Study area, data collection and identification of variables for analysis

The study was carried out in the Rio Ibáñez Commune located within the Region Aysén del General Carlos Ibáñez del Campo of Chile, 48°16′00″ S, 71°56′00″ E (Figure 1). It is one of the biggest regions in Chile (14 % of the country), with almost one million ha devoted to farming, especially cattle and sheep-farming for meat and wool¹³. The study region is cold steppe with average temperature 6.4 ºC and precipitation of 612 mm. The minimum temperature can fall to -37 ºC.

Figura 1: Location of the study area within the Rio Ibañez Commune (Region Aisén del General Carlos Ibáñez del Campo of Chile) 

In 2012, 28 producers with bovine presence were interviewed for the “Nodo Río Ibañez” programme, the object of which was to transfer technological knowledge to sheep and cattle producers. They were chosen at random from a total of 230 of the study area (SA), ie 12 %¹⁴. The questionnaire used included the following aspects of the livestock production unit (LPU) survey: (i) Sociological; (ii) Infrastructures, facilities and farm size; (iii) Management variables (feeding, reproduction and health); and (iv) Commercialization and economic variables. From data obtained of this questionnaire, a database with a total of 55 variables was drawn up, being 30 quantitative and 25 qualitative. Of the qualitative variables, 22 are binary (yes= 1 or no= 0 answers), two have three options and one has four options. Each option of these last variables has been converted into binary. All variables are showed in Tables 1 to 4.

Statistical analysis

The data on the variables were subjected to multivariate analysis in two stages: principal component analysis (PCA) and cluster analysis (CA)¹⁵. PCA transforms the data on the diagnostic variables studied into a small set of new synthetic variables - principal components (PC) - with little loss of information. The purpose of PCA is therefore to reduce the number of variables and thus the dimensions of the problem^16,17. The method used for PCA was the optimal scaling analysis which is used when the variables analyzed are both qualitative and quantitative variables¹⁵.

Before the multivariate analysis the number of variables was reduced excluding those with a low variability coefficient (< 50 %) and therefore little discriminatory capacity¹⁸, as were those which correlated with others considered by the authors to be more important for defining the production system¹⁷. During the process of reducing variables, 31 of the 44 variables used in the study were discarded. Therefore, 13 variables were used to perform the discriminant analysis which was started checking the appropriate number of PCs, as well as the variables attached to these PCs, in order to obtain the best and most synthetic explanation of the existing variability. In order the PCs to be sufficiently representative of the set of variables, the eigenvalues were required to be greater than 1¹⁶.

After PCA, the farms were classified by K-means type CA. In this case, this type of CA is better than the hierarchical, because authors know what the number of cluster do not can be high. The PCs obtained in the first part of the multivariate analysis were used for the CA instead the conjoint of variables¹⁷. Finally, once the different clusters are obtained, they can be described and afterward compared using one-way ANOVA for each of the original variables. With this analysis the multivariate analysis is confirmed¹⁵. For all variables were calculated the mean and standard error. In the case of qualitative variables the mean coincides with the relative frequency of value 1, ranging the presented results from 0 to100 %. For variables with significant differences, the multiple comparison post hoc test was applied (Duncan). All the statistical analyses were carried out with the IBM SPSS Statistics 20 statistical package¹⁹.

Principal component analysis (PCA) and cluster analysis (CA)

From the optimal scaling analysis (PCA), three principal components were obtained, being the eigenvalues 3.125 for the first PC, 2.158 for the second and 1.210 for the third. The relative variances were 34.7, 24.0 and 13.4 respectively, explaining 72.1 % of the total variance. The first PC named ‘Farm size’ includes the variables total area, number of cows, total costs and total sales. The second PC named ‘Farm management’ includes the variables person in charge of the farm, farmer makes hay and farmer fertilises meadows. The third PC named ‘Producer´s age and Number of ewes’ includes these two variables. The eigenvectors (weights) for each of the nine variables according to the three PCs are shown in Table 5.

As a result of the subsequent cluster analysis four clusters were found: C1, C2, C3 and C4, containing 5, 11, 5 and 7 LPU respectively. Figure 2 shows the distribution of the LPU among the three principal components (PC).

