Opportunity cost of natural forest management in the Arenal-Huetar Norte Conservation Area, Costa Rica

Background: Part of the success of forest conservation programs is due to the economic sustainability they can provide to owners of forest resources, and how these management mechanisms can be used within an increasingly aggressive productive landscape matrix. However, there are currently no precise or up-to-date data on the economic relationships between land uses and their respective productive activities. This study designed a model to evaluate the opportunity cost of natural forest management, taking as a reference the primary productive activities that take place within the Arenal-Huetar Norte Conservation Area, in Costa Rica. Methods: Protability data from 24 sites in natural forests with a forest management plan approved by the State Forest Administration was used, as well as geographic and productive information on alternative land uses. Results: Based on these data, an opportunity cost map was generated which shows a marked segregation of the forests into two main areas: a) a high-opportunity cost area, located south of the study area; and b) a medium-low opportunity cost area, to the center-north of the study area. Conclusions: It is concluded that ideal areas for timber harvesting are currently restricted to places far from the market, and with low opportunity costs (ranging between ≤ $0 ha -1 year -1 and $500 ha -1 year -1 ).


Background
Forest conservation as a development strategy should not be seen as a secondary option for land use, but as a growing need in contemporary societies. Despite improvements in forest ecosystems in terms of both area and management mechanisms (FAO 2016), deforestation and degradation, even today, continue to cause the loss of biodiversity and natural habitat. This dynamic has also increased the production of greenhouse gases and the climate crisis, resulting in a decrease in the quality of life of communities that depend directly or indirectly on products and resources of forests (Andrew and Bariweni 2018) From the point of view of land use competitiveness, tropical forest ecosystems have been vulnerable in recent decades, and have been characterized by low pro tability compared to other alternative land uses, which makes them less likely to remain standing. Discussions of biodiversity conservation must therefore begin with an understanding of the ecological, social and economic pillars upon which sustainable forest management (SFM) are based. To improve decision-making, it is necessary to have indicators that effectively measure these pillars, making it possible to locate forests with SFM within a productive matrix (Adamowicz 2003).
In this case, the most important information related to the economic pillar has to do with pro tability. This may be evaluated using the conceptual-economic framework of Opportunity Cost (OC), which provides tools to analyze patterns of behavior, allocation and effective use of scarce resources (Case and Fair 1997;González 2000;Sullivan et al. 2004). In simple terms, OC is what is lost by having chosen another option. In the speci c case of SFM, OC can be evaluated as the net income per hectare per year or the net present value that is no longer received for not harvesting the forest sustainably, but rather doing nothing with the forest or investing in other land uses such as agriculture or livestock (although Costa Rican environmental legislation prevents land use change, it continues to occur in practice) (Louman et al. 2001;Kniivilä and Saastamoinen 2002;Vega-Araya 2014;Andrew and Bariweni 2018).
The country has the potential for forest management in an area of approximately 867,590.4 ha, which is equivalent to 32.8% of the national forest cover (Camacho 2014). Despite this, the yield of wood from the management of mature forest has decreased, contributing only 6.1% of the total volume consumed locally (Barrantes and Ugalde 2019). Given this situation, some authors have acknowledged that there is little interest on the part of forest owners in adopting SFM, due to its low pro tability, problems in maintaining a cash ow with constant income, and better returns from other economic activities. These factors contribute to the displacement of forests by other alternative uses, through illegal mechanisms Meza 2008;Camacho 2015;de Camino et al. 2016;MINAE 2018).
The low pro tability of SFM compared to alternative land uses is an alarming sign. It is therefore vitally important to identify and compare the incomes generated by different uses of the land, and how differences in the income they produce affect producers' decisions to invest in SFM (or not). This economic and decision-making approach has not been frequently used in the region, and no data or body of literature has been produced in the last 10 years that allows the creation of economic clusters according to different land uses which compete with forests. This study is intended to ll this gap, and presents a model for calculating the opportunity cost of SFM in the Arenal-Huetar Norte Conservation Area of Costa Rica, based on a set of geographic and economic data of SFM and principal alternative activities. This empirical data is used to provide a graphic and geospatial perspective on OC trends, and shows that to ensure biodiversity conservation it is necessary to design and adjust schemes differentiated according tithe OCs for each region.

