*2.3. Statistical Analysis*

The semi-structured interview provided information on 41 variables, from which were selected the most representative variables in the study. The data processing was done with IBM SPSS Statistics software version 20 (Armonk, NY, USA).

The variables were chosen according to the value of the coefficients of variation (CV) for each initial variable, selecting the variables whose CV was greater than or equal to 50% of variability in terms of the mean [21].

The multivariant method of multiple correspondence analysis (MCA) was applied to the selected variables. This method analyzes the relation between categories of quantitative variables so their dimension can be reduced. Finally, Ward's method of hierarchical cluster analysis was applied to define homogeneous groups of SFLs in the study area according to their similarities for the variables ultimately selected [22].

#### *2.4. Validation of Typologies and Work Lines*

This study was conducted within the framework of the project "Program for training and technological transfer for a better application of Law 20.283, aimed at small landowners in the VI Region." In this project, the problems, needs, and training of the local communities were discussed with local leaders, and the SFL typologies were obtained. However, forest activities were proposed to eachlandowneraccordingtothecharacteristicsofthepropertyandincompliancewithLaw20.283.

 In addition, each typology was validated and the landowners' problems and future lines of work were identified by consulting a panel of experts described in Table 1.


#### **3. Results and Discussion**

#### *3.1. Spatial Distribution of Small Forest Landowners in the Region*

The analysis of the maps, supplemented with information from territorial institutions, initially identified 420 small forest landowners in the region, which represented 14% of the total SFLs, according to the 2013 Agroforestry Census. The 420 small forest landowners were natural persons and not companies or associations. In the interview it was verified that the property met the requirements established by Law 20.283 in its definition of the characteristics of the small forest landowner. This verification eliminated any landowner who did not fulfill any of the following requirements: the property did not have native forest, the area was larger than 200 ha, the landowners did not have the deed to the property, or the information collected in the survey was incomplete. Finally, the number of SFLs was reduced to 211. Average area of native forest in this sample was 18.25 ha, and the sampling relative error was 16%.

A useful map was generated for the subsequent classification of the producers based on the number of properties that meet the initial criteria for selection and analysis for each commune.

Table 2 shows the community to which the SFLs belong, the associated area of native forest and the number of landowners. Figure 3 shows their location. As can be seen in Figure 3, no landowners can be seen in the eastern and western zone (Coastal Range and Los Andes Range) of the map as the forests in these zones are not sclerophyllous.


**Table 2.** Spatial distribution of small forest landowners and area of native forest at the commune level.

**Figure 3.** Spatial distribution of small forest landowners (SFL) in the Libertador General Bernardo O'Higgins Region.

#### *3.2. Identification of Landowner Groups*

### 3.2.1. Initial Variables

The choice of variables was made according to their discriminant power, based on their coe fficient of variation and prior studies. Of the total of 14 variables selected, four correspond to quantitative variables and ten to categorical variables. The selected quantitative variables were age of the head of household (AG), property area (PA), area of native forest (NFA), and per capita income (PCI). The selected qualitative variables were type of animal unit (AU), economic activity of the head of household (EA), forest status (FS), education of the head of household (ED), infrastructure (IN), extraction of forest products (FP), main problem (MP), main productive subsystem (MPS), training subject (TR), and source of water for agriculture and livestock (WS).

3.2.2. Results of the Multiple Correspondence Analysis

The multiple correspondence analysis separated three dimensions that explain 60% of the total variance, each with similar inertia values of between 18.4% and 21.3%.

Table 3 shows the discriminatory capacity of each variable in each dimension. As can be seen, Dimension 1 is mainly explained by productive variables (main productive subsystem, type of animal unit, and source of water for agriculture and livestock); Dimension 2 is explained by property variables (property area and area of native forest); and Dimension 3 is explained by social variables (age, main economic activity, and education of the head of household).


**Table 3.** Discriminant measures for each variable.

#### 3.2.3. Grouping of Landowners

The cluster analysis, through Ward's method, identified small landowners with similar characteristics in the three dimensions (eight variables) obtained from the MCA—property, productive, and social. Based on this, four homogeneous groups of producers were formed: elderly SFLs, retired SFLs, largest SFLs, and middle-aged SFLs, as shown in Table 4.

**Table 4.** Description of typologies based on variables obtained from the multiple correspondence analysis (MCA).


The reduction of the sample size from 420 SFL to 211 implied a smaller sample size for each typology and a greater sampling error. The highest standard deviation (SD) for the different variables

and typologies in Table 6 corresponded to largest SFLs, with 15%, while the maximum SD for the other typologies was 5% for the elderly SFLs, 8.5% for retired SFLs, and 9.2% for middle-aged SFLs. In further studies, it will be necessary to increase the sample size of the SFL with the largest properties.

The characterization of producers was supplemented with information on specific variables related to forest management, agricultural production, and characteristics of the family groups obtained in the survey in order to refine the profile of each type of producer and observe some of the typical problems in each group (see Tables 5 and 6).



