Castro-Costa, Andreia, Caja, Gerardo, Eymard, Alexandra, Dhumez, Ophelie, Tessier, Joseph-Alexandre, Giger-Reverdin, Sylvie, Universitat Autònoma de Barcelona [Barcelona] (UAB), Modélisation Systémique Appliquée aux Ruminants (MoSAR), Institut National de la Recherche Agronomique (INRA)-AgroParisTech, Universitat Autònoma de Barcelona (UAB), and AgroParisTech-Institut National de la Recherche Agronomique (INRA)
Eight rumen cannulated dairy goats in early lactation (3.7 ± 0.2 kg/d) from the INRA-AgroParisTech experimental farm (Thiverval-Grignon, FR), were provided with wireless bolus sensors (KB1001 Kahne, Auckland, NZ) of pH and temperature to study the relationship between intake and subclinical acidosis for 35 d. After adapting to a TMR control diet (CO, 20% concentrate) for 12 d, goats were brusquely changed to an acidogenic diet (AC, 50% concentrate) for 23 d. Diets were fed ad libitum twice daily (a.m. 1/3, p.m. 2/3) according to milking intervals. Rumen pH and temperature data were captured every 15 min and intake measured every 2 min by weighing scales. Rumen samples were collected (h 0, 1, 2, 4 and 6 post feeding) to measure pH by pH-meter before (d 8 and 11) and after (d 13, 14, 15, 16, 20, 26 and 34) the change. One sensor failed and was discarded, the rest of data being modeled by logistic regression with Solver of Microsoft Excel. Data were analyzed by MIXED for repeated measurements and GLM procedures of SAS. Values of pH-meter vs. sensors correlated (r2 = 0.86; P < 0.01) and were used for sensor recalibration. Mean rumen pH varied markedly by hour and diet; on average, it was higher in CO vs. AC (6.34 ± 0.06 vs. 6.10 ± 0.03; P < 0.001). Despite the high concentrate percentage of AC diet, rumen pH was shortly under pH 6.0. Feed intake reached plateaus during the day when pH was closer to the a.m. or p.m. nadirs, and correlated negatively (r2 = 0.77 to 0.87; P < 0.01) during the periods in which pH dropped. Correlations between rumen temperature and pH were very poor (r2 < 0.1), except for the nightly resting period (r2 = 0.93; P < 0.001), the rumen being slightly colder in AC goats (CO vs. AC, 39.73 ± 0.09 vs. 39.61 ± 0.09°C; P < 0.001). Temperature and pH data from sensors fit logistic models (r2 = 0.97 to 0.99; P < 0.001). Pattern of pH logistic models and time spent under pH 6.0, allowed us to classify the goats as sensitive (3/7, 43%) or tolerant (4/7, 57%) to acidosis, which was related to individual feeding behavior. In conclusion, daily intake measurement and wireless sensors proved to be useful for monitoring rumen function, which allow for an individual separation of sensitive and tolerant goats to rumen acidosis.