Viability of directly sown paddy rice

Out of the 34 different DS blocks we only kept 6 for the study, all using the broadcast method. We kept any treatments with over 66% desired sprouting coverage, and decided Treatments with less than 66% desired sprouting coverage were considered seeding failures and typically would not be considered a viable stand by a farmer. For this reason, these plots were removed from the experiment. Coverage was defined as an area within 8 inches of a rice plant. For comparison, in many directly sown operations, there are over 5 plants per square foot.

In 2016 the paddies were extremely dry, with no adequate pond water and only 1 rain event in May, three weeks after planting. 2 blocks were successful, with 75% & 76% emergence density. These blocks were the only ones to be adequately moist and had standing water for much of May. They were in the lowest paddy. 2 other blocks in 2016 had around 25% coverage, and, the other 20 blocks, had 0-10% coverage. Our large scale sowing with the broadcaster and row seeder used in two different paddies was also abandoned due to poor germination.

The farm did receive sufficient rain in 2017, yet only in the broadcast treatments did we get any successful results. The 4 successful ones were found in areas that were higher up and not flooded, but saturated and mulched. Two other blocks, both mulched and broadcast from 2017 had around 50% coverage and everything else was below 10%, with more than half having 0% germination. None of the rice in flooded areas, nor planted underground had a germination rate above 25%. With a wet spring we were unable to keep many of the mulched row treatments from having standing water.

The rice took, on average, 18 days in 2015 and 27 days in 2017 to sprout at 50% desired coverage as shown in table 1. It took the rice between 160 (2015 avg) and 184 (2017 avg) GDD for the rice to emerge. 2015 was the year before we started this SARE Project. N=1 because we were using one block, which was 50 x 25. The higher GDD accumulation at planting, probably indicative of higher soil temperatures, could have contributed to the rice coming up earlier in 2015. In 2016, which was the first year of the study the rice took 27.5 days for our successful treatments to get 50% coverage with a GDD accumulation of 279. This much higher number could be explained by lack of moisture causing a delay in germination and/or emergence.

As germination was our biggest issue, in the fall of 2017 we tried more germination trials in three small paddies, all kept moist and mulched, as shown in Table 2. The data shows that with average air temperatures above 60 degrees the rice comes up in roughly 2 weeks and decreases after that, not emerging in the final trials. We had above 100% emergence coverage in the first three trials. The last trial, with temperatures well below 50 degrees had no germination. We conducted analysis of variance on the time to emergence data for the first three planting dates. The last date was excluded because the rice did not emerge. Analysis of variance showed a statistically significant difference in the number of days to 50% emergence among the three planting dates (p=0.00011). Emergence was fastest for the rice planted Sept. 27 and slowest for the rice planted Oct. 8. Figure 1 shows the relationship between emergence and growing degree days for this trial.

Figure 1. Effect of growing degree days on emergence of fall-sown, direct-seeded Hayayuki rice, planted Sept. 6, Sept. 7, Oct. 6, and Oct. 26, 2017.

Why did we have success in 2015, the fall trials of 2017 and not the growing seasons of 2016 and 2017? I believe this was mainly due to temperature in 2017. In 2017 the sprouted rice was planted into much colder air temperatures and most likely ground temperatures than 2015, as shown in Table 1. Out of the broadcast treatments more flooding actually seemed to lead to a decrease in sprouting success, and only the mulched & non-flooded rice did well. The directly sown rice treatment that was buried and also the spring flooded treatment would probably be colder in the spring when ground water is moving through the paddies. In the spring it would make sense for the lightly mulched rice treatment to be the warmest. The rice was sown after soaking and when at least 25% had sprouted, so perhaps most of that rice died in environments below 50 degrees, rather than just halt growth, which would explain why only the heavily seeded broadcast came up and the row seeded rice did not. Furthermore maybe the remaining rice that didn’t germinate didn’t die and stayed dormant for a while and then also emerged slower, as suggested by the third trial in our fall data, from Table 2, which shows slower emergence in colder temperatures. Lastly, emergence coverage does not equate to 90% emergence rate. For example in the broadcast treatment we sow the rice very densely, 80 lbs./acre and a 90% emergence coverage could be achieved with less than 25% emergence rate, meaning much of the rice could have died due to cold temperatures and we still could have got a stand worth keeping.

