions.
Warming × weediness effects on forage dry matter yieldTo evaluate how levels of weediness (weedy or weeded) and levels of warming (control or constant warming) affect forage dry matter yield, we used linear mixed-effects models using the lme function in the nlme package in R (Pinheiro et al., 2022), with warming treatment and weediness, as well as their interaction, as fixed effects and block as a random intercept.
RESULTS Site environmental conditionsPrecipitation and temperature at our site varied considerably between the two years of our study (Figure 2). Year 1 was on average approximately 0.37°C cooler than the 30-year normal, while the second year of the study did not differ from the 30-year normal when averaged over the entire year. However, in both years, deviation from the monthly 30-year normal temperature varied considerably over the year. For example, November 2020 was 2°C warmer, while May, July, and September 2021 were 2.5, 1.4, and 3.8°C cooler than the 30-year normal, respectively. In Year 2 (September 2021–September 2022), October, December, and March were 2.3, 2.5, and 1°C warmer than the 30-year normal, respectively, while November, January, and April were 1.4, 1.8, and 1.6°C cooler than normal, respectively.
Enlarge this image.FIGURE 2. Average daily (A, B) air temperature and (C, D) precipitation at our research site (obtained from NRCS National Water and Climate Center, 2023) over the course of (A, C) Year 1 and (B, D) Year 2 of our study. (A, B) The black line is the observed daily average air temperature over the course of our study, and the red line is the 30-year normal average daily temperature at our site. (C, D) The black bars are daily precipitation in Year 1, the blue dashed line is the observed monthly total during our study, and the red solid line is the 30-year normal of monthly total precipitation.
Precipitation also varied widely between the two years of our study (Figure 2). Compared with the 30-year normal for our study location, Year 1 had overall greater precipitation than normal, while Year 2 was drier than normal. For example, in between the first and last harvest timepoints, we had 45 cm greater precipitation in Year 1 compared with Year 2. In Year 1, precipitation was 1 cm lower than the 30-year average for the month of June; however, we experienced 2.7, 3.7, and 17.9 cm greater precipitation than normal for July, August, and September, respectively. In Year 2, precipitation at our study site was 2.5, 4.4, 1.5, and 8.6 cm lower than the 30-year normal for June, July, August, and September, respectively.
Effects ofThe OTCs (in the constant warming treatment) raised the air temperature in the alfalfa–orchardgrass canopy by an average of 0.45°C in Year 1 (F1,1307 = 569.6; p < 0.001) and 0.72°C in Year 2 (F1,1011 = 618.7; p < 0.001; Appendix S1: Table S3), which was associated with 160.5 and 189 greater GDD accumulation in Year 1 and Year 2, respectively (Figure 3A,B). However, the warming effect of the OTCs varied over the course of the year (Figure 3C,D). In Year 1, when averaged over the entire month, the OTCs increased the average daily temperature by more than 0.9°C from February through April, as well as in the month of August. In Year 2, the greatest warming occurred in August, where monthly average air temperatures within the OTCs were elevated by 1.8°C above the control, while from January through April, the average temperature was increased by more than 0.85°C.
Enlarge this image.FIGURE 3. Growing degree-day accumulation throughout the project duration in (A) Year 1 (2020–2021) and (B) Year 2 (2021–2022); and average monthly difference in temperature between the constant warming and control treatments for (C) Year 1 (2020–2021) and (D) Year 2 (2021–2022). OTC, open-top chamber.
The OTCs increased the daily maximum temperature more than the daily average temperature or the daily minimum (Appendix S1: Table S3). For example, averaged across the entire year, the OTCs increased the average minimum temperature by 0.5 and 0.8°C in Year 1 and Year 2, respectively, while they increased the average maximum temperature by 1.4 and 2.0°C in Year 1 and Year 2, respectively. When the daily maximum temperature was averaged by month, we found maximum temperatures were 3°C greater within the OTCs in March, April, and August of both years, while in August of Year 2, the daily maximum temperature was increased by approximately 6.5°C when averaged throughout the month (data not shown). In both years, there were multiple days in which the OTCs caused extreme increases in the daily maximum temperature. For example, the OTCs increased daily maximum temperatures by more than 7°C on three days in April 2021, and one day in August 2021. In Year 2, we found seven days in late March through April and three days in August in which the OTCs increased the daily maximum temperature by more than 8°C. One of these days, April 2, 2022, the OTCs increased the daily maximum temperature by 10.8°C.
