This study was conducted in the province of Misiones (29,801 km²) in northeast Argentina (Figure 1). The climate in Misiones is subtropical humid without a marked dry season. Precipitation occurs year round, reaching an annual average of 2000 mm. The average annual temperature is 21 °C. The provincial territory contains 800 tributary streams of the Paraná and Uruguay rivers, which are the main rivers of the Del Plata watershed, and the northern boundary follows the Iguazú River which includes the famous Iguazú waterfalls.

Misiones includes the largest remaining continuous area of Atlantic Forests (~10,000 km²) within the Upper Paraná Atlantic Forest ecoregion [29]. This is an ecoregion of global importance for conservation due to its high level of biodiversity and endemism. The ecoregion as a whole is classified as “endangered” due to relatively high levels of habitat loss and lack of adequate protection [30], placing it on Conservation International’s Hotspots list [31,32] and World Wildlife Fund (WWF) “Global 200” list of ecoregions for conservation priority [33]. Although socioeconomic factors have driven increased deforestation in Misiones in the last decades [34,35], economic development in the province has lagged behind most of the other Argentinean provinces [36], leaving a relatively large portion of its area still covered by forests (~49%). Misiones has acquired a “green” profile as it is the Argentine province with the largest percent of protected territory (~17%), although these protected areas have different levels of effective management. Since the creation of the Iguazú National Park as the first protected area in Misiones in 1934, 67 new protected areas have been created in the province under different jurisdictions. In addition, other initiatives and legal tools have been developed in the province to support conservation and sustainable development. For example, in 1985 the Ministry of Ecology and Renewable Natural Resources of Misiones (MERNR) was created and oversaw the creation of the Urugua-I Provincial Park in 1988, the largest protected area in Misiones (840 km²) and established as “environmental compensation” for the reservoir created by the construction of the Urugua-I dam . The area under protection was also strongly expanded by the creation of the Yaboti Biosphere Reserve in 1993, which includes a provincial park as a core protected area with no development and various private properties as buffer areas, with a total area of 268 km²; this area alone represents 52% of the current protected area in the province. However, the capacity of law enforcement varies among the administrations within the province and, as a consequence, among protected areas [37]. In addition, other initiatives and legal tools are being used in the province to support conservation and sustainable development, such as Green Corridor Provincial Law 3631 and the Landscape Planning Provincial Law, known as the “Ordenamiento Territorial” for Misiones. The Green Corridor Law promotes sustainable development in the area between the National Park Iguazú and Cuña-Pirú Valley to prevent the isolation of protected areas. The Landscape Planning Law is a provincial ratification to the National Law 23,331 (known as the “Forest Law”) which intended to regulate the protection, enrichment, restoration, utilization, and management of native forests and the environmental services they produce. The Landscape Planning Law categorizes remnant forest in three classes: category I or “red”—areas where no productive activities are permitted (224 km²); category II or “yellow”—areas where sustainable activities are allowed if no forest cover is lost (967 km²); and, category III or “green”—areas of low conservation value, which can be converted to other uses (448 km²).

In addition to native forests, Misiones includes four major land-use classes: modern agriculture, forest plantations, subsistence agriculture, and pastures [34]. Modern agriculture is mainly dominated by perennial crops, particularly yerba mate and tea. Forest plantations are mainly Pinus and Eucalyptus. Between 1973 and 2006, the area in forest plantations increased from 1 to 11% of the province, replacing native forests, agriculture and pastures [34]. Subsistence agriculture (i.e., mixed use) is carried out by small producers mainly dominated by tobacco and annual crops. Cattle grazing occurs in native grasslands in the southwestern departments and in pastures created in previously forested areas in the central and northern departments [34]. In addition, cattle grazing occurs within forests, which involves the clearing of the understory to enhance grass productivity. Another important human activity impacting natural ecosystems is infrastructure development. Although important negative impacts associated with dam construction has been shown [38,39], the existence of flowing rivers in the Atlantic Forests region has made it attractive for dam development, with hundreds of dams constructed and proposed over the past 50 years [38]. Dams are built principally for hydropower, water storage, flood control or recreation. There are currently two dams in Misiones’ rivers downstream, Urugua-I and Yacyretá, and two others have been proposed, Corpus Christi and Roncador/Parambi. Other important dams exist upstream on the Paraná (e.g., Itaipú) and Iguazú Rivers, and the Garabi dam is proposed downstream of the Uruguay River.

