The first phase of our methodology involves defining the study area, gathering all the secondary information, and defining the field campaigns necessary for the study. At this stage, it is also important to identify the existence of previously established protected areas in the study area as the criteria for representativeness and complementarity depend on this information.

Defining conservation objectives serves several purposes. For example, it facilitates the selection of the minimum area necessary to represent and ensure the persistence of conservation values (Linke et al., 2011). These objectives may be established considering the economic constraints related to investment in actions, or they may simply be a product of the specific goals of an organization or group of organizations (Téllez et al., 2011). In this methodological framework, the conservation objectives are set as input criteria for prioritization (see Step 2.17), allowing conservationists to address specific questions or interests.

Following the proposal by Riato et al. (2020) to select conservation actions (see Step 2.16) using the integrity criterion, our framework uses micro-basins and sub-basins as the scales of analysis. We based our scales on basins as they are appropriate units for the SCP of freshwater ecosystems (Tognelli et al., 2019; Dorji et al., 2020; Linke and Hermoso, 2022; Nogueira et al., 2023). This stage of our proposal consists of automatically delimiting these units using a digital terrain elevation model (DEM), which allows us to obtain the units of analysis and the river segments between two consecutive nodes or within the units. This approach produces a graph representation of the study area that defines the connectivity between the units, which facilitates efficient cumulative calculations (important aspects for Steps 2.8, 2.11, and 2.17).

According to Higgins et al. (2021), conservation values refer to various aspects, such as species, ecosystem services (cultural, provisioning, supporting, and regulating), as well as cultural and spiritual values that are important to a particular community. In our methodology, it is possible to consider one or several of these elements; the number of values to consider will depend on their level of relevance in the selected study area. At the end of the process, conservation values are condensed into a numerical index that we have defined as an index of importance (in Step 2.13).

KEAs are essential characteristics for the maintenance of freshwater ecosystems and, therefore, for the conservation values (Higgins et al., 2021). Existing research has identified the hydrological regime, sediment transport, water quality, physical structure, and connectivity as the main determinants of the physical habitat and biotic communities of freshwater ecosystems (Poff et al., 1997; Castello et al., 2013; Zeiringer et al., 2018; Higgins et al., 2021). In turn, these characteristics shape the social dynamics of communities that depend on these ecosystems for their livelihoods. Therefore, this methodology prioritizes these KEAs for analysis.

Threats, on the other hand, are factors that generate stress on KEAs—e.g., climate change, deforestation, pollution, and dam construction (Poff et al., 1997; Castello et al., 2013; Dudgeon, 2014; Alho et al., 2015; Arthington et al., 2016; Zeiringer et al., 2018)—resulting in ecosystem degradation (Higgins et al., 2021). Therefore, the KEAs, the threats to the KEAs, and the sources of the threats must be selected depending on the study basin. This step is crucial because these decisions will determine the indexes, models, or analysis tools to be used in the study.

A freshwater group is a set of planning units, or micro-basins, that possess similar KEAs. In this sense, freshwater groups can be understood as collections of habitats within a basin. Our proposed framework employs simple characteristics and indexes for each KEA, allowing for precise differentiation of areas with unique KEAs without requiring extensive amounts of information for their configuration. The freshwater groups are of great importance in our proposal because we understand them as the freshwater habitats present in the basin. Additionally, the rarity and representativeness criteria are derived from these groups.

Our methodology considers integrity as the capacity of a basin to support and maintain the broad range of ecological processes and functions essential for both biodiversity sustainability and the resources and services that the basin provides to society (Flotemersch et al., 2016). Based on this definition, our framework uses a set of indexes to assess how the identified threats (in Step 0) may impact the KEAs in the study basin, in turn affecting ecosystem integrity.

The selection of models and analysis tools is based on the indexes defined for the freshwater and integrity groups. The selected tools must be sensitive to the threats identified in each KEA. Our framework is flexible and allows for the adoption of different approaches and strategies, such as a conceptual approach (Thornbrugh et al., 2018), machine learning (Giri et al., 2019), empirical models, or a combination of these (Einheuser et al., 2013). Moreover, it is possible to select individual tools for each KEA or to use the same tool to model multiple KEAs.

