The future of growing rice in Vermont: Managing for climate change

Germination trials

The seeds were pre-soaked in water between 6 and 8 days (Picture 1) and subsequently sown in a germination soil mix following the protocol in the Rice Growing Manual (Akaogi 2009). At JM in Shelburne, the seed trays were covered and placed in a heated greenhouse. At WSRF in Moretown, the seed trays were covered and placed in unheated cold frames. The seeds took ~7 days to germinate and germination rates varied among different varieties of rice (Figure 1). Matsumae had the highest germination rate at both farms. Over 50% of M202 seeds germinated in heated greenhouses at JM, whereas less than 30% of M202 seeds germinated in cold frames at WSRF. Hayayuki germination rates were quite low at JM (less than 50%) compared to WSRF (90%). Finally, Akitakomachi had the lowest germination rates at both farms. Interestingly, the germination rates were significantly higher at WSRF in Moretown for all but 1 strain (i.e., California strain M202), indicating that cold frames are sufficient for germinating rice varieties from northern Japan and heated greenhouses are unnecessary for successful on-farm rice germination.

We measured seedling height for an indirect estimate of seedling vigor for the Shelburne trials. It is important to note that some strains are bred to grow taller than other strains, so seedling height may has a slight bias as an estimate of vigor. Nevertheless, M202 and Hayayuki had the most vigorous seedlings followed by Matsumae, and Akitakomachi exhibited the lowest germination rates and seedling vigor (Figure 2).

Early season (May – June) estimates of plant vigor - WSRF

Early season plant vigor was estimated for plants prior to panicle formation on each farm, WSRF(Zone 4a) and JM (Zone 4b). Rice is a grass that produces tillers (stems) throughout the growing season. Most tillers develop into reproductive panicles, which mature into fruits containing the seeds (grains). We used tiller measurements to assess early growth vigor among rice plants at each farm. We estimated average early-season plant vigor by multiplying total number of tillers by tiller height to yield a representative numeric value for overall plant vigor (Hill Bermingham 2010). We used analysis of variance (ANOVA) statistical tests to assess differences in plant vigor among water treatments and rice varieties. Individual t-tests were used to detect differences between pairs if the ANOVA model was significant.

We analyzed differences in early season plant vigor among rice varieties by building an ANOVA model with both rice variety and farm. The entire model detected significant differences in plant vigor (P < 0.0001). Early season plant vigor differed among rice varieties (P = 0.04) and between the 2 farms (P < 0.0001). Rice plants of all varieties at WSRF were significantly more vigorous in the early season compared to JM (Figure 3).

The difference in fertilization frequency between the 2 farms most likely drove differences in early season plant performance. Young seedlings at WSRF were fertilized with organic fertilizers, but seedlings were not fertilized at JM. Organic, high-N fertilizers (e.g., duck and chicken manure at WSRF and fish emulsion at JM) were used to fertilize to the plants. Fertilizers were first applied at WSRF when the seedling’s leaves began to yellow, and fertilizer was applied once every 2 weeks or as needed prior to transplant. Given these results, rice seedlings should be fertilized as seedlings prior to transplant into buckets or paddies. It should also be noted that Ben Falk at WSRF has years of experience growing rice, and it was the first year of rice farming at JM in Shelburne.

One of the major goals of the project was to determine whether climatic differences between the 2 farms would result in performance differences among rice varieties. JM Plants and Produce is located in the Champlain Valley of Vermont where temperatures tend to be significantly warmer than central Vermont were Whole Systems Research Farm is located in Moretown. At JM, Hayayuki was the most vigorous variety early in the season, followed by Matsumae and M202 (Figure 3). Akitakomachi was the least vigorous variety and differed significantly from all varieties (Akitakomachi – Hayayuki: P = 0.02; Akitakomachi – M202: P = 0.009; Akitakomachi – Matsumae: P = 0.004). On the other hand, Matsumae was the most vigorous variety, followed by Hayayuki, at WSRF. The least vigorous variety at both farms was Akitakomachi.

We detected significant differences in estimates of early plant vigor among water treatments at JM. The control plants were significantly more vigorous compared to the saturated and low treatment plants (P = 0.004; Figure 4). The average plant vigor for the 3 water treatments are as follows: control = 63.0; low = 45.1, and saturated = 43.1. This finding correlates with the notion that water retains a significant amount of heat and creates a much milder environment for rice early in the season (Akaogi 2009). Plant vigor between low and saturated treatment plants did not differ significantly (P = 0.74). However, plant vigor between control plants and low water plants differed (P = 0.005), as well as between control plants and saturated plants (P = 0.002). There were no significant differences in early plant vigor among experimental water treatments at WSRF (P = 0.81).