Interestingly, there were time periods during which the OTCs decreased the average daily temperatures compared with the control, which occurred in May of both years during which the OTCs decreased the average daily temperature by 0.59 and 0.37°C in Year 1 and Year 2, respectively (Figure 3C,D). The OTCs also decreased the average daily temperature by 0.2°C in July of Year 1. Again, these differences were driven more by the OTCs affecting the daily maximum temperature, because the monthly average of the daily maximum temperature in May was 2.1 and 1.7°C lower within the OTCs in Year 1 and Year 2, respectively.
Canopy temperature in the fluctuating warming treatment differed from the control and constant warming, both when the OTCs were on and during winter when the OTCs were removed (Appendix S1: Figure S1). Overall, however, when the OTCs were on the plots, temperatures within the forage canopy of the fluctuating warming treatment was similar to the constant warming treatment (Appendix S1: Table S3), and during the periods of winter when the OTCs were removed, the canopy temperature in the fluctuating warming was similar to the control (Appendix S1: Figure S1). For example, during the periods of winter when the OTCs were removed in the fluctuating warming treatments, the canopy temperature was ~1.15°C lower than the constant warming, and ~0.08°C higher than the control. During the rest of the experiment when the OTCs were on the plots in the fluctuating warming treatment, the fluctuating warming was 0.02 and 0.2°C lower than the constant warming in Year 1 and Year 2, respectively, and 0.4 and 0.5°C warmer than the control in Year 1 and Year 2, respectively.
Experiment 1: Warming effects on alfalfa and chickweed frost injuryAs we hypothesized, over both years of our study, we found decreased alfalfa frost injury in the constant warming compared with the control (Figure 4, Table 1; χ2 = 8.2; p = 0.017), and these effects did not vary by year (χ2 = 4.1; p = 0.129). The alfalfa plants within the fluctuating warming had similar levels of frost injury as the control plots.
![View Image - FIGURE 4. (A) Alfalfa and common chickweed frost injury rating (mean ± SE) measured on 15 plants in April of both Year 1 (2020–2021) and Year 2 (2021–2022) in the control, constant warming (constant), and fluctuating warming (fluctuating) treatments. Means with different letters are significantly different at α [less than] 0.05. (B) Examples of frost injury on alfalfa plants along the 0–1 frost injury scale used to assess frost injury in both alfalfa and chickweed, with 0 indicating no injury and 1 indicating plants that are completely dead and brown. Photo credits: Ashley Isaacson.](https://media.proquest.com/media/hms/THUM/8/AFVuY?_a=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&_s=pEDQJ6VfpwDc9lpmiam%2BJMxQhjg%3D "View Image - FIGURE 4. (A) Alfalfa and common chickweed frost injury rating (mean ± SE) measured on 15 plants in April of both Year 1 (2020–2021) and Year 2 (2021–2022) in the control, constant warming (constant), and fluctuating warming (fluctuating) treatments. Means with different letters are significantly different at α [less than] 0.05. (B) Examples of frost injury on alfalfa plants along the 0–1 frost injury scale used to assess frost injury in both alfalfa and chickweed, with 0 indicating no injury and 1 indicating plants that are completely dead and brown. Photo credits: Ashley Isaacson.")
Enlarge this image.FIGURE 4. (A) Alfalfa and common chickweed frost injury rating (mean ± SE) measured on 15 plants in April of both Year 1 (2020–2021) and Year 2 (2021–2022) in the control, constant warming (constant), and fluctuating warming (fluctuating) treatments. Means with different letters are significantly different at α [less than] 0.05. (B) Examples of frost injury on alfalfa plants along the 0–1 frost injury scale used to assess frost injury in both alfalfa and chickweed, with 0 indicating no injury and 1 indicating plants that are completely dead and brown. Photo credits: Ashley Isaacson.
