Aims: Freshwater ecosystems have complex spatial and temporal connections that support multiple processes essential for life. Since two of the most critical impacts on riparian ecosystems is the regulation of river flows and water pollution, we aimed to investigate whether a dam and mining activities on the Ranchería River affect hydrochory.

Location: Tropical dry forest of La Guajira, Colombia; a region in need of water.

Methods: We evaluated hydrochory and compared seed richness, abundance and composition with the unaltered Cesar River over two seasons. We sampled both upstream and downstream of the dam and the mining site. At these same sites, we quantified levels of turbidity and 15 chemical compounds that affect water quality.

Results: There was significant spatial heterogeneity of the seed communities sampled along both rivers, with particularly distinct communities collected downstream from the open pit mine. We also found a significant effect of the dam on seed composition and abundance in the Ranchería River. Seed assemblages, but not seed numbers, differed between seasons and rivers. We also found significantly higher levels in chemical compounds and turbidity downstream of the mine compared to upstream.

Conclusions: There is a significant effect of the dam and the open pit mine on the Ranchería River in Colombia. The alteration of natural flows, environmental conditions and water quality affects hydrochory, a key ecological function of the river. Urgent measures are imperative to address these challenges effectively and ensure the long-term sustainability of the ecosystem.

Tropical dry forests are among the most floristically rich ecosystems and have high rates of endemism (Gentry 1982); however, they are also one of the most threatened in the world (Janzen 1988, Portillo-Quintero and Sánchez-Azofeifa 2010, IUCN 2012). It is estimated that about 54% of the area covered by this ecosystem is distributed in South America (Ceccon 2001, Portillo-Quintero and Sánchez-Azofeifa 2010, Blackie et al. 2014). In Colombia, this ecosystem crosses six bioregions that vary widely in climate, soils, and anthropogenic pressures; four of these regions are in a critical state of fragmentation, with an average maximum patch area of only 100 ha (Ariza et al. 2014, Pizano et al. 2017). One of these, the La Guajira region, has a long history of anthropogenic alteration; since colonial times, the establishment of large and intensive crops promoted land use change and the alteration of rivers (Guerra Curvelo and Egurrola Hinojosa 2015). The Ranchería River, one of the most important in the region, has been one of the most affected since colonization and has since undergone repeated modifications (Guerra Curvelo and Egurrola Hinojosa 2015). Currently, due to severe anthropic changes, the river’s flow has decreased sharply, causing several economic, social and environmental problems (Guerra Curvelo and Egurrola Hinojosa 2015).

Coal mining activities began 46 years ago in the region, transforming and threatening human and environmental health (Urrea and Calvo 2014, Moreno Rodríguez and Montero Torres 2016, Galvis 2018) by affecting ecosystem services (Environmental Justice Atlas 2015). One of the critical problems associated with open pit coal mining is acid mine drainage (AMD). Mineral exploitation generates a high level of dissolved solids, such as bicarbonate, chlorides, sodium calcium sulfur, magnesium and iron in drainage waters (Akcil and Koldas 2006, Goswami 2015). Additionally, there is an excess of nitrates and phosphates, nutrients strongly associated with the growth of algae, which generate a strong decrease in oxygen in the water (Castro et al. 1998, Talukdar et al. 2016). These impacts make the water extremely acidic and very high in minerals, thus affecting the biodiversity and life cycles of hundreds of organisms in wetlands near the mines (Castro et al. 1998, Akcil and Koldas 2006, Munnik 2010, Goswami 2015, Talukdar et al. 2016, Pat-Espadas et al. 2018). It has been reported that water acidification generates modifications in seed morphology by damaging specialized structures for buoyancy and viability, which also affects seed germination and establishment (Fulbright 1988, Fargašová 1994). Additionally, the decomposition of nutrients, microorganisms and mycorrhizae generates strong transformations in the ecological conditions of soils and water (Fulbright 1988, Fargašová 1994, Clark 1997). It has been reported that the accumulation of salts or heavy metals together with water acidification completely inhibit the seed germination process (Fulbright 1988, Fargašová 1994, Clark 1997, Pérez-Fernández et al. 2006, Reichman et al. 2006, Donggan et al. 2011, Pat-Espadas et al. 2018). Coal mining also has major physical (mechanical) effects, including altering and diversion of water courses and banks and the disconnection between surface and groundwater flows (Seybold et al. 2004, Anawar 2013). These can completely change water and soil conditions and the composition of seed species transported by the river (Ghose 2004, Pandey et al. 2014). This, in turn, affects seed establishment sites and leaves soils poorly drained, eroded and poor in organic matter (Ghose 2004, Seybold et al. 2004, Donggan et al. 2011, Anawar 2013, Pandey et al. 2014).

