Introduction
By understanding the spatial ecology of a particular species, researchers can assess its distribution based on its biological requirements (Brown, Stevens, & Kaufman, 1996). These attributes are especially important in species with conservation concerns such as otters, canids, crocodylians and other Neotropical predators (Hernández-Romero, Botello López, Hernández García, & Espinoza Rodríguez, 2018), providing relevant information for the development of action plans and conservation strategies (Boitani & Powell, 2012; Balaguera-Reina et al., 2016). Geostatistical approaches such as hot and cold spots (Getis-Ord Gi) and clusters and outliers (Anselin Local Moran I) analyses combined with environmental variables relevant for species, allow researchers to define and prioritize key areas based on the understanding of how species use space (Nelson & Boots, 2008). This is important, especially in impacted or transformed areas, allowing the development of well-informed conservation and management planning.
The Neotropical Otter (Lontra longicaudis Olfers, 1818) is a semi-aquatic mammal species that ranges from Mexico to Argentina (including northeastern Brazil) and is currently categorized as “Near Threatened” by the IUCN Red List (Kruuk, 2006; Rheingantz, de Menezes, & de Thoisy, 2014; Rheingantz et al., 2018). This species inhabits a large variety of climates and ecosystems including fast-flowing rivers, lakes, marshes, rocky seacoasts, mangroves, and wetlands (Kruuk, 2006; Almeida & Ramos Pereira, 2017; Rheingantz, Santiago-Plata, & Trinca, 2017; Gomez, Túnez, Fracassi, & Cassini, 2014; Andrade, Arcoverde, & Albernaz, 2019). However, despite its wide distribution, L. longicaudis has been poorly studied and a vast majority of information regarding its biology and natural history is still unknown (Larivière, 1999; Kruuk, 2006; Andrade-Ponce & Angarita-Sierra, 2017; Rheingantz et al., 2018). This is also the case for Colombia, where despite the L. longicaudis widespread distribution [river networks in the Caribbean, Pacific, Andean, Amazonian, and Orinoco regions up to 3 110 m.a.s.l. (Alberico, Cadena, Hernández-Camacho, & Muñoz-Saba, 2000, Trujillo et al., 2016, Andrade-Ponce & Angarita-Sierra, 2017), the species still lack of some basic ecological information and it is considered of special interest for research and conservation assessments (González-Maya et al., 2011; Trujillo et al., 2016; Botero-Botero, Delgado, & Gamboa, 2017; Camp, 2018; Pinillosa, Pérez-Torres, & Botero-Botero, 2018).
Colombia has the third largest installed hydropower capacity in South America, producing a total of 11 837 MW per year (IHA, 2019). The construction of regulated river systems along the country has increased in the last 30 years, mainly due to the need to increase the national installed energy capacity (Bresney, 2016). However, little is known regarding the impact of these anthropogenic developments over wild population, especially cryptic species like the Neotropical Otter. Understanding these impacts are required for the development of technical-wise management planning (Quadros, 2012; Palmeirim, Peres, & Rosas, 2014; Calaça & de Melo, 2017; Restrepo, Botero-Botero, Puerta Parra, Franco Pérez, & Guevara, 2018). We assessed the distribution and habitat use intensity of L. longicaudis in La Miel I hydroelectric power dam reservoir and its influence area (up and downstream) in the Department of Caldas, Colombia.
Materials and methods
Pre-sampling: In 2014, we visited La Miel I hydroelectric power dam headquarters to define areas, transects, sampling periods, and all the logistics in agreement with permits granted by the dam operator (ISAGEN). We surveyed the upstream, reservoir, and downstream waterscape areas evaluating the topography, accessibility, and navigability of water bodies, assessing restrictions and adjustments to be done to the standardized method defined for the study of otter’s distribution (Reuther et al., 2000; Groenendijk et al., 2005). Based on these surveys, we defined 10 sites across three main waterscape areas: (1) upstream [La Miel River before Samaná bridge (LMR-SB), Tasajos River (TR), Manso River (MR), Moro River (MrR), San Luis River (SLR), Fraguas Stream (FS) and Santa Barbara Stream (SBS)], (2) reservoir [Amaní reservoir (AR)], and (3) downstream [La Miel River La Isla bridge (LMR-IB) and La Miel River La Palmera-San Miguel (LMR-SM)] to be assessed by boat and foot (Fig. 1). The former method was used in two of the 10 sites (AR and LMR-SM) due to topographic (mean slope 44 º, slippery substrate) and navigability (deep waters) conditions, moving at slow speeds (< 10 km/h) close to the shoreline. The latter method was used in the remaining eight sites [LMR-SB, TR, MR, MrR, SLR, FS, SBS, and LMR-IB] due to navigability conditions (rocky-shallowed areas), following linear transects along both shorelines of streams and rivers in opposite directions.