TABLE 1 Results from generalized linear mixed-effects model analyzing the effects of open-top chamber treatments (constant and fluctuating warming) compared with the control on alfalfa and chickweed frost injury.
| Effect | χ2 | df | Pr(>χ2) |
| Alfalfa | |||
| Treatment | 8.17 | 2 | 0.017 |
| Year | 7.51 | 1 | 0.006 |
| Treatment:Year | 4.1 | 2 | 0.129 |
| Chickweed | |||
| Treatment | 13.5 | 2 | 0.001 |
| Year | 5.66 | 1 | 0.017 |
| Treatment:Year | 8.33 | 2 | 0.016 |
We found that constant and fluctuating warming effects on chickweed injury varied by year (Figure 4, Table 1; χ2 = 8.3; p = 0.016). In Year 1, we found greater chickweed frost injury in the control treatment compared with the constant and fluctuating warming treatments; however, contrary to our expectations, we found no difference between the constant and fluctuating warming. In the second year of the study, chickweed frost injury was relatively low, and we found no differences in chickweed injury across any of our treatments.
Experiment 1: Warming effects on alfalfa, orchardgrass, and total forage yieldOTC effects on forage dry matter yield varied across harvest timepoints within a year and between the two years of our study (Figure 5, Table 2). We found support for our hypothesis that warming from OTCs would increase forage yield early in the season. At the first harvest timepoint in both years, both the constant and fluctuating warming treatments increased total forage yield compared with the control (Figure 5A,B). In the first year of the study at the first harvest timepoint, we observed over 60% greater total forage yield in the constant and fluctuating warming treatments compared with the control, which resulted largely from ~65% greater alfalfa dry matter yield within the warming treatments. In Year 1, we also observed a 72% increase in orchardgrass dry matter yield in the constant warming compared with the control at the first harvest timepoint. In the second year of the study, we found an ~20% increase in total forage yield at the first harvest timepoint within both warming treatments. However, we saw ~39% lower cumulative alfalfa dry matter yield in the fluctuating warming treatment, yet this was compensated by 50% greater orchardgrass dry matter yield in both OTC treatments.
Enlarge this image.FIGURE 5. Dry matter yield of alfalfa, orchardgrass, and the total forage (alfalfa + orchardgrass; mean ± SE) in the control, constant warming (constant), and fluctuating warming (fluctuating) treatments in both Year 1 and Year 2 at each cutting timepoint (T1 through T4, plus the sum of all harvests). Letters indicate treatments that are significantly different from one another within a species and year (years were analyzed separately, and we used a Tukey's honestly significant difference test for pairwise comparisons).
TABLE 2 Results from linear mixed-effects model analyzing the effects of open-top chamber treatments (constant and fluctuating warming) on alfalfa, orchardgrass, and total forage yield in both Year 1 (2020–2021) and Year 2 (2021–2022).
| Effect | Year 1 | Year 2 | ||||
| df | F | p | df | F | p | |
| Alfalfa | ||||||
| Treatment (Trt) | 2, 6 | 2.57 | 0.16 | 2, 6 | 0.829 | 0.481 |
| Timepoint (Tpt) | 3, 27 | 17.4 | <0.001 | 3, 26 | 15.5 | <0.001 |
| Treatment:Timepoint | 6, 27 | 2.71 | 0.03 | 6, 26 | 2.26 | 0.069 |
| Orchardgrass | ||||||
| Treatment | 2, 6 | 12.8 | 0.01 | 2, 6 | 3.4 | 0.102 |
| Timepoint | 3, 27 | 8.14 | <0.001 | 3, 26 | 39.9 | <0.001 |
| Treatment:Timepoint | 6, 27 | 3.08 | 0.02 | 4, 26 | 2.78 | 0.032 |
| Total forage | ||||||
| Treatment | 2, 6 | 12.98 | 0.007 | 2, 6 | 1.6188 | 0.274 |
| Timepoint | 3, 27 | 5.96 | 0.003 | 3, 26 | 53.134 | <0.001 |
| Treatment:Timepoint | 6, 27 | 9.67 | <0.001 | 6, 26 | 2.1951 | 0.076 |
We found contrasting effects of OTCs on forage yields across the second harvest timepoint of both years (Figure 5C,D). As with the first harvest timepoint, in Year 1, we found that constant and fluctuating warming increased alfalfa dry matter yield (a 32% increase) in the second harvest timepoint (Treatment × Date, F = 2.64; p = 0.037). However, we found no effect of OTC treatments on orchardgrass dry matter yield, nor on total forage yield. In contrast, at the second harvest in Year 2, neither the alfalfa nor orchardgrass dry matter yield differed in the constant or fluctuating compared with the control.
