Available online at www.sciencedirect.com

(2008) 96–107www.elsevier.com/locate/livsci

Livestock Science 116

Do labour productivity and preferences about work load distributionaffect reproduction management and performance in pig farms☆

G. Martel a,b,⁎, J.-Y. Dourmad a, B. Dedieu b,⁎

a INRA, Agrocampus Rennes, UMR1079, Système d'élevage nutrition animale et humaine, F-35000 Rennes, Franceb INRA Transformation des Systèmes d'Elevage, UMR1273, Metafort, F-63122 Saint-Genès-Champanelle, France

Received 4 April 2007; received in revised form 10 September 2007; accepted 11 September 2007

Abstract

Increases in labour productivity are essential factors, as well as technical effectiveness, for the competitiveness of pig farming.However, the preferences of farmers for controlled (i.e. limited) daily working hours or available days for vacation also increase.The objective of this study was to explore how these preferences about work might be associated to specific combinations ofpractices or affect performance. The study was carried out by direct investigation of the stockbreeders. Data analysis used factorialanalysis to identify relationships between practices, labour productivity, sow productivity and work load distribution. Resultsshowed independence between sow productivity and labour productivity. Three independent types of preferences about work loaddistribution were identified: the limitation of density of daily work, the avoidance of insemination activities during the weekendand the avoidance of farrowing supervision during the weekend. These preferences about work load distribution were mainlyrelated to weaning, oestrus detection and insemination techniques. A relationship was also seen between farrowing and cross-fostering techniques, and labour and sow productivity. Results suggest that preferences about work load distribution influence thechoice of reproduction practices without influencing performance. Finally, concerning the labour productivity, it was linked withsome specific techniques at farrowing but the results also indicated that it was mainly related to the size of farrowing batches.© 2007 Elsevier B.V. All rights reserved.

Keywords: Farm management; Work planning; Survey; Factorial analysis; Reproductive performance

1. Introduction

Asmost of the other animal productions, pig productionis facing pressures on the farmer work. The first pressure ison labour effectiveness which is an essential factor, withtechnical effectiveness, of pig farming competitiveness:the average number of sows per AnnualWorkUnit (AWU)

☆ This paper was presented at the 58th annual meeting of the EAAP.⁎ Corresponding authors. Dedieu is to be contacted at Tel.: +33 4 73

62 40 38; fax: +33 4 73 62 41 18. Martel, UMR SENAH, Domaine dela Prise 35590 Saint Gilles, France.

E-mail addresses: gilles.martel@rennes.inra.fr (G. Martel),Benoit.Dedieu@clermont.inra.fr (B. Dedieu).

1871-1413/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.livsci.2007.09.012

is continuously increasing (by 80% between 1990 and2006 in France; IFIP, personal communication). Thesecond pressure is relative to the work load distribution, inrelation with the evolution of social norms and values inagriculture. The farmers expect something else from theirwork than the “peasant toil” where private and work livesare closely mingled (Barthez, 1986; Jean et al., 1988). Thefarmers now have various attitudes toward work loaddistribution including the control of dailyworking hours orthe ability to free some days for vacation or weekends(Guillaumin et al., 2004). The development of wage-earning and the reduction of the agricultural family work-force reinforce also the necessity of taking into account the

97G. Martel et al. / Livestock Science 116 (2008) 96–107

duration and distribution of work. Although studies onwork organisation are rare in swine breeding, studiesundertaken in other sectors of production (ruminants)show that stockbreeders adapt their livestock practices, aswell as farm equipment and the composition of the labourforce, in their search of new coherences between issues ofcompetitiveness and issues of preserved time (Dedieuet al., 2006). More specifically, technical adjustments canbe occasional (concerning a specific unit or period of theyear) or profound, such as modifications in the livestockmanagement strategies, notably reproduction strategies(Cournut and Dedieu, 2005). In pig production, thepractices at stake refer to the tasks which have a periodicrhythm (Le Borgne et al., 1994) (Fig. 1A) in relation withthe reproductive cycle of a farrowing batch (i.e. a group ofsowsmanaged together). Themajor element tomanage thedistribution of periodic tasks from 1 week to another is thenumber of farrowing batches in the herd (Caugant, 2002)(Fig. 1B). But the relationship between reproductionmanagement practices on the one hand and control of thedaily work duration and preserved weekends on the otheris less documented. This paper deals with the changes ofwork in pig farms, either the research of increased labourproductivity or the control of work load distributionwithinthe week. We hypothesise that the increase in labour

Fig. 1. Periodical tasks occurring in the reproduction unit of a pig farm. The popposed to daily tasks such as feeding the animals (Le Borgne et al., 1994).represents one farrowing batch over the full reproductive cycle. Usual days ofeach task are written under the line. B) Periodical task organisation at the herdfarrowing batches, in the 3-week BFS there are 7 farrowing batches and in t

productivity might have induced a simplification ofreproduction management or have consequences on theherd productivity. We also hypothesise that, as in the otherproduction systems, the pig farmers have variouspreferences when considering the work in odd hours andin weekends. These various preferences may be related todifferent reproduction managements and may affect theherd productivity. To test these hypotheses, we havecarried out a survey with stockbreeders who had variousbatch farrowing systems and herd size, so presumably alarge variety of management practices and laboureffectiveness. The data collected were supplemented bythe retrieval of performance data from the Technical SowHerd Management System (TSHMS).