Study area: La Miel I hydroelectric power dam is located between 05°33′38″ N and 74°53′12″ W to an elevation of 446 m.a.s.l. covering the municipalities of Norcasia, Samaná, Victoria, and La Dorada, Department of Caldas, Colombia. It is a gravity dam with a reservoir wall of 188 m high that can storage 571 million m3 water with a surface area of 1 220 ha. This dam receives waters from La Miel and Moro rivers (contributing to 70 % of the total input water), which are the main tributaries of the Magdalena River middle basin. The dam was built between 1997 and 2002, with the main purpose of hydroelectric power generation with an installed capacity of 396 MW and an annual generation of 1 146 GWh. This area has a mean annual temperature of 27 °C, average precipitation of 3 500 mm per year, and relative humidity between 60 and 100 % (Rojas-Morales, Marín-Martínez, & Zuluaga-Isaza, 2018). This region corresponds to the so-called humid tropical or equatorial zonobiome of the Colombian Andes central range formed by hilly landscapes with steep slopes and areas dominated by pastures and pastures in early regeneration (0-2 years) with some isolated patches of riparian, primary and secondary forest (mainly on the headwaters; Rodríguez, Armenteras, Morales, & Romero, 2006).
Surveys and data analyses: We carried out diurnal surveys across standardized transects between 2014 and 2018 looking for records (sightings, faeces, tracks, and dens) that indicate L. longicaudis presence. Each year, surveys were done within 12 consecutive days over three seasonal sampling periods [February - March (low rainy season), July (dry season), and November (high rainy season); 36 sampling days per year] assessing the upstream, reservoir, and downstream waterscape areas (Table 1). We attributed a site as positive when we found at least one otter record and as negative otherwise (Riesco et al., 2020). Each positive record was georeferenced (Garmin 62sc) and variables such as the number of records (individuals, faeces, tracks, or dens), date, riverscape area (upstream, reservoir, downstream), and marking substrate (trunk, rock or latrine) were registered. In all cases, four expert members (3 biologists and 1 local fisherman) participated in the field expeditions; paired members were ubicated oppositely within the boat to have a full vision of the prospected areas during the travels. When surveying by foot, the same team divided forces, two per side of the river, following fixed linear transects throughout the river shore.
Riverscape area | Sampling site | Transect length (km) | Sampling method |
Upstream | La Miel River before dam Samaná bridge (LMR=SB). | 2.5 0.7 | By foot |
Tasajos River (TR) | 2.7 0.6 | By foot | |
Manso River (MR) | 5.0 0.0 | By foot | |
Fraguas stream (FS) | 2.6 1.0 | By foot | |
Santa Bárbara stream (SBS) | 2.7 1.0 | By foot | |
Moro River (MrR) | 2.5 0.8 | By foot | |
San Luis River (SLR) | 2.7 0.8 | By foot | |
Reservoir | Amaní reservoir (AR) | 116.4 23.5 | By boat |
Downstream | La Miel River Muro- La Isla bridge (LMR-IB) | 2.6 0.5 | By foot |
La Miel River La Palmera-San Miguel (LMR-SM) | 35.5 1.6 | By boat |
Data were analyzed descriptively, classifying them by record type (sighting, faeces, track or dens), estimating the frequency of use by sites, areas, and years via contingency tables (Pearson’s χ2). We also performed two geostatistical analyses, the optimized hot spot analysis (Getis-Ord Gi) and the cluster and outlier (Anselin local Moran’s I) analysis via ArcGIS 10.7 (ESRI, 2018), to define statistically significant areas (p < 0.05) with high and low record aggregations (hot and cold spots) within a threshold (5 km), as well as, specific sites with high and low spatial autocorrelation (high - high and low - low clusters).
Results
We sampled a total of 875 km in a five-year span across the three main waterscape areas (upstream -103 km, reservoir -582 km, and downstream -190 km). Transects by foot and by boat were in average 2.9 ± 1 km and 75.9 ± 45.4 km, respectively across areas and years (Table 2). These variations in transect lengths across space and time were mainly due to dam water oscillations (~30 m between the highest and lowest water mark) caused by precipitation (weather season) and the mechanical water release and storage depending on the hydroelectrical power requirements, which influenced walkable shorelines availability.