We had the opportunity to determine differences between rice plants grown in 5-gallon buckets versus rice grown in paddies at WSRF. We found interesting results, as plants grown in buckets were significantly more vigorous early in the season compared to plants grown in paddies (Table 3). Plants grown in buckets, on average, had a plant vigor value of 366.0, where plants grown in paddies had a plant vigor value of 293.8. Recall that plant vigor is the products of number of tillers and tiller height and is a metric that estimates early season plant size. Black, 5-gallon buckets may provide rice plants with warmer conditions early in the season compared to paddies at WSRF. Also, some buckets contained wind-dispersed Azolla, a nitrogen-fixing water fern that grows in the paddies. The buckets with Azolla were noted in the dataset, and no differences were found in early season plant vigor between buckets containing Azolla and those not containing Azolla. Azolla was removed from all buckets after these early season measurements were collected to keep the experiment controlled.

Late season (July – September) estimates of plant vigor – JM Plants and Produce

In mid-July at both farms, the non-reproductive tillers began maturing into reproductive culms that produce panicles, the inflorescences of grasses that produce a cluster of spikelets (i.e., grass flowers) that, once pollinated, mature into the grain. At this time, we set up bird netting at both farms to prevent birds from eating the grain.

We gathered data on panicle height (i.e., measured from the base of the plant to the top of the panicle) on July 19 2012 solely at JM to determine whether there were differences in the growth of reproductive plants among water treatments and rice varieties. Significant differences in late season panicle height were detected among water treatments (P < 0.0001; Figure 5), as control plants had the largest panicles, followed by saturated treatments plants and low water treatment plants. We analyzed each pair of water treatments (i.e., C – L, S – C, and S – L) with t-tests, and each pair was significantly different from one another (all pairs, P < 0.0001).

Throughout the month of September, we counted the number of panicles at JM for a late season estimate of plant vigor. The vast majority of low water treatment plants did not survive throughout the season and/or did not produce panicles. Akitakomachi did not have mature panicles as of our last data collection on September 25, 2012, but all other rice varieties had mature panicles. Akitakomachi is a variety cited to be late-maturing in the Northeast (Takeshi Akaogi, personal communication). Matsumae and M202 tended to produce the most panicles at the end of the season compared to Hayayuki. However, there were no significant differences among the 3 rice varieties that produced panicles at JM (P = 0.19). Number of panicles produced by rice plants in the 3 experimental water treatments differed significantly (P = 0.02). Surprisingly, the low water treatment plants produced the most panicles (average number of panicles: saturated = 14.8; control = 14.2; low = 24.8). This result may be driven by the fact that all surviving low water treatment plants were the Matsumae variety, which had the largest number of panicles on average (Table 1).

Grain yield at JM

Rice farmers are most interested in the grain yield. We grew an estimated 2500 lbs per acre of rice in buckets at JM, which is substantial but significantly less than the 6500 lbs per acre yielded at the Akaogi Farm in southern Vermont (Akaogi NESARE Report 2008, available by request through NESARE). We used 240 square feet of space at JM, and 168 square feet of plants produced viable and harvestable grain. We harvested all stems from plants at JM in late September 2012. At harvest time at JM, we noted a large amount of browsing by field mice, and we also observed a small amount of birds eating the grain because they flew under the netting to access the seed. Harvested stems were transported to the Plant Biology laboratory at the University of Vermont to score seed set and seed weight.

We used analysis of variance (ANOVA) to assess differences in total grain yield and grain weight among the 3 water treatments (i.e., control, saturated, and low) at JM. We did not detect significant differences in total grain yield (P = 0.50) or grain weight (P = 0.27) among water treatments at JM. It is significant to note that, once again, there were absolutely no differences in grain yield between the control and saturated plants (P = 0.95), indicating again that rice grown in the Northeast does not require a high water table to flourish throughout the season and produce ample grain.

Matsumae and Hayayuki varieties had the highest yields at JM, as well as WSRF (results presented below). These results, coupled with the germination and seedling vigor data presented in Figures 1 and 2, indicate that Hayayuki and Matsumae are the 2 rice varieties best suited to the Northeastern climate of the 4 strains that were tested in this experiment. However, the California strain M-202 also yielded a significant amount of grain at JM in Shelburne, VT in the warmer Champlain Valley. These results indicate that California rice varieties (e.g., M202) may be less tolerant of cold temperatures and should be avoided in colder, higher elevation regions of the Northeast but can be grown with success in warmer regions, such as the Champlain Valley. The Akaogi’s trialed 25 varieties of rice in collaboration with Cornell University in 2008 (Akaogi NESARE Report 2008, available by request through NESARE), but only 2 strains were California strains (M-16 and M-201). Farmers in warmer regions of the Northeast should freely experiment with growing less cold-hardy rice varieties outside of the northern Japan region, as has been traditionally grown in the state.