Figure 2: Clustered farms plotted according to the principal components 1-2 and 1-3 

Finally it was applied ANOVA to differentiate the effects of the groups obtained in each of the original variables, thus confirming the multivariate analysis. Results for each variable are showed on Tables 2 to 5.

Characteristics of the LPU clusters

Below are described the principal characteristics of the four clusters obtained:

C1 (5 farms): Middle aged producers with high proportion of women. Moderately low LPU size and low economic results. Deficient farm management. Producers are women living alone in 80 % of cases. Herds have only 17 cows but 84 ewes (mean of farms having sheep). Farms have 123 ha of grazing, having the higher stoking rate (1.2 livestock units, LU/ha). The mean of the farm profit is 4,043 € per year. The farm management is deficient in relation to animal feeding (farmer do not makes hay, do not fertilises meadows) and to the reproduction.

C2 (11 farms): Young producers with medium proportion of women. Medium LPU size and economic results. Medium farm management. Producers are women living alone in 45 % of cases or husband runs the farm in 46 % of cases. Herds have 33 cows and only 16 ewes (mean of farms having sheep). Farms have 154 ha of grazing area. The mean of the farm profit is 9,089 € per year, having relatively low costs. The farm management is moderately good in relation to animal feeding and deficient to the reproduction.

C3: Elderly producers with medium proportion of women. Relatively high LPU size and good economic results. Moderately good farm management. Producers are women living alone (20 %) or running the farm because the husband works in mining (20 %), in the rest of cases (60 %) the husband runs the farm. Herds have 52 cows and 49 ewes (mean of farms having sheep), the higher quantity of ewes together the cluster 1. Farms have 593 ha of grazing area, the highest of all clusters, being the stocking rate the lower (0.1 LU/ha). The mean of the farm profit is 13,044 € per year, having the highest sales (17,333 € per year) but also the highest cost by consequence of the hired labour costs (2,400 € per year). The farm management is good in relation to animal feeding and is medium in relation to the reproduction.

C4: Elderly producers with medium proportion of women. Low LPU size and very poor economic results. Moderately deficient farm management. Producers are women living alone (14 %) or running the farm because the husband works in mining (29 %), in the rest of cases (57 %) the husband runs the farm. Herds have only 10 cows and 20 ewes (mean of farms having sheep), the lower quantity of ewes but they are present in 86 % of farms. Farms have 81 ha of grazing area, the lowest of all clusters. The mean of the total costs, the total sales and the farm profit are the lowest of the all clusters (523, 3,264 and 2,740 € per year respectively). The farm management is moderately good in relation to animal feeding and deficient in relation to the reproduction.

Variables linked to sociological aspects

Clusters C3 and C4, which include the largest and smallest farms respectively, contain the oldest farmers (around 60 yr), who also have the lowest level of education (60 and 86 % respectively only attended primary school). That coincides with the results of a study done in the Center of Chile³. In this sense, one cause of the migration of the young population in Chile is the need for education²⁰. Nevertheless it is encouraging to see that in cluster C2 which includes the greatest number of farms (11) there are younger farmers (39 yr old average) with a high level of education (82 %) compared to other countries. In many areas of Latin America (LA) as in Ecuador with goat production²¹ or in Mexico (Veracruz and Morelos) with cattle^1,22, the education of farmers is low, around 80 % have only primary education. Concerning the training courses done by farmers, the proportion in C3 (100 %) and C4 (29 %) are the higher and the lower respectively. In general, specialized training is low in the agricultural and rural areas of LA^2,23.

As for the territorial base, 64 % are owned by the farmer, but there is only one farm in C2 that had leased pastures. It should be noted that, except in C3, there is a proportion of farms, around 15 %, that are in the process of inheritance and that in C3 40 % of farms belong to the family group. This last has a positively influence to avoid the division of the farm when an inheritance process is come. In the central area of Chile³ and in Colombia²⁴, ownership of land by farmer is the most common option. In other countries as Ecuador with goat production systems, the proportion of owner is low (22 %)²¹.

The farm work is generally done by the family members, except in C3, group of large farms, where 60 % have hired hand. This situation is similar in small or medium-sized farms of less favoured regions^2,6.