2. Materials and methods

2.1. Work and practices analysis

Livestock practices are described in the Livestock FarmingSystem approach (Gibon et al., 1999) by their technical content,their justification and their effect on herd performances.Following previous studies on work organisation in livestockfarms (Dedieu et al., 2006)we refer to the ergonomics frameworkto clarify the relationship between livestock practices and work.Livestock practices refer to “prescribed” tasks i.e. to a protocol

eriodical tasks refer to non daily tasks which have regularity. They areA) Periodical task organisation at the farrowing batch level. The lineeach task are written under the task name. Usual time intervals betweenlevel. In the 1-week Batch Farrowing System (BFS), there are 20 to 21he 4-week BFS there are 5 farrowing batches.

Table 1Characteristics and productivity criteria of farms from the studysample and from the Technical Sow Herd Management System(Brittany data)

Items Study sample Brittany

Average Standarddeviation

Average Standarddeviation

Number of sows 306 226 210 157Number of sowsper batch

26.0 14.1 – –

Number of AWUa

affected to sows1.65 1.13 – –

Sows per AWUa 196 63.0 – –Piglets weaned perweek per AWUa

88.1 32.6 – –

Age of weanedpiglets (days)

24.1 3.4 25.1 2.8

Piglets weaned persow per year

28.0 1.7 26.2 2.4

Live born pigletsper litter

13.1 0.5 12.5 0.8

Weaned piglets per litter 11.3 0.6 10.7 0.8Piglet mortality(% of total born)

20.1 4.4 21.2 5.6

Age at firstfarrowing (days)

377 11.5 373 19

Weaning to oestrusinterval (WOI) (days)

5.33 0.68 5.10 0.7

Weaning to conceptioninterval (WCI) (days)

7.90 1.99 8.90 4.80

Fertility rate atfirst oestrus

87.3 6 85.1 10

Number of littersper culled sow

5.3 0.6 5.0 1.0

Age at culling (months) 34.3 3.1 34.2 5.3

a AWU: Annual Work Unit.

98 G. Martel et al. / Livestock Science 116 (2008) 96–107

(insemination for instance) specifying the procedure (oneartificial insemination every 12 h or one every 24 h for eachsow), the different procedures depending for example on theanimal condition or parity. The other face of the practice is the“working activity” defined as the interaction between theprescribed task and the operator(s) (Leplat, 1994). In reality,the prescribed tasks are executed by the labour force withinworking sessions. Aworking session is a time interval devoted tosome specific activities. The number of sessions devoted toreproduction (within a day when reproductive activities occur, orwithin a period) is thus variable between labour forces and,because of this labour force, can be more or less adjusted to otherwork peaks. The number of sessions of periodic workingactivities is one major factor of the duration of the periodic work.

The work combines different temporal coherences (Madel-rieux and Dedieu, submitted for publication). Notably, inanimal productions, tasks can be divided into daily tasks(which are repeated each day) such as feeding, and non dailytasks, some of them being flexible (they can be postponedwithin a period), others being rather inflexible (they have to bedone at a predetermined moment (of the week, of the period).The periodical tasks of reproduction that we are studying inthis paper refer to this last category.

2.2. Sampling and surveying the farms

Farmers and farms were selected according to four criteriathat we assumed could highlight different reproductionmanagement methods and/or different preferences about laboureffectiveness and work load distribution. The criteria were 1)number of sows in a farrowing batch (less or equal to 20 vs.more), 2) number of employees on the farm (no employee, onepartial or full time employee or several employees), 3) number offarrowing batches in the herd (from 4 to 21, corresponding tointervals between farrowing batches of 5 to 1 week(s)respectively), and 4) lactation length (21 or 28 days). Twenty-five farrowing-to-finishing pig breeders from the main swineproduction area in France (Brittany) were selected in partnershipwith three producer groups between February and June 2006.

2.3. Survey contents — Technical Sow Herd ManagementSystem data

Surveys were semi-directed and approached 1) farm historyand characteristics, 2) labour force and its distribution over thefarm activities, and 3) practices, protocols, procedures, workingsessions and calendar over weeks during farrowing supervision,cross-fostering, weaning, oestrus detection and insemination.Technical data fromTSHMS included all reproductive parameters(Table 1).

2.4. Management of collected data

The analysis used three types of variables: structural, practicesand performances. Farmer practice data were categorised with the“knowledge formalisation” grid proposed by Girard et al. (2001)into “practical procedures modalities” referring to the technical

themes cited above. This method taken from “knowledgeengineering” consists in building “series of dichotomic attributes”,defined by extreme situations encountered in the studied cases andthen to identify intermediate situations. (Fiorelli et al., 2007). Eachpractical procedure modality summarizes proximate chaining ofelementary acts. For example, “Provide much help to piglets”includes drying the piglets, milking of sow in order to manuallyfeedsmallpiglets, keepingpiglets awayfromthesowuntil thereareenough piglets born, and heating with UV lamps. The othermodality “Provide some help to piglets” includes only drying thepiglets and heating. Practical procedure modalities with implica-tions on the number of working sessions within the day (“numberof oestrus detection per day”) or their distribution over the week(“weaning in order to avoid insemination during weekend”) wereidentified to analyse the preferences about work load distribution.