Grain yield at WSRF

We multiplied the average number of seeds per panicle by the average number of panicles per plant to estimate yield for our 4 rice varieties under 3 experimental water treatments. Data was collected only from rice plants that were not browsed by animals. There was a large degree of end-of-season browsing by farm-raised ducks at WSRF.

The experimental design at WSRF allowed us to compare grain yield in bucket-raised rice plants compared to paddy-raised plants. We found that bucket-raised rice plants yielded significantly more grain compared to the paddy plants (average total seed production per plant: bucket = 927.2; paddy = 257.5; P < 0.0001; Figure 6). Although it was not directly tested, there could be a variety of reasons why bucket-raised plants outperformed paddy plants, including, but not limited to increased temperatures in buckets and/or lower weed pressure in buckets. Farmers that have interest in growing rice but have not yet constructed paddies, we suggest germinating ~5 g of seed for each variety of interest. Then, transplant 2-3 plants per 5-gallon bucket. This is a reasonable, cost-effective, and successful method of growing seed stock for the first year of rice production.

We used analysis of variance (ANOVA) to assess differences in total grain yield among the 3 water treatments (i.e., control, saturated, and low) at WSRF. There were marginally significant differences among the 3 treatments (average total seed production per plant: control = 673.1; low = 427.2; saturated = 749.3; P = 0.06; Figure 7). However, when individual t-tests were used to make paired comparisons between the average grain yield between pairs of water treatments, some significant patterns emerged from the dataset. The “saturated” and “low” water treatments had significantly different grain yields (S – L; P = 0.05), and the “control” and “low” water treatments were marginally different (C – L; P = 0.06). However, the “saturated” and “control” water treatments were not significantly different from one another in terms of total grain yield (S – C; P = 0.65), indicating that rice farming in the Northeast need not be as water-intensive as is normally practiced on-farm. In fact, plants raised in saturated water conditions produced more seeds, on average, compared to control plants, although this was not a statistically significant difference. However, many Northeastern farmers utilize water to flood rice paddies as a method of weed control. Flooded paddies create anaerobic conditions that release large amounts of methane gas, a potent greenhouse gas, into the atmosphere, therefore exacerbating the effect of climate change. We recommend alternative methods of weed suppression but realize the labor-intensive repercussions due to the lack of mechanization for rice farming in our region.  

Even though rice does not require flooded soils, many farmers in the Northeast use flooding as a means of weed control. However, flooding the soil creates anaerobic conditions and results in a build-up of methane, one of the most potent greenhouse gasses that accelerate global warming. Rice paddies are considered one of the most important anthropogenic sources of atmospheric methane that contributes to climate change, as cited in the Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventories Reference Manual (1996). Flooding paddies is an unsustainable method of weed control. Flooded conditions boost early growth for the rice plants but are unnecessary past the early season (mid-June). Sustainable water use management strategies for rice farmers is a top priority of the International Rice Research Institute (IRRI), and they are investigating management strategies for rice farmers to cope with the effects of climate change (Wassmann, 2007). To date, the IRRI recommends that farmers establish efficient irrigation infrastructure, coupled with water-saving techniques. Good nutrient management with alternate wetting and drying are also recommended as a sustainable management practice (IRRI, 2010). Nonetheless, Vermont rice farmers are concerned about weed control, so more time and money must be directed toward determining effective weed control methods such as the use of cover crops. Winter cover crop species for rice in the Northeast have yet to be determined, but there is research on cover crops in the southern U.S. (McClung et al. 2013) that could be applied to our region.

The only rice varieties that survived the entire season and were not significantly browsed by animals at WSRF were Hayayuki and Matsumae, which had significantly different total grain yields (P = 0.001). Matsumae thrived in drought conditions (average total seed production per plant: control = 812.0; low = 1297.6; saturated = 795.2; Figure 8). Hayayuki, however, cannot tolerate low water conditions, but was highest yielding in saturated water conditions (average total seed production per plant: control = 580.6; low = 302.9; saturated = 726.4; Figure 9). Furthermore, Matsumae performed equally well in control and saturated water conditions, indicating that less water-intensive practices are sufficient for high grain yields with these particular varieties.

We did not quantify percent sterility per panicle post-harvest as a measure of cold-resistance, but the Hayayuki plants grown in low water conditions were largely sterile, whereas the M202 plants grown in saturated and control water treatments yielded full, dense grains at JM. We do not have any data, quantitative or anecdotal, on percent sterility from WSRF.

