Redalyc.Influence of production factors on feed intake and feed

Redalyc.Influence of production factors on feed intake and feed

Semina: Ciências Agrárias ISSN: 1676-546X [email protected] Universidade Estadual de Londrina Brasil Abércio da Silva, Caio; Ketilim Novais, Ali...

237KB Sizes 0 Downloads 0 Views

Recommend Documents

Feed intake, feed efficiency,
Feed conversion ratio and Kleiber ratio had a strong negative correlation (-0.88). In concluded, increasing Kleiber rati

Management factors affecting mortality, feed intake and feed
The aim of this study was to determine the effect of animal management and farm facilities on total feed intake (TFI), f

feed intake and feed conversion ratio in purebred - ARCC Journals
Data on feed Intake and feed conversion ratio of 17 purebred, 18 two-breed and. 27 three-breed cross bunnies from 5 to 1

Genetic variance components for residual feed intake and feed
Genetic variance components for residual feed intake and feed conversion ratio and their correlations with other product

Feed Intake, Body Weight, Average Daily Gain, Feed Conversion Ratio
Abstract: The objective of this study was to investigate the effect of feeding varying levels of Phane meal. (Imbrasia b

Redalyc.Evaluation of animal performance, feed intake, and economic
Evaluation of animal performance, feed intake, and economic losses in sheep experimentally infected with Trypanosoma viv

Effect of Salinity on Feed Conversion Rate, Feed Conversion
Mar 6, 2017 - To cite this article: Mohammed Shaker Al_Khshali. Effect of Salinity on Feed Conversion Rate, Feed Convers

Evaluation of animal performance, feed intake, and economic losses
Evaluation of animal performance, feed intake, and economic losses in sheep experimentally infected with Trypanosoma viv

NET FEED INTAKE OF GROWING BULLS AND - Ag Professional
performance, body composition, and fertility traits. Net feed intake was positively correlated phenotypically with feed

Feed intake and growth performance of goats supplemented - Ainfo
Jul 2, 2014 - supplemental soy waste have a lower total dry matter intake, feed conversion ratio, and feed cost per kilo

Semina: Ciências Agrárias ISSN: 1676-546X [email protected] Universidade Estadual de Londrina Brasil

Abércio da Silva, Caio; Ketilim Novais, Aliny; Silva dos Santos, Rita de Kássia; Pierozan, Carlos Rodolfo; da Silva Agostini, Piero; Gasa Gasó, Josep Influence of production factors on feed intake and feed conversion ratio of grow-finishing pigs Semina: Ciências Agrárias, vol. 38, núm. 2, marzo-abril, 2017, pp. 997-1007 Universidade Estadual de Londrina Londrina, Brasil

Available in: http://www.redalyc.org/articulo.oa?id=445750711039

How to cite Complete issue More information about this article Journal's homepage in redalyc.org

Scientific Information System Network of Scientific Journals from Latin America, the Caribbean, Spain and Portugal Non-profit academic project, developed under the open access initiative

DOI: 10.5433/1679-0359.2017v38n2p997

In uence of production factors on feed intake and feed conversion ratio of grow- nishing pigs In uência dos fatores de produção sobre o consumo de ração e a conversão alimentar de suínos em crescimento e terminação Caio Abércio da Silva1*; Aliny Ketilim Novais2; Rita de Kássia Silva dos Santos2; Carlos Rodolfo Pierozan2; Piero da Silva Agostini3; Josep Gasa Gasó4 Abstract The aim of this study was to identify and quantify, through mathematical models, the production factors of grow- nishing (GF) phases that in uence the daily feed intake (DFI) and feed conversion ratio (FCR) in pigs. Sixty- ve GF farms were evaluated between 2010 and 2013, linked to a cooperative system located in the western Parana State, Brazil, representing 463 batches, with a mean of 642.79 ± 363.29 animals per batch, equalling approximately 300,000 animals. Forty production factors were considered that related to management, sanitation, installations and equipment, nutrition, genetics and environment on the farms. The DFI was in uenced by the barn’s position relative to the sun (P = 0.048), initial body weight (P < 0.0001) and nal body weight (P < 0.0001). It was observed that the FCR was in uenced by the barn’s position relative to the sun (P = 0.0001), the use of humidi ers/misting (P = 0.03), the presence of composters (P = 0.006), trees on the sides of barns (P < 0.045), the initial body weight of the pigs (P<0.0001) and duration of the grow- nishing phase (P < 0.0001). The variables selected in the models explained approximately 44 and 20% of the total variance in the DFI and FCR, respectively, demonstrating that this resource is a good tool for interpreting the factors related to the parameters evaluated. Key words: Management. Multilevel modelling. Swine.