2.5. Data analysis

Data analysis was performed in three steps with factorialanalysis methods. These methods provide several factors built

Table 3Practices on farrowing and cross-fostering observed in the sample

Items Modality A Modality B Modality C

FarrowingPharmacological

productsFew (8) Many but few

sows (8)Intensive (9)

Help to piglets Some (15) Much (10)Manual

assistanceSeldom (10) Sometimes (8) Frequently (4)

Anticipatefarrowingdifficulties

No (18) Yes (7)

Cross-fosteringHyperprolificity

managementWithin batch(8)

Betweenbatches (9)

Several extrabatch possibilities(7)

Number ofpiglets undergilts

Less thanunder sows(3)

Equal to sows(9)

More than undersows (13)

Cross-fosteringwithin parity

No (19) Yes (6)

Reduce litterweightvariability

No (8) Yes, 1 litter ofsmall piglets (8)

Yes for all litters(9)

Fostering/productive

No (11) Yes (14)

99G. Martel et al. / Livestock Science 116 (2008) 96–107

from active variables and describing their distribution. Thefirst step aimed at studying the relationship between sowperformance, herd productivity and labour effectiveness. Thelabour effectiveness could be estimated with the labourproductivity (Tirel, 1989) which was calculated as the numberof sows present in the herd divided by the number ofequivalent full time worker (AWU criterion) declared by thefarmer as being involved in the management of sows andpiglets until weaning. The analysis consisted of a PrincipalComponents Analysis method (PCA) (Escofier and Pagès,1998) using the variables defined in Table 1. The second stepconsisted of an analysis of farmers' motivations concerningwork load distribution. Variables shown in Table 2 were usedin a Correspondence Factorial Analysis (CFA) (Escofier andPagès, 1998), which was followed by an ascendant hierarchi-cal classification (AHC) in order to define classes ofmotivations.

The last step consisted of an analysis of farmers' practices,which was performed using a Multiple Factorial Analysis(MFA) (Escofier and Pagès, 1998). The choice of MFA wasmade to allow analysis of relationships both within andbetween groups of active variables. This step was divided inthree subparts: the first one focused on farrowing and crossfostering practices (Table 3) and the second on weaning,oestrus detection and insemination (Table 4). The third subpart

Table 2Working rhythm observed in the sample

Items Modality A Modality B Modality C

Principal dayof farrowing

Wednesday(7)

Wednesday/Thursday (9)

Thursday(9)

Weaning in order toavoid inseminationduring weekend

No (15) Yes (10)

Weaning in orderto avoid farrowingduring weekend

No (19) Yes (6)

Weaning in order tomanage dailywork time

No (20) Yes (5)

First day ofoestrus detection

Friday (13) Sundayevening (5)

Monday(7)

Do detections stop ona specific day inthe week?

No (20) Yes (5)

Number of detectionsessions per day

One (10) Several (15)

Number of detectionsessions per week

1 to 7 (9) 8 or more(14)

Number of detectionand AI sessionsper week

2 to 7 (5) 8 to 11 (10) 12 ormore (8)

Change in workinghours for farrowingsupervision

None orfew (14)

More than 12 hof work somedays (11)

Number of cases is shown in brackets.

history

Number of cases is shown in brackets.

included the four groups of active variables (Tables 3 and 4).All the analyses performed at this step had also two groups ofsupplementary variables corresponding to variables used instep 1 and the classes resulting from step 2.

All analysis was made using SPAD v6.5 software (SPAD,2006).

3. Results

3.1. Step 1: Sow and labour productivity

The PCA identified three main factors that explained31, 19 and 15% of the variability. They are presented inTable 5. The first factor was related to the productivityof sows. It opposed the number of piglets weaned persow per year to piglet mortality between birth andweaning. The second factor was related to sow fertilityand opposed low fertility and low number of litters perculled sow on one hand to high fertility and high numberof litters per culled sow on the other. The third factorwas linked to labour productivity measured as thenumber of sows and piglets per AWU. These resultsshowed no relationship between sow productivity,fertility and labour productivity.

Table 4Practices on weaning, oestrus detection and insemination observed inthe sample

Items Modality A Modality B Modality C

Weaning and Oestrus detectionDay of weaning Wednesday

(13)Thursday (12)

Type of weaning Alltogether(18)

Split weaning(4)

Sows and pigletsmove on differentdays (3)

Techniques tohelp oestrus

None (8) Usepharmaceuticals(9)

Sows groupedafter weaning (8)

Boar number fordetection

One (14) Several (11)

InseminationWeeklymodificationsto protocol

None (8) One (11) Several (6)

Number ofinseminationsper sow perday

One (7) One or two (12) Two (6)

Distinctionbetween giltsand sows

No (17) Yes (8)

Boars can mate No (19) Yes (6)Source of semen Purchased

(15)On farm (10)

Number of cases is shown in brackets.

100 G. Martel et al. / Livestock Science 116 (2008) 96–107

3.2. Step 2: Work load distribution

The first three factors of the CFA explained 23, 17and 14% of the variability. They are presented inTable 5. The first factor was relative to the load ofperiodic tasks during a day. It opposed on the one handpractices with only one session of oestrus detection in aday, a limited number of insemination sessions in aweek and little modification to working hours forfarrowing supervision, and on the other hand, practiceswith two or more sessions of oestrus detection andinsemination per day, and more working hours forfarrowing supervision. The second factor mainlyexpressed options chosen by the farmers to avoid theoccurrence of farrowing during the weekend. The thirdfactor opposed practices on their capacity to limitoestrus detection and insemination activities during theweekend. Based on these factors, the AHC categorisedfarmers into four classes (Fig. 2). The first class wasmainly defined on the first factor and was qualified aspractices aiming at the “limiting the density of periodictasks during the day” (summarised as “limitation ofdaily tasks” in Fig. 2). The second class was mainly

defined on the second factor and was qualified as“practices adapted to the sows' biological rhythm”. Thefarmers started to detect oestrus on Sunday evening inorder to be as precise as possible on the onset of oestrus.They also remained on the farm to supervise farrowingthat were not induced by pharmacological products. Thethird class was mainly defined on the third and thesecond factor and was qualified as practices attemptingto “avoid insemination tasks on weekends”. The lastclass was mainly defined on the first and second factorand corresponded to practices with many periodicworking sessions per day (“priority to work that has tobe done in order to achieve the prescribed tasks”) andpractices attempting to avoid farrowing during theweekend. This class is summarised as “priority to work”in the following text.