Sixteen years have passed since the diversion of the Rancheria River for the construction of the El Cercado dam, causing several environmental impacts including the fragmentation of the river network and the alteration of the flow regime, the generation of a new ecosystem (reservoir), and impeding the downstream movement of aquatic organisms, sediments and nutrients (Contraloría General de la República 2010, Corpoguajira 2020). Various studies report reduction in hydrochory and decreased floristic diversity downstream of dams (Nilsson et al. 1989, Nilsson and Jansson 1995, Jansson et al. 2000, Andersson and Nilsson 2002). For example, Brown and Chenoweth (2008) found a 90% reduction in abundance and an 84% reduction in seed species richness in sampling downstream of a dam; they also report that the germination of seeds collected downstream of the dam was very low. Likewise, the decomposition of biomass accumulated in dams can generate water toxicity and eutrophication of the ecosystem, affecting the germination of seeds downstream of the dam (Merritt and Wohl 2006). Therefore, based on the above, we hypothesized that damming and mining of the Ranchería River will affect the seed community composition flowing with the river; for example, a marked decrease in richness and abundance of seeds transported downstream of the dam and the mining area.

In order to test the effects of the anthropogenic alterations of the Ranchería River on hydrochory, 1. we assessed the composition, richness, and abundance of the seed community transported by the Ranchería, both upstream and downstream of a dam and of the coal mine;  2. we then compared these parameters to zones with similar physiographic features in the nearby Cesar River, which has undergone much less intense modification. We conducted these survey activities during both the dry and rainy season, which are highly contrasting in this tropical dry forest ecosystem. We expected distinct composition and higher richness and abundance of seeds in the Cesar than in the Ranchería River, with particularly marked differences at sites downstream of the dam and mine. Furthermore, 3. to complement ecological results and to investigate the effects of the coal mine on the physical and chemical characteristics of the Ranchería River water, we analyzed physical-chemical water parameters in this River. Understanding the role of rivers and mechanisms that determine the geographic range of seed dispersal is of vital importance to determine the intensity of human impacts on wetland transformation, and may point to ways to successfully restore biodiversity and their lost ecosystem services (Richter and Stromberg 2005).

There was significant spatial heterogeneity of the seed communities sampled along both rivers, with especially strong separation of the community collected downstream from the open pit mine. We also found a significant effect of the dam on seed composition and abundance in the Ranchería River. Seed community assemblages, but not seed numbers, were significantly different between seasons and rivers. We also found significantly higher turbidity, chloride, carbonate, iron, nitrate and sulfate levels downstream of the mine compared to upstream samples. There is a clear effect of the dam and the open pit* *mine on the Ranchería River in Colombia. The alteration of natural flows, environmental conditions and water quality triggers a radical change in the composition, abundance and structure of the riparian plant communities and affect key ecological functions of the river. Urgent measures are imperative to address these challenges effectively and ensure the long-term sustainability of the ecosystem.

2.2.1. The flow pattern of the Ranchería River

In order to characterize the flow pattern of the Rancheria River, we calculated mean monthly flows and total mean flows (m³s^-1) for each of four hydrological stations located along the Ranchería river for a period of 13 to 15 years (dataset: IDEAM 2023) according to data availability (see Fig. 2a, b). Hydrological stations were located close to our sampling zones (Appendix S1a, b). The location of the stations from upstream to downstream of the river were: Caracolí (located before the dam), El Silencio (located after the dam), Pte. Guajiro (located before the mining area), and Cuestecitas (located immediately after the mining area) (Appendix S1a, b).

2.3 Selection of sampling zones and sites

To evaluate the effect of the dam on seed dispersal in the Ranchería River, we collected seeds transported along the Cesar River, as well as from upstream and downstream sections of both the dam and mining zones of the Ranchería River. The Cesar River, unaffected by major anthropogenic modifications, was used as a control. Two sampling zones in the Cesar River were selected based on their similar physiographic traits -such as altitude, geographic and environmental characteristics, and orientation- to the upper and middle zones of the Ranchería River. In total, we designated five sampling zones: the Ranchería upper zone (UR), located upstream of the dam; the Ranchería middle zone (MR), located downstream from the dam: the Ranchería lower zone (LR), located downstream of the mine area; and two zones in the Cesar River, the Cesar upper zone (UC) and the Cesar middle zone (MC) (Fig. 1a; Appendix S1a). Within each zone, we chose three sampling sites (Fig. 1a, Appendix S1a), separated from each other by at least 200 m. We placed three seed traps at each site (i.e., nine traps per zone).