Fifty three (53) % of the farmers running the farms are women (80 % in C1). This proportion is higher than the Chilean average (24 %)²⁵; although in Chile there is a large increase in women engaged in agriculture²³. There is also a high proportion of women farmers in milk goat production on the Peruvian coast², but that is in general infrequent in other parts of the world as in the state of Veracruz (Mexico)¹ where 94 % are men.

The number of family members is higher than 3.4 (mean 4.2). That is similar to other areas of meat ruminant production^21,22. In the SA, 86 % of the farmers believe that their farm will continue in the hands of future generations. This value indicates that in the SA the problem of depopulation is not as serious as in rural areas in Chile in general²³. However, in C1 only 40 % of farmers believe that their farms will continue. In this cluster 80 % of farms are running by an alone women with no other income in the family unit than the livestock activity.

Variables regarding to infrastructure, facilities and farm size

Almost all the farms in the study have electricity (93 %); this value is similar for meat goat production systems of Ecuador²¹. The drinking water supply is very deficient, only 7 % of farms; in other studies done in LA, as in the meat goat production systems of Ecuador are found better results²¹. The level of facilities is medium (2.8 out of 4); it is higher in the cluster C3 (3.4) where farms have higher size farms.

Sheep are reared in 16 of the 28 LPU studied, while all have cattle. The number of ewes per farm is low (mean 18) ranging in farms having sheep from 16 (C2) to 84 (C1). The number of cows per farm is 28 ranging from 10 (C4) to 52 (C3). Consequently, the cattle production is much more important than the sheep production. The same is true for most of LA, as in the states of Veracruz and Morelos (Mexico)^1,22; however in the central area of Chile the ovine production predominates with respect to the bovine³. In the present study, the grazing area is quite high (the stocking rate is only 0.5 LU/ha). In C3 there is the higher grazing area (593 ha); in this group there is also the lowest stocking rate (0.1 LU/ha) and the highest surface of scrubs and bushes (110 ha). Farms of the ovine and bovine systems of central Chile have a grazing area slightly lower than in the SA (160 ha) and a similar LU (34)³, so the stocking rate is slightly high. The stocking rate in the SA is also higher than in other zones of LA, as in the state of Veracruz (Mexico)²² (around 1 LU/ha)¹, in the state of Morelos (Mexico) (8 LU/ha with cattle partially in cowshed) and in la Pampa, Argentina, with extensive cattle systems for fattening (1.8 LU/ha)²⁶.

Variables regarding to farm management

A high proportion of farmers in SA produce hay in the farm (75 %), except in C1 where no farmer does it, although the stocking rate in this group is the higher (1.2). However they purchased eight bales of hay per LU and year in average (25 kg each, therefore 200 kg). The amount of concentrate purchased per LU is considerably lower than that of hay (26 kg per LU and year in average). Considering the whole of the purchase of concentrate and hay, in C1 farmers purchase the lower quantity and in C4 the higher (19 kg and 6 bales and 31 kg and 9 bales respectively). In C3 the whole of the purchase is similar to in C2 (medium). However, the amount purchased in C3 might have been lower than in C2 if grazing management would have been better for two raisons: in C3, 60 % of farmers grow oats for grain while this percentage is very low for the other groups and in this group, as said, there is the lowest stocking area (0.1 vs 0.5 of study mean). Anyway in general in the SA, animal feeding is based on extensive grazing; as in the whole of central and southern Chile, in the critical periods of winter and summer the pastures do not provide sufficient nutrients²⁷. When pasture is scarce (in summer), herds are moved up to land at higher altitude, involving a short migration of maximum 10 km (Table 4). This activity is practiced in other parts of the world with extensive livestock as Mexico²⁸, Peru² or Spain⁷. The pasture shortage at some times of the year occurs in many marginal areas¹⁶. Instead, in the state of Veracruz (Mexico) in dual-purpose cattle farms, there is a group of business or specialist farms which produce grass and forage from improved meadows¹. The climate in this zone enables farmers to obtain grass and forage over a longer period of the year than in the SA.