3.3. Step 3: Management of reproduction

3.3.1. Subpart a: Farrowing and cross-fosteringPractices performed at these working sessions could

be associated on two factors which explained 16 and15% of the variability. They are presented in Table 5.The first factor was relative to the frequency ofinterventions on animals. At one end of the axis, theanalysis grouped frequent interventions on animalsincluded frequent manual assistance and frequent useof pharmaceuticals and several possibilities for cross-fostering piglets including sows from other farrowingbatches and artificial suckling machines. This set ofpractical procedures is positively correlated with thesow productivity (R=0.53). At the other end weregrouped practices with minimal use of pharmacologicalproducts on sows and limited manual assistance duringfarrowing, fewer piglets under gilts, and cross-fosteringwithin the batch with consideration for piglet weight.These procedures were linked to higher piglet mortality(R=0.50). The second factor established a gradient ofinterest towards piglet survival. On one side, we foundpractices that offer considerable help to new-bornpiglets and utilised multiple criteria methods for thecross-fostering of surplus piglets (in relation with weightof piglets, number homogeneity of the litter size andparity of the sow). This set of practices was related tohigh sow and labour productivity (R=0.37 and 0.58respectively) and large farrowing batches. On the otherside, practices providing only limited help to new-bornpiglets were found. They used simplified criteria forcross-fostering that occurred only within the batch anddid not depend on parity status or on weight of piglets.These practices tend to be associated with higher pigletmortality (R=0.12).

101G. Martel et al. / Livestock Science 116 (2008) 96–107

3.3.2. Subpart b: Weaning, oestrus detection andinseminations

The variability was mainly described by two factorsexplaining 21 and 15% of the variability respectively.

Table 5Description by the modalities of each criterion of the two extremities of the

Factor 1

Step 1: Sow and labour productivity “Sow productivity”Low High

Piglet weaned/sow/year XPiglet mortality between birth and weaning XFarrowing rateLitters/culled sowSow/AWUPiglet weaned/week/AWUStep 2: Work load distribution “Daily workload”

Limited Priority to wNumber of oestrus detection/day One SeveralNumber of detection and insemination

sessions/week2 to 7 –

Change in working hours for farrowingsupervision

Few More

Principal day of farrowing – –Weaning to avoid insemination

during weekend– –

Weaning to avoid farrowingduring weekend

– –

First day of oestrus detection – –Do oestrus detections stop in the week? – –Step 3a: Around farrowing “Frequency of interventio

Low HighPharmacological products Few IntensiveHelp to piglets – –Manual assistance – FrequentlyAnticipate farrowing difficulties No YesHyperprolificity management Within batch Several extra

possibilitiesNumber of piglets under gilts

compare to sowsLess –

Reduce litter weight variability Yes, for alllitters

Yes, 1 litterpiglets

Step 3b: Around inseminations “Adjustment of protocolsWOI Parity

Day of weaning Thu WedType of weaning – All togetherTechniques to help oestrus – –Boar number for detection One SeveralWeekly modifications to protocol Several NoneNumber of inseminations per sow per day 1 or 2 –Distinction between gilts and sows No YesBoars can mate – NoStep 3c : Global analysis “Adjustment of protocols

WOI Parity

The factors names (in bold) were given a posteriori. In step 1, the “X” sicorresponding extremities of the factor.

They are presented in Table 5. The first factor dif-ferentiated oestrus detection and the practical protocolsof inseminations and their adjustment indicators. Atone extremity, the analysis found a protocol that adjusts

factors identified at each analysis

Factor 2 Factor 3

“Sow fertility” “Labour productivity”Low High Low High

XX

XX

“Farrowing and weekend” “Insemination andweekend”

ork No matter Avoid No matter Avoid– – – –– – – –

– – – –

Thu Wed/Thu – –Yes No No Yes

No Yes – –

– Sun Fri Sun or Mon– – No Yes

ns” “Interest towards pigletsurvival”Little High– FewSome Much– –No Yes

batch Withinbatch

Between batches

– –

of small No Yes, 1 litter of smallpiglets

” “Focus”Oestrus InseminationThu WedSplit –Grouping HormonesSeveral One– –1 2– –Yes No

” “Interest towards pigletsurvival”

“Frequency ofinterventions”

Little High Low High

gn indicates that the higher values for the criteria are related to the

Fig. 2. Distribution of farmers (circles) on factors 1 and 2 of the CFA on preferences about work load distribution. The size of each circle point representshow well the farmer is described by these factors. The classes issued from AHC are the numbered areas. Class definitions are written in the areas.