August (our third harvest timepoint) is traditionally when cool-season forages experience the “summer slump.” We hypothesized that warmer temperatures from OTCs would decrease forage yields during the August summer slump period (Duiker et al., 2023). Lower forage yield at the third harvest timepoint only occurred in Year 2, in which the constant warming decreased total forage yield by ~60%, which resulted both from a 77% decrease in alfalfa dry matter yield and a 55% decrease in orchardgrass dry matter yield (Figure 5E,F). Total forage yields in August were greater within the fluctuating warming treatments in Year 1, which was due to a 45% increase in alfalfa dry matter yield because orchardgrass did not differ across treatments. In Year 1 at the third harvest timepoint, we found no difference between the constant warming and control in total forage, alfalfa, or orchardgrass dry matter yields.
In both years, we found no effect of either warming treatment on total forage yields in the final harvest timepoint (Figure 5G,H). However, in both years, alfalfa dry matter yield was lower in the constant warming compared with the control. In contrast, the effects of warming treatments on the orchardgrass yield in the final timepoint varied between years. In the first year, orchardgrass dry matter yields did not differ across warming treatments, while in the second year, orchardgrass dry matter yields were higher in both constant and fluctuating warming compared with the control.
When the forage dry matter yield was summed across all harvests for the entire establishment year, we only found an effect of OTC treatments in Year 1, but not in Year 2 (Figure 5I,J). In Year 1, we found that the constant warming increased total forage and orchardgrass yield across the establishment year but had no effect on alfalfa. In both years, the annual sum of alfalfa, orchardgrass, or total forage dry matter yield in the fluctuating warming treatment did not differ from the control or constant warming treatment.
Experiment 1: Warming effects on chickweed and other weed biomassChickweed biomass was only present on the first cutting date of both years. Neither OTC warming treatment affected chickweed biomass at the first harvest timepoint (Figure 6; Appendix S1: Table S4). We also found no effect of OTC warming treatments on the biomass of all other weeds (Figure 6; Appendix S1: Table S4). Total weed biomass at the first harvest timepoint averaged over 1100 and 717 kg ha−1 in Year 1 and Year 2, respectively. However, after the first harvest timepoint, weed biomass significantly declined to less than 75 kg ha−1 at the second timepoint, and was negligible (less than 6 kg ha−1) at later harvest timepoints. While we found no effect of OTC treatments on total weed biomass, we did find a negative association between weed biomass and total forage dry matter yield at the first harvest timepoint (Appendix S1: Figure S2).
Enlarge this image.FIGURE 6. Chickweed biomass or biomass of all weeds (including chickweed; mean ± SE) at the first harvest timepoint in late May of both Year 1 and Year 2 in the control, constant warming (constant), and fluctuating warming (fluctuating) treatments. No significant differences were found when comparing treatments in either year. NS, not significant.
To more thoroughly examine whether winter annual weeds varied in their suppression of forage yield at the first harvest timepoint in spring, in Year 2, we added two treatments in which we removed all weeds at two timepoints, once in November and again in April. However, despite these weeding events causing large differences in weed biomass at the first harvest timepoint in Year 2 (Appendix S1: Figure S3), we found no difference in forage dry matter yields between the weedy and weeded treatments, nor did we find any interaction between weeded treatments and warming treatments (Figure 7).
Enlarge this image.FIGURE 7. Dry matter yield of alfalfa, orchardgrass, as well as the total forage (alfalfa + orchardgrass; mean ± SE), in the control and constant warming (constant) treatments that were either weedy (not weeded at all plus included added chickweed plants) or weeded (no added chickweed and all weeds removed once in fall and once in early spring) in Experiment 2 at the first cutting date. Letters indicate treatments that are significantly different from one another using Tukey's honestly significant difference test for pairwise comparisons.