Resumo Objetivou-se neste estudo identi car e quanti car, através de modelos matemáticos, os fatores de produção presentes em unidades de crescimento e terminação (CT) de suínos que in uenciam os parâmetros consumo diário de ração (CDR) e conversão alimentar (CA). Foram avaliados o histórico produtivo de 65 granjas de CT entre os anos de 2010 e 2013, vinculadas a um sistema cooperativo localizado na região oeste do Estado do Paraná, Brasil, representando 463 lotes com média de 642,79 ± 363,29 animais por lote, totalizando aproximadamente 300.000 animais. Foram considerados 40 fatores de produção relacionados ao manejo, sanidade, instalações e equipamentos, nutrição, genética e ambiente. Observou-se que o CDR foi in uenciado pela posição dos barracões em relação ao sol (P = 0,048), pelo peso de entrada (P < 0,0001) e de saída (P < 0,0001). Para a variável CA a posição dos barracões em relação ao sol (P = 0,0001), o uso de umidi cadores/nebulizadores (P = 0,03), a presença de composteira (P = 0,006) e de árvores nas laterais dos barracões (P = 0,045), o peso de entrada (P < 0,0001) e a duração da fase de CT (P < 0,0001) in uenciaram o parâmetro. As variáveis selecionadas nos modelos explicaram aproximadamente 44 e 20% da variância total do CDR e CA, respectivamente, sendo este recurso uma boa ferramenta para interpretar os fatores relacionados com os parâmetros avaliados. Palavras-chave: Manejo. Modelagem multinível. Suínos. 1 2 3

4 *

Prof. Dr., Departamento de Zootecnia, Universidade Estadual de Londrina, UEL, Londrina, PR, Brasil. E-mail: [email protected] Pesquisadores, UEL, Londrina, PR, Brasil. E-mail: [email protected]; [email protected]; [email protected] Pesquisador, Universidad Autonoma de Barcelona, UAB, Cerdanyola del Valles Catalunha, Espanha. E-mail: [email protected] hotmail.com Prof., Universidad Autonoma de Barcelona, UAB, Cerdanyola del Valles Catalunha, Espanha. E-mail: [email protected] Author for correspondence Received: May 21, 2016 – Approved: Sept. 30, 2016

Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

997 ar./a r. 2017

Si va,

. A. da et a .

Introduction In Brazilian pig production, the growing and nishing (GF) phases are predominantly managed by pig farmers linked to cooperatives and private integrations (DAGA et al., 2008), accounting for over 80% of the total (ABCS, 2015). Despite the high concentration of producers linked to these complexes, GF farms hold management and nutrition and health programmes to little uniformity and usually have facilities and equipment with different characteristics (MAES et al., 2004; OLIVEIRA et al., 2007). Consequently, these conditions may contribute to the differences in growth performance of animals between farms. Information about the impact of these factors on performance characteristics is scarce because few studies approach this issue (AGOSTINI et al., 2014). Less is known about the effects of the interaction between various factors that may in uence feed ef ciency in pigs (DOUGLAS et al., 2015). According to Heck (2009), it is important to have the domain and to act assertively in the main factors that in uence the development of pigs in the growing and nishing phases for these steps to achieve high production costs (VAN HEUGTEN, 2010). Mathematical models represented a tool that could help to understand and quantify the biological phenomena or factors involved in animal production (POMAR, 2014). This procedure could improve animal performance, the cost of production and provide an ef cient utilization of feed according to each speci c situation (TEDESCHI et al., 2005). This resource allows a joint assessment of the effects of one or more factors of production on a particular livestock parameter (VILLALBA MATA, 2000), which provides knowledge about the operation as a whole (GIBON et al., 1999), in contrast to experimental studies in which only a small number of factors are considered, which contributes to limitations in knowing what factors affect the ef ciency in the systems of GF pigs (DOUGLAS

et al., 2015). However, the use of mathematical models can support decisions regarding demand in production systems with limited information (data used in the model), which is a normal scenario on a farm (TEDESCHI et al., 2005). Although there is evidence of the feasibility of this approach to allow an holistic view of GF farms in Brazil, this approach was recently applied and evaluated, involving over 93 farms with 683 batches and approximately 495,000 animals (PIEROZAN et al., 2016). The factors that most in uenced the daily feed intake and feed conversion in GF units were the number of pigs per pen, the feeder model and the origin and sex of animals submitted for fattening. However, it is important to consider that the information obtained is not similar between companies or regions, and also, they underwent changes with the development of this activity. Furthermore, they were able to expose the factors that have an effect on performance. Therefore, the objective of this study was to identify and quantify, using mathematical models, the impact of intrinsic and extrinsic factors of production present in GF farms linked to a cooperative system of the Western region of Paraná State, Brazil, on the parameters of daily feed intake and feed conversion. These factors display other critical points involved in predicting production rates and increases the perception of the most important factors that in uence the performance of GF units.

Materials and Methods The production history of 463 batches of pigs in GF phases (642.79 ± 363.29 animals per batch) (mean ± standard deviation), totalling approximately 300,000 animals from 65 GF farms linked to a cooperative system located in the Western region of the Paraná State (Brazil), were evaluated between 2010 and 2013. The batches of pigs in the nursery output weighed 21.72 ± 1.31 kg and were kept in the GF phases until they achieved a slaughter weight of 113.28 ± 4.26 kg.

