Jeff Phillips directed harvest of the Indiana Prairie Farmer/Precision Planting plots at the Throckmorton Purdue University Research Farm near Romney a couple of weeks ago. The trial was replicated four times. However, a big rain shortly after planting heavily influenced about 15 of the 108 individual plots within the trial. The chore was deciding how to handle them to still get good data from the experiment.
"I tried to identify the ones from looking at them the day we harvested that were obviously hit by wet weather," he explains. Phillips is Extension ag educator in Tippecanoe County. Typically, stalks were of smaller diameter, shorter and ears were smaller where water stood. Although the ponded areas didn't look that much lower in topography when planted, rain certainly found a place to sit when it arrived less than a week after planting.
Phillips was still able to run statistics on the plot, since there were four replications. In most cases, an individual treatment combination was only lost once, so there were still three reps remaining. There are methods that ag statisticians can use to still get credible data from such a situation, he notes.
The trick, of course, was knowing exactly which plots should be marked as stunted and kicked out of the data. In the end, Phillips ran the statistics both ways, with the stunted plants in, and with them out, although he believes data with them excluded will be more meaningful.
Despite the ponding, yellowing, slower growth and heavier foxtail pressure, the ponded plots responded to better weather and an accurate herbicide application to produce better than expected. The lowest –yielding individual six-row plot in the study, in the stunted area, was 99 bushels per acre. The highest yielding six –row plot in the study was about 240 bushels per acre.
The rub came when Phillips reviewed notes taken earlier in the season, at about V5 to V7 growth stage. "One of the plots that I marked to exclude was rated OK on the visual inspection in late June," he says. "I checked the yield and is was 201 bushels per acre. "Obviously, I just went one plot too far in deciding which ones not to include in statistics. Based on the summer observations, we put that plot back into the study. We also used the notes to add another plot. The good thing was that we had harvested and weighed all plots, whether we were thinking of excluding them or not.
"It's very helpful to know how someone viewed the plots earlier in the summer, he says. "Once you combine yield results worth those observations, it helps explain more things."
Nick Linder, of the farm crew, harvested the plot. He captured an image on the yield monitor screen that pretty well detailed how various treatments compared in the plot. While true weights from a weigh scales were used instead of the yield monitor numbers, the graph was still helpful in showing relative weaknesses or strengths from one plot to another.