2014 | 2015 | 2016 | 2017 | 2018 | ||||||||||||
RA | SS | N | SL (km) | N/km | N | SL (km) | N/km | N | SL (km) | N/km | N | SL (km) | N/km | N | SL (km) | N/km |
US | LMR-SB | 37 | 3.0 | 12.3 | 74 | 3.0 | 24.7 | 31 | 2.4 | 12.9 | 1 | 2.5 | 0.4 | 9 | 1.4 | 6.6 |
TR | 12 | 3.2 | 3.8 | 34 | 3.3 | 10.3 | 24 | 2.4 | 10.0 | 5 | 2.5 | 2.0 | 10 | 1.9 | 5.2 | |
MR | 1 | 5.0 | 0.2 | 26 | 5.0 | 5.2 | 1 | 5.0 | 0.2 | 6 | 5.0 | 1.2 | 5 | 5.0 | 1.0 | |
MrR | 33 | 2.7 | 12.4 | 3 | 3.3 | 0.9 | 20 | 1.7 | 11.8 | 11 | 3.1 | 3.5 | 6 | 1.5 | 4.0 | |
SLR | 26 | 1.8 | 14.2 | 3 | 3.2 | 0.9 | 16 | 3.4 | 4.7 | 11 | 3.3 | 3.3 | 6 | 1.9 | 3.2 | |
FS | 16 | 3.5 | 4.6 | 41 | 3.3 | 12.4 | 24 | 2.8 | 8.6 | 19 | 2.0 | 9.5 | 1 | 1.2 | 0.8 | |
SBS | 10 | 3.3 | 3.0 | 10 | 3.5 | 2.9 | 35 | 2.3 | 15.2 | 2.3 | 0.0 | 2 | 1.9 | 1.0 | ||
R | AR | 223 | 142 | 1.6 | 289 | 142 | 2.0 | 137 | 96 | 1.4 | 26 | 101 | 0.3 | 58 | 101 | 0.6 |
DS | LMR-IB | 10 | 2.3 | 4.3 | 4 | 2.3 | 1.7 | 6 | 3.1 | 1.9 | 13 | 3.1 | 4.2 | 17 | 2.1 | 8.1 |
LMR-SM | 40 | 38 | 1.1 | 20 | 35 | 0.6 | 51 | 34.2 | 1.5 | 36.1 | 0.0 | 9 | 34 | 0.3 | ||
Total | 408 | 204.8 | 2.0 | 504 | 203.9 | 2.5 | 345 | 153.3 | 2.3 | 92 | 160.9 | 0.6 | 118 | 146.7 | 0.8 |
We registered a total of 1 496 records from which faeces had the highest number (95.4 %, N = 1 467), followed by sightings, dens (1.7 %, N = 26 and 25, respectively), and tracks (1.3 %, N = 19). The number of records varied across sampling years, with the highest presence occurring during 2015 (reservoir = 163 > upstream = 116 > downstream = 14), followed by 2014 (reservoir = 121 > upstream = 107 > downstream = 29), 2016 (upstream = 93 > reservoir = 64 > downstream = 26), 2017 (upstream = 44 > reservoir = 21 > downstream = 19), and 2018 (reservoir = 32 > upstream = 29 > downstream = 16). However, considering only faeces (the most abundant record) standardized by km, the highest frequency of use was reported upstream in LMR-SB (29.6 faeces/km), followed by FS (20.5 faeces/km), TR (13.6 faeces/km), and MR (10.4 faeces/km) in 2015. In contrast, we did not register any records across SBS (upstream) and LMR-SM (downstream) in 2018. Likewise, the downstream area (LMR-IB and LMR-SM) showed the lowest frequency of use during the whole study (2014-2018; Table 2). We found significant differences by riverscape areas, sampling periods, and years with respect substrate preference (rocks vs trunks) being the former preferred over the latter in all cases (rock N = 813 vs. trunk N = 501; Pearson’s χ2 = 988.10 and 300.60, d.f. = 144 and 108, p < 0.001; Fig. 2).