In terms of fertilisation of meadows as well as soil analyzes, farmers of C1 do this activities in lower proportion that in the rest of groups. As mentioned before, in C1 the proportion is also the lowest for haymaking, so in this group the farmers used the most deficient technology. In the whole of the study only 32 % use leguminous plants. This is in agreement with the fact that pasture is an activity that is little carried out in Chile's pastoral livestock systems²⁹.

There is little control over reproductive management; the males are separated from the females outside the breeding season in only 14 % of cases for sheep and 21 % for cattle. However, in C4 they are separated in 40 % of cases (although without significant differences with other groups). In sheep meat systems in Colombia, the period of stay of males with females and mating protocols depend on the availability of facilities and the number of males per herd²⁴. In the cattle production of Ecuador, farmers do not plan the mating process⁴.

As aid in results of study, 93 % of the farmers desparasite their animals, but only 29 % vaccinate. In the SA these proportions are 76 % and 34 % respectively in 2011³⁰. In other parts of LA most farmers desparasite animals, for example in states of Morelos and Veracruz (Mexico)^22,28 or in sheep systems of Colombia²⁴. To the contrary, vaccinations are less frequent and depend heavily on the technification of farms²⁸. Concerning the collection of farm data there is little difference between groups, in average 36 % of farmers kept production records and 14 % economic records. Technical and economic management is also poor in other countries of LA such as Mexico²⁸ and in Spanish goat pastoral systems¹⁶. Farmers received some advice from the State, but only 21 % also receive private advice.

Commercialisation and economic variables

In studied farms the animals are always sold after weaning. Calves are sold to fatteners or in other areas, which results in a loss of potential income. The animal sales as well as the feed purchase are made by farmers through a unique association. Selling weaned animals is common practice in Chile³¹; as well as other countries - for example both calves and lambs in Spain⁷, lambs in Veracruz, Mexico²⁸; and calves in Chiapas, Mexico^32,33. Cattle are almost all sold, although some are kept for self-provisioning on the farm; in contrast an important part of the produced lambs are kept for family consumption. This type of self-provisioning is common in Chile³¹, just as in the state of Veracruz (Mexico)²⁸. Farmers sold a mean of 0.76 calves per cow and year (the lowest value is 0.64 in C1) and only 0.35 lambs per ewe and year (the highest is 0.52 in C1), there being no significant differences. In Chile's more intensified systems, lamb production can reach 1 lamb / ewe and year³.

Finally economic variables are discussed. It is interesting to compare the values of costs, sales and profit per LU in order to analyze the economic efficiency, and also the farm profit (i.e. the family incomes). The mean of total feed cost per LU is 40 €/yr, ranging from 31 and 48 €/yr in C1 and C4 respectively, without significant differences between groups, being medium values in C2 and C3 (40 and 39 €/yr respectively). This agrees with the discussion about the management variables. With respect to the hired labour cost is very high in C1 and C3 (900 and 2,400 €/yr respectively). These high values are due to a different raison for each two groups: in C3 farms had a much larger size than in the others groups and in 80 % of cases of C1 an alone woman runs the farm. By consequence, in C1 and C3 there are the highest total cost per LU (91 and 88 respectively being 70 €/yr the mean of study). In Colombian sheep production²⁴, the cost of labor is important with respect to the cost of feeding on farms that have fewer animals and the opposite occurs in those with more animals; but in general in the systems of production of ruminants the feeding is the most important cost²⁶.

The total sales per LU ranged from 248 and 316 €/yr in C1 and C2 respectively; in C3 and C4 the value are similar that in C2 (307 and 290 €/yr respectively). These in general low values could be due, among other factors, to the familiar consumption of a part of produced animals, especially of lambs, or to the deficient farm management (reproduction or feeding). This last aspect is particularly important in C1 where as mentioned there is the most deficient feeding management of study (at the same time there are a high stocking rate, no production of hay and little feed purchase). In the Mexican state of Morelos²² with a similar number of dual-purpose cows per farm (45), the total income is similar to that found in the present study; however as the stocking rate area of pasture in the Morelos study is very small (only 9 ha/farm), the production costs are much higher.