We used the Marxan 2.43 conservation decision support software to select a network of priority conservation areas [40]. The Marxan analysis aims to achieve representation of conservation targets (i.e., species occurrences, ecological systems, and physical features) while also factoring in human impacts to the landscape and existing land-use scenarios. Marxan uses a heuristic algorithm to select portfolios of planning units according to conservation targets, goals, and other various software settings. This analysis involves the definition of: (1) conservation targets, in our case habitat and ecosystem services (i.e., carbon storage, soil retention, and water yield); (2) desired conservation goals for each ecosystem service; (3) spatial units of analysis, or “planning units”; (4) a cost surface, which in our case represents accumulated threats to ecosystem services.

The Marxan tool seeks to find the optimal network of PU’s that reach the specified conservation goals for targets in the study area while minimizing cost, as tallied from the selected PUs. In our study, habitat and ecosystem services were conservation targets, conservation goals were 50% of each target, and cost was the PU’s summed threat from the threat surface. We ran Marxan with the following parameters: 100 runs; boundary length modifier (BLM) of 100; 1,000,000 iterations with 10,000 temperature thresholds. The BLM is a parameter that controls the level of planning unit aggregation in the selected network by weighting the boundary length in the total solution cost [40]. A larger BLM causes Marxan to minimize boundary cost by finding compact, aggregated groups of planning units in the solution. We determined the BLM using a method described in Game and Grantham, 2008 [40]. The Penalty Factor (PF) is the importance a user puts on each target meeting its goal and it also influences the selection of planning units by ensuring that goals are met .As such, the PF can be thought of as way of distinguishing the relative worth of different conservation targets and how important it is to have them fully represented [40]. We defined PF for each target by experimenting in an interactive fashion with different scenarios based in methodology described in Game and Grantham, 2008 [40]., with our final setting at: 5 for the three ecosystem services from InVEST; and for habitat, 10 for forest in biological corridors, 5 for forest, 4 for plantations, and 2 for plantations in biological corridors. The “Status Layer” or “Status Identifier” is an optional input in Marxan that provides a scenario where the planning units that are protected areas or special interest areas are to be “locked in” or fixed in the final PU network solution [40]. We did not use this option, and so the PUs selected in our solutions are included because of their efficiency in reaching the target goals and with the minimal costs.

The Marxan analysis produces two outputs for PUs: (1) the summed solution and (2) the best solution. The summed solution is the sum of the number of times PUs are included in all solutions from the total runs of the software, which is a measure of the each PU’s “irreplaceability”—in our case ranging from 0 to 100 (low to high irreplaceability). The best solution includes only the PUs selected in the run with the lowest total cost (i.e., threat). To test the spatial distribution and overlap of the priority sites for each combination of targets, we calculated Spearman’s correlations between the “irreplaceability” across all planning units.

To evaluate the contribution of existing protected areas, management zones and potential biological corridors in conserving habitat and ecosystem services, we tallied the targets within PUs selected by Marxan’s best-solution, and we calculated the percentage of services occurring in existing protected areas (System of Natural Protected Areas of Misiones), the Biosphere Reserve Yaboti, biological corridors, as well as the rest of provincial territory (i.e., the matrix).

The network of planning units selected by Marxan for carbon and water were much patchier than the networks for habitat and soil, considering both the best and summary solutions (Figure 4).

This result was expected because spatial connectivity of selected PUs was favored by habitat due to our inclusion of biological corridors in the penalty factor, while PU connectivity of soil retention was influenced by topography, where high values were concentrated in mountainous areas. Beyond that, the Marxan solutions show the importance of the largest remnants of native forest in Misiones and their role in providing ecosystem services (Figure 1, Figure 4). The prioritized planning units for targets differ among themselves (Figure 4), showing areas where they overlap completely (i.e., a “win-win” situation) to areas with no overlap. Although protected areas were not considered in the Marxan solutions, PUs in these areas were often selected, particularly for habitat and carbon. The solutions for all four targets included the central highlands and Yabotí region as prioritized areas (Figure 4), which are not protected, yet can be considered win-win areas in providing multiple ecosystem services. Both areas are included in the Green Corridor and have a “yellow” category in Misiones’s Landscape planning document. Nevertheless, some studies have shown that the relatively low protection levels in these regions have historically had negative effects on biodiversity [2,3]. The Iguazú National Park and its surrounding areas in the north of Misiones were not selected for water and soil retention (Figure 4). This could be due to the relatively lower elevation and less precipitation of the North region, factors which lead to lower levels of modeled soil retention and water yield than the other areas in the province. Although these protected areas and connecting corridors were not ideal in terms of complete overlap of services, they encompass relatively high levels of native habitat and carbon stocks, and provide other ecosystem services not considered in this study (e.g., tourism opportunities near Iguazú Falls and Puerto Iguazú). One important landscape function of the northwest region is the connection with the Brazilian Iguaçu National Park across from Iguazú Falls, which contains the largest remnant Upper Paraná Atlantic Forest patch in Brazil and also provides high levels of protection for biodiversity [2,3].