Our study examined how in situ warming affected alfalfa frost injury, alfalfa and orchardgrass yield, and the abundance of a winter annual weed, chickweed (S. media), during the establishment year in an alfalfa–orchardgrass mixture. We used OTCs to impose the warming treatments in the field. Our results support our hypothesis that warmer temperatures decrease frost injury, and that temperature fluctuations during abrupt cold spells eliminate any benefits from warmer temperature. However, we found that the effect of warmer temperatures on chickweed frost injury varied between years. We also found support for our hypothesis that warmer temperatures would increase forage yields early in the growing season (at the first harvest timepoint); however, the effects of warmer and more variable temperature on forage yield throughout the remaining harvests were inconsistent. We found no effect of warming treatments on the abundance of chickweed, nor all other weeds found within our study. Finally, although weeding reduced chickweed abundance during the second year, chickweed abundance had no effect on total forage yield within and among warming treatments.
OTCs within our study system increased daily temperature within the alfalfa–orchardgrass canopy by between 0.45 and 0.72°C, which is reflective of the level of additional warming predicted by the IPCC by 2050 (IPCC, 2022). While the level of warming we found with the OTCs is less than the 1.5–2.0°C reported by studies using OTCs in the Arctic tundra (Hollister & Webber, 2000; Post & Pedersen, 2008), it is on par with warming achieved by OTCs in studies within northern climates of the continental United States. For example, in a separate experiment done close to our study site in central Pennsylvania, OTCs increased air temperature by an average of 0.31°C (Keller & Shea, 2022), while in an agroecosystem in Montana, OTCs increased temperature by approximately 0.8°C (DuPre et al., 2022). We note that the strongest temperature increases we observed in the OTCs were for maximum temperatures over daily and monthly timescales, with less warming for minimum temperatures. This pattern is consistent with warming that occurs during the day due to the OTC trapping solar energy. However, it diverges from long-term trends of more rapidly increasing minimum, or nighttime, temperatures occurring in both the northeastern United States and other areas of the globe (Cox et al., 2020; Vose et al., 2005), highlighting a limitation of the OTC method in replicating minimum temperature increases that are already occurring and are projected to continue with climate change.
It is unclear why we observed some time periods throughout the year during which the OTCs decreased canopy air temperatures compared with the control. The shading and transpirational cooling within the alfalfa–orchardgrass canopy may have buffered the extent to which OTCs increased forage canopy temperature. Variation in solar radiation is a dominant factor influencing the degree to which OTCs increase internal air temperature. However, when examining solar radiation at the site during our study, we did not find that months with an average decrease in temperatures within the OTCs compared with the control had lower solar radiation (data not shown). An alternative and more likely explanation is that greater forage productivity within the OTCs resulted in a cooling effect from increased transpiration (Ayeneh et al., 2002). This interpretation is supported by the patterns in forage dry matter yield we observed over the growing season. In both Year 1 and Year 2, we saw the greatest increase in forage dry matter yield within the OTCs at the first harvest timepoint, which occurred in the last week of May (Figure 7), and in both years, we saw the greatest decrease in canopy temperatures within the OTCs during that month. Congruently, the month of May typically featured sufficient precipitation such that forage productivity was unlikely to be limited by soil moisture, which would increase the potential for cooling via evapotranspiration.
Warmer temperature effects on alfalfa and chickweed frost injury in the establishment yearWe found support for our hypothesis that the elevated temperatures within the OTCs would decrease alfalfa frost injury. However, periodically removing the OTCs when temperatures dropped below freezing within the fluctuating warming treatment eliminated any benefit to the alfalfa plants that would have otherwise resulted from warmer temperatures throughout the rest of the winter. It is important to note that even though winter is warming faster than any other season, changes in the polar jet stream may lead to more frequent incursions of extreme cold Arctic air over temperate areas (Cohen et al., 2019; Francis & Skific, 2015). The resulting “winter weather whiplash” (Casson et al., 2019) may exacerbate temperature oscillations that result in severe reductions in the winter survival of cold-sensitive perennial forage crops (Fernandez et al., 2020). Greater winter weather variability, combined with reduced snow cover and a shortening of the autumn hardening period, is projected to increase both the frequency and severity of winter kill in perennial forage crops in future years (Bélanger et al., 2002; Hristov et al., 2018).
