The study was conducted in the Argentinian Gran Chaco (Fig. 1), a vast and heterogeneous lowland region characterized by predominantly flat terrain, with elevations ranging from sea level in the east to nearly 2,900 m a.s.l. in the southwest (Naumann 2006). Due to complex climatic, edaphic, and vegetational gradients, the region is broadly divided into two major ecoregions: the Dry Chaco and the Humid Chaco (Olson et al. 2001).

In the Dry Chaco (Fig. 1A, B), the xerophilous and semi-deciduous forests reach their greatest extent and continuity, harbouring the highest representation of characteristic species such as A. quebracho-blanco and Schinopsis lorentzii (Griseb.) Engl. that dominate the upper tree layer. This forest also includes other lower-growing trees such as Sarcomphalus mistol (Griseb.) Hauenschild, as well as a wide variety of trees and shrubs, with a notable presence of “algarrobo” species (Neltuma spp.). These areas support drought-adapted forests, which decreases in height and species richness towards the south (Naumann 2006). The central-northern Dry Chaco comprises a sedimentary plain dissected by the Juramento-Salado-Pasaje, Bermejo, and Pilcomayo rivers, which flow from northwest to southeast and form shallow valleys flanked by lateral ridges (Erize et al. 1993). Originating in the Andean foothills, these rivers form fluvial megafans, extensive floodplains and wetlands (Orfeo 1999; Suárez et al. 2010; Venturini and Krepper 2013; Thalmeier et al. 2021). The Pilcomayo River is noted for its extremely high sediment load and seasonal fluctuations, leading to pronounced sediment deposition, avulsion, and channel infilling, all contributing to a dynamic megafan landscape (Lamenza et al. 2019; Fernández et al. 2025). The climate is subtropical continental and features South America’s thermal maximum (heat pole), recording some of the highest temperatures on the continent (Moglia and Giménez 1998; Naumann 2006). In the southwestern Dry Chaco, large endorheic basins such as the Salinas Grandes—one of the largest salt flats in South America (Dargám 1994)—and the Salí-Dulce System, which drains into the Mar Chiquita Lake, support a vast wetland complex (Naumann 2006; Bucher 2019). High evaporation rates lead to soil salinization, strongly influencing vegetation patterns. Nonetheless, diverse shrublands occur, and less saline soils support typical xerophytic Chaco forests. The climate remains warm and subtropical. The western boundary of the Dry Chaco follows the Sub-Andean and Pampean Sierras, where topographic heterogeneity creates favourable microclimatic conditions for subtropical forests that transition into montane grasslands and high-altitude vegetation.

The Humid Chaco is characterized by a warm subtropical climate. Due to the predominance of impermeable or poorly permeable soils, the region experiences extensive seasonal flooding, driven by both local rainfall and overflows from major rivers. Its hydrology is dominated by the Paraná and Paraguay rivers. The Paraná River forms a complex, multi-channel alluvial system with meandering thalweg and a sandy bed (Scarabottia et al. 2025). The Paraguay River supports one of the largest continuous continental floodplains in South America (Wantzen et al. 2024). Despite apparent geological homogeneity, extreme topography, and a relatively uniform climate, the Humid Chaco supports a wide array of vegetation types at small spatial scales, including hydrophilic forests, savannas, wetlands, and xerophytic shrublands (Naumann 2006). The “Monte Fuerte” or “Quebrachal” constitutes the most important floristic community. This forest is dominated by Schinopsis balansae Engl. and, to a lesser extent, A. quebracho-blanco. Overall, it is a semi-xerophilous forest with deciduous trees, occurring mainly in high and intermediate topographic positions.