In 2016 I believe moisture and temperature were limiting factors, which could explain why it was the least successful year, why the rice took so many more GDD to emerge and why it only came up in the blocks with adequate moisture.

In the fall we also compared directly sown from a soil media of potting soil and a 50/50 mix of the paddy clay and potting soil, to see if that effected emergence success and duration. Using a two-tailed t-test, we found there was a statistically significant difference between the number of days to 50% germination for rice started in a mix of clay and potting soil and compared to rice started in potting soil alone (p=.03347). When started in potting soil, rice germinated almost three days earlier than when started in a mixture of clay and potting soil (Table 4).

In the fall we compared our three most successful directly sown blocks with three transplanted blocks, both of the Hayayuki variety. Drawing on data collection methods from other Rice SARE grants, we looked at tillers, viable tillers and average height of tallest three tillers. Numerically, the number of tillers per plant (average of 10 plants per plot) was higher for transplanted rice than for direct-seeded rice (36 for transplanted, 24 for direct-seeded). Similarly, the number of viable tillers (average from 10 plants per plot) was higher for transplanted rice than for direct-seeded rice (26 for transplanted, 17 for direct-seeded. However, neither of these was statistically significant (p=0.14208 for tillers, p=0.20515 for viable tillers). Tiller height, measured as the average of the tallest 3 tillers for each of 10 plants per plot, was nearly equal between the two treatments (34 for transplanted, 33 for direct-seeded). This was also not statistically significant.

The main purpose of our grant was to compare directly sown rice to transplanted rice, and compare direct sowing methods, though with such poor germination success we were not able to gather adequate data.

Table 5 compares labor for directly sown rice to transplanted rice. The labor for the rice soaking and drying is longer for the transplanted rice due to labor purposes; it requires different batches, as it cannot all be planted in the fields on the same day nor all sown in flats on the same day.   We used an Earthway Seeder for broadcasting the rice and then used our rice straw for mulch. A light layer of mulch is necessary to keep the birds out, if flooding isn’t available, at least at Wild Folk Farm. In 2016 we observed Red-Winged Blackbirds digging into the ground and eating much of our seed on barren ground. Using human power, the harvest for transplanted rice takes substantially less time, as the plants are bushier, require less sickle strokes, and are easier to transport and bunch. If we were to use a combine, the labor would not be any different, assuming the rice does not lodge. Other studies have indicated directly sown rice has a higher propensity to lodge. Much of this labor can vary greatly, whether it be from human power or from machine use. We have found that drier & harder paddies can take 4x as long to transplant and weed as wet soft ones.

This huge variability between and within paddies and for weed pressure, as well as logistical difficulties led us to stop tracking weeding labor. Our observations as well as other farming data suggest substantially higher weed pressure for the directly sown crops vs. transplanted ones, but no numbers are represented in this study for that.

Table 5 and 6 also show much higher costs for transplanted rice. Calculating the cost for high tunnel space can be difficult because there is a wide variability in structures, different opportunity cost for the high tunnel space, if existing structures exist and the rate of attrition of those structures. Directly sown rice also requires more mulch and rice seed, which would be for cost effective for an existing rice farm.

Table 7 shows the raw data for yields as well as extrapolations with 100% desired coverage. Our fall germination location all had over 100% desired coverage and our average coverage for spring sowed blocks was 84%. When extrapolated out to 100% coverage, directly sown rice yields 9% lower than the transplanted method of Hayayuki rice. The actual average emergence coverage of 84% had 23.5% lower yields than the TP Haya. For an accurate comparison of directly sown rice to transplanted rice it makes more sense to use a longer season rice variety (Yukihikari), that yields more, and better represents the advantage of starting earlier when using the transplanting method. This longer season variety yielded 19% more than the shorter season varieties and 38.4% more than the directly sown rice and 26% more than the extrapolated yields of 100% emergence coverage for directly sown rice.