998 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

In

ence o prod ction actor on eed inta e and eed conver ion ratio o

The work model followed the methods by Agostini et al. (2014) in three stages. The rst was the choice of the variables of livestock interest, which is the most important production factor in the cooperative complex (Tables 1 and 2). Subsequently, a procedure that offered reliability, agility and ef ciency in collecting the information was established. More details on the rst two steps are described in the following paragraphs. The third stage involved the development of a system that would ensure the representativeness of the data collected in the cooperative. To ensure representation, information from all of the farms belonging to the cooperative was collected. Furthermore, as opposed to Agostini et al. (2014), data were collected in the present study from a larger number of batches per farm (approximately 7.12 batches per farm). The variables were selected from recent scienti c studies. The experiences of the proposing team and of the cooperative’s staff were divided into two groups: “dependent” and “independent” variables. The “dependent” variables corresponded to continuous variables, such as daily feed intake (DFI) and feed conversion (FC). The total feed intake per animal was calculated as the total amount of feed (in kilograms) delivered to each batch during the GF period, minus the amount of feed (in kilograms) remained in the silos when the animals were sent to slaughter, the result of which was divided by the number of pigs marketed. The DFI per animal was calculated using the results of total consumption per animal, divided by the average number of days in which the animals remained in the GF unit. The FC was obtained by dividing the total feed intake of each batch by the difference between the total kilograms of pigs sent to slaughter and the total kilograms of pigs that entered at the GF batch.

ro - ini in pi

The “independent” variables evaluated included three continuous variables: initial weight, nal weight and duration of GF phases. The initial weight corresponded to the average live weight of pigs when entering the farm, in GF units, and the nal weight and the average live weight at slaughter were both expressed in kilograms. The duration of the GF phases represented the period, in days, that the animals remained in the GF unit. Forty “independent” categorical variables were evaluated (Table 1), accounting for the factors of production, including issues linked to facilities, health status of the herd and aspects related to the nutritional, feed and animal managements. An Excel spreadsheet was used as the basis for carrying out statistical analyses of the data collected, which were divided into two phases: exploratory analysis and models development. In the exploratory analysis phase, the data were submitted to a descriptive analysis of categorical variables performed through a frequency study using the SAS FREQ procedure (SAS Inst., Inc., Cary, NC, USA, version 9.2) (Table 1). The descriptive analysis of continuous variables was performed using measures of central tendency (mean and median) and dispersion (standard deviation, quartiles and amplitude) through the SAS MEANS procedure (Table 2). Continuous variables were submitted to statistical evaluations to assess the normality of their distributions through the SAS UNIVARIATE procedure. For the analysis of all the variables, the batch was considered the experimental unit, de ned as the group of piglets that were coming out of the nursery phase and entering the GF unit, being held to slaughter. All batches were managed as all-in allout system.

999 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

Si va,

. A. da et a .

Table 1. Occurrence percentages in each category for the production factors (independent categorical variables) studied in 463 batches of growing and grow- nishing pigs. Variable Reform of facilities from 20091 Presence of shallow pools in pens1 Humidi ers/misting in farm2 Presence of composters in farm2 Trees around the barns of pigs2 Barn’s position relative to the sun3 Semester of pig placement on the farm4 Number of animals placed4 Number of barns in the farm4 Stall age4 Type of feeder4 Presence of slurry tank4 Labor force in the farm4 Number of pigs per pen5 Building material/barns5 Type of drinker5 Water source provided to animals5 Type of material for the roofs of the barns5 Type of material for the water pipes5 Type of oor material of the pens5 Electricity supply to the barns5 Presence of waste lagoons in the farm5 Presence of ventilation fans in the barns5 Presence of exhaust fans in the barns5 Agricultural areas close to the farm5 Number of feed uses per period5 Different feeds according to the sex5 Feed form5 Use of shock with antibiotics management5 Routes used to administer antibiotics5 Programmes used in the farm5 Number of employed genetic5 Animal breeds used5 Baths housed in the farm are sexed5 Sex segregation in pens5 Location of animal origin5 Sex of animals housed5 Enzootic pneumonia, Glasser’s disease5 Ileitis, meningitis, erysipela5

Percentage of batches (%) in each category Yes (52.58%); no (47.42%) Yes (44.72); no (55.28%) Yes (35.21%); no (64.79%) Yes (86.18%); no (13.82%) Yes (73.00%); no (27.00%) Opposite (36.29%); parallel (63.71%) Summer/autumn (49.12%); winter/spring (50.88%) <500 (30.02%); 500-1,000 (36.29%); >1,000 (33.69%) One (28.08%); two or more (71.92%) <10 years (68.94%); >10 years (31.06%) Linear dump (58.75%); others (41.25%) Yes (57.45%); no (42.55%) Unfamiliar (13.61%); familiar (86.39%) <20 (95.82%); 20-40 (2.86%); >40 (1.32%) Masonry (97.19%); wood (1.73%); mixed (1.08%) Nipple (98.06%); water cup (1.94%) Well/headwater (64.60%); treated water (35.40%) Clay (55.76%); others (44.24%) PVC pipe (91.41%); hose (7.32%); mixed (1.27%) Concrete (100%) Yes (100%) Yes (100%) No (100%) No (100%) Yes (73.43%); no (26.57%) Four (92.22%); ve (7.78%) No (100%) Pelleted (100%) Yes (100%) Water and feed (100%) Ractopamine and immunocastration (100%) Three (100%) Hybrids: Pietrain/Landrace/Large White (100%) No (100%) No (100%) SPU6 (68.77%); farrow-to- nish units (31.23%) Mixed (100%) Yes (100%) Yes (100%)