2.4 Seed sampling

Our seed traps were made following the model proposed by Vogt et al. (2004) and used in Esper-Reyes et al. (2018) from PVC pipe with a cloth funnel and cap. The cloth forming the funnel and the outer part of the cap consisted of synthetic fiber gauze with a mesh size of 0.5 mm, and the inner part consisted of gauze with a mesh size of 0.1 mm. Thirty traps were constructed as follows: each trap consisted of a section of PVC pipe measuring 15.24 cm in diameter × 30 cm long and a 7.62 cm PVC pipe coupling was fitted with caps at the end of each trap (Fig. 1b).

Two field trips were conducted to the study areas, one in July and the other in December 2019. The trips were planned based on logistical considerations to ensure the feasibility of carrying out sampling in the field. According to the climate regime, July is the driest month of the shorter dry season, and December is a transitional month between the end of the rainy season and the beginning of the longer dry season (Arregocés et al. 2024). Thus, in the subsequent analyses we used 'transitional' season for December and 'dry' season for July. The mean monthly flows obtained in hydrographs show a contrast between the two dates, with December having a higher river flow because of the end of the more extended rainy period of the year. At each sampling site the traps were placed hanging from rocks and surrounding vegetation, to remain hidden to prevent removal by local people. Traps were suspended so that half of the mouth of the trap was submerged in the river (Fig. 1b). These remained suspended in the river for 48 hours, after which the material collected was placed in resealable plastic bags. Likewise, we sampled the flowers/fruits of herbs, shrubs and trees that were within 50 m of the riverbank for approximately 100–200 meters. This material was pressed and stored for later identification, in order to be used as support for the identification of the seed species found in the traps. The material that was collected by the traps was placed in drying trays and kept in the sun for fourth days. After this time, rocks, leaves and large dry branches were discarded, and the seeds were separated by using a Hubbard #548 six-screen sieve. In some cases, such as Ficus spp., fruits were considered the dispersal unit rather than individual seeds. Finally, the selected material was placed in labeled paper bags for later identification at the Colombian National Herbarium, at the National University of Colombia, in Bogotá, Colombia. The material was dried in an oven until all the moisture was removed, on average for two days. Once dry, the seeds of each of the samples were separated using a stereomicroscope. Subsequently, these were grouped into similar morphospecies and identified to the finest possible taxonomic level, with the help of taxonomists and the support of specimens from the fruit and seed collection of the Colombian National Herbarium (COL).

2.5 Water sampling

On November 24, 2020, we collected water samples from the Ranchería River at two sampling points within each of the three zones (UR, MR and LR), with the sampling points spaced more than 200 m apart. Additionally, a fourth zone upstream of the mine (U-mine) was included to allow for a more thorough evaluation of the mine´s impact. All samples were kept cold and taken to the laboratory and analyzed the day after collection. A total of fifteen chemical parameters of the water were analyzed in the laboratory following the regulations of the Institute of Hydrology, Meteorology and Environmental Studies of Colombia, IDEAM (Nancy Flórez Garcia S.A.S. Laboratories; IDEAM Accredited Laboratory; Appendix S2). To measure water turbidity—an indicator of how much light is absorbed by suspended particles or particulate contaminants—we used a highly accurate portable HI 98703 Hanna Instruments Turbidimeter, which meets or exceeds EPA standards (USEPA Method 180.1 for wastewater and Standard Method 2130 B for drinking water). The instrument measures the turbidity of a sample in the range of 0.00 to 1000 nephelometric turbidity units (NTU), with NTU indicating the number of particles per unit volume, or a specific light scattering coefficient.

2.6 Statistical analysis

2.6.1. The seed community transported by the rivers

We determined the composition, richness and abundance of the seeds dispersed by the Ranchería and Cesar Rivers in the sampling zones detailed above. Species rank-abundance curves were generated to observe the relationships between richness, relative abundance and evenness of the collected seed species. Rank-abundance curves can be fitted to various types of models (Wilson 1991). This analysis is useful not only because they help reveal similarities and/or distinctions of seed communities between sampling sites and climatic seasons, but also dominance, and this is likely to have different effects on the dynamics of riparian ecosystems. To evaluate the fit of the different curve models, we used the “radfit” function of the “vegan” package in the R statistical program (Oksanen et al. 2022).

