Synthetic feedlot cattle populations for the High Plains cattle feeding area

Abstract

Objective

Our objective was to develop synthetic mul-tivariate normal populations of feedlot steers and heifers of various types using data from 2 cattle feeding compa-nies and to make these synthetic population data publicly available.

Materials and Methods

Close-out data from lots of cattle from 20 feedlot locations owned by Cactus Feed-ers and Five Rivers Cattle Feeding (10 locations for each company) were made available for analysis. Lots that en-compassed 6 groups of cattle were evaluated, including Native steers (3,483 and 2,777 lots for Cactus Feeders and Five Rivers, respectively); Native heifers (2,440 and 1,941 lots); Mexican steers (553 and 456 lots); Mexican heifers (132 and 174 lots); Beef × Dairy steers (474 and 71 lots); and Beef × Dairy heifers (285 and 36, lots). The SIM-NORMAL procedure of SAS was used to create multivari-ate normal populations with the following variables: initial BW, final BW, DMI, dead cattle excluded (deads-out) and dead cattle included (deads-in) ADG, and HCW. The F:G was calculated for each lot from the simulated DMI and ADG values, and DP was calculated from simulated final BW and HCW.

Results and Discussion

Means and standard devia-tions for simulated population data were nearly the same as the original close-out data for each group. Considerable variation was evident for minimum and maximum values in the synthetic populations compared with the original data, particularly for the Beef × Dairy and Mexican-origin cattle, reflecting the smaller numbers of observations in these datasets and use of the normal distribution to gener-ate the synthetic data. Indeed, abnormally low (Mexican cattle) and negative (Beef × Dairy cattle) values for initial BW were observed, indicating the need for careful data curation when sampling these synthetic populations.

Implications and Applications

Potential applica-tions for these data include economic evaluations, simu-lations related to nutrient requirement recommendations or environmental impacts of cattle feeding, and statistical applications like determining appropriate sample sizes for field experiments. The 6 synthetic populations (10,000 lot-level observations per population) we created are available for download at https://www.depts.ttu.edu/afs/burnett_center/synthetic-populations.php. We intend to periodi-cally update these files to provide more robust values, par-ticularly for the Beef × Dairy and Mexican-origin popula-tions.