Aspidosperma quebracho-blanco is a long-lived, slow-growing tree species that can reach up to 20 meters in height. The species produces abundant pollen and anemochorous (wind-dispersed) seeds. Juvenile individuals often remain in an immature vegetative phase for over a decade (Barchuk and Díaz 1999, 2000). Flowers are protandrous, which helps reduce self-pollination, though selfing can still occur. Flowering begins in late winter and may continue through late summer, with pollination primarily mediated by nocturnal moths (Lin and Bernardello 1999). Fruits mature approximately one year after ovule fertilization and release seeds through dry, dehiscent capsules. Although seeds are short-lived, they exhibit high germination rates during the first few weeks after dispersal. In addition to sexual reproduction, A. quebracho-blanco also propagates clonally via epirrhizal shoots, contributing to its persistence in heterogeneous environments (Barchuk et al. 2006; Torres Basso 2014).

The genetic dataset used in this study includes 20 populations sampled throughout the Argentinian Gran Chaco (Suppl. material 1), originally published in Almirón et al. (2022). This dataset was constructed with four primer combinations used for AFLP analysis and generated a total of 505 bands in the 600–1500 bp range.

Individuals were regrouped according to watersheds and watershed systems (Table 1). To assess genetic variability, the following descriptive statistics were calculated for each group: total number of bands (TNB), number of exclusive bands (NEB), number of alleles (Na), effective number of alleles (Ne), expected heterozygosity with the correction factor for small samples for dominant markers (uHe), Shannon index (Sh), percentage of polymorphic loci (%P). All genetic variability data were processed using GenAlEx v.6.3 (Peakall and Smouse 2012). Polymorphic information content (PIC) was calculated as PIC = 2ƒ(1 - ƒ) (Roldán-Ruiz et al. 2000), where ƒ represents the frequency of present bands in the gel and 1 - ƒ the frequency of absent bands. PIC was averaged over the fragments for each primer combination. For dominant markers, PIC values range from 0 (monomorphic) to 0.5 (highly informative) and are classified as low (0–0.10), moderate (0.10–0.25), high (0.25–0.40), and very high informativeness (0.40–0.50), according to Serrote et al. (2020). The pairwise genetic differentiation among watersheds and watershed systems was estimated using analysis of molecular variance AMOVA (Weir and Cockerham 1984; Excoffier et al. 1992) based on Rho_ST estimates (Meirmans and Van Tienderen 2013). Both calculated in Genodive v.3.05 (Meirmans 2020). The significance of Rho_ST for individual loci was tested by 999 allele permutations among samples. Also, gene flow (Nm) was estimated by software program POPGENE v.1.31 (Yeh et al. 1999).

To assess the influence of geographic distance on genetic variability, the relationship between Nei’ genetic distance (Nei 1973; Suppl. material 2) and geographic distance (km-based; Suppl. material 3) was analysed using a Mantel test (Mantel 1967). This test was performed in GenAlex v.6.3 (Peakall and Smouse 2012), with 9,999 permutations.

Potential barriers to gene flow between populations were identified using Barrier v.2.2 (Manni et al. 2004), which employed Monmonier’s maximum distance algorithm (Monmonier 1973). This program uses the Voronoï tessellation that represents a polygonal neighbourhood for each sample (population) constituted of those points, on the plane, that are closer to such sample than to any other one. Then the algorithm chooses the edge with the largest associated distance in this triangulation network and starts with an edge that extends across the next adjacent edge associated with the largest distances. Additional sections are added to the barrier by repeating the steps until the barrier reaches the outer edge of the network or meets another barrier (Manni et al. 2004). For the analysis of the barriers, matrix pairwise populations Rho_ST were employed. The detected barriers were then compared with topographic features, rivers and watersheds within the Gran Chaco using QGIS v.3.16.15 (QGIS Development Team 2021).

To assess the spatial distribution of genetic diversity, the Shannon index (Sh) values per population from Almirón et al. (2022; Suppl. material 1) were used. The Shannon index was selected as it is particularly sensitive to expected variability loss due to small population sizes and colonization events. Compared to other metrics such as expected heterozygosity (He) and allelic richness, the Shannon index provides more informative insights for conservation genetics, especially when comparing populations with identical allele counts (Sherwin et al. 2017).