Due to the large degree of browsing of the grain at both farms, we did not yield a significant amount of grain to process. Therefore, no rice was threshed and dehulled. Instead, all grain was stored as seed stock for planting in 2013 and 2014. The rice grown at WSRF was kept on-farm, whereas the grain yielded from JM Plants and Produce was collected and brought to the Plant Biology laboratory in Jeffords Hall at the University of Vermont. Some seeds from JM were used in a Vermont Science Fair project discussed in the “Outreach” section below, or disseminated to farmers and homeowners interested in growing rice at the February 2012 Rice Growing Panel at UVM. Some seed (< 20 g) is currently being stored by Dr. Laura Hill Bermingham in the Plant Biology laboratory in Jeffords Hall at UVM.

Biodiversity survey at WSRF

We sampled rice paddies and holding ponds at WSRF for aquatic wildlife in August 2012. The major findings were that no wildlife was detected in the rice paddies, but it was inferred that paddies provide suitable habitat for amphibians in early spring (April/May). There are 3 holding ponds at WSRF of variable age (3 – 7 years old). The lower and upper ponds housed mainly larvae from the following insect families: Diptera, Odonata, and Hemiptera. In addition, many isopods and crayfish were noted in each pond. Some green frogs were observed in all the ponds. The oldest pond (7 years old) housed the greatest diversity of aquatic wildlife, including Eastern newts, water boatman, green frog tadpoles, isopods, dipterans, and odonate larvae. More detailed species data and substrate data on each pond can be found in the Biodiversity Survey Database (pp. 19-20). These data will be complimentary to the wildlife assays conducted at the Akaogi Farm in West Westminster, VT. We were unable to estimate the index of biotic integrity (IBI) based on crayfish, fish, and amphibian assemblages due to the low sample sizes of aquatic wildlife at WSRF.

Akaogi, T. and Akaogi, L. 2009. Rice Growing Manual for the Northeast USA. Provided by NESARE office, Burlington VT.

Hill Bermingham, L. (2010). Deer herbivory and habitat type influence long-term population dynamics of a rare wetland plant. Plant Ecology, 210: 359 – 378

Hoffman, M.P. and Smith, D.L. Feeding our great cities: Climate change and opportunities for agriculture in eastern Canada and the Northeastern US. A publication from Cornell and McGill Universities:
https://www.ca.uky.edu/.../US-CanadaClimateChangeWhitePaper.pdf

Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventories Reference Manual, 1996: http://www.ipcc-nggip.iges.or.jp/public/gl/invs1.html

International Rice Research Institute (IRRI) website, http://irri.org/. Accessed October 16, 2011.

McClung A.M. et al, 2013. Integrating choice of variety, soil amendments, and cover crops to optimize organic rice production: http://www.ushrl.saa.ars.usda.gov/research/publications/publications.htm?SEQ_NO_115=297056

Wassmann, Reiner. 2007. Coping with Climate Change. Rice Today, July – September issue, p. 11-15.

Vermont Hardiness Zone Map, www.gardentimeonline.com/VermontHardinessZoneMap.html, accessed January 29, 2014.

• Germination rates of rice varieties from both WSRF and JM farms
• Seedling height as an estimate of plant vigor. All data was collected in Shelburne VT
• Germinated rice seedlings in propagation trays prior to transplant. Photo taken by Troy Griffis
• Early season (May-June) plant vigor among water treatments at JM Plants and Produce (JM) in Shelburne, VT
• Average late season (July-August) panicle height among water treatments at JM Plants and Produce (JM) in Shelburne, VT
• Average grain yield (average seed production) of rice grown in 5-gallon buckets (B) compared to paddies (P) at Whole Systems Design Research Farm (WSRF) in Moretown, VT
• Average grain yield of rice among 3 experimental water treatments at Whole Systems Design Research Farm (WSRF) in Moretown, VT
• Average total grain yield for Orzya sativa ssp. japonica cv. “Matsumae” at Whole Systems Research Farm in Moretown, Vermont
• Average total grain yield for Orzya sativa ssp. japonica cv.“Hayayuki” at Whole Systems Design Research Farm in Moretown, Vermont
• Table 1
• Differences in early season (May-June) plant vigor among rice varieties at the 2 Vermont farms
• Rice after transplant into control paddies at Whole Systems Research Farm. Photo taken by Ben Falk
• Aquatic wildlife database
• Rice producing grain in control paddies at Whole Systems Research Farm. Photo taken by Ben Falk
• Pre-soaked Hayauki seed with radicles (i.e., embryonic roots) emerging from the seed coat, indicating that seeds are ready to germinate in germination mix. Photo taken by Troy Griffis