Variables included only in the nal regression model for dependent variable daily feed intake (DFI); 2Variables included only in the nal regression model for dependent variable feed conversion (FC); 3Variables included in the nal regression model for two dependent variables (DFI e FC); 4Variables initially considered for statistical analysis but not included in the nal models; 5 Variables rejected for statistical analysis due to low variability between categories; 6Specialized piglet production unit. 1

1000 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

In

ence o prod ction actor on eed inta e and eed conver ion ratio o

ro - ini in pi

Table 2. Measures of central tendency and dispersion for the dependent and independent continuous variables selected for the nal models. Variable Initial weight (kg) Final weight (kg) DGF (day) DFI (kg pig-1) FC (kg kg-1)

Nº of batches 463 463 463 463 463

Mean 21.72 113.28 109.60 2.11 2.52

SD 1.31 4.26 3.48 0.10 0.08

Minimum 15.01 96.03 100.00 1.84 2.28

1st quartile 21.43 110.28 107.00 2.04 2.47

Median 21.94 113.04 110.00 2.11 2.52

3rd quartile 22.38 115.90 112.00 2.17 2.57

Maximum 24.65 126.09 119.00 2.46 2.78

SD = standard deviation; DGF = duration of growing- nishing phases; DFI = daily feed intake; FC = feed conversion.

The models were tted based on the variables coded in the rst phase by mixed linear regression using the SAS MIXED procedure with the effect of the farm (primary) and batch linked to the farm considered as random factors, using the restricted maximum likelihood method (REML) for the estimation of variance components. The comparison of the goodness of t of the nal model was based on the proportion of variance explained by the different models, using the coef cient of determination (R2) as a parameter. Initially, in the second phase, a single regression model was used for each variable as a xed effect for each single dependent variable. The independent variables with P ≤ 0.20 were selected for the multivariate analysis. Pearson and Spearman correlations were performed between the independent variables in the multivariate model to avoid multicollinearity between continuous variables and confounding problems between categorical variables. When two variables had a high correlation coef cient (absolute value ≥0.60), only one was used in the multivariate analysis. The choice between them was made by comparing the P values in the univariate analysis, and additionally evaluating their biological relevance with respect to the dependent variable. Subsequently, all independent variables selected in the univariate analysis were submitted to the procedure “stepwise”, where all the factors with P<0.05 were kept in the nal multivariate model. Fixed-effect testing was based on the F-test with denominator degrees of freedom approximated

by the Satterthwaite’s procedure. Signi cant interactions (P<0.05) between the variables in the multivariate model were tested and included. After obtaining the models for each dependent variable, the residues were plotted against the predicted values to check the homogeneity of variances and the presence of outliers. All of the factors with P<0.05 in the nal models for each of the two dependent variables (DFI and FC) were considered statistically signi cant.

Results Table 1 shows the percentages of the occurrence of independent categorical variables (factors of production) studied in GF farms. The factors that were initially considered for statistical analysis and later added in the multiple linear regression model for the dependent variable, DFI, were “reform of facilities from 2009” and “presence of shallow pools in pens”. The factors added into the multiple linear regression model for the dependent variable, FC, were “humidi ers/misting in farm”, “presence of composters in farm” and “trees around the barns of pigs”. The factor, “barn’s position relative to the sun,” was subjected to the initial statistical analysis and was added in two models, for DFI and FC. The DFI per pig was 2.11 ± 0.10 kg (mean ± standard deviation) (range 1.84 to 2.46 kg) (Table 2). Multiple linear regression analysis indicated that the DFI was in uenced by the barn’s position relative to the sun (P = 0.048), initial weight (P<0.0001) and nal weight (P<0.0001) (Table 3). 1001

Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

Si va,

. A. da et a .

When the barn’s position on the farm was contrary relative to the sun, the DFI of pigs increased approximately 0.9% (Table 3). For each additional kilogram of initial weight in animals, the DFI increased by 16 grams (relative to average), and for each kilogram of nal weight, DFI increased by 13 grams. The variance of the model without predictors (null model) and for the model with predictors (full model) for the DFI are shown in Table 4. The total variance in the DFI null model was 0.00993, where

0.00144 (14.5%) was observed between farms and 0.00849 (85.5%), was observed between batches linked to a farm. After the variables were included in the multivariate model, the variance for the DFI was reduced to 0.00552, which indicated that approximately 44.4% of the total variance related to the DFI was explained by the variables included in the full model. The percentages of variance that were elucidated between farms and between batches linked to a farm using the full model for DFI were 52.3 and 43.0%, respectively.

Table 3. Multiple linear regression model to estimate the effects of production factors on the daily feed intake (DFI) in 463 batches. Factors Intercept Reform of facilities from 2009 Presence of shallow pools in pens Barn’s position relative to the sun Initial weight Final weight

Category

Estimate (s.e.)

--Yes No Yes No Opposite Parallel -----

0.234 (0.105) 0.014 (0.009) 0 0.013 (0.009) 0 0.018 (0.009) 0 0.016 (0.003) 0.013 (0.001)

Low 0.028 -0.005 ---0.006 --0.000 --0.009 0.011

95% CL Upper 0.439 0.033 --0.033 --0.037 --0.021 0.015

P-value 0.02 0.16 --0.18 --0.048 --<0.0001 <0.0001

s.e. = standard error; CL = con dence level.