102 G. Martel et al. / Livestock Science 116 (2008) 96–107

the procedures with respect to the weaning-to-oestrusinterval. It was defined by i) several modifications of theinsemination procedures (number of inseminations andnumber of hours between detection and inseminations)changing with the day, ii) moving sows and piglets ondifferent days at weaning and iii) no differentiation inthe management of gilts and sows. At the otherextremity we found a management based on animalparity that uses different insemination procedures forgilts and sows, but without any modification within theweek. This factor was not related to productivity orfertility of sows. The second factor opposed practicesfocused on oestrus detection and those focused oninseminations. Practices focused on oestrus detectionconcerned split weaning, constitution of groups of sowsto facilitate the occurrence of oestrus and the use ofmore than one boar to detect oestrus. Practicesoptimised for insemination facilitated oestrus bypharmaceutical products, detected oestrus of sowswith one boar and did two inseminations a day. Thisset of procedures was linked to higher values ofweaning-to-oestrus and weaning-to-conception intervals(R=0.38 and 0.42 respectively).

3.3.3. Subpart c: Synthetic analysis on reproductionmanagement

The diversity in reproduction management wasexplained by three main factors identified by theMFA, contributing for 12, 11 and 10% to explaining

the variability. They are presented in Table 5. The firstfactor mainly concerned groups of variables aboutweaning, oestrus detection and insemination and couldbe interpreted as the first factor of the “subpart b”analysis. It mainly discriminated the management basis,either according to the weaning-to-oestrus interval (“daybasis”) or to the parity of females. Work loaddistributions “trying to avoid inseminations on week-ends” and “follow the sows' biological rhythm” wereclose to the day-basis management whereas themanagement on the basis of female parity was linkedto the “priority to work” modality (Fig. 3). No relationwas found with productivity data.

The second factor could be identified mainly as thesecond factor resulting from the “subpart a” analysis andestablished a gradient of interest towards piglet survival.In addition to the high interest in piglet survival, farmersmore often used semen collected from their own boars toinseminate sows in oestrus twice per day. This set ofpractices was related to high labour and sow productivity(R=0.40 and 0.43 respectively), and high fertility criteria(R=0.34). High interest in piglet survival was also linkedto the “priority to work” modality and to the largefarrowing batches. On the other hand, practices provid-ing only limited help to new-born piglets were found.Semen bought from an AI centre was used more oftenwhile sows with delayed oestrus were mated by the boar.These practices tend to be associated with higher pigletmortality between birth and weaning (R=0.17).

Fig. 3. Distribution of reproduction management techniques on factors 1 and 2 of the MFA in step 3c. Shape size is related to its proximity to thisfactorial plan.

103G. Martel et al. / Livestock Science 116 (2008) 96–107

The third factor was related to the first factorresulting from the “subpart a” analysis and translatedthe frequency of interventions on animals. Practiceswith frequent interventions on animals were related tosplit weaning and several modifications to inseminationprotocol during the week. This set of practices ispositively correlated with sow productivity (R=0.40).Limited interventions on sows were correlated with abetter fertility rate (R=0.34). This factor had no relationto work parameters.

In conclusion, this analysis indicated:

(i) Few links between practices in relation withinsemination and practices in relation with farrow-ing. Farmers surveyed may achieve a day basismanagement for insemination protocol, “high inter-est” procedures for piglet survival and low frequencyof interventions for farrowing and cross fostering.

(ii) Referring to the preferences about work loaddistribution, the “priority to work” modality wasrelated to frequent inseminations and considerablehelp to piglets, whereas “follow the sow biologicalrhythm” and “try to avoid insemination duringweekend” modalities were only linked to the “day-basis management” of inseminations. The “lowdensity of tasks per day” modality is not linked to

these factors despite its position on Factors 1 and 2(Fig. 3).

(iii) The labour and the sow productivity as well as thelarge batch sizes were related to high level of helpto sows and piglets during farrowing and cross-fostering.

4. Discussion

This work refers to the research on livestock farmingsystems which study the relationships between humanobjectives and constraints, farmer practices and herdoperations and performances (Gibon et al., 1999). One ofthe most used methods is the survey of farmers. But thismethod needs to take into account the sample studied andthe quality of the data collected (Gibon, 1994). It is alsonecessary to comment the relation between practices andperformances before discussing the results obtained.

4.1. Sample

To study the relationships between labour productivity,farmer practices and sow productivity, there was a need toobtain a sample of farms with high variability in theseareas. Variability in the productivity per AWU isimportant and equivalent to that observed by Le Moan

104 G. Martel et al. / Livestock Science 116 (2008) 96–107

et al. (2005). The wide range of practices observed is alsosimilar to that described in previous studies (Holyoakeet al., 1995; Orgeur et al., 2000; Corrêa et al., 2002).Variability in performances is slightly lower and meanvalues are higher in our sample than in the geographicalarea of study probably due to the selection of farmerscarried out by the producer groups. Our study confirmsthat swine production, often perceived as uniform is infact diverse as regards to farmer practices, concrete workand performances, in agreement with the studies on“farming styles” (Commandeur, 2003, 2006).

4.2. Methodology and indicators

The validity of the AWU is always discussed becausethis unit underestimates the working time of farmers andnotably its variability when farmers declare themselvesas “full time” workers (Lacroix and Mollard, 1991). Thepresent study analyses the work load of farmers, not thelabour time, by self evaluation of weekly labour time.Although self evaluation of time repartition is notsuitable to study the annual labour time, it providespertinent information on weekly work distribution(Lacroix and Mollard, 1991).

