On several occasions, I have interacted with farmers testing a product or technology simply by comparing one plot against another. In the world of statistics, this approach means that the “treatments” had no replications. To some readers, the lack of replication is not an issue. To readers with some background in statistics, this approach is insufficient to test a product or a technology. Therefore, the key question is why replicating treatments is important when performing research.
Our research group at Virginia Tech University has performed a series of experiments evaluating the maturity at harvest of triticale on the nutritional composition and digestibility of the silage. Based on the literature, we should expect a higher concentration of crude protein, a lower concentration of neutral detergent fiber (NDF), and a higher NDF digestibility when we harvest the small-grain grass at the boot stage of maturity than when we harvest the small-grain grass at the soft-dough stage of maturity. However, our oversight of field or soil variation has proven that those differences in composition and digestibility are not always there.
In 2021, for example, we seeded a single 25-acre field with triticale. The goal was to obtain two triticale silages (i.e., boot stage versus soft-dough stage), so we split the field into two halves and harvested the second half 26 days after harvesting the first half. So far, everything seemed straightforward. To our surprise, however, the two resulting silages had a similar (i.e., almost identical) nutritional composition. How could that be possible if the literature says that silages of different maturities should differ?
Seeking answers to our unexpected outcome, we screened satellite images of the field and observed the variation in the field (see photos). Without us knowing, the variation in the field had coincided with how we split the field for harvesting. So, what has occurred? In statistical terms, the (lack of) effect of maturity at harvest was confounded with the variation in the field, and this confounding occurred, at least partially, due to a lack of replication.
Did field variation affect the nutritional composition of the silages? We cannot provide a clear answer to that. However, in 2022 and 2023, we seeded the same triticale variety in the same field and harvested at almost the exact days. Differently, however, we divided the field into four quarters and harvested two of the quarters at each stage of maturity. By implementing this approach, we could finally obtain two silages of different nutritional compositions.
As a take-home message, studies without replication can be a starting point toward an answer. However, to find a solid answer, replication of the treatments is a critical tool for research and innovation.
Satellite images from the field where triticale was grown and harvested. Images were captured on 03/14/22 (left), 04/23/22 (center), and 05/18/22 (right). Half of the field was harvested on 04/29/22 (red area of the image on the right), and the other half of the field was harvested on 05/25/22 (green area of the image on the right).