Table 4. Variance observed between farms and between batches within farms and percentage of variance explained by the variables included in the full model for daily feed intake (DFI). Variance observed Full model2 Null model1 0.00144 (14.50%) 0.00068 (12.43%) 0.00849 (85.50%) 0.00484 (87.57%) 0.00993 (100%) 0.00552 (100%)

Effect Farm Batch (Farm) Total

Variance explained 52.32% 43.00% 44.36%

Model without predictors; 2Final model, with predictors.

1

The FC per pig/batch was 2.52 kg ± 0.08 (range 2.28 to 2.78 kg) (Table 2). Multiple linear regression analysis showed that the FC was in uenced by humidi ers/misting in the farm (P = 0.03), the presence of composters at the farm (P = 0.006), trees around the barns (P = 0.045), the barn’s position

relative to the sun (P = 0.0001), initial weight (P<0.0001) and duration of GF phases (P<0.0001) (Table 5). When the farms had humidi ers/misting in their barns, the FC improved approximately 0.7% (Table 5). The animals from pig farms with a composter

1002 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

In

ence o prod ction actor on eed inta e and eed conver ion ratio o

facility had an FC that was 1.2% better than the animals from farms that did not have this feature. The pigs housed on farms with trees planted around the barns had an FC that was 0.7% better than those on farms without trees. The FC was 1.2% worse for the animals kept in farms with barns built in

ro - ini in pi

the opposite direction of sunrise-sunset. For each additional kilogram of initial weight for the animals that arrived on the GF farm, the FC worsened by 0.022 units (relative to average), and for each additional day that the animals remained on the GF period, the FC declined by 0.005 units.

Table 5. Multiple linear regression model to estimate the effects of production factors on the feed conversion (FC) in 463 batches. Factors Intercept Humidi ers/misting in farm Presence of composters in farm Trees around the barns of pigs Barn’s position relative to the sun Initial weight Duration of GF phases

Category

Estimate (s.p.)

--Yes No Yes No Yes No Opposite Parallel -----

1.531 (0.133) -0.017 (0.008) 0 -0.029 (0.010) 0 -0.017 (0.008) 0 0.030 (0.008) 0 0.022 (0.003) 0.005 (0.001)

Low 1.269 -0.032 ---0.049 ---0.034 --0.015 --0.017 0.003

95% CL Upper 1.792 -0.002 ---0.008 ---0.000 --0.045 --0.027 0.007

P-value <0.0001 0.03 --0.006 --0.045 --0.0001 --<0.0001 <0.0001

s.e. = standard error; CL = con dence level; GF = growing and nishing.

The model without predictors (null model) for FC had a total variance of 0.00602, where 0.00050 (8.32%) was observed between farms, and 0.00552 (91.68%) was observed between batches linked to the farm (Table 6). Once the variables were included in the multivariate model, the variance of

FC was reduced to 0.00482, which indicated that approximately 19.9% of the total variance related to FC was explained by the variables selected for the full model. The percentage of variance between farms and batches linked to a farm, using the full model for FC, was 38.1 and 18.2%, respectively.

Table 6. Variance observed between farms and between batches within farms and the percentage of variance explained by the variables included in the full model for feed conversion (FC). Variance observed Null model1 Full model2 0.00050 (8.32%) 0.00031 (6.42%) 0.00552 (91.68%) 0.00451 (93.58%) 0.00602 (100%) 0.00482 (100%)

Effect Farm Batch (Farm) Total 1

Variance explained 38.12% 18.23% 19.88%

Model without predictors; 2Final model, with predictors.

1003 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

Si va,

. A. da et a .

Discussion It was observed that the variables DFI (P = 0.048) and FC (P = 0.0001) (Tables 3 and 5) were affected by the barn’s position relative to the sun, increasing the DFI and decreasing the FC of animals housed in barns positioned crosswise relative to sunrisesunset. When the barn was built in this direction, there was direct sunlight on the sides of the barn for the majority of the day. This could have increased the temperature inside of the barn, causing a negative effect on animal performance. The decrease in the DFI is the most effective mechanism to reduce heat stress (COLLIN et al., 2001) and appears to be essential in regulating the body temperature of pigs (RENAUDEAU et al., 2011). Through this strategy, the resulting heat production of digestive and metabolic processes related to food intake was minimized (MANNO et al., 2006). However, there was an increase in DFI for housed pigs, which could have been related to the environment where consumption should decrease. There is a possibility that other unmeasured factors may have affected the results, such as the formulation of feed. The decline in FC coincided with the results from an earlier study by Kiefer et al. (2009), who observed that this parameter declined in pigs between 30 and 60 kg maintained in the heat compared to pigs in a thermoneutral environment. Renaudeau et al. (2011) reported a worsening of 0.2 kg kg-1 in the FC of pigs weighing 50 kg at a temperature 36 °C compared to pigs maintained at 30 °C. The factor of humidi ers/misting contributed to an improvement of the FC in animals (P = 0.03) (Table 5). Growing pigs produced a larger quantity of heat because of a high metabolic rate. Moreover, pigs in the GF phase have a dif culty in expelling heat because of an increase in the fat layer. This contributes to heat stress, resulting in energy consumed intended to control homeothermy with consequent worsening of the livestock performance index (NÄÄS; JUSTINO, 2014). The management of the environment on pig farms needs to ensure that the concentration of pollutants are minimized