The profit per LU ranges from 156 to 254 €/yr in C1 and C2 respectively. The farm profit of the study is not very high, ranging from 2,740 to 13,044 €/yr in C4 and C3 respectively; however it is much higher than obtained in the cattle production of Chiapas (Mexico) (923 €/yr); having a mean of seven cows per farm and a profit per LU of 130 €/yr³³. In C4 the profit per LU is almost so high than in C2 (234 €/yr) but the very low quantity of LU (13 in average) lead to a very low farm profit (family income). This very low income obtained in C4 is compensated in 29 % of the cases with the income that the husband of the woman farmer obtains working in the mine. In C1, the farm profit is also quite low (4,043) by consequence of the low profit per LU and the deficient farm management which lead, as said, to a low proportion of farmers believing that the farm will continue for future generations (only 40 %). In C2 the farm profit is medium (9,089 €/yr) although the profit per LU is the highest because the LU value is also medium (34). Instead, in C3, the farm profit is the highest although the profit per LU is medium (220 €/yr) because the LU value is also the highest (56). Consequently in C2 and C3 there is a moderately adjusted economic activity, being more adjusted in C2 due mainly to in C3 the hired labour cost is very high. In 20 % of cases of this last group, this high hired labour cost could be explained by the fact that the husband works in the mine (thus complementing family income).

The majority of farmers (89 %) only had public loans from the Intituto de Desarrollo Agropecuario (INDAP) to pay for investments³⁴. Only in the larger farms (C3) which have better facilities some farmers also hold loans from private entities (40 %).

Strategies for improving production systems

The stoking rate could increase in C3, even goats could be introduced into the farms of this group to take advantage of scrubs and bushes (110 ha)¹⁶. In the other groups it is not clear whether it is desirable to increase the stocking rate, especially in C1 (where the current value is 1.2 LU/ha). In order not to diminish the size of the farms, the farmers should try to accelerate the inheritance process and try to increase the farms belonging to the family group (as in C3). It must be studied the convenience in some farms (those with better pastures) the production of milk destined to produce cheese for the market.

Farmers in all the clusters need to improve their water infrastructure, since in 68 % of the cases they had only water of spring, river or lake. In all clusters, although less in C3, facilities should be improved; for instance better stables could favorise the giving births.

In all clusters it would be necessary to plan the mating periods, so that when the majority of mothers have the highest nutritional needs (during lactation), there is a greater availability of natural pasture¹⁶. In general, efforts should be made to increase productivity and therefore the number of animals sold, especially in C1⁸. Also, farmers should increase the cultivated area destined to obtain hay or grain, especially in C1 where no farmer has these crops¹⁶. The quality of pasture should also be improved (using more leguminous plants) in C2 and C4 and especially in C1; also in this last group would fertilize the pastures. In terms of sanitary management of livestock, it should be improved, especially by carrying out more vaccinations; the Servicio Agrícola y Ganadero of Chile (SAG) increased from 2013 the aid to apply vaccines and deworming to cattle in the Commune of Rio Ibáñez (SA)³⁵.

In the present study, the INDAP (Instituto de Desarrollo Agropecuario) offered the majority of advice and training services and likewise the loans (Tables 4 and 5). As all marginal areas, these services (whether public or private) are essential¹⁶. In addition, among the objectives of INDAP there are other aspects that are of interest for the improvement of the farms of this study³⁴: installation of young and women as farmers, marketing (including the added value of productions), off-farm activities (rural tourism and handicrafts)¹ and promotion of associationism in different aspects. The work of farmers in other economic activities could improve the family incomes; but, it is not always clear whether this employment is complementary and compatible with their own farm work or could suppose a threat and a loss of commitment, which could eventually lead them to abandon the farm⁷.

To improve commercialisation the only existing association needs to become more active, or other associations need to be created^8,33. It could be studied whether breeders could participate in some way in fattening weaned animals^7,32 or at least could have an animal typing centre that would allow them to obtain better sales prices. Sending the animals away for fattening is also an obstacle to obtain differentiated quality products sold if possible in short commercial circuits^12,32,34. In this sense, studies about the meat quality according the animal feeding, have been done in Chile¹⁰. Policies and institutional arrangements should be strengthened to favorise the farm production commercialization⁸; the creation of an Protected Area: public-private inter-institutional networks capable of promoting rural development processes⁶⁾ could be interesting to solve this commercialisation problem.