Study area
The research was carried out in the Arenal-Huetar Norte Conservation Area (AHNCA), located between the Las Haciendas River in Upala and the Sarapiquí River in La Virgen de Sarapiquí. It is bordered to the north by Nicaragua, to the west by the Cordillera de Guanacaste, to the east by the Sarapiquí and Toro Amarillo rivers, and to the south by the canton of Naranjo. It covers 13% of the national territory (6,724.67 km 2 ), and is the region with the highest levels of use of natural forest in the country (SINAC and SIREFOR 2011, 2012. The farms used for the analysis are privately owned and located in the cantons of San Carlos and Sarapiquí. The study was oriented towards the creation of an opportunity cost model for the districts of Florencia, Cutris, and Pital de San Carlos, as well as Cureña de Sarapiquí (Figure 1). The model focuses on SFM activities, and also evaluates other primary productive activities that take place in land adjacent to forests.

Base information sources
The opportunity cost model was based on four sources of base information: Data processing, cleaning, and management Spreadsheet applications and Geographic Information Systems (GIS) were used when working with the data, which followed the sequence below: Distribution of vegetation coverage within the study area In this step, polygons derived from the vector layers of the National Forest Inventory were used as the unit of geographic analysis (Ortíz-Malavassi et al. 2013). This data set contains attributes for different types of vegetation cover, of which the following were used in this analysis: mature forest, forest plantations, pastures and agricultural crops.
Determination of alternatives for primary agricultural production Using information generated by the 4th National Census of Agriculture 2014(INEC 2014, data were obtained on the area (ha) planted for annual, permanent, forest, ornamental and pasture crops; as well as the area dedicated to livestock activities (meat, milk and dual-purpose) within the zones of pasture.
Determination of pro tability of primary agricultural production activities Pro tability data for primary production activities was obtained from several secondary sources (ONF[1]; Calvo and Somarriba 1998; -Méndez 2018). In the case of timber harvesting, the nancial indicator called Forest Value (FV) (Zúñiga-Méndez 2018) was used for those forest units under forest management with an area of between 50 ha -100 ha. All data were indexed to 2014 and the microeconomic indicator called Equivalent Annual Value (EAV) was used to standardize them. It expresses the net pro t ($ha -1 year -1 ) that a producer would receive from dedicating his entire life to a given activity, assuming that the conditions for carrying it out it remain constant. EAV is expressed as follows: [Please see the supplementary les section to view the equations.] (1) Where: EAV = Equivalent Annual Value, in $ha -1 year -1 . NPV = Net Present Value, in $ha -1 LEV∞ = Land Expectation Value, in $ha -1 i = discount rate (for this procedure an interest rate of 5.2% is used for all productive activities) t = shift, production cycle Categorization of primary agricultural production activities according to vegetation coverage and weighted average pro tability for each class This procedure consisted of crossing information about vegetation cover categories with different primary production activities. Annual and permanent crops were classi ed as agricultural crops; forest crops were placed in the category of forest plantations; cattle ranching was placed in the category of pastures; and timber harvesting was placed in the category of mature forest vegetation cover. Once this procedure was de ned, the average pro tability of each of the vegetation cover categories was calculated, weighting the EAV of the productive activities by the geographical extent of the areas in which they took place.

Calculation of the adjustment factor for agricultural values
Values for this factor were obtained using data from the ONT's Agricultural Values platform, which is de ned in terms of agricultural areas. The purpose of this factor was to adjust pro tability values according to the speci c production conditions and characteristics of each geographic unit, including market access, urbanization, public services, infrastructure, land use, capacity for use, area, regularity, slope, access roads and hydrology. To determine this factor, equation (2) was used.
[Please see the supplementary les section to view the equations.]  2014). However, the components used in the original source were modi ed to obtain an estimate of pro tability closer to the socioeconomic conditions that exist at a given location, and then use it as the basis for calculating the OC of the geographical unit with respect to its neighbors.
Pro tability vc = weighted pro tability by type of vegetation cover, in $ha -1 year -1 . AF AV = adjustment factor for Agricultural Values.
Determination of the opportunity cost of timber harvesting in natural forest with respect to the alternative productive activities in the vicinity Opportunity cost was evaluated as the difference between the pro tability of the vegetation cover minus the pro tability of the timber harvest. The result should be interpreted as what forests owners forfeit if they choose to dedicate their land to forest management rather than to agriculture, livestock or forest plantations. This procedure was performed for all categories, as follows: [Please see the supplementary les section to view the equations.] Where: OC VC = opportunity cost by type of vegetation cover, in $ha -1 year -1 .
After calculating OC VC , a Neighborhood Analysis was carried out; this is a GIS process that permits the evaluation of the behavior of a variable (in this case opportunity cost) around a speci c location. To do so, a raster layer (cells of 270 x 270 m) with opportunity cost information for different types of vegetation cover was used. In addition, the Focused Statistics tool was used, which allows obtaining a raster data matrix in which each output cell value is a function of the values of the input cells that are around it. A neighborhood or donut shape was used as the neighborhood limit, whose area lies between two circles of different sizes, and which determines the nal value of each cell (Figure 2). The resulting value represents the average opportunity cost of the activities carried out in the vicinity of the unit being analyzed. [