and the thermal environment is optimized to maximize production ef ciency (BANHAZI et al., 2008). This objective can be achieve through misting, which allows evaporative cooling and is considered the most ef cient system for air cooling (NÄÄS; JUSTINO, 2014). The results of this study were related to ndings of Choi et al. (2010), who showed the importance of environmental control through ventilation/misting installations for pigs by comparing automatic with manual ventilation systems. Berton et al. (2015) also observed a worsening FC (P<0.05) for animals in the GF phase that were subjected to climatic variations (uncontrolled environment). The results reinforce the importance of a system for controlling the thermal sensation of animals, improving thermal comfort and performance. Regarding the factor of trees around the barns of pigs, this factor is associated with environmental bene ts. The best FC results (P<0.045) were in favour of farms that adopted wooden sides (Table 5). Studies demonstrated a direct relationship between the FC pigs that were not found; however, the planting of trees on the sides of the barn reduces the direct impact of solar radiation on animals (DIAS et al., 2011), while minimizing the temperature inside the barns on hot days. Keeping animals in conditions within their thermoneutral zone maximizes performance (MILLER, 2012). Another factor with a positive response on the FC was the use of composting on the farm (P = 0.006) (Table 5). An animal carcass is a great source of pathogens, drug toxins and other chemicals, which must be eliminated or reduced to safe values to minimize their potential risk (BERGE et al., 2009). Thus, the composting of animals has be incorporated to prevent the transmission and spread of an infection (KALBASI et al., 2005) and to reduce the pathogenic microorganisms to appropriate levels (BERGE et al., 2009). Therefore, it is considered a resource that shows positive results on environmental and health aspects (SOTO et al., 2010). Thus, it is possible to assume that

1004 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

In

ence o prod ction actor on eed inta e and eed conver ion ratio o

ro - ini in pi

farms will have to incorporate the use of composters to directly and indirectly promote the health of the herd as a whole, affecting animal performance; therefore, in an improved sanitary condition and with less adversity, the dietary nutrients would be used primarily for the development of the animal and less for immune responses, thereby improving the FC.

housed in GF units with a high average weight, a large number of pigs reach their slaughter weight earlier, but the lighter animals require more time to reach a slaughter weight, which is of interest to the company. The worsening of the FC for the entire batch is a result of an increased deposition of fat and less lean gain in animals with a higher initial weight.

With respect to the continuous variables initial weight and nal weight, there was an increased DFI when the initial weight of pigs and the weight of the nisher pigs were higher (Table 3), similar to results obtained by Pierozan et al. (2016). In a comparison between light and heavy piglets at birth, Wolter et al. (2002) observed that, within any GF period, piglets who were heavy at birth had a higher daily weight gain and increased DFI that piglets who were lighter at birth. Thus, it was assumed that the animals entering the GF phases at a heavier weight also ended this phase with greater weight and consequently have a higher feed intake during the whole period of housing. Thus, there is a clear synergistic relationship in the initial and nal weights with the DFI (SILVA, 2010).

The value of 44.4% of the total variance explained by the DFI model was close to that achieved by Pierozan et al. (2016) (50%), but dissimilar to the value obtained by the Agostini et al. (2014) (62%). For total variance explained by the FC model, 19.9% was observed in the present study, which was a closer value to that obtained by Agostini et al. (2014) (24.8%) and distant from that observed by Pierozan et al. (2016) (64%). All of these studies used a multiple linear regression model to quantify the factors that act on pig performance characteristics. The differences between the explained variance values could be attributed to differences between the variability of the factors included in the full models ( nal models with predictors).

The independent variables “initial weight” and “duration of the GF phases” in uenced the dependent variable FC, which declined with an increasing initial weight of pigs and of the period in which the animals were kept on the farm (Table 5). Until reaching 56-63 kg, the pigs had an increasing lean meat deposition, and from this point forward, they were heavier and older, causing a decline in the FC because of a reduction in the deposition rate of lean tissue and an increase in fat deposition (BÜNZEN et al., 2014). Considering that the pigs enter the GF phases heavier when reaching their slaughter weight, seven days before then animals that enter lighter (WOLTER et al., 2002), batches with high initial weight spent more time than necessary in the farm until the lighter pigs reached slaughter weight, causing the FC to worsening for the batch as a whole. A possible explanation for the results of this study is that when animals are

The factor, “barn’s position relative to the sun,” had signi cance for both performance variables (DFI and FC), which could be explained because the factors included in each model were different (except the factor aforementioned), unlike the study by Pierozan et al. (2016) in which all the factors included in the linear regression models for DFI and FC were the same and were signi cant for both parameters.