We found the highest number of sightings in the downstream waterscape area at LMR-SM (N = 11) followed by the AR (N = 2). Conversely, FS (N = 2) had the highest number of sightings upstream followed by LMR-SB and SLR (N = 1). The number of sightings was also higher in 2015 with 10 observations in the reservoir. Finally, we did not record any sighting across the studied area in 2014. Regarding dens, we registered a total of 25 over the study period with the highest records found upstream in LMR-SB (N = 4) and SBS (N = 2) in 2014, AR (N = 2) in 2015, LMR-SM (N = 3) and AR (N = 1) in 2016, and FS (N = 1) and LMR-SM (N = 2) in 2017. The average dimensions of dens were 2.7 ± 2.7 m high and 2.9 ± 2.9 m wide with a horizontal distance to the water line of 2.6 ± 1.2 m, a canopy cover of 46 ± 27.7 % and a slope of 41 ± 15.5 °.
An initial exploratory analysis showed a large agglomeration of data at the upstream and reservoir waterscape areas compared with the downstream waterscape area across all years. From these, 2017 had the highest number of sightings and the lowest number of faeces recorded comparatively speaking with other years (Fig. 3). Geostatistical analyses corroborate this exploratory analysis defining the upstream waterscape area as a hot spot (Gi Z-score = 4.46, p < 0.001) and the reservoir as a cold spot (Gi Z-score = -2.69, p = 0.01; Fig. 3). This means the upstream area had a higher number of records with a homogeneous variance which may indicate a constant presence of the species and a higher number of individuals inhabiting the area. This also is true for the reservoir (i.e., a cold spot), but with a much smaller number of records, indicating a constant presence of L. longicaudis with a lower number of individuals. Finally, even though downstream waterscape areas had a variety of records (faeces, tracks, and dens) and the largest number of sightings across the study, it was not statistically significant (Gi Z-score = -0.11, p = 0.48; Fig. 4), implying that we could not identify patterns in the number of records spatially registered, which can be explained by the use of this area as a passing by rather than a residential area.
A more in-depth spatial analysis (Anselin Local Moran’s I cluster and outlier analysis) showed higher spatial autocorrelation with the high number of records in the reservoir AR (high - high cluster; LMi Z-score = 5.63, p = 0.001) and high spatial autocorrelation with a low number of records upstream in LMR-SB and TR (low - low cluster; LMi Z-score = 2.12, p = 0.001). This means that even though the upstream waterscape area was defined as a hot spot area, it is not all but some sites such as LMR-SB and MR more likely to host resident populations of the Neotropical Otter. Interestingly, this analysis recovered a low - low cluster downstream (LMR-SM) that the hot spot analysis could not identify (Fig. 4).
Discussion
This is one of the first studies in Colombia that sheds some light on the spatial ecology, habitat, and intensity of use of the Neotropical Otter in an anthropogenically regulated system. Unfortunately, there are no available data regarding L. longicaudis ecological aspects in La Miel River basin before, during, or after damming that allow us to compare our results with. The only available data include anecdotal records and information from local fishermen across the river (Castaño & Corrales, 2010; Restrepo et al., 2018). Thus, we presented the first technically driven information regarding L. longicaudis collected La Miel I hydroelectric power dam and influence areas after 12-years of commercial operation.
We found that L. longicaudis preferred the upstream riverscape area based on a larger number of faeces and dens found, allowing us to identify this area as a hot spot (Fig. 4). This is likely due to the presence of habitat characteristics such as a rocky bed and forest coverage provided by riparian vegetation, which positively influences the species presence providing protection and shelter above water level to ensure cubs survival avoiding flooding and predators (Santos & Reis, 2012; Rheingantz et al., 2017; Rheingantz et al., 2018), positive influencing the establishment of permanent otter populations. Downstream riverscape areas (LMR-SB and LMR-IB) had the lowest frequency of positive records but also one of the highest number of sightings and dens recorded, which might be related to mobility in response to species requirements such as food availability and refuge. However, special attention should be paid to records found in this area since the downstream waterscape has a high degree of anthropogenic pressure, which can likely impact the reproductive role of the species due to den availability and riparian habitat are essential for the survival of Neotropical Otters (Trujillo et al., 2016).
The low intensity of use (frequency of faeces/km) registered in the reservoir (AR) could be derived from topographic and bathymetric features such as water depth, the width of the water body, flooding oscillation (~30 m between the highest and lowest water peak) and shoreline and rocks availability (Foster-Turley, Macdonald, & Mason, 1990; López-Casas, Jiménez-Segura, & Pérez-Gallego, 2014). It has been registered that L. longicaudis tends to prefer shallow over deep water bodies due mainly to depth issues, reducing otter’s fishing capacity (hunting efficiency) and prey availability (Neotropical Otters prefer prey with benthic habits; Foster-Turley et al., 1990, Somers & Nel, 2004; Kruuk, 2006; Pedroso, Marques, & Santos-Reis, 2014; Rheingantz et al., 2017), implying that systems as the Amaní reservoir (AR) are less suitable for Neotropical Otters than upstream and downstream waterscape areas.