Ultimately, frost injury may decrease plant survival, thereby reducing alfalfa abundance in the subsequent spring (Bélanger et al., 2002; Castonguay et al., 2006). We found a weak, nonsignificant correlation (p = 0.089) between alfalfa frost injury and alfalfa dry matter yield at the first harvest timepoint (Appendix S1: Figure S4), suggesting that while frost injury may have affected alfalfa productivity, it is likely that factors other than frost injury were equally or more important. For example, while alfalfa frost injury may explain reduced alfalfa yield in the fluctuating warming treatment in Year 2, it is unlikely to be the sole reason for reduced yields, because we did not see the same effect in the control despite similar levels of frost injury. A more likely explanation is that increases in orchardgrass biomass within the fluctuating warming treatment in Year 2 suppressed alfalfa growth.
We hypothesized that significant alfalfa frost injury in the control or fluctuating warming treatments would benefit winter annual weeds, such as chickweed, at the first harvest timepoint (Cici & Van Acker, 2009). However, while we saw a pattern of greater chickweed biomass in the control and fluctuating warming in Year 2, the high variability within treatments precluded us from detecting any differences statistically. Overall, we saw limited evidence that chickweed abundance was strongly influenced by frost injury because we saw no association between chickweed frost injury and biomass, likely because chickweed plants exhibited less frost injury compared with alfalfa plants (Appendix S1: Figure S5).
Winter annual weeds, such as chickweed, can be especially problematic in fall-planted perennial forages in the establishment year, because perennial forages tend to be slower growing and less competitive compared with winter annual species (Hall et al., 1995; Peters & Peters, 1972). The fact that we found no difference in forage yield between our weeded and weedy treatments suggests that weed competition was not a major factor influencing forage yield in Year 2. However, over the two years combined, we did see a negative association between weed biomass and total forage productivity at the first harvest timepoint (Appendix S1: Figure S2), suggesting that overwintering weeds may have suppressed total forage productivity early in the season.
Warmer temperature effects on total forage dry matter yield throughout the establishment yearPrecipitation and temperature varied widely between the two years of our study, which may explain the differences we found in treatment effects on forage yield between the two years (Thivierge et al., 2016). For example, we found a greater benefit from warmer temperatures within the OTCs in Year 1, the year that was comparably cooler and wetter. In contrast, the summer of Year 2 was drier than the 30-year normal, and precipitation at our study site in Year 2 was nearly 36 cm less from March through September compared with Year 1 (Figure 2). It is likely that drought stress eliminated any benefit that would have occurred from warmer temperatures within the OTCs and likely explains why we found declining forage yields throughout the warmest summer period. Other studies examining the effect of climate change on alfalfa productivity have also found that increasing temperatures are likely to lead to increases in alfalfa yield, but only when water is not limiting. For example, in Australia, climate models predict that potential decreases in precipitation will greatly reduce any benefit of warmer temperatures for alfalfa productivity (Pembleton et al., 2016). Across both years, the total annual forage yield was within the range of average forage yields for alfalfa grown globally in rain-fed conditions, which was reported to be between 5 and 17 ton DM ha−1 (Moot et al., 2012), and in Year 1, yields (~10–11 ton ha−1) within warming conditions are similar to yields reported in both China (Feng et al., 2022) and southeastern Australia (Pembleton et al., 2016).
We hypothesized that OTCs would be most likely to cause yield declines during the warmest period of the summer, in August for our region, thereby exacerbating the “feed gap” that typically occurs when hot and dry summer conditions limit the growth of cool-season perennial forage species (Moore et al., 2009). We found support for this hypothesis only in Year 2, when precipitation was well below the 30-year normal, and total forage yield in the constant warming treatment was ~77% lower compared with the third harvest timepoint in Year 1. Orchardgrass yield was especially sensitive to the dry summer months in Year 2, in which we saw reduced yields in constant warming at both the second and third harvest timepoint, as well as a reduction in yield in the fluctuating warming at the third harvest timepoint. Temperature stress may reduce growth and tillering of orchardgrass, especially when precipitation is limiting, which could be why we saw a reduction in orchardgrass yield within the OTC treatments (Blaker & Jung, 1968; Wolfe et al., 2018).
