This study clarified the growth patterns and developmental changes of the backfat thickness (BF) of Okinawa Agu gilts regarding their management. Five nonlinear growth models (Logistic, Gompertz, von Bertalanffy, Richards, and Janoschek) were used to describe growth patterns. Weekly, body weight (BW) data from birth to 34 weeks of age were collected. BF at the P2 position from 4 to 34 weeks of age was recorded every 4-5 weeks. Among 23 gilts pigs, 14 were Agu pigs, and nine were Landrace pigs. Animals were allowed fed ad libitum to 16 weeks of age and then reared under restricted feeding (1.3 kg/day for Agu gilts : 2.5 kg/day for Landrace gilts) to 34 weeks of age. Comparison among the five models revealed that the Richards model was the best fit for the growth data based on adjusted R-square (AdjR2) and Akaike’s information criterion (AIC) for both breeds. The Agu gilts showed smaller mature BW, early maturing rate, and earlier weeks of age at the inflection point than Landrace gilts. BF development in Agu gilts was rapid during 4-16 weeks of age under ad libitum feeding but was suppressed after 16 weeks of age under restricted feeding. The breed differences of BF were observed from 8 weeks of age, and Agu gilts had thicker BF than Landrace gilts.
In Japan, nitrate-related nitrogen (sum total value of NO2--N, NO3--N and 0.4 times of NH4+-N concentration) in effluent of wastewater treatment facility is regulated by Water Pollution Control Law. Real time monitoring of nitrate-related nitrogen in the effluent would be important for its stable reduction. This study attempted to establish simple nitrate-related nitrogen estimation method using portable water quality measuring devices. Analysis of data obtained from 14 farms suggested that both multiple regression model and decision tree model with pH and EC or pH and refractive index as explanatory variables would be applicable for the estimation of the nitrate-related nitrogen. Ammonium nitrogen was predictable by EC or refractive index. Simple colorimetric pH measurement with BTB reagent was applicable instead of pH electrode method in the case of decision tree model.