**Table 6.** General observations for each typology.



**Table 6.** *Cont.*

By typifying the small forest landowners in the region, it was possible to recognize their characteristics and the problems that need to be tackled in a specific way for each situation. This diagnostic is essential for designing programs or lines of work. The detailed analysis of the variables revealed the similarities and di fferences between the four typologies of small forest landowners in the Libertador General Bernardo O'Higgins Region.

Similar behavior was observed in the characteristics of the head of household, namely advanced average age and low education level. Most landowners in the typologies elderly SFLs, retired SFLs, and largest SFLs are illiterate. Middle-aged SFLs, the younges<sup>t</sup> group (average 52 years), have completed at least the basic education level, and 63% of the university graduates interviewed belong to this group.

The main economic activity among elderly SFLs and largest SFLs is full-time work on the property, and the livestock subsystem accounts for the largest proportion of household income. This is followed by the agricultural subsystem, whose yield is used for the consumption of the family group. In the case of the retired SFLs, the producers' income comes mainly from their retirement, and although the main economic activity in the middle-aged SFLs typology is full-time work on the property (41%), 39% of the producers carry out activities outside the property on either a full-time or seasonal basis. They are also quite isolated. The closest paved road is at least 10 km away, which hinders their access to markets and services.

The size of the property is the main di fference between typologies from the economic point of view. SFLs in the largest SFLs typology have the largest properties (117 ha) on average, with an average of 93 ha of native forest. The largest SFLs typology has the lowest number of SFLs, with 15 out of a total of 221, indicating the dispersion of properties in the native forest, which is a disadvantage for forest management.

This is the only group that rents mechanical traction equipment; it also has the highest per capita income, at between \$238 and \$426, and the highest number of owners who extract forest products, including firewood, charcoal, and boldo leaves. These are the only producers who allocate part of the products extracted from the forest for commercialization, and 50% have received training in the field of forestry, although the total number of trained people is close to the other typologies. This typology has the highest number of family members involved in the managemen<sup>t</sup> of the property, which ensures their future engagemen<sup>t</sup> with forest managemen<sup>t</sup> and production. However, there are no data on the volume of the products extracted for any of the typologies. Firewood is used as a source of energy for the home and for cooking, which explains the interest among the SFLs in the restoration of native forests, as this would represent a source of savings [9,23,24].

Regarding the status and managemen<sup>t</sup> of the native forest, the landowners have very little training in forest management, with only 27 SFLs trained in the subject of forestry. The largest SFLs typology, with the highest average size, has the highest proportion and the best forest training. The only forest activity the landowners perform is clearing. According to the answers in the survey, they consider this activity to be the correct way to manage the forest. This lack of knowledge of forest management, coupled with the fact that none of the native forest properties had a forestry plan, leads to a substantial rate of intervention in the forest, which causes little regeneration and soil erosion problems. In addition, since property passes down through families, many new landowners are not entered in the property registry. This is one of the main problems when applying for subsidies and reduces the capacity for production, forest conservation, and water managemen<sup>t</sup> [8,9]. These problems also have a global impact, as these landowners play an important role in preventing the deforestation of the native forest, which has grea<sup>t</sup> potential for carbon mitigation [25].

In this context, there is a unanimous need for all the typologies to implement sustainable managemen<sup>t</sup> plans whose activities could be subsidized (Law 20.283 on the Recovery of the Native Forest and Forest Development), and for training in forest management, the role of the native forest, and the market for and commercialization of timber and non-timber forest products.

This would represent a first step toward the recovery of these forests, increase water availability, and encourage forest development among landowners [6]. The activities identified according to the landowners' typology and the use of the forest, validated by the panel of experts, are shown in Table 7.


**Table 7.** Eligible activities proposed by typology.

The SFLs' only type of organization is the neighborhood association to which they belong. There is no kind of SFL community forest managemen<sup>t</sup> in the region that promotes the conservation of forests, ensures the landowners' income from the use of the forest, and improves the governance of the managemen<sup>t</sup> [26,27].

There is therefore a lack of resources to transform these SFLs into the main agents of their development, intervening in the decision-making processes together with other stakeholders (technicians, administration, policy-makers), and avoiding the generation of policies at the territorial level that have a generalizing tendency and whose impact is unclear [23,28]. This would increase the e fficiency of their own potential and of the public and private initiatives that a ffect their development [9,28].

As a line of work for local management, we therefore propose the creation of social innovation and knowledge dissemination networks that make it possible to design, operate and assess strategies to stimulate innovation among the landowners. These networks are systems of informal interrelation that can be easily disassembled and recombined, encouraging non-hierarchical relations of trust between their members (SFLs, other landowners, technicians, Administration), and may endure over time [29]. In the case of the SFLs in the VI Region, after characterizing the landowners and identifying the forestry technology practices and innovations to be adopted, the key local actors can be identified due to their greater knowledge of the forestry activity or higher social prestige. These key local actors catalyze the processes of dissemination and adoption of innovations [30].