Current area and distribution of vegetation cover by canton
The distribution of vegetation cover by district is not dominated by a speci c type of vegetation. Mature forest was the only type of coverage that dominated two of the four districts analyzed. Thus, mature forests cover more than 43% (36,636.41 ha) of the total area of Cutris, 39% (14,192.45 Fig. 3a). At the same time, an inverse relationship can be observed between the areas covered by pasture and by mature forests; the greater the extent of pastures, the smaller the area of mature forest, and vice versa. This condition is most clearly seen in Florencia, where the percentage of pastures is 43% (8,469.16 ha) and that of mature forest is 10% (1,922.55 ha). In the case of Agricultural Crops, the dynamics are different. In Cureña, only 2% (846.65 ha) of the district is dedicated to agriculture; in Cutris 10% (8 772.80 ha), in Pital 19% (7,211.15 ha), and in Florencia 28% (5, 556.90 ha) (Table 1, Fig. 3b).

Primary Agricultural Production Activities
The productive activities or crops that occur in the study area are mainly made up of agricultural crops, pastures, and forest plantations (Table 2). Cassava is the annual crop with the largest planted area, covering just over 2,000 ha (12% of agricultural crops), mainly in the district of Pital. Other annual crops, such as rice and corn, only cover 2% of the area studied. The permanent crop with the largest planted area is pineapple, with 9,673.36 ha, or 54% of the total area covered by permanent crops, which is planted primarily in Pital and Cutris. It is followed by sugar cane, with 4,063 ha (23% of the area covered by permanent crops), found mostly in Florencia and Cutris. Other products such as heart of palm, oil palm, banana and cocoa, together account for only 6% of the total area covered by permanent crops. The major forest plantation crops are Gmelina arborea and Tectona grandis, which covers 2,362 ha (39% of the area included in the forest plantation category), and 1,193 ha (20% of the category) respectively, as well as Vochysia guatemalensis, which covers 854 ha (14% of the category). year -1 to $651.6 ha -1 year -1 . The district with the highest income from forest plantations is Florencia, followed by Cutris (Table 3). In the case of pastures, pro tability ranges from $316.2 ha -1 year -1 to $630.2 ha -1 year -1 , which is lower than what is obtained from agricultural crops and forest plantations, but higher than what is obtained from mature forest ( Table 3). The average pro tability of forest management in mature forest was estimated using Forest Value (FV), which corresponds to a forest management unit with an area between 50-100 ha; this gure was $33.4 ha -1 year -1 [26], the lowest pro tability of any type of land use. 3.4. Opportunity cost of timber harvesting in natural forest with respect to alternative productive activities The opportunity costs of SFM are greatest when it is compared to agricultural crops, in which case OCs range from $1,474.0 ha -1 year -1 to $2,512.1 ha -1 year -1 (Table 4). Comparisons with forest plantations showed intermediate opportunity costs, with the economic pressure caused by timber harvesting ranging between $406.7 ha -1 year -1 and $618.2 ha -1 year -1 . Finally, comparisons with pastures produced OC estimates that range from $282.8 ha -1 year -1 to $596.8 ha -1 year -1 , much lower than those of agricultural crops. In practical terms, OCs of SFM are higher in places where productive activities are carried out close to markets, service providers and infrastructure; these factors favor the production and marketing of products, increasing the pro tability of the activities. This suggests that in those cases where conditions are not favorable for carrying out a productive activity, the OC of SFM is lower, and SFM is therefore a more competitive use of land.
It is necessary to clarify that only values of primary productive activities were considered when estimating OCs, leaving aside industrial activities, agro-industrial processing, trade and services, real estate developments, etc. Had these activities been taken into account, OC ranges would have been higher, mainly in those places where there is a signi cant presence of such activities, such as the district of Florencia. Additionally, an inverse relationship is observed between the area occupied by a given type of plant coverage and the opportunity cost of productive activities. This is clear in the case of pastures, which cover the largest total area, but at the same time exert the least economic pressure on the forest, since they have the lowest OC. The opposite happens in the case of agricultural crops that, despite having the smallest planted areas, involve products that generate high incomes. This implies a higher OC and consequently, a greater pressure on timber harvesting (which is made evident by the observed increase in area dedicated to pineapple in the region). Table 4 Opportunity cost ($ha − 1 year − 1 ) for each type of vegetation cover relative to timber harvesting in natural forest. AHNCA, Costa Rica.