Also, a horizontal interpolation analysis was performed and then was compared with Gran Chaco watersheds and watershed system using QGIS v.3.16.15 (QGIS Development Team 2021). This approach generated a 3D surface graph based on a Delaunay triangulation-based connectivity network among populations (Brouns et al. 2003). The graph was constructed using Alleles in Space (AIS) v.1.0 (Miller 2005). Genetic Landscape Interpolation Editor / Viewer using a distance weighting parameter α = 1 and a grid area of 50 × 50. In this graph, the X and Y axes represent geographic coordinates and the Z axis corresponding to genetic distances between sample points. Pairwise individual genetic distance was calculated for AIS as the percentage/proportion of mismatched loci between individuals i and j in a data set. Specifically, the genetic distance between individuals i and j is obtained from

D i j = ∑ k = 1 n d k n

where dk takes a value of 1 if individuals i and j contained different alleles at locus k and 0 if individuals contained the same allele at locus k.

Finally, to evaluate the impact of climate variables on A. quebracho-blanco populations, a Principal Component Analysis (PCA) was performed using Infostat v.2017 (Di Rienzo et al. 2017). This analysis incorporated genetic population data, climate variables previously selected by Almirón et al. (2022) and elevation (m a.s.l) (Suppl. material 1). For the variable selection, the original environmental dataset was subjected to a Principal Component Analysis, and the scores of each component were used as new predictor variables. Components explaining more than 95% of the total variance were retained. A pairwise Pearson correlation analysis was then applied to identify variables with correlations of r ≥ 0.75 and to keep eight uncorrelated bioclimatic variables: BIO 2 (Mean diurnal range), BIO 3 (Isothermality), BIO 5 (Maximum temperature of warmest month), BIO 6 (Minimum temperature of coldest month), BIO 8 (Mean temperature of wettest quarter), BIO 14 (Precipitation of driest month), BIO 18 (Precipitation of warmest quarter), and BIO 19 (Precipitation of coldest quarter). The elevation data was obtained from the respective layer from WorldClim v.2.1 (Fick and Hijmans 2017; Suppl. material 1).

The average PIC values obtained for the four primer combinations were 0.23 for EcoRI-AAG/MseI-AAA, 0.18 for EcoRI-AAG/MseI-AAT, 0.16 for EcoRI-ACA/MseI-AAG, and 0.32 for EcoRI-AAG/MseI-AAG. The analysis of genetic variability revealed highest values in the Corrientes River, areas of rivers and streams in Salta and Formosa tributaries of the Paraguay River, and several watersheds of Salinas Grandes. In contrast, the Pasaje or Salado River watershed exhibited the lowest genetic variability. Exclusive bands were detected in the Corrientes and several watersheds of Salinas Grandes. At the watersheds system level, the Paraguay River System showed the highest levels of genetic variability, while the Mar Chiquita System had the lowest values. Exclusive bands were also observed in the Paraná and Serrano systems (Table 1).

Rho_ST coefficient values indicated moderate genetic structure among watersheds and low genetic structure among watershed systems (Table 2). Among watersheds, the Santa Lucía and Pasaje or Salado rivers exhibited the highest pairwise genetic differentiation (Rho_ST = 0.27), whereas the Paraná and Middle and Lower Bermejo rivers exhibited the lowest (Rho_ST = 0.02; Suppl. material 4). Gene flow (Nm) ranged from 2.30 (Santa Lucia and Pasaje or Salado Rivers) to 12.05 (Salí-Dulce River and Oriental Skirt of Ambato) values for watersheds (Suppl. material 5) and 7.06 (Paraguay and Mar Chiquita) to 17.26 (Paraná and Serrano) values for watershed System (Suppl. material 6).