Once the models and analysis tools have been chosen, this stage also includes collecting and processing the information needed to build the models, configuring and running the selected models, and calculating the indexes corresponding to the river and integrity groups.

In this step, the defined attribute indexes are used to group the planning units to identify freshwater clusters. Different clustering approaches can be used to achieve this, including hierarchical, partitional, grid, density-based, or model-based methods (Saxena et al., 2017). The choice of method will depend on the characteristics (qualitative or quantitative) of the attribute indexes selected in Step 2.6. Our goal in identifying freshwater group clusters is to generate connected corridors of high–conservation value micro-basins that host the greatest possible diversity of freshwater groups (representativeness criterion) and include the rarest freshwater groups (rarity criterion).

In our methodology, rarity is a measure of the uniqueness or scarcity of a given freshwater group within the study basin, or the proportion or area occupied by that group compared with the rest of the groups present in the basin. Rarity is a criterion in the prioritization process, as described in Step 2.17. Thus, in this phase, we construct an index that reflects the rarity of each freshwater group in the basin using a formula that considers the proportion or relative area occupied by the group in relation to all the groups present. This process produces a numerical index that captures the rarity of each group, enabling the subsequent comparison and prioritization of the groups in the basin.

Because impacts on aquatic ecosystems are cumulative, we seek to foster connectivity in the prioritized micro-basins due to the importance of connectivity for freshwater ecosystems (Saura et al., 2017; Herrera-Pérez et al., 2019). Therefore, in this step, we construct an index that reflects the degree of connection between two micro-basins to prioritize corridors of connected micro-basins linking headwater rivers (see Step 2.17). This index is based on the proximity between micro-basins, measured from their centroids or across their river segments. Key references for the construction of this index were Hermoso et al. (2011, 2018), Cattarino et al. (2015), Wohl (2017), and Dorji et al. (2020).

In our proposal, integrity plays a fundamental role in both prioritization (see Step 2.17) and the definition of conservation opportunities (see Step 2.16). Both steps require a single index ranging from 0 to 1, where 0 represents low integrity and 1 represents high integrity. Therefore, this step involves constructing an integrity index that groups the indexes defined in Step 2.7 and calculated in Step 2.8. This can be accomplished using aggregation methods, such as geometric or arithmetic aggregation (Juwana et al., 2012), or multiplicative approaches, such as those used by Thornbrugh et al. (2018).

The concept of importance in our methodology is closely related to the quantity of conservation values present in a micro-basin. As described in the prioritization section (see Step 2.17), our methodology seeks to maximize importance while achieving the conservation target. In this phase, we construct a numerical index ranging from 0 to 1, where 0 denotes the total absence of conservation values and 1 represents a high concentration of conservation values. One possible approach is to assign a relative weight (between 0 and 1) to each conservation value identified in Step 2.4 and calculate a weighted average of these values. This would provide an index that reflects the overall importance of the conservation values present in the micro-basin.

Our methodological framework defines representativeness as a measure of the presence of freshwater groups in the network of existing protected areas in the study area. The representativeness index can be constructed by considering the area of the freshwater group and the area covered within the protected areas (Duarte et al., 2016). If there are no protected areas in the study area, this index will have a value of 0 for all the micro-basins.

Complementarity in our methodology refers to the inclusion in the priority portfolio of those freshwater groups that are not represented in protected areas. This concept is closely related to that of representativeness. Therefore, our methodology incorporates this criterion as a constraint in the prioritization process (see Step 2.17) to ensure that the priority portfolio includes existing protected areas in the study area. If there are no protected areas in the study area, this criterion is not considered in the prioritization.