Our analyses showed a selectivity pattern of use regarding the space across the study area likely influenced by no measured variables such as food availability, anthropogenic disturbance, and habitat conservation. Both the optimized hot spot analysis and the cluster and outlier analysis showed a more constant presence in higher numbers of Neotropical Otters in upstream areassuch as LMR-SB, MR, FS and TR (hot spots and high - high cluster areas). In contrast, areas such as the reservoir (AR, cold spot) and downstream (LMR-SM; low - low cluster) had a constant presence of Neotropical Otters, but with a comparatively lower number of individuals even though the drastic flooding oscillation (~30 m between the highest and lowest water peak) in the reservoir.
We could not find any statistically significant hot-cold spots downstream even though the great variety of records (faeces, tracks, and dens) and the highest number of sightings recorded in the area. However, we did identify a low - low cluster site (LMR-SM), which was highly related to the sector where dens were found. This is important to highlight because we do not know how changes in the river dynamic have impacted Neotropical Otter populations in this area. Unfortunately, no data regarding Neotropical Otter populations were collected prior to the dam’s construction, so we cannot confidently say whether the low number of records reported downstream derived as an effect of the change in the water dynamic (lotic-to-lentic transition). Molecular approaches using codominant markers such as microsatellites and SNPs could bring information regarding gene flow and L. longicaudis population genetic structure that help to quantify this effect, which should be a priority to define management and conservation strategies.
Faeces counting has been a traditional non-invasive method used in numerous investigations because they are easily detected in the natural environment (Kasper, Bastazini, Salvi, & Grillo, 2008; Rheingantz et al., 2011; Franco et al., 2013; Biffi & Williams, 2017; Grajales-García et al., 2019). Faeces are generally deposited in conspicuous places along the shore of water bodies and can be clearly differentiated from those dropped by other carnivores (Delibes-Mateos, Díaz-Ruiz, Caro, & Ferreras, 2014; Juarez-Sanchez, Blake, & Hellgren, 2019). However, the detection probability of faeces decreases in prolonged rains periods since they can be easily rinsed-off (Medina-Vogel & González-Lagos, 2008; Delibes-Mateos et al. 2014). Across our five-year study, we had high precipitation values in the last two years influenced by the “La Niña” phenomenon, which likely influenced the number of faeces recovered in our study. This implies the low numbers recorded in 2017 and 2018 could not derive from a reduction in the number of individuals present in the study area but rather a sampling bias caused by high precipitation levels.
Riversides areas such as LMR-SM are exposed to greater anthropogenic pressure due to livestock and agricultural activities. However, based on our findings, these areas are still used by L. longicaudis, corroborating what has been reported in the literature regarding Neotropical Otters tolerating certain degrees of disturbance (Kruuk, 2006; Navarro-Picado et al., 2017; Rheingantz et al., 2017; Rheingantz et al., 2018). Similarly, the LMR-IB is another site highly impacted by river flow change, having a low flow rate that increases slightly during the rainy season. This variable could have a long-term direct impact upon food availability for the Neotropical Otter, which could permanently influence the establishment of the species since it does not meet the optimal conditions and resources necessary for its survival.
Anthropogenic activities such as the construction of hydroelectric dams can significantly affect Neotropical Otter populations also generating drastic reductions in prey availability and local diversity due to habitat modification and physicochemical changes of water (Santos & Reis, 2012; Palmeirim et al., 2014; Pedroso et al., 2014; Calaça, Faedo, & de Melo, 2015; Rheingantz et al., 2018; Costa-Braga, Rossi, & Srbek-Araujo, 2019). Therefore, long-term monitoring is required to design sound conservation measures that can reduce impacts derived of these megaprojects, protecting the biodiversity as well as ecosystem functions (see Bresney, 2016; Angarita et al., 2018; IHA, 2019).
Ethical statement: Authors declare that they all agree with this publication and made significant contributions; that there is no conflict of interest of any kind; and that they followed all pertinent ethical and legal procedures and requirements. All financial sources are fully and clearly stated in the acknowledgements section. A signed document has been filed in the journal archives.