The study stresses the work load distribution offarmers. Though preferences about work load distribu-tion are considered as a critical point to understand thechanges that operate in livestock management (Dedieuet al., 2006), there is no simple method to characterize it.Our approach gives importance to the number ofworking sessions, to the tasks carried out duringweekends and to the justifications given by the farmersto build, on the basis of the concrete practices data andthe factorial analysis, the classes of preferences aboutwork load distribution. This method is more approxi-mate than monitoring the working activities system(Madelrieux et al., 2006). It was performed on the sowreproduction unit only, excluding all the other parts ofthe farm and without consideration of the work loaddistribution between workers and its changes if one ofthem is missing (Hostiou et al., 2007). Moreover, onlythe farm manager was surveyed and he answered for allthe worker of the sow reproduction unit. So this is apartial view of a large field but it is the first attemptwhich allows us to stress the link between management,practices and work load distribution.

4.3. Practices and productivity

The analysis of practices also confirmed some of therelationships between practices and sow performancesfound in literature. Helping sows at farrowing (step 3a,

factor 1) or helping piglets to survive (step 3a, factor 2)reduces piglet mortality, whereas accurate practices ofoestrus detection improve the weaning-to-oestrus inter-val. In literature, Le Cozler et al. (2002) found a positiveeffect of farrowing supervision improving on thenumber of piglets born alive and Holyoake et al.(1995) and White et al. (1996) were able to decrease themortality of piglets before weaning by improving thenew-born management. The relationships between sowproductivity and “help to sows” and “help to piglets”modalities (step 3a) suggest a cumulative effect of thesepractices on piglet survival. Despite its effect on sowproductivity, not all farmers provide what was defined inthe result section as considerable help to sows andpiglets. This fact can be explained by diversity of theexpectations about the farm and by the capacity toprocure this help. Indeed, considerable help to pigletswas often procured in large farrowing batches becausefarrowing supervision is facilitated and it is easier toapply improved cross-fostering techniques.

The effects of oestrus detection procedures describedin step 3b— several boars for oestrus detection, groupingsows between weaning and oestrus detection and splitweaning — on weaning-to-oestrus interval have beenstudied by Signoret et al. (1975), Schmidt et al. (1985) andVesseur et al. (1997), respectively, and they found areduction of the weaning-to-oestrus interval when theseprocedures were used. However the effect of groupingsows is not found in all studies (Kemp et al., 2005).Moreover some farmers want to avoid early oestrus, forinstance those who do not detect oestrus on weekends.These reasons could explain why these practices are notapplied by all farmers.

4.4. First hypothesis: Is labour productivity related tospecific practices and herd productivity?

The first step of the analysis highlights the indepen-dence between sow productivity and labour productivityin the reproduction unit. This means that farms with ahigh level of labour productivity can also have a highlevel of sow performances. This result agrees with LeMoan et al. (2005). There is also independence betweensow productivity and sow fertility rate which means thatdespite a low fertility rate, some farmers obtain a goodproductivity of sows. This focuses on the diversity ofpossible managements to obtain the same resultscombining levels of fertility, oestrus-insemination inter-val, prolificacy and piglet mortality.

The third step of the analysis suggests a relationshipbetween labour productivity and, batch size and reproduc-tion practices, notably farrowing supervision and cross-

105G. Martel et al. / Livestock Science 116 (2008) 96–107

fostering, which are related to the herd productivity. Therelationships between these parameters can be partiallyexplained by an economy of scale. Indeed the surveyedfarmers frequently indicated that the farrowing supervisionof a few sows needed as much time as the supervision of amuch larger group. The possibilities of piglet cross-fostering are also more numerous in a large batch ofsows and thus makes it possible to set up practices forimproving the survival of piglets, resulting in an increase insow productivity.

4.5. Second hypothesis: The pig farmers have variouswork preferences, especially when considering the workin odd hours and in weekends

The first factor of diversity in sow herd managementstrategies and operations is the batch management(number and size of the farrowing batches) which isassociated with replacement management. This fact hasbeen verified in the study sample (analysis not shown).The choice of number of farrowing batches composingthe herd is argued by the distribution of the periodictasks between weeks and by the labour force manage-ment. Some batch managements (for example 4-weekbatch farrowing systems) have some weeks withoutperiodic tasks enabling holidays to be taken, whereas1-week batch farrowing systems have all the periodictasks each week allowing specialisation of workers(Caugant, 2002) (Fig. 1B). But our study indicates thatthe batch management is independent of the preferencesabout the work in odd hours and in weekends. At thislevel, our results make it possible to distinguish severalpreferences and are compliant with the tendencies ob-served in dairy farming (Cournut and Dedieu, 2005). Theways to adapt the work load distribution could include i)limiting the amount of periodic working sessions which isan indicator of daily intensity (the days these activitiesoccur), ii) avoiding activities during the weekend, eitherinsemination or farrowing supervision.

The data collected during the survey also allow theanalysis of the relationships between preferences aboutwork load distribution and labour force composition. Itsuggests that the preferences about spare time (by day orfor the weekend) are more frequently associated withfamily farms whereas farms with several employeesgive the “priority to work”. It can be assumed thatstockbreeders without employees having to work everyday try to avoid periodic tasks during the weekend, suchas farrowing supervision or insemination. The stock-breeders with employees can organise labour rotationsand are then able to keep protocols that are more stableover time.