We followed the proposal of Riato et al. (2020) to define conservation opportunities, including protection, restoration, and sustainable management, using the integrity index estimated at the micro-basin and sub-basin levels. The process begins by contrasting the indexes on a 2D scatterplot (both indexes should be scaled between 0 and 1, where 0 represents low integrity and 1 represents high integrity). Then, four quadrants are defined, centered on the 0.5 value of each index. If both the micro-basin and sub-basin indexes show values greater than 0.5, the ecosystem is in good condition; thus, it would be best to consider a protection action because doing so would require minimal intervention. If, instead, the index at the sub-basin scale is greater than 0.5 but the index at the micro-basin scale is less than 0.5, it would be more appropriate to consider restoration as the best conservation opportunity because a healthy sub-basin can support this action. However, if both indexes have values below 0.5, sustainable management would be the best option. In this case, protection and restoration would require considerable effort and resources because they would not be supported by good conditions at either scale.

SCP aims to maximize the representation of conservation objects. In our case, it involves identifying the network of micro-basins that maximizes conservation values (importance) while i) promoting the highest representativeness of freshwater groups, ii) connecting the largest number of rare freshwater groups, iii) including micro-basins with high integrity, and iv) generating corridors of connected micro-basins. To achieve this, we relied on the proposal by Hermoso et al. (2011) to formulate the following optimization objective function O f : O f = ( ∑ M i c r o − b a s i n I C V ) + ( ∑ I n t e g r i t y p e n a l t y − I ) + ( ∑ R a r i t y p e n a l t y I R ) + ( ∑ R e p r e s e n t a t i v i t y p e n a l t y I R e p ) + ( ∑ C o n n e c t i v i t y p e n a l t y − C I )

We propose using the integrity, rarity, representativeness, and connectivity indexes as penalties (with equal weights) along with a constraint that ensures the selection of micro-basins located within existing conservation areas. Thus, micro-basins with lower integrity, higher abundance of a freshwater group, and representation in existing conservation areas will be penalized more significantly.

The conservation portfolio is composed of the priority areas resulting from the optimization process and the conservation opportunities identified from the 2D scatterplot proposed by Riato et al. (2020), which must be addressed to achieve the established conservation objectives.

For the application of our methodological framework, we considered the Caquetá River basin, which covers an area of 148,763 km² and represents 31% of the Colombian Amazon biome (Figures 2A–C). This region has an altitudinal gradient of 4,200–60 masl (Figure 2D), generating a spatially variable annual precipitation range of 790–4,924 mm. The climate is characterized by two rainy seasons per year in the mountain zone (March–May and October–December) and one season in the central and foothill zones (April–June). These climatic conditions foster a wide range of cold to warm tropical environments.

The Caquetá River is 1,400 km in length and collects the waters from other important rivers, such as Orteguaza (696 m³/s), Caguán (1,142 m³/s), Yarí (2,138 m³/s), Cahunarí (986 m³/s), and Mirití Paraná (654 m³/s), reaching an average annual flow of 10,100 m³/s. Approximately 13% of its drainage area is wetlands, of which 18% have been intervened on or transformed by anthropic actions (Ministerio de Ambiente Desarrollo Sostenible, 2021). Currently, 21% of the basin is protected by 10 natural parks (Figure 2C) (Parques Nacionales Naturales de Colombia, 2020).

The characteristics of the Caquetá River basin support biodiversity in its freshwater ecosystems. More than 400 fish species have been recorded in the different tributaries (Celis-Granada et al., 2022). The Caquetá River is also an important migratory corridor for 23 species, including turtles (e.g., Podocnemis expansa), fish (e.g., Brachyplatystoma rousseauxii), and dolphins (e.g., Sotalia fluviatilis) (He et al., 2018; 2021; Caldas et al., 2023). Cultural richness is also a distinctive feature of this area, which is home to Indigenous, peasant, and Raizales communities (Agencia Nacional de Tierras, 2023). Indigenous communities have a greater presence in the territory, with 96 legally constituted Indigenous reserves occupying 43% of the total area and bringing together peoples such as the Murui-Muinane, Yucuna, Andoque, Inga, and Coreguaje, among others (Agencia Nacional de Tierras, 2023). There is also recorded evidence of uncontacted isolated communities in this region (Seifart and Echeverri, 2014; Walker et al., 2016; Walker and Hamilton, 2019).

