Fatores que afetam o consumo de matéria seca e modelos preditivos para novilhas em pré-parto e vacas secas leiteiras

• Ano de defesa: 2022
• Autor(a) principal: Donadia, Andrea Beltrani
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Mato Grosso; Brasil; Faculdade de Agronomia e Zootecnia (FAAZ); UFMT CUC - Cuiabá; Programa de Pós-Graduação em Ciência Animal
• Palavras-chave em Português: Meta-análise; Modelagem; Sistema nutricional; CNPQ::CIENCIAS AGRARIAS::MEDICINA VETERINARIA; Meta-analysis; Modeling; Nutritional system
• Link de acesso: http://ri.ufmt.br/handle/1/5565

A meta-analysis was performed to elucidate the factors that affect dry matter intake (DMI) and to develop predictive models of DMI for prepartum heifers and dry dairy cows, using only animal or animal and dietary characteristics as inputs. Accuracy and precision comparisons were also performed with existing models (NRC, 2001; CNCPS, 2003; and NASEM, 2021), using an independent dataset. The dataset (peer-reviewed papers) was randomly divided into two subsets. The first subset was used to develop the models (140 peer-reviewed papers, 141 experiments, n = 466 treatment means; 473 to 834 kg body weight (BW); 1 to 97 days relative to calving (DRC); 2.50 to 3.97 body condition score (BCS); 6.60 to 18.50 kg/d of DMI; 224 to 651 g/kg of neutral detergent fiber (NDF; DM); 272 to 1000 g/kg forage (DM). The second subset was used to evaluated the accuracy and precision of the models (96 peer-reviewed papers, 97 experiments, n = 323 treatment means; 398 to 861 kg BW; 1 to 84 DRC; 2.05 to 4.25 BCS; 5.40 to 19.90 kg/d of DMI; 280 to 666 g/kg NDF(DM); 300 to 1000 g/kg forage (DM). We developed six models divided into three groups: I) Model I (linear) and (nonlinear) were BW and DRC were used as predictor variables; II) Model II (linear) and (nonlinear) were adjusted with BW, forage, NDF and DRC as predictor variables; and III) model III (linear) and (nonlinear) were adjusted with the all variables used in models II with parity inclusion. BW was the main animal factor that affected DMI. Forage and NDF were the only diet associated variables that was significant (P<0.05) in the DMI prediction model for dry dairy cows and prepartum heifers. Between proposed models: The linear and nonlinear Models I predicted DMI with the highest (P<0.05) root mean square prediction error (RMSPE; 16.90 ± 0.18% and 16.96 ± 0.18 % DMI observed). Linear and nonlinear models III for dry cows predicted xii DMI with lower (P<0.05) RMSPE (12.27 ± 0.24% and 11.98 ± 0.23% observed DMI). Linear model III for dry cows is similar to linear model II (RMSPE = 12.27 ± 0.24% and 12.69 ± 0.20% observed DMI). Among the previously published models, only the NASEM (2021) for prepartum heifers and dry cows (RMSPE = 17.42 ± 5.14 and 16.32 ± 0.56% DMI observed) and CNCPS (2003) (RMSPE = 17.03 ± 0.18 % DMI observed) had similar prediction (P≥0.05) to our linear and nonlinear models I. NASEM (2021) model for prepartum heifers presented predicted DMI similar (P>0.05) RMSEP (17.42 ± 5.14% of DMI observed) to linear and nonlinear models III for prepartum dairy heifers (14.53 ± 3.69% and 10.21 ± 3.11% of observed DMI). The CNCPS (2003) model is similar (RMSEP 17.03± 0.18% of observed DMI) to the linear model III for prepartum heifers. We recommend the following models to predict DMI for prepartum heifers and dry dairy cows: linear model III (DMI (kg/d) = 6.3800+ 0.0089 × BW – 0.0000080 × Forage × NDF + (Heifer = 0 or cow = 1.1239 or all =0 .3932) + 0.0303 × DRC) and nonlinear model III (DMI (kg/d) = (6.6908 + 0.0074 × BW – 0.0000093 × Forage × NDF + (Heifer = 0 or cow = 2.2949 or all =1.9951)) × e0.0035 × DRC)). Models with animals and diet as inputs improve DMI prediction. Our models can contribute to improve the prediction of dry matter intake for prepartum heifers and dry dairy cows.