4.6. Third hypothesis: Preferences about work loaddistribution are related to practices, labour productivityand herd productivity

The results suggest a strong link between thepreferences concerning spare time during the weekendand reproduction practices, especially at weaning,oestrus detection and insemination. The farmers thatfollow “the sow biological rhythm” manage theirinsemination protocol in agreement with scientificknowledge (the longer the weaning-to-oestrus interval,the shorter the interval between the onset of oestrus andovulation, so the interval between two inseminations hasto be reduced as weaning-to-oestrus interval increases—Kemp and Soede, 1996). This reproduction protocol isalso often used by farmers trying to avoid inseminationon weekends. When the first oestrus detection occurs onMonday morning the sows detected in oestrus may havebeen in oestrus for 0 to 48 h. So they use a specificprocedure of insemination for these sows that can beclose to their ovulation onset and some other proceduresfor sows detected later in the week. The protocol with theanimal parity criteria is coherent with the differentiationof oestrus profile between gilts and sows (Martinat-Botteet al., 1995; Steverink et al., 1999). This type ofmanagement is associated with large farrowing batchsize and with farmers with the “priority to work”modality. Our hypothesis is that the high number ofsows to inseminate involves a simple inseminationprotocol and distinction between gilts and sows is theeasiest way to manage it.

According to the relationship observed betweenlabour productivity and the practices around farrowing,these results show that classes of preferences about workload distribution and labour productivity do not dependon the same practices but each one is in relation withreproduction management.

Surprisingly, the number of interventions on sowsand piglets at farrowing and for cross-fostering (thirdfactor of the step 3c) does not depend on any modality ofpreferences about work load distribution, labour pro-ductivity or sow productivity. Two hypotheses can bemade. The first supposes that the level of intervention ispositively related to the occurrence of productivityproblems in the farm. The second hypothesis is that thisfactor returns to another dimension of work: themeaning given by the farmer to his craft and his relationwith the animals. Commandeur (2003) proposed aclassification of the stockbreeders farming styles. Shedescribed the “craftsmanship” practices characterized ona techno-economic dimension by frequent interventionson their animals in order to obtain the best productivity

106 G. Martel et al. / Livestock Science 116 (2008) 96–107

and the “stockmanship” socio-economic approach thatprefers to allow sows to follow nature's course even ifthis results in lower productivity.

5. Conclusion

The reorganisation of labour, the pressure oncompetitiveness, the new values and norms that emergein the farming communities are movements that not onlyaffect the socio-economical component of livestockfarming. They affect also the livestock management andnotably the reproduction management of sow herd units.Our study confirms the diversity in what is oftenpresented as a uniform world of production. Diversity ofthe reproduction management, diversity of workpreferences form the entrepreneurial point of view (theefficient use of manpower is important) to someconsiderations on the liveability of the farming activitieswhere weekend and preferences about daily work loadmust be taken into account in the management itself.

This study suggests that labour productivity waslinked to the size of the farrowing batches through ascale economy. But this paper describes more thediversity in the operational achievement of reproductionmanagement. It suggests that this management can be acritical point to consider when changes in preferences asto work load distribution occur and confirms therelationship with sow productivity. The preferencesabout work load distribution were related to specificinsemination protocols whereas sow productivity wasmainly related to farrowing and cross-fostering man-agement. These results have to be considered foroperational support provided by producer groups. Thework load distributions identified confirm that they arecomposed of different time scales (day and week) andthat farmers can look at their work load distributionwhereas others have work as their priority. Thesepreferences do not influence the capacity of farmers toobtain high sow productivity. The last is in relation withfarrowing management capacity that is partially linkedto the size of the farrowing batches. Studying the wayfarmers with small batch size can improve their capacityto provide help to sows and piglets during farrowingwith respect to the preferences about work loaddistribution could be a line for further investigations.

Acknowledgements

We wish to thank the farmers, pig producer groups(CECAB, Pigalys and LT) and their technicians (M.Herlédan, P. Rouxel, J. Lossouarn andB.Bernicot) for theirkind collaboration and M. Commandeur, F. Casabianca

(INRACorte) and S. Cournut (Reper—Metafort) for theiradvices.

References

Barthez, A., 1986. Du labeur paysan au métier d'agriculteur : l'élaborationstatistique en agriculture. Cah. Econ. Sociol. Rurales 3, 45–72.

Caugant, A., 2002. La conduite en bandes en élevage porcin. EDEBretagne & Chambres d'Agriculture de Bretagne, Rennes. 28 pp.

Commandeur, M., 2003. Styles of Pig Farming. PhD, WageningenUniv. 400 pp.

Commandeur, M., 2006. Diversity of pig farming styles: understandinghow it is structured. Neth. J. Agric. Sci. 54, 111–127.

Corrêa, M.N., Lucia Jr., T., Afonso, J.A.B., Deschamps, J.C., 2002.Reproductive performance of early-weaned female swine accordingto their estrus profile and frequency of artificial insemination.Theriogenology 58, 103–112.

Cournut, S., Dedieu, B., 2005. Simplification des conduites d'élevage enbovins laitiers. Cah. Etudes Rech. Francoph., Agric. 14, 541–547.

Dedieu, B., Servière, G., Madelrieux, S., Dobremez, L., Cournut, S.,2006. Comment appréhender conjointement les changementstechniques et les changements du travail en élevage? Cah. EtudesRech. Francoph., Agric. 15, 506–513.

Escofier, B., Pagès, J., 1998. Analyses factorielles simples et multiples,3rd edition. Editions Dunod, Paris. 284 pp.

Fiorelli, C., Dedieu, B., Pailleux, J.Y., 2007. Explaining diversity oflivestock farming management strategies of multiple-job-holders:importance of level of production objectives and role of farming inthe household. Animal. 1, 1209–1218.

Gibon, A., 1994. Dispositifs pour l'étude des systèmes d'élevage enferme. In: Gibon, A., Flamant, J.C. (Eds.), The Study of LivestockFarming Systems in A Research and Development Framework.EAAP Publication, vol. 63. Wageningen Pers, Wageningen,pp. 410–422.

