Mobley, M. L., Kruse, A. S., & McNickle, G. G. (2022). Pisum sativum has no competitive responses to neighbors: A case study in (non)reproducible plant biology. Plant Direct, 6( 10), e411. https://doi.org/10.1002/pld3.411
Date of this Version
Bayesian hierarchical analysis, below ground competition, meta-analysis, pisum sativum, plant-plant competition, reproducibility
Plant–plant competition is ubiquitous in nature. However, studying below ground behavior of roots has always posed certain difficulties. Pea (Pisum sativum L.) has become a common study species for questions about how plant roots respond to neighboring plant roots and barriers in soil. However, published results point in several different directions. This has sometimes been interpreted as pea having sophisticated context dependent responses that can change in complex ways depending on its surroundings, but it could also just point to small statistical power resulting in type I or II statistical errors. To explore this further, here, we combine the result of five new experiments with published results to examine 18 unique experiments from 10 different studies and 6 cultivars of pea for a total of 254 replicate plants. We used a Bayesian hierarchical meta-analysis approach to estimating the likely effect size from the available data, as well as quantify heterogeneity among different experiments, studies and cultivars. The posterior distributions show that, at the coarsest possible scale of total root production, it is unlikely that P. sativum root growth is influenced by either neighbors or pot volume that varies primarily by depth. We find no evidence of publication bias and conclude that this is simply due to statistical sampling error and the scientific method combined with frequentist statistics operating as intended. We suggest that further work on pea should consider repeating experiments that reported finer scale root plasticity at the rhizosphere scale or consider exploring different pot geometries such as volume that varies by depth or width. We also suggest that more diversity in study species are needed to better understand the neighbor-volume response.