Conclusions Under the tested conditions, the multiple linear regression model for DFI was able to identify that there was an increased daily feed intake by pigs in the GF phases when the following conditions were met: I) the position of the barns in farm was opposite to the direction of sunrise-sunset (daily feed intake 0.9% higher); II) the initial weight of the animals of the batch was greater; and III) the nal weight 1005

Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

Si va,

. A. da et a .

of the animals in the batch was greater. The model obtained for FC identi ed improvement in this index when the following conditions were met: I) the farms had humidi ers/misting in their installations (FC 0.7% better); II) the farms had composter facility on the farm (1.2%); and III) the farm had trees planted around the barn of pigs (0.7%). The FC worsened under the following conditions: I) the position of the barns at the farm was opposite to the direction of sunrise-sunset (FC 1.2% worse); II) the initial weight of the animals in the batch was greater; and III) the animals spent more days on the farm before they could be transported to slaughter. These results can contribute to the improvement of the indices evaluated by actions on the factors involved, and this mathematical feature is a tool to support technical decisions on the farm.

Acknowledgements First, we thank the cooperative participants, who trusted us with the data from their farms. We also thank the public research project funded by the Spanish Ministry of Education (AGL 2011-29960), in which this study was developed.

Referências AGOSTINI, P. S.; FAHEY, A. G.; MANZANILLA, E. G.; O’DOHERTY, J. V.; BLAS, C.; GASA, J. Management factors affecting mortality, feed intake and feed conversion ratio of grow- nishing pigs. Animal, Cambridge, v. 8, n. 8, p. 1312-1318, 2014. ASSOCIAÇÃO BRASILEIRA DOS CRIADORES DE SUÍNOS – ABCS. Panorama Setorial da Suinocultura. Dossiê detalhado do setor suinícola. [S.l.]: Sede Brasília, 2015. Disponível em: . Acesso em: 30 ago. 2016. BANHAZI, T. M.; SEEDORF, J.; RUTLEY, D. L.; PITCHFORD, W. S. identi cation of risk factors for suboptimal house conditions in Australian Piggeries: Part 3. Environmental Parameters Journal of Agricultural Safety and Health, St. Joseph, v. 14, n. 1, p. 41-52, 2008. BERGE, A. C. B.; GRANVILLE, T. D.; MILLNER, P. D.; KLINGBORG, D. J. Methods and microbial risks associated with composting of animal carcasses in

the United States. Journal of the American Veterinary Medical Association, Ithaca, v. 234, n. 1, p. 47-56, 2009. BERTON, M. P.; DOURADO, R. C.; LIMA, F. B. F.; RODRIGUES, A. B. B.; FERRARI, F. B.; VIEIRA, L. D. C.; SOUZA, P. A.; BORBA, H. Growing- nishing performance and carcass yield of pigs reared in a climatecontrolled and uncontrolled environment. International Journal of Biometeorology, Nova Iorque, v. 59, n. 8, p. 955-960, 2015. BÜNZEN, S.; APOLÔNIO, L. R.; SILVA, M. A. Técnicas de manejo e alimentação para melhoria da conversão alimentar. In: FERREIRA, A.H.; CARRARO, B.; DALLANORA, D.; MACHADO, G.; MACHADO, I. P.; PINHEIRO, R.; ROHR, S. (Org.). Criação de suínos: teoria e prática. Brasília: ABCS, Integrall, 2014. p. 686-697. CHOI, H. L.; SONG, J. I.; LEE, J. H.; ALBRIGHT, L. D. Comparison of natural and forced ventilation systems in nursery pig houses. Applied Engineering in Agriculture, St. Joseph, v. 26, n. 6, p. 1023-1033, 2010. COLLIN, A.; VAN MILGEN, J.; DUBOIS, S.; NOBLET, J. Effect of high temperature on feeding behaviour and heat production in group-housed young pig. The British Journal of Nutrition, Cambridge, v. 86, n. 1, p. 63-70, 2001. DAGA, J.; CAMPOS, A. T.; FEIDEN, A.; KLOSOWSKI, E. S.; CÂMARA, R. Análise da adequação ambiental e manejo dos dejetos de instalações para suinocultura em propriedades na região oeste do Paraná. Engenharia Agrícola, Botucatu, v. 27, n. 3, p. 1-11, 2008. DIAS, A. C.; CARRARO, B. Z.; DALLANORA, D.; COSE, F. J.; MACHADO, G. S.; MACHADO, I. P.; PINHEIRO, R.; ROHR, S. A. Manual brasileiro de boas práticas agropecuárias na produção de suínos. Brasília: ABCS, 2011. 140 p. DOUGLAS, S. L.; SZYSZKA, O.; STODDART, K.; EDWARDS, S. A.; KYRIAZAKIS, I. Animal and management factors in uencing grower and nisher pig performance and ef ciency in European systems: a metaanalysis. Animal, Cambridge, v. 9, n. 7, p. 1210-1220, 2015. GIBON, A.; SIBBALD, A. R.; THOMAS, C. Improved sustainability in livestock systems, a challenge for animal production science – Introduction. Livestock Production Science, Amsterdam, n. 61, p. 107-110, 1999. HECK, A. Fatores que in uenciam o desenvolvimento dos leitões na recria e terminação. Acta Scientiae Veterinariae, Porto Alegre, v. 37, p. 211-218, 2009. Suplemento 1.