Recognizing the global importance of the Amazon River basin for the conservation of freshwater biodiversity and the communities that depend on it, The Nature Conservancy has prioritized this area in its conservation vision for 2030 (The Nature Conservancy, 2022). As part of this commitment, they have embarked on a conservation planning process to establish a roadmap for their future work, thereby helping to preserve the biodiversity and the values it supports. To achieve this, The Nature Conservancy developed a conservation plan for the Amazon River basin with the following objectives, which we have considered for the Caquetá River basin.

• Conserve 80% of the main rivers (large and very large; flow >100 m³/s), ensuring connected corridors that remain functional/healthy.

The materialization of these objectives by The Nature Conservancy would contribute to Target 3 of the United Nations Convention on Biological Diversity: to protect and effectively manage 30% of the world’s terrestrial, inland waters, and coastal and marine areas by 2030 (ONU, 2022).

To delimit the micro- and sub-basins of the Caquetá River basin, we used the Shuttle Radar Topography Mission DEM with a 90-m resolution (Jarvis et al., 2008). We processed the DEM using ArcGIS 10.7 and ArcHydro Tools. To delimit micro-basins, we used an area accumulation threshold of 2.8 km². However, in areas where the river profile showed a steeper slope than the adjacent segments, known as knickpoints, we performed an additional segmentation. For this subdivision, we considered the geomorphological conditions of the areas and the habitats they provide (Ross et al., 2001) and used the algorithm proposed by Hayakawa and Oguchi (2006) to detect the knickpoints. We defined the sub-basins according to the delimitations established by the environmental authority, Corporación para el Desarrollo Sostenible del Sur de la Amazonia (Corpoamazonía), so the proposed conservation portfolio can be effectively integrated with territorial planning instruments.

We considered the following conservation values for the Caquetá River basin:

Species: i) We obtained information on 55 species of fish, 47 species of amphibians, and 102 species of aquatic birds from the collaborative BioModelos system of the Alexander Von Humboldt Institute (Velásquez-Tibatá et al., 2019). ii) We considered the distribution of four species (Pteronura brasiliensis, S. fluviatilis, Tapirus terrestres, Tapirus pinchaque) in danger of extinction as reported by the IUCN (2022). iii) Data on megafauna species and their migratory corridors were taken from Caldas et al. (2023) and He et al. (2018), respectively.

Ecosystem services: Recreation and tourism were integrated based on the distribution of recreation person-days, which we generated using the InVEST Visitation: Recreation and Tourism model (Natural Capital Project, 2022).

Cultural areas: Ninety-six Indigenous reserves were included from two areas of the Raizales community and one peasant reserve, using the information reported by Agencia Nacional de Tierras (2023).

Spiritual areas: Sacred and spiritual sites for Indigenous communities were defined according to Organización Nacional de los Pueblos Indígenas de la Amazonia Colombiana (2017), considering the importance they represent for the cosmovision of Indigenous peoples (The Nature Conservancy and The Amazon Conservation Team, 2019).

In Amazonian freshwater ecosystems, including the Caquetá River basin, deforestation is primarily driven by agricultural and livestock expansion. These activities are also sources of phosphorus and nitrogen pollution resulting from the fertilization of pastures and crops. Other sources of pollution include oil and gas concessions and legal and illegal mining activities, the latter of which is primarily associated with mercury contamination from gold extraction (Castello et al., 2013). According to Diaz et al. (2020), Colombia is among the countries that use the greatest amount of mercury to produce one ton of gold (4.19 Hg/ton). Moreover, climate change has reduced precipitation and increased temperatures in the Amazon (Killeen and Solórzano, 2008). The Caquetá basin does not currently have hydropower development modifying water and sediment flows or disrupting the river network. Table 1 summarizes the main threats and the sources that we considered in our analysis, according to the selected KEAs.

Here, we describe the indexes selected for each KEA in the case study.

Hydrological regime: We considered the components of the hydrological regime (magnitude, frequency, duration, timing, and rate of change) using the following hydrological signatures of the Caquetá River (McMillan, 2020; 2021): i) mean annual flow (magnitude), ii) high flow duration (duration), iii) frequency of peak flow (frequency), iv) slope of the flow duration curve (rate of change), and v) mean half flow date (timing) (Table 1).

