The American Heritage dictionary defines drought as "a long period with no rain, especially during a growing season." My dictionary has an identical definition, but also includes: "for example, NW Minnesota in 1988, 1989, and 2006." Without question, we are experiencing some very hot, dry weather this year, but it is questionable just how much this weather will influence final soybean yields. A lot will depend on a field's available soil water at planting and timing of in-season precipitation. Conjecture is commonplace, but we lack good evidence from designed experiments in NW Minnesota to address these questions. This does not mean that "...we have no idea," just that our ideas may lie some distance from reality. Other researchers in the Midwestern USA have determined that soybean water use (i) is closely related to stage of plant development; (ii) increases rapidly during vegetative growth from about V3 to V6; (iii) reaches a maximum plateau during R1 to R6; and (iv) declines rapidly as the plant matures. The crop's susceptibility to an insufficient water supply is especially acute during early to peak flowering and late flowering to early pod development, i.e., prior to the beginning of rapid seed filling (R5). This brief article presents some likely consequences of this year’s dry weather on final soybean yields, based on limited historical data and some knowledge of soybean growth and development.

The Soybean Breeding and Genetics Project of the University of Minnesota has conducted varietal trials at the Northwest Research and Outreach Center (NWROC) at Crookston for over 40 years and has often included check varieties for extended periods in their work. One such variety, McCall, was included from 1983 through 2002 in Uniform Regional Trials, providing yield data for both 'wet' and 'dry' years. Using this yield data, and weather data collected at the NWROC for the same period, we attempted to derive an equation that would describe McCall yields based on rainfall records. A better 'equation' would have included measures of available soil moisture at planting, but these values are seldom documented for most producer’s fields.

We regressed yield each year on two NWROC yearly rainfall records. The first 'record' is referred to as a 'recharge' and includes the total rain received in Sept. and Oct. of the previous year and the total April and May rain of the current year, i.e. water 'stored' in the soil profile just prior to increasing demand by the current crop (X₁). The second 'record' is the total, 'in-season,' rain received during June and July (X₂), a period of rapid soybean growth and development. August rain was not included for two reasons: (1) McCall yields from 1983 to 2002 were not related to August rain; and (2) once the soybean plant has advanced to stage R5 to R5.5 (which 'usually' occurs about the first to second week of August in NW Minnesota with MG 00 varieties), there is no further increase in node or pod number that contributes to final yield. Although soybean can compensate for fewer seeds by producing larger seeds, there are physical restrictions on how large. Thus, rain we receive in August will help us realize the 'potential' established by 'stored' and 'in-season' rain, but will not likely increase that potential. It is emphasized, however, that McCall is an early maturity variety (MG 00.7) and that later varieties which normally reach R5 later in August may benefit from August rains.

Some of the results of our analyses are presented in Table 1 and graphically in Fig. 1, 2 and 3. From 1983 to 2002, grain yields of McCall closely followed (reasonably close) the total amount of June-July rainfall (Fig. 1A). Rain received during rapid soybean growth and development likely promoted increases in node and pod number and, thus, yield potential. About 40% of the total variation in yield could be described just by June-July rainfall (Fig. 1B). The relationship between August rainfall and McCall yield, given in Fig. 2A and 2B, suggests that August rain did not increase yield, although it may have prevented some yield loss. Again, later maturing varieties may have responded more to August rain, provided they didn’t freeze in early September.

Using 'stored water' and 'in-season' rain together in our regression improved our ability to describe McCall yield (Fig. 3A and 3B). Nearly 60% of the total variation in yield could be described by the equation listed in Fig. 3B. The regression coefficients suggest that each additional inch of 'in-season' precipitation increased yield about 2.6 bushels, whereas each additional inch of 'stored water' increased yield about 1.6 bushels. Using the equation given in Fig. 3B and rainfall records pertinent to 2006, the predicted value (17.8 bu acre^-1) and the upper and lower 95% confidence limits suggest that our 2006 McCall yield at the NWROC will be between 10.3 and 25.3 bu acre^-1, with an average of 17.8 bu acre^-1. This does not necessarily mean that the yields on your farm are going to be in this same range. You probably have a range of relative maturities, stored soil moistures, and in-season rains just in your own fields. The impact of this year's drought on soybean yields in NW Minnesota and on your farm with potentially higher yielding varieties will be measured later, but it will be directly related to the yield potential established by the time the plant reaches growth stage R5.

Just out of curiosity, and with total disregard for sound science, I wondered if the approach used to describe McCall yields could be 'stretched' (quantum leap; do not do this at home) to describe Polk county yields for the same time period. The results of this journey to the dark side with a candle are presented in Table 2 and Fig. 4. The predicted value (21.2 bu acre^-1) and the upper and lower 95% confidence limits suggest that our 2006 Polk county yield will be between 16.4 and 25.9 bu acre^-1, with an average of 21.2 bu acre^-1. Interesting .... but.

I acknowledge that this approach is, of course, very risky. Using weather data obtained at one location as representative of an entire (large) county is close to ridiculous and only one possible source of error. Others include previous crops, soil types, varieties, planting dates, etc. On the other hand, maybe it's closer than I think.

Table 1. Measured yield, X₁, X₂, predicted yield, and upper and lower 95% confidence limits of the predicted yield of McCall soybean grown at the NWROC from 1983 through 2002.

The prediction equation was: Yield = 0.34 + 1.61X₁ + 2.63X₂.

Fig 1. At the NWROC from 1983 to 2002, increasing June - July rainfall likely increased node, flower and pod production prior to R5 and, consequently, McCall grain yield potential.

Fig 2. Increasing August rainfall seldom increased McCall grain yields, although it may have prevented yield losses during drier years.

Fig 3. McCall grain yields at the NWROC from 1983 to 2002 were closely related to the amount of accumulated precipitation prior to planting and June - July rainfall.

Table 2. Polk county average yield, X₁, X₂, predicted yield, and upper and lower 95% confidence limits of the predicted yield of soybeans grown in Polk county from 1983 through 2002.

Fig 4. Polk county average soybean yields from 1983 to 2002 and yields predicted using NWROC weather records for the same time period.

printer-friendly PDF version