1006 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017

In

ence o prod ction actor on eed inta e and eed conver ion ratio o

ro - ini in pi

KALBASI, A.; MUKHTAR, S.; HAWKINS, S. E.; AUVERMANN, B. W. Carcass composting for management of farm mortalities: a review. Compost Science & Utilization, Emmaus, v. 13, n. 3, p. 180-193, 2005.

POMAR, C. The utilization of mathematical models to improve swine production. In: CONGRESSO LATINOAMERICANO DE NUTRIÇÃO ANIMAL, 6., 2014, Estância de São Pedro. Anais... Estância de São Pedro: [s.n.], 2014. p. 23-26.

KIEFER, C.; MEIGNEN, B. C. G.; SANCHES, J. F.; CARRIJO, A. S. Resposta de suínos em crescimento mantidos em diferentes temperaturas. Archivos de Zootecnia, Córdoba, v. 58, n. 221, p. 55-64, 2009.

RENAUDEAU, D.; GOURDINE, J. L.; ST-PIERRE, N. R. A meta-analysis of the effects of high ambient temperature on growth performance of growing- nishing pigs. Journal of Animal Science, Champaign, v. 89, n. 7, p. 2220-2230, 2011.

MAES, D.; DUCHATEAU, L.; LARRIESTRA, A. J.; DEEN, J.; MORRISON, R. B. Risk factors for mortality in grow- nishing pigs in Belgium. Journal of Veterinary Medicine B, Berlin, v. 51, n. 7, p. 321-326, 2004. MANNO, M. C.; OLIVEIRA, R. F. M.; DONZELE, J. L.; OLIVEIRA, W. P.; VAZ, R. G. M. V.; SILVA, B. A. N.; SARAIVA, E. P.; LIMA, K. R. S. Efeitos da temperatura ambiente sobre o desempenho de suínos dos 30 aos 60 kg. Revista Brasileira de Zootecnia, Viçosa, MG, v. 35, n. 2, p. 471-477, 2006. MILLER, T. G. Swine feed ef ciency: in uence of temperature. [S.l.]: IPIC, 2012. 2 p. Available a t : < h t t p : / / l i b . d r. i a s t a t e . e d u / c g i / v i e w c o n t e n t . cgi?article=1005&context=ipic_factsheets>. Accessed at: 30 ago. 2016. NÄÄS, I. A.; JUSTINO, E. Sistemas de climatização parcial e total em granjas de suínos. In: FERREIRA, A. H.; CARRARO, B.; DALLANORA, D.; MACHADO, G.; MACHADO, I. P.; PINHEIRO, R.; ROHR, S. (Org.). Criação de suínos: teoria e prática. Brasília: ABCS, Integrall, 2014. p. 896-905. OLIVEIRA, J.; GUITIÁN, F. J.; YUS, E. Effect of introducing piglets from farrow to nish breeding farms into all-in all-out fattening batches in Spain on productive parameters and economic pro t. Preventive Veterinary Medicine, Amsterdam, v. 80, n. 4, p. 243-256, 2007. PIEROZAN, C. R.; AGOSTINI, P. S.; GASA, J.; NOVAIS, A. K.; DIAS, C. P.; SANTOS, R. K. S.; PEREIRA JÚNIOR, M.; NAGI, J. G.; ALVEZ, J. B.; SILVA, C. A. Factors affecting the daily feed intake and feed conversion ratio of pigs in grow- nishing units: the case of a company. Porcine Health Management, Barcelona, v. 2, n. 1, p. 1-8, 2016.

SILVA, C. A. Consumo dos suínos. In: VIEIRA, S. L. (Org.). Consumo e preferência alimentar dos animais domésticos. Londrina: Midiograf, 2010. v. 1, p. 204-265. SOTO, F. R. M.; COELHO, C. P.; BENITES, N. R.; BONAMIN, L. V.; MELVILLE, P. A. Elimination of Escherichio coli from piglet carcass using an on-farm composting system. Veterinary Record, London, v. 166, n. 18, p. 564-565, 2010. TEDESCHI, L. O.; FOX, D. G.; SAINZ, R. D.; BARIONI, L. G.; MEDEIROS, S. R.; BOIN, C. Mathematical models in ruminant nutrition. Scientia Agricola, Piracicaba, v. 62, n. 1, p. 76-97, 2005. VAN HEUGTEN, E. Growing- nishing swine nutrient recommendations and feeding management. In: MEISINGER, D. J. (Ed.). National swine nutrition guide. Ames: Iowa State University Press, 2010. p. 80-95. VILLALBA MATA, D. V. Construcción y utilización de un modelo estocástico para la simulación de estrategias de manejo invernal en rebaños de vacas nodrizas. 2000. Tesis (Doctoral en Ciencia y Tecnologia Agraria i Alimentaria) – Departament de Producció Animal. Universitat de Lleida, Lleida. WOLTER, B. F.; ELLIS, M.; CORRIGAN, B. P.; DeDECKER, J. M. The effect of birth weight and feeding of supplemental milk replacer to piglets during lactation on preweaning and postweaning growth performance and carcass characteristics. Journal of Animal Science, Champaign, v. 80, n. 2, p. 301-308, 2002.

1007 Se ina:

i ncia A r ria , ondrina, v. 38, n. 2, p. 997-1008,

ar./a r. 2017