Considering correlation of planning unit irreplaceability among targets as a measure of spatial similarity, carbon and habitat were the most similar, followed by soil and habitat, and carbon and soil (Table 2). This was expected because habitat and carbon were both very dependent on the land cover map. On the other hand, the selected network for soil retention is related to topography, where steeper areas also had remnants of forest cover. The irreplaceable of PUs for water conservation had little overlap with those from the other targets, with correlation of 0.17, −0.09, and −0.10 with soil retention, habitat and carbon, respectively (Table 2). The spatial incongruence between the PU network for water yield and the other three PU networks can be explained by the positive relationship between forested areas and habitat and carbon, and negative relationship between forested areas and the water availability due to increasing evapotranspiration. The region of Yabotí has benefited from high average precipitation in last decades, which explains its high value in water yield despite being mostly forested.

Figure 5 summarizes the provision of habitat and ecosystem services by each PU. The map shows the sum of the best solutions from the four targets, with a value of 4 indicating complete overlap of all targets (Figure 5b).

The “hotspots” of multiple overlapped targets reveals the effectiveness of the largest protected areas (i.e., Urugua-I and Iguazú National Parks) and Yabotí. As mentioned, Yabotí is a Biosphere Reserve with specific rules for sustainable management and where multiple uses are permitted. Other hotspots do not have any legal protection in the province. For example, the central highlands conserve many ecosystem services simultaneously (Figure 4), yet all PUs with 3 to 4 selected targets are in the “red” or “yellow” categories of the Landscape Planning Law, showing the importance of taking efficient management decisions in the yellow category, in particular. The Iguazú and Urugua-I National Parks are important regions, with relatively high levels of protection, yet they tended to have PUs with only 2–3 selected ecosystem services. However, these national parks have other important functions. For example, Iguazú National Park, a natural UN world heritage [68] (hich has the famous Iguazú water falls recently selected as one of the natural wonders of world [69] (This is an important tourist resource, not only for the province but for the country as a whole. Iguazú N.P. and Los Glaciares in Patagonia are the two main national parks that, with their revenue, support the other 32 National Parks and Reserves and 6 Natural Monuments in Argentina.

Although the majority of protected areas were efficient in terms of incorporating best-solution planning units with multiple ecosystem services, this analysis revealed some less-efficient protected areas. For example, the Foerster Provincial Park in the northeast of Misiones is a protected area with a poor design in terms of its convoluted shape (i.e., more edge to interior habitat) and its location in a region disconnected from other large protected areas, and under increased threat due to fast land-use changes near its borders [70]. This has led to Foerster P.P. to suffer the biological consequences of isolation and degradation. The surrounding threats are the main reason why the protected area was not selected by the Marxan analyses, despite its forested area (Figure 3). For this reason, biota and fauna in Foerster P.P. depends on the connection with Urugua-I P.P., where there are significant conservation efforts [71]. However, this result also requires land planners to consider the costs and benefits of protecting this area, and whether there are other non-protected regions with more potential benefits (e.g., highland region) [72].

Of the total PUs in Misiones (58,480), 32% had none of our conservation targets selected by the Marxan best solutions, while other PUs provided one (34%), two (17%), three (11%) and four (6%) targets selected by best solutions, respectively (Figure 5a).

