Effect of stocking density and effective fiber on the ruminal bacterial communities in lactating Holstein cows

Overstocking can be a major issue in the dairy cattle industry, leading to negative changes in feeding and resting behavior. Additional stress imposed and alterations in feeding behavior may significantly impact the rumen microbiome. The rumen microbiome is responsible for the successful conversion of feed to usable energy for its host. Thus, understanding the effects of stocking density on the rumen microbiome is imperative for further elucidation of potentially negative consequences of overstocking in dairy cattle. This study implemented a Latin Square design accounting for four pens of cattle and four treatment periods so that all treatment combinations were assigned to every pen during one period of the study. Two treatment factors, including two levels of physically effective neutral detergent fiber, achieved with addition of chopped straw, and stocking density (100% vs. 142%) of freestalls and headlocks, were combined and tested within a factorial treatment design. Within each pen, three or four cannulated cows (n = 15 total) were sampled for rumen content on the final day of each treatment period. Each treatment was randomly assigned to a single pen for a 14-day period. The V1-V3 hypervariable regions of the 16S rRNA gene were targeted for bacterial analyses. Variables with approximately normally-distributed residuals and a Shapiro-Wilk statistic of >= 0.85 were analyzed using a mixed model analysis of variance with the GLIMMIX procedure with fixed effects of feed (straw vs. no straw), stocking density (100% vs. 142%), and the interaction of feed x stocking density, and random effects of pen, period, feed x stocking x pen x period. Pen was included as the experimental unit in a given period and the sampling unit as cow. Variables included Shannon's Diversity Index, Faith's phylogenetic diversity index, chao1, observed OTU, and Simpson's evenness E as well as most individual taxa. Data were analyzed in SAS 9.4 utilizing the GLIMMIX procedure to perform mixed model analysis of variance. If data were not normally distributed, a ranked analysis was performed. No differences were observed in a-diversity metrics by fiber or stocking density (P > 0.05). Beta diversity was assessed using weighted and unweighted Unifrac distances in QIIME 1.9.1 and analyzed using ANOSIM. No differences were observed in weighted (P = 0.6660; R = -0.0121) nor unweighted (P = 0.9190; R = - 0.0261) metrics and R values suggested similar bacterial communities among treatments. At