Opportunity cost according to neighborhood analysis
Using cartographic units (graphic elements associated with a territorial unit existing in reality), a set of values was generated between ≤ $0 ha -1 year -1 and ≥ $4,000.0 ha -1 year -1 ; representing the OC of adjacent productive activities for a speci c cartographic unit (Fig. 4). These, in turn, were classi ed into three zones: high OC, medium OC and low OC.
Sites with high OCs (ranging from $1,000.0 ha -1 year -1 to ≥ $4,000.0 ha -1 year -1 ) are located in the southern parts of Cutris and Pital, and almost all of the district of Florencia. It is assumed that these values are in uenced by a high presence of agricultural crops and pastures, which have production chains that generate high returns, as is the case for products such as pineapple, cassava, plantain and sugar cane. Agriculture and livestock predominate in this region due to conditions of infrastructure, services, and access to markets that provide more productive scenarios for these activities. These areas are indicated by red colors in the Opportunity Cost Map (Fig. 4).
The medium OC sites (with values of $100.0 ha -1 year -1 to $1,000.0 ha -1 year -1 ) are colored light green, yellow, and light orange on the Opportunity Cost Map. Low OC sites (which range from ≤ $0 to $100.0 ha -1 year -1 ) are located in the central and northern part of the study area and are colored dark green in the Opportunity Cost Map (Fig. 4). The low OC activities that are carried out in these regions are de ned by three conditions: 1) they are crops with low pro tability; 2) they are crops that are established in places which are di cult to access, and are biophysically and geographically limited; 3) the land is suitable for forestry, which restricts the activities to be carried out within the forest to timber harvesting.
Using the estimates and cartographic mapping of OCs, it was determined that economic expectation is lower where large areas of forest are located. Seventy-three percent (73%) of mature forest is in the low OC zone, while 82% of the area dedicated to agricultural crops is located in the high OC zone.

Discussion
Timber harvesting takes place in two clearly de ned areas, in terms of opportunity cost: 1) highopportunity cost areas, and 2) low-opportunity cost areas. The values of OC vary according to the productive dynamics of the region, which are mainly related to the predominantly agricultural landscape (agriculture and livestock). In recent years, agricultural practices have optimized the production of some crops, with pineapple, sugar cane, and oil palm experiencing the greatest productive advances (OECD and FUNDEVI 2017), and many of these crops displacing traditional products and other land uses, such as mature forest (Morales and Rodríguez 2010; Sierra et al. 2016;OECD and FUNDEVI 2017).
In the Huetar Norte Region, net deforestation is decreasing, and it is possible that the region is beginning to enter a period of transition from exploiting forests towards a more highly dynamic use of the soil, even though more than 50% of net deforestation in Costa Rica took place between 1987 and 1997 (Sierra et al. 2016). This evidence re ects divergent behavior in land use: although mature forests cover the largest areas in the region, agricultural crops and pastures for livestock dominate the region in economic terms.
Meanwhile, the use and consumption of natural forest wood has uctuated, and has provided a very low percentage of the total of local wood consumed in the last ve years -only 6.1% of the volume used at the national level (Barrantes and Ugalde 2019).
Under current conditions, forest management would automatically have a high opportunity cost for producers in the southern regions of the study area, making it di cult to compete with other land uses with higher economic yields. SFM is possible in areas far from the market and where site conditions limit agricultural production. Our analysis shows that forests under active forest management are currently located in areas of low opportunity cost. Priority should be placed on focusing efforts in these areas, where there are greater opportunities for adoption of sustainable forest management. Not only are they located in the regions with the highest concentration of mature forest, but this type of activity could be more competitive with respect to alternative uses. In other words, with the introduction of forest management in marginal zones, forest owners would be receiving economic bene ts which are more competitive with those of other land uses .

Conclusions
From an investor's point of view, the approach used in this study makes it possible to show graphically where sustainable forest management is feasible and where it is not. The OC model shows the losses that producers face when they decide to invest in SFM instead of other types of land use such as cattle, agriculture or forest plantations. These OCs could be used as a reference to calculate compensation for owners who practice SFM in their forests, or who receive Payment for Environmental Services, with opportunity cost payments differentiated according to region. The methodology used here assists in determining the economic performance of land use (the economic and environmental components of SFM) (Naidoo and Adamowicz 2006;Adams et al. 2010;Bryan et al. 2011;Nakajima et al. 2017). However, the adjustment required for implementing a differentiated compensation associated with OC would force a paradigm shift in the current institutional framework in Costa Rica -changes for which the country may not be ready for or interested in carrying out.   AHNCA, Costa Rica.