The majority of PUs within Protected Areas and Yabotí had 2–3 and 3–4 targets selected by the Marxan best solutions, respectively (Figure 6a). In other areas of conservation interest, biological corridors and plantations, and in the surrounding matrix, the majority of best-solution PUs contained no ecosystem services. These results lend support to existing protected areas in Misiones, and particularly the Yabotí Biosphere Reserve, as these areas were selected repeatedly by Marxan for sustaining multiple targets. Of the selected PUs, 27.5% of those that provide four (1,044 PUs) and 25% of those that provide three (1,563 PUs) targets, respectively, were inside protected areas (Figure 6b). Similarly, 38% (1,432) of the selected PUs with four and 22% (1,379) with three targets, respectively, were inside Yabotí (Figure 6b). As a Biosphere Reserve, Yabotí include a core protected area (i.e., Esmeralda P.P. of 316 km²) and a buffer area of private proprieties of 2,050 km² which must conform to the United Nations multiple-use management principles. Currently, many conflicts exist between provincial government and land owners, who exert political pressure for more flexible laws or revised land conversion permits. However, our results show that this is a priority area for conservation of ecosystem services. In addition, the area is a critically important for providing habitat for many threatened species [2,3,66]. Finally, it was created on the territory of 15 of the Mbyá-Guaraní indigenous groups [73], providing community resources and opportunities for cultural preservation. Our results thus provide important information that should be considered in formulating holistic management decision in Yabotí.

Although there is widespread agreement on the importance of conserving natural areas for species habitat and associated ecosystem services, achieving these dual goals is often absent from the planning and design of protected areas or multi-use zones, and once established, the funding for management of these areas is almost always inadequate. It is estimated that the Argentinean investment in national parks is currently$8.5/ha, with ~60% coming from governmental funds [74]. At the province level, the investment is less well known. While private initiatives are very heterogeneous in objectives and financing, some private reserves have an estimated $20 to $440/ha/year in operating costs [75]. However, other private reserves do not invest in management and only maintain the reserves “on paper”. A major reason for the underfunded state of protected areas is that the benefits to society are often grossly underestimated, and the immediate and maintenance costs of protection appear larger in comparison. Understanding and quantifying ecosystem services provided by areas in a landscape can play a key role in increasing the efficiency of conservation investments and in attracting additional funding. One key example is the economic valuation of carbon storage in response to global climate change. Voluntary markets and international initiatives, such as the United Nations Reduce Emissions from Deforestation and forest Degradation (REDD) program [76], seek to pay credit nations for maintaining their carbon stored in vegetation (i.e., not deforesting). Argentina is not currently a REDD participant, yet has developed a national readiness plan to join when funding permits. As REDD or other valuation mechanisms mature, there is a clear opportunity in Misiones to receive payments for its carbon stored in forests – money that could be deployed to strengthen and expand forest protection, as well as build political commitment to multiple conservation goals.

Our Marxan-based analysis selects areas that conserve the most ecosystem services at the lowest threat. Another analytical approach is to differentiate areas with perceived or historical risk of deforestation in the conservation planning process [77], and include these in programs of Payments for Environmental Services (PES) as a tool for conserving natural resources [78]. This approach could be considered contradictory with our Marxan-based analytical approach, i.e. select areas with least risk instead of greatest risk. However our analysis permits the identification of general priority areas across the region that may require different tools for conservation, one of which could be PES. For example, Yabotí and zones of Urugua-I P.P. are identified as a priority area for conservation of multiple ecosystem services by our analysis. While Urugua-I P.P. is a Protected Area with problems related to effective management, Yabotí is an area of intensive pressure for land conversion. A better and effective management policy could be the way to conserve Urugua-I, and a PES could be a tool for conservation in Yabotí.

Well-managed protected areas or areas under effective PES programs alone may not be sufficient for long-term species conservation. These areas need connectivity through the matrix to support wide-ranging species, gene flow and future migration routes in the face of climate change. In Misiones, an independent analysis identified areas of critical connectivity as biological corridors [29]; however, since these corridors span degraded habitat and matrix lands, they lack the overlapping ecosystem services (Figure 6) that would give them priority in our proposed planning process. How can conservation prioritization include ecosystem services without sacrificing the broader spatial context or other long-term considerations? There is a serendipitous example in Misiones. A local conservation non-governmental organization (Conservación Argentina) is working to recover matrix lands between Foerster P.P. and Urugua-I P.P., with the main goal of providing connecting habitat for species migrating between protected areas—a single ecosystem service. Although our analyses do not select Foerster as a priority area, mainly due to surrounding threats and isolation, other considerations could be reason for conserving this protected area. In the case of carbon, increasing biomass in these forests could provide additional economic value and revenues, such as through REDD-type initiatives, thus sustaining corridor reforestation efforts and elevating these areas in priority over time. Thus, planners and decision-makers should thus be mindful of broader landscape considerations, such as connectivity, and seek ways to boost their social and economic valuation through time.