Sediment transport: We estimated the sediment transport capacity under bankfull conditions as an index for this KEA (Table 1).

Water quality: For Amazonian environments, three types of water have been defined in terms of chemical composition (Ríos-Villamizar et al., 2013): i) blackwater rivers, ii) clearwater rivers, and iii) whitewater rivers (see Table 1).

Physical structure: Considering the importance of biotic productivity (Venarsky et al., 2018) and available habitats in freshwater ecosystems (Flores et al., 2006; Buffington and Montgomery, 2022), we selected the following indexes to assess the study area’s physical structure: i) the morphological configuration of the river, which can be confined or unconfined, and ii) the specific stream power, which can be capacity- or supply-limited (Table 1).

Connectivity: Connectivity is an essential attribute of freshwater ecosystems (Saura et al., 2017; Herrera-Pérez et al., 2019), particularly for fish that perform migratory movements in the Caquetá River basin, such as the B. rousseauxii catfish (Córdoba et al., 2013). For this KEA, we have selected the dendritic connectivity index, as proposed by Cote et al. (2009) (Table 1).

Step 3.9

details how we used these attribute indexes to identify freshwater groups.

Hydrological regime and sediment transport: Based on the hydrological alterations concept by Poff et al. (2010), we created an index that allows us to evaluate the impact of the climate change threat on the hydrological regime and sediment transport. The mass flow index ( I H R or I S T , according to the considered variable) assesses the average percentage variation of the p percentiles of the duration curve (5, 10, 15, … , 95) of streamflow Q or sediment transport Q S between a historical condition h and a climate change condition c c : I H R = 1 N ∑ p = 1 N p Q p , c c − Q p , h Q p , h I S T = 1 N ∑ p = 1 N p Q s p , c c −   Q s p , h Q s p , h where N p is the number of percentiles of the streamflow or sediment flow duration curve.

Water quality: We considered the following water quality index I W Q (Orjuela and Lopez, 2011; Akhtar et al., 2021) as representative of the river’s chemical composition in our study area: I W Q = ∑ d = 1 N d W d × S W Q d ∑ d = 1 N d W d = 1 where S W Q d is the value of the sub-index for water quality determinant d , which has a value between 0 and 1; N d is the number of water quality determinants considered; and W d is the weighting factor. We used equal weights for each determinant with a value of 1 / N d . The subscripts for each considered water quality determinant are presented in Table 2.

In Table 2, S S is the suspended solids concentration (mg/L); X is the concentration of pathogenic organisms (NMP/100 mL); T N is the total nitrogen concentration (mg/L), which includes organic nitrogen, ammoniacal nitrogen, and nitrates; T P is the total phosphorus concentration (mg/L), which includes organic and inorganic phosphorus; O M is the organic matter concentration (mg/L); D O is the dissolved oxygen concentration (mg/L); O s is the saturation oxygen concentration (mg/L); and T H g is the total mercury concentration (mg/L), which includes elemental, divalent, and methyl mercury. For total mercury, we considered a binary sub-index with two categories (good = 1 and bad = 0), according to the permissible limit (0.001 mg Hg/L) defined by Ministerio de Salud y Protección Social and Ministerio de Ambiente Desarrollo Sostenible (2007) for Colombia.

Physical structure: We adapted a stressor mapping index I P S from Flotemersch et al. (2016) by applying a geometric aggregation method (Juwana et al., 2012): I P S = ∏ z = 1 N z g z S z S z , max 1 w z ∑ z = 1 N z W z = 1 where S z is the observed value of stressor z in a micro-basin, S z , max is the maximum value of stressor z at the micro-basin level in the entire study area (where S z / S z , max varies from 0 for unaltered to 1 for maximum impact), N z is the number of stressors affecting the physical structure of the ecosystem, g z is a mathematical function of a single variable that describes the degree of impact caused by stressor z , and W z is the weighting factor. We used equal weights for each stressor with a value of 1 / N z .

