Prediction of associative effects between forages and supplements in goats

Abstract

Equations to predict associative effects between basal forages and supplemental feedstuffs for goats were developed to improve diet formulation with the Langston University Interactive Nutrient Calculation system (LINC, http://40.65.112.141/). A literature survey of goat nutrition studies with ad libitum forage intake with or without supplementation was conducted, resulting in a database with 135 treatment mean observations, representing measures from 503 animals and derived from 26 publications. The database was divided into three datasets based on forage CP concentration (Low: < 6%, Moderate: 6 – 10%, and High: > 10%). The datasets were used to develop equations addressing positive and negative associative effects. Change in forage ME intake relative to metabolic BW (MEIMBWFOR; kJ/kg BW^0.75) due to supplementation was predicted based on potential variables of supplement ME intake also scaled to metabolic BW (MEIMBWSUP; kJ/kg BW^0.75), forage OM digestibility (OMDIGFOR; %), CP concentration (PTCPFOR; %), and NDF concentration (PTNDFFOR; % DM), supplement CP concentration (PTCPSUP; %) and ME concentration (MECSUP; MJ/kg), and quadratic functions of these variables (MEIMBWSUP2, OMDIGFOR2, PTCPFOR2, PTNDFFOR2, PTCPSUP2, and MECSUP2). Model development for each dataset was conducted by two analytical methods, stepwise regression and the Least Absolute Shrinkage Selection Operator (LASSO). Equations employing both methods were developed, with stepwise regression accounting for greatest variation (Low: MEIMBWFOR = 937 + 0.604 × MEIMBWSUP − 0.0018 × MEIMBWSUP2 – 0.105 × PTNDFFOR2; ARSq = 0.871, Moderate: MEIMBWFOR = 1732 − 0.579 × MEIMBWSUP + 40.9 × PTCPFOR − 62.4 × OMDIGFOR + 0.601 × OMDIGFOR2; ARSq = 0.826, and High: MEIMBWFOR = 51.0 − 0.00162 × MEIMBWSUP2 + 8.42 × PTNDFFOR; ARSq = 0.628). Similar variables were selected with both analytical methods for each dataset, but variables selected differed among datasets. Although goodness of fit measures were relatively high for each dataset, they ranked Low > Moderate > High, suggesting greatest robustness for Low and complexity in influencing factors for High. In conclusion, equations to predict associative effects between basal forages and supplements consumed by goats developed for basal forage with low, moderate, and high CP concentrations should be useful in diet formulation tools such as LINC. Future research should consider a wider array of conditions such as animal physiological status (e.g., growing and adult), type of production (e.g., milk and meat), and carbohydrate composition of supplements.