To assess the relationships between seed community composition, zone, season and river, we performed non-metric multidimensional scaling (NMDS) ordinations on Bray-Curtis dissimilarities, permuted analysis of variance (PERMANOVA) and permuted multivariate analysis using the functions metaMDS, adonis2 and betadisper, respectively from the vegan package in R (Oksanen et al., 2022; R Development Core Team, 2023). The square root-transformed abundance data was used as the input matrix. Plant species with < 3 seeds were not included in the analysis. Bray-Curtis dissimilarity matrices were visualized using non-metric multidimensional scaling (NMDS) plots. For PERMANOVA analyses, the occurrence of 0 seeds for some sites precludes a hierarchy analysis using constraining permutations with the adonis2 procedure. To overcome this limitation, we incorporated into the data matrix a “dummy species” (i.e., to set a minimum abundance of one for all sites), as recommended by Clarke et al. (2006).

2.6.2. Spatial variation of seed transport in relation to sampling zones

PERMANOVA was used to test for significant differences in seed communities collected from the rivers across five sampling zones, with permutations constrained by season.

2.6.3. Effects of the dam on seed communities transported by the river

To evaluate the impact of the dam on the seed community of the Ranchería River, we used a PERMANOVA nested design considering four sampling zones (the upper and middle zones of the Cesar and Rancheria rivers). We assessed the effect of river (Cesar, Ranchería), season (dry, transitional), zone (upstream and midstream, with the latter representing downstream of the dam for Rancheria River), and the two-factor interactions: season×river; season×zone, and river×zone. We assumed that a significant river×zone effect would indicate an effect of the dam on hydrochory. The permutation was constrained with the strata option in adonis2 procedure by a “plot” factor with four levels resulting from the river/zone combination (i.e., Cesar-upper, Cesar-middle, Ranchería-upper, Ranchería-middle). The test was performed with 9999 permutations.

2.6.4. Effects of the dam on seed abundances transported by the rivers

We constructed a mixed nested model to evaluate the impact of river (main effect) and zone (subplot effect) on the total abundance of seeds, representing the cumulative sum across all species. The model incorporated a 'plot' factor as a random effect, comprising four consecutive levels (1 to 4), each corresponding to unique combinations of season and river (transitional-Cesar, transitional-Ranchería, dry-Cesar, dry-Ranchería). Additionally, season was included as a blocking factor in the model. This approach allowed for a comprehensive examination of the effects of river, zone, and their interactions while controlling for seasonal variations. As indicated in the multivariate analysis, it was hypothesized that if the dam influenced seed transport, a statistically significant river×zone interaction would be anticipated. To find the best (simplest) model we used backwards stepwise selection, starting with the most complex model (including all terms and the specified interaction) and dropping each term sequentially, retaining only terms whose removal significantly worsened the model fit (p < 0.05). The total number of seeds (the response variable) was first transformed to square root of y plus 0.5 to meet the assumptions of normality and homogeneity of variance of the model residuals. To perform the mixed model, we used the nlme library of R including the varPower option to model the variance (Pinheiro and Bates 2023). In all cases, we report mean and standard error values for untransformed data.

2.6.5. Effects of the mine on seed communities transported by the river

In order to evaluate the potential effect of the mine on the seed community composition we performed pairwise comparisons (PERMANOVA) of the sampling zone downstream of the mine on the Ranchería River to the upstream and midstream sampling zones on both rivers. To control family-wise error rate following multiple comparisons, P-values were adjusted using the Benjamini-Hochberg procedure (Benjamini and Hochberg 1995). All permutation tests were carried out with 999 permutations.

2.6.6. Effects of the mine on seed abundances transported by the Rancheria River

We then assessed whether there was a difference in the total number of seeds transported by the Rancheria River between the two seasons upstream versus downstream of the mine in the Ranchería River, using an ANOVA (Type III sum of squares) for the two main factors and their interaction. Prior to analysis, we transformed the response variable by square root + 0.5 to achieve normality and homogeneity of variances. To fit the statistical model, we used ‘car’ package in R statistical software (Fox and Weisberg 2019).

2.6.7. Effect of the mine and the dam on water quality

To test whether there were effects of the mine and the dam on water quality, we used a PERMANOVA analysis with a planned orthogonal comparison model, applying the reverse Helmert coding (i.e., comparing each level with the mean of the previous levels: downstream of the mine (LR) vs mean of the three upstream sampling zones (UR, MR, U-mine); upstream of the mine (U-mine) vs mean of upstream and downstream of the dam (UR, MR), and downstream (MR) vs upstream of the dam (UR). To perform the analysis, we calculated a Euclidean distance matrix on the data transformed by log(x+1). The PERMANOVA was performed with 9999 permutations. To visualize the arrangement of chemical water compounds, Principal Component Analysis (PCA) was employed. Before PCA, sample adequacy was confirmed through the Kaiser-Meyer-Olkin (KMO) Test for Sampling Adequacy. For PCA analysis we used prcomp function in the “stat” package and biplot from the “vegan” package in the R statistical program.