The stressors we refer to are anthropogenic disturbances that degrade ecosystems and, therefore, their functions (Flotemersch et al., 2016) (e.g., human-caused forest fires, agricultural land use, urban areas, and road density). To calculate the I P S index, we used the stressors listed in Table 2. The technical details of the configuration, inputs, and outputs of the models used can be found in the Supplementary Material.

Connectivity: We constructed a longitudinal connectivity index I C : I C = L B L A where L A is the length of the shortest path from a micro-basin to the mouth of the analyzed basin, without considering the presence of barriers that may generate a disconnection, and L B is the length of the shortest connected path between a micro-basin and the mouth of the analyzed basin, considering the presence of barriers in the channel. In this case, the length is measured from the mouth to the first barrier encountered. Table 1 presents a summary of the integrity indexes by KEA. The implementation of the indexes described here (including scaling and ranges) is detailed in Step 3.12.

To estimate the two groups of considered indexes—freshwater group indexes and river integrity indexes—we configured a set of models and analysis tools for each KEA. The purpose and objectives of these tools are outlined in the following sections. The Supplementary Material provides the technical details of the configuration, inputs, and outputs of the models.

To identify the freshwater groups in the Caquetá River basin, we clustered the micro-basins based on the similarity of the KEAs using the indexes defined in Step 3.6 (Table 1). For this process, we used an agglomerative hierarchical clustering method and selected the inconsistency coefficient as the metric for determining the clusters (Saxena et al., 2017). During the clustering process, we categorized the indexes according to their ranges of variation, as shown in Table 5, and used these categories as the basis for clustering to define the freshwater groups.

We calculated the rarity I R of a micro-basin belonging to freshwater group i as (Duarte et al., 2016): I R = − ln a i A where a is the total area of the micro-basins belonging to freshwater group i , and A is: A = ∑ i = 1 N g a i where N g is the total number of freshwater groups. High values indicate a unique habitat in the basin, and low values indicate a common habitat. The Supplementary Material provides additional details for the rarity index calculation.

To calculate the connectivity index, we constructed a directional upstream adjacency matrix (Wohl, 2017) that includes, for each unit, the degree of connectivity defined by the proximity between micro-basins, measured using the length of the main channel of each micro-basin. As an example, consider that the study area is segmented into five micro-basins, each of which has a river segment with a length of 5 km. The directional upstream adjacency matrix for the micro-basins would be represented by A . If we calculate the river segment lengths from one micro-basin to all upstream micro-basins, we obtain the matrix of accumulated distances, B . By estimating the maximum distance for each micro-basin, we obtain the matrix of maximum distances, C . The connectivity index C I would then be given by 1 − B m , n − C m , n / B m , n , where values near 1 indicate high connectivity and values close to 0 indicate low connectivity. The Supplementary Material offers additional details about the connectivity index calculation.

For the Caquetá River basin, we developed a global integrity index I for the micro-basins and sub-basins based on the integrity indexes I k i selected for each KEA k i = H R , S T , W Q , P S , C in Step 3.7. We used a geometric aggregation method to integrate I k i (Juwana et al., 2012; Thornbrugh et al., 2018). We assigned equal weights w k i to all integrity indexes by KEA, considering that each is crucial for the integrity of freshwater ecosystems. Therefore, each I k i was assigned a weight equal to 1 / K I , where K I is the number of KEAs selected in Step 3.5. The mathematical expression of I is: I = ∏ k i = 1 K I I N k i 1 w k i I N k i = m * I k i , r + α r min − m * min I k i , r m = α r max − α r min max I k i , r −   min I k i , r

In this equation, we normalize each I k i using a linear scaling method. However, this normalization is performed differentially by range r . The values of α r min and α r max correspond to the scaling limits used in each range: min I k i , r represents the minimum threshold of I k i , r in range r , and max I k i , r reflects the maximum threshold of I k i in range r . In the end, I has values ranging from 0 to 1, where 1 reflects high integrity and 0 reflects low integrity. Table 6 presents the four ranges considered for the scaling of each I k i .