As we hypothesized, we found the greatest increase in forage biomass from warmer temperatures in late spring at the first harvest timepoint, when ambient temperatures were cool, and precipitation generally was not limiting. We expected that alfalfa would benefit more from the warmer temperatures within the OTCs over winter compared with orchardgrass because orchardgrass has a greater winter hardiness compared with alfalfa (especially alfalfa with a fall dormancy of 5) (Belanger et al., 2006). However, we only saw an increase in alfalfa productivity in the constant warming in Year 1, despite the constant warming increasing GDD by only 7% in Year 1 (compared with 10% in Year 2). Despite this, we saw a consistent increase in orchardgrass dry matter yield in the constant warming across both years. However, we only saw an increase in alfalfa productivity in the constant warming in Year 1, despite the constant warming increasing GDD by only 7% in Year 1 (compared with 10% in Year 2).
CONCLUSIONWarmer temperatures within OTCs consistently decreased alfalfa frost injury and increased total forage yield at the first harvest timepoint in spring but had variable effects at later harvest dates. Overall, the warmer temperature within the OTCs increased forage yield only in 2021, the year with greater precipitation. In contrast, precipitation in 2022 was low, and therefore, we saw limited yield benefits from warmer temperatures, especially during the summer slump in which yield declines were exacerbated by warmer temperatures within the OTCs. As global temperatures continue to rise in the future, it is likely that yield loss of perennial forages, such as alfalfa and orchardgrass, will be exacerbated in years without adequate precipitation. Therefore, exploring heat- and drought-tolerant forage alternatives will increasingly be important in the future, especially where access to irrigation is limited.
It is important to note that we included only one cultivar each of alfalfa and orchardgrass within our study to evaluate the effect of warmer and more variable temperature on alfalfa frost injury and forage yield. However, a common adaptation to climate change at higher latitudes may be a change in cultivars and their associated fall dormancy and winter hardiness characteristics (Thivierge et al., 2023). Already farmers are shifting toward cultivars with a higher fall dormancy rating (indicating less dormancy) and lower winter hardiness. Farmers are also doing an additional harvest at the end of the growing season to take advantage of the increased growth resulting from reduced dormancy under cooler temperatures. Therefore, future research should evaluate the effect of warmer and more variable temperatures on multiple cultivars and number of harvests per year to determine how weather variation may interact with management to affect crop productivity and stability.
The results of our study also provide valuable insights into the effects of OTCs on site environmental conditions, as well as frost injury of alfalfa and chickweed, forage yield, and weed biomass. OTCs are one of the most cost-effective forms of increasing temperatures in the field, but they are not without their drawbacks, such as inhibiting wind or causing inconsistent effects on temperature throughout the year. Within our study, it is likely that the warmer temperatures within OTCs increased plant productivity over fall and spring, thereby increasing transpiration cooling and decreasing temperature within the OTCs in May. Additionally, we observed the greatest increase in maximum daily temperature and less warming for minimum temperature, which is inconsistent with current trends and future projections of more rapidly increasing nighttime minimum temperature. It is important for future researchers interested in using OTCs to understand their limitations.
AUTHOR CONTRIBUTIONSAll authors contributed to the conception and design of this experiment. A. Isaacson and C. Stamplis collected the data, while A. Isaacson and C. J. Lowry analyzed the data. A. Isaacson and C. J. Lowry led the writing of the manuscript, with all other authors providing insightful and critical comments, and approving the final version of this manuscript.
ACKNOWLEDGMENTSWe thank John Wallace for his helpful comments on this manuscript. We also thank a team of undergraduate researchers for their assistance in the field, as well as with sample processing. Finally, we thank the entire research farm staff at the Russell E. Larson Agricultural Research Center for assistance with agronomic activities. Support for this research was provided by the U.S. Department of Agriculture–National Institute of Food and Agriculture, including Agriculture and Food Research Initiative grant No. 2021-09936 and Hatch Appropriations under Project No. 1025327-PEN04759.
CONFLICT OF INTEREST STATEMENTThe authors declare no conflicts of interest.
DATA AVAILABILITY STATEMENTData (Lowry, 2024) are available from Scholarsphere:
© 2024. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Smith, Richard G 3 ; Stamplis, Christopher 1 ; Lowry, Carolyn J 1
1 Department of Plant Science, The Pennsylvania State University, University Park, Pennsylvania, USA
2 Earth Systems Research Center, Institute for the Study of Earth, Oceans, and Space, University of New Hampshire, Durham, New Hampshire, USA
3 Department of Natural Resources and the Environment, University of New Hampshire, Durham, New Hampshire, USA