RESUMO
Premise: Most traits are polygenic and most genes are pleiotropic, resulting in complex, integrated phenotypes. Polyploidy presents an excellent opportunity to explore the evolution of phenotypic integration as entire genomes are duplicated, allowing for new associations among traits and potentially leading to enhanced or reduced phenotypic integration. Despite the multivariate nature of phenotypic evolution, studies often rely on simplistic bivariate correlations that cannot accurately represent complex phenotypes or data reduction techniques that can obscure specific trait relationships. Methods: We apply network modeling, a common gene co-expression analysis, to the study of phenotypic integration to identify multivariate patterns of phenotypic evolution, including anatomy and morphology (structural) and physiology (functional) traits in response to whole genome duplication in the genus Brassica. Results: We identify four key structural traits that are overrepresented in the evolution of phenotypic integration. Seeding networks with key traits allowed us to identify structure-function relationships not apparent from bivariate analyses. In general, allopolyploids exhibited larger, more robust networks indicative of increased phenotypic integration compared to diploids. Discussion: Phenotypic network analysis may provide important insights into the effects of selection on non-target traits, even when they lack direct correlations with the target traits. Network analysis may allow for more nuanced predictions of both natural and artificial selection.
RESUMO
PREMISE: Physiological processes may vary within leaf laminae; however, the accompanying heterogeneity in leaf venation is rarely investigated because its quantification can be time consuming. Here we introduce accelerated protocols using existing software to increase sample throughput and ask whether laminae venation varies among three crop types and four subspecies of Brassica rapa. METHODS: FAA (formaldehyde, glacial acetic acid, and ethanol)-fixed samples were stored in ethanol. Without performing any additional clearing or staining, we tested two methods of image acquisition at three locations along the proximal-distal axis of the laminae and estimated the patterns of venation using the program phenoVein. We developed and made available an R script to handle the phenoVein output and then analyzed our data using linear mixed-effects models. RESULTS: Beyond fixation and storage, staining and clearing are not necessary to estimate leaf venation using phenoVein if the images are acquired using a stereomicroscope. All estimates of venation required some manual adjustment. We found a significant effect of location within the laminae for all aspects of venation. DISCUSSION: By removing the clearing and staining steps and utilizing the semi-automated program phenoVein, we quickly and cheaply acquired leaf venation data. Venation may be an important target for crop breeding efforts, particularly if intralaminar variation correlates with variation in physiological processes, which remains an open question.