The ranges used correspond to integrity index categories: low integrity 0 ≤ I < 0.25 , medium integrity 0.25 ≤ I < 0.5 , high integrity 0.5 ≤ I < 0.75 , and very high integrity 0.75 ≤ I ≤ 1 .

We constructed an importance index I C V that groups v sub-indexes, which are estimated for each conservation value C V v defined for the basin. This index ranges from 0 to 1, where a value of 1 represents units of high importance. I C V = ∑ v = 1 N c v C V v

For fish, amphibians, reptiles, and waterbirds, we used the normalized values of the rarity-weighted richness index R W R I (Williams et al., 1996; Abell et al., 2011). This index counts the number of species in a micro-basin and weights each species by the inverse of the number of micro-basins it occupies: R W R I = ∑ s = 1 S p i 1 N s C V R W R = R W R I i R W R I i , max where S p i is the number of species in micro-basin i , N s is the total number of micro-basins occupied by species s , R W R I i is the observed value of RWRI in micro-basin i , and R W R I i , max is the maximum R W R I value in the entire study area.

For the remaining conservation values (see Table 7) we constructed a sub-index C V , where X i , w is the observed value of conservation value w in micro-basin i and X w , max is the maximum value of conservation value w recorded at the micro-basin level in the entire study area: C V w = X i , w X w , max

Table 7 summarizes the sub-indices for the selected conservation values in the Caquetá River basin and presents the numerical ranges for each conservation value. Note that all values are scaled from 0 to 1 using the equation above. The Supplementary Material details the calculation of the conservation values sub-indexes.

The representativeness index I R e p of micro-basin i that belongs to freshwater group k is defined as: I R e p = 1 A k ∑ i = 1 I a k , w A k = ∑ i = 1 I a k where a k , w is the area of micro-basins belonging to freshwater group k and contained within an existing conservation area w , and a k is the area of all micro-basins belonging to freshwater group k (Duarte et al., 2016). The details of the representativeness index calculation can be reviewed in the Supplementary Material.

To ensure complementarity in prioritization, we selected the areas of the natural parks as reported by Parques Nacionales Naturales de Colombia (2020). The Caquetá River basin currently contains 10 conservation areas representing 21% of the total area of the basin (Figure 2).

We used MATLAB Release 2019b to construct the 2D scatterplot of the Caquetá River basin to identify conservation opportunities. This platform allowed us to analyze and visualize the data and generate graphical outputs.

We solved the optimization problem for the Caquetá River basin by using the prioritizr package v8.0.2.1 (Hanson et al., 2023) to find an accurate, optimal solution in conjunction with Gurobi solver v10.0, which provides solvers based on integer linear programming. This combination of packages is computationally efficient compared with packages such as Marxan, which generates nearly optimal solutions (Beyer et al., 2016).

Figure 3 presents the spatial units of analysis obtained for the Caquetá River basin. We delineated 30,028 micro-basins and 320 sub-basins for the study area (Figure 3A). Considering the large number of delimited micro-basins, Figure 3 shows example distributions of the micro-basins within the sub-basins of the Mamansoya and Peneya Rivers, located in the middle and upper part of the Caquetá River basin, respectively (see Figure 3B). Approximately 61% of the defined micro-basins are smaller than 5 km², and only 2.4% are larger than 15 km² (see Figure 3C).

We identified 518 freshwater groups with unique ecological characteristics (Figure 4); 28.4% of these are found in large and very large rivers (flow >100 m³/s), and 71.6% are present in medium and small rivers (flow ≤100 m³/s). Thirty percent of the identified groups are in the mountainous zone, which encompasses 11% of the entire Caquetá River basin. We also determined that 70% of the basin is occupied by only 4.2% of the identified freshwater groups.

Figure 5A provides a spatial representation of the importance index in the Caquetá River basin, highlighting that the micro-basins that make up the main waterways were assigned the highest importance values. This is reasonable considering that these tributaries are migratory corridors for approximately 23 species of megafauna, including S. fluviatilis (He et al., 2018; 2021; Caldas et al., 2023), and other important species, such as B. rousseauxii (Córdoba et al., 2013). Moreover, the basin’s structure does not currently contain river network interruptions.