Gibon, A., Sibbald, A.R., Flamant, J.C., Lhoste, P., Revilla, R.,Rubino, R., Sørensen, J.T., 1999. Livestock farming systemsresearch in Europe and its potential contribution for managingtowards sustainability in livestock farming. Livest. Prod. Sci. 61,121–137.

Girard, N., Bellon, S., Hubert, B., Lardon, S., Moulin, C.-H., Osty, P.,2001. Categorising combinations of farmers' land use practices: anapproach based on examples of sheep farms in the south of France.Agronomie 21, 435–459.

Guillaumin, A., Kling-Eveillard, F., Marty, M., 2004. Attentes deséleveurs laitiers d'Aquitaine en matière de qualité de vie et deconditions de travail. Rencontres Rech. Ruminants 11, 208.

Holyoake, P.K., Dial, G., Trigg, T., King, V.L., 1995. Reducing pigmortality through supervision during the perinatal period. J. Anim.Sci. 73, 3543–3551.

Hostiou, N., Dedieu, B., Pailleux, J.-Y., 2007. Le salariat en élevageporcin et les régulations du travail. Journ. Rech. Porc. Fr. 39,193–198.

Jean, N., Lacroix, A., Maamoun, M., Mollard, A., 1988. Durée etintensité du travail des agriculteurs dans la crise économique. Actescommun.— Inst. Natl Rech. Agron., Econ. Soc. Rural. 3, 45–82.

Kemp, B., Soede, N.M., 1996. Relationship of weaning-to-estrusinterval to timing of ovulation and fertilization in sows. J. Anim.Sci. 74, 944–949.

Kemp,B., Soede, N.M., Langendijk, P., 2005. Effects of boar contact andhousing conditions on estrus expression in sows. Theriogenology 63,643–656.

107G. Martel et al. / Livestock Science 116 (2008) 96–107

Lacroix, A., Mollard, A., 1991. Mesurer le travail agricole. Del'enregistrement à la reconstitution analytique. Cah. Econ. Sociol.Rurales 20, 27–46.

Le Borgne, M., Lebris, B., Quillien, J.-P., 1994. Temps de travaux:enquête dans 89 ateliers porcins de Bretagne. EDE Bretagne &Chambres d'Agriculture de Bretagne, Rennes. 30 pp.

Le Cozler, Y., Guyomarc'h, C., Pichodo, X., Quinio, P.-Y., Pellois, H.,2002. Factors associated with stillborn and mummified piglets inhigh-prolific sows. Anim. Res. 51, 261–268.

Le Moan, L., Quinio, P.-Y., Le Cozler, Y., Donet, P., Sallard, R.,Bartolomeu, D., Le Borgne, M., 2005. Know-how and workingorganization in pig farming. In: Bonneau,M., Bourgoin,M., Bonneau,A., Bitteur, A. (Eds.), International Workshop on Green PorkProduction. INRA and Agrocampus Rennes, Rennes, pp. 153–154.

Leplat, J., 1994. Collective activity in work: some ways of research.Trav. hum., 57, 209–226.

Madelrieux, S., Dedieu, B., submitted for publication. Qualificationand assessment of work organization in livestock farms.

Madelrieux, S., Dedieu, B., Dobremez, L., 2006. ATELAGE: unmodèle pour qualifier l'organisation du travail dans les exploita-tions d'élevage. Prod. Anim. 19, 47–58.

Martinat-Botte, F., Bariteau, F., Forgerit, Y., Macar, C., Poirier, P.,Terqui, M., 1995. Synchronization of oestrus in gilts withaltrenogest: effects on ovulation rate and foetal survival. Anim.Reprod. Sci. 39, 267–274.

Orgeur, P., Salaun, C., Roux, T., Venturi, E., Dividich, J., 2000.L'adoption et le sevrage ultra— précoce : une stratégie pour éleverles porcelets en surnombre. Journ. Rech. Porc. Fr. 32, 143–149.

Schmidt, W.E., Stevenson, J.S., Davis, D.L., 1985. Reproductive traitsof sows penned individually or in groups until 35 days afterbreeding. J. Anim. Sci. 60, 755–759.

Signoret, J., Baldwin, B.A., Fraser, D., Hafez, E.S.E., 1975. Thebehaviour of swine. In: Hafez, E.S.E. (Ed.), The behaviour ofdomestic animals. Baillière Tindall, London, pp. 295–329.

SPAD, 2006. SPAD: Software for Data Mining, Data Analysis andText Mining. 6.5 Release. SPAD, Paris. www.spad.eu.

Steverink, D.W.B., Soede, N.M., Groenland, G.J.R., van Schie, F.W.,Noordhuizen, J.P.T.M., Kemp, B., 1999. Duration of estrus inrelation to reproduction results in pings on commercial farms.J. Anim. Sci. 77, 801–809.

Tirel, J.C., 1989. Extensification? Intensifications? Quelle productivitépour l'agriculture? C. R. Acad. Agric. 75 (2), 19–36.

Vesseur, P.C., Kemp, V., den Hartog, L.A., Noordhuizen, J.P.T.M.,1997. Effect of split-weaning in first and second parity sows onsow and piglet performance. Livest. Prod. Sci. 49, 227–285.

White, K.R., Anderson, D.M., Bate, L.A., 1996. Increasing pigletsurvival through an improved farrowing management protocol. Can.J. Anim. Sci. 76, 491–495.