High importance was assigned to the middle and lower zones of the Caquetá River basin, given the significant concentration of cultural values stemming from the presence of the Murui-Muinane, Yucuna, Andoque, Inga, and Coreguaje peoples (who occupy 43% of the entire basin area). These ethnic groups maintain a profound connection with the territory, as, in accordance with their worldview, they conceive the world as an entity encompassing the spiritual, material, and social dimensions. Consequently, they regard rivers as vital sources for their communities, providing them with sustenance through artisanal fishing and facilitating transportation within and beyond their territories (The Nature Conservancy and The Amazon Conservation Team, 2019). Additionally, areas of high importance were identified in the Andean region due to species abundance and the high provision of ecosystem-cultural services associated with recreation and tourism activities.

We found that 46.7% of the identified freshwater groups in the Caquetá River basin are located within the 10 existing protected areas I R e p > 0 . Of these areas, the Serranía de Chiribiquete National Natural Park stands out as it covers 13.1% of the entire basin area and hosts 23% of all identified groups, equivalent to 49.1% of the represented groups (Figure 5B, brown areas). We also discovered that 4.2% of the identified groups are considered rare (Figure 5C, red areas) because they are uniquely present within the basin. Out of these rare groups, 41% are found within protected areas. The representation levels of the identified groups are as follows: 29.8% have representation below 25%, 26% have representation between 25% and 50%, and 42.2% have representation above 50% (see Table 8).

Figure 5D illustrates the graphical scheme of the connectivity index for the Caquetá River basin as an adjacency matrix. In this matrix, the asymmetry to the left indicates the direction of the connections between the micro-basins, showing the flow of water and interrelationships from the upper areas (upstream) to the lower areas (downstream).

We found that 5.5% of the micro-basins presented low integrity values, 14.8% showed medium integrity values, 33.7% exhibited a high level of integrity, and 46% had a very high integrity rating (Figure 6A). The lowest integrity values were recorded in the upper part of the Caquetá River, especially in the Orteguaza and Caguán River basins (Figure 6A), which face the highest number of threats to the KEAs. Upon analyzing the integrity in the main channels of these rivers, we observed dispersed behavior with an increasing trend as the accumulated flow increases (Figure 6B), which translates into an increase in integrity downstream. However, in the lower part of the main channel of the Caquetá River, integrity values tend to decrease despite remaining high. Of the sub-basins, 1.7% had low integrity, 16.2% had moderate integrity, 33.4% exhibited high integrity, and 48.7% received a very high integrity rating (Figure 6C).

The 2D scatterplot (Figure 6D) showed that the best opportunity to conserve 77.4% of the basins is through protection actions. This action shows a primarily continuous spatial pattern in the middle and lower zones of the basin, whereas the pattern is more scattered in the upper zone. Restoration presents a better opportunity in 4.7% of the basin, specifically in the upper, piedmont, and upper-middle zones. According to the integrity values, sustainable management should be implemented in 17.9% of the basin. The pattern of this action is primarily continuous and is present in the piedmont and upper-middle zones of the basin (Figure 6E). Furthermore, the results reveal that the established protection areas in the basin present the best opportunity for conservation, which we anticipated.

The spatial distribution of the prioritized portfolio is mainly concentrated in existing protected areas, specifically in the Serranía de Chiribiquete National Natural Park (Figure 7A), which is consistent with the selected criteria. The prioritized portfolio shows high connectivity between micro-basins, which generates corridors, as well as between existing protected areas. It is noteworthy that the prioritized portfolio includes the Caguán River despite its reduced integrity values in some sections. The portfolio managed to represent 96% of the freshwater groups identified in the large and very large river categories and 74% of the freshwater groups identified in medium and small rivers. According to the defined conservation objective, the prioritized portfolio achieves representation of 80.1% of all the freshwater groups identified in the Caquetá River basin. We found that 91.9% of the prioritized micro-basins present opportunities for conservation through protection, 3% through restoration, and 5.1% through sustainable management (Figure 7B).