The Mantel test revealed significant positive correlation between Nei’ genetic distance (Suppl. material 2) and geographic distances (Suppl. material 3) across the studied A. quebracho-blanco populations (R = 0.175; p < 0.001).

Using Monmonier’s algorithm, eight putative barriers to gene flow were identified across the study area (Fig. 2), with four located in each ecoregion the Humid Chaco and the Dry Chaco. In the Humid Chaco, Barrier A corresponds to the Santa Lucía River, separating populations 1 (Corriente River watershed) from population 2 (Santa Lucía river watershed). Barrier B aligns with the Paraná River and isolates population 3 (Middle Paraná River watershed) from the remaining populations of the Paraná River System. Both Barriers C and D isolate population 4 (areas without major rivers or streams) from other Humid Chaco populations and from those of the northeastern Dry Chaco populations, respectively. These barriers are in northeastern Argentina. Notably, in the Dry Chaco, in the north of Argentinian Gran Chaco, Barrier D follows the course of the Teuquito stream, a part of the Bermejo River megafan, separating populations 8 and 9 (Middle and Lower Bermejo watershed) from populations 5, 6, and 7 (rivers and streams of Salta and Formosa watershed). In the centre of Argentina, Barrier E is associated with Dulce River and separates populations 10, 11, and 12 (Salado-Pasaje River watershed, Paraná River System) from central-western populations. Barrier F coincides with Ancasti hills and Barrier G aligns with the Salinas Grandes, isolating population 16. Finally, Barrier H coincides with the Sierras Chicas, separating populations 18 and 19. These last barriers were in the southwest of the Argentinian Gran Chaco.

Horizontal interpolation analysis (Fig. 3) revealed areas of genetic divergence among populations from different watersheds and watershed systems. In the western region, genetic divergence was observed between populations 8 and 9 from 10 to 12—associated with the Pasaje or Salado River watershed of the Paraná River system. Populations from the Serrano system, specifically 16, 17, and 20 (located on the eastern hillside of Ambato watershed), were different from populations 18 and 19 from the Salinas Grandes watershed. In the northeast and eastern regions, divergence was detected between all populations analysed: 1 (Corriente River), 2 (Santa Lucía River), 3 (Middle Paraná River), and 4 (areas without major rivers or streams) of the Paraná River System. In turn, these populations are different from those of the Paraguay system. In the last system, greater genetic distances were observed between populations 5, 6, 7 (areas of rivers and streams of Salta and Formosa) from population 8 and 9 (Middle and Lower Bermejo).

The principal components analysis (PCA) revealed that the first two components accounted for 78.7% of the total variation (Suppl. material 7). The contribution of each bioclimatic and topographic variable to the components is provided in Suppl. material 8. The first principal component, which explained 44.6% of the variance, is positively correlated with precipitation in driest month (BIO 14), warmest quarter (BIO 18), and coldest quarter (BIO 19), and negatively correlated with average diurnal temperature range (BIO 2) and elevation (m a.s.l.). Lower PC1 values were associated with populations from Bayesian groups I (populations 10–17 and 20),and II (populations 18 and 19), as well as group III populations from the Dry Chaco (populations 5–9), all of which exhibited medium to low genetic variation. In contrast, higher PC1 values correspond to group III populations in the Humid Chaco (populations 1–4), which showed higher levels of genetic variability. The second principal component (PC2), accounting for 34.1% of the variance, was positively correlated with isothermality (BIO 3), maximum temperature of warmest month (BIO 5), and mean temperature of wettest quarter (BIO 8). Populations from Bayesian groups I (western Dry Chaco) and III (northern and central Dry Chaco) tended to occupy the upper section of the PCA plot, whereas population 17 (group I), populations 18 and 19 (group II, southern Dry Chaco), and group III populations from the Humid Chaco (populations 1–4) were associated with lower PC2 values.

Genetic variability and structure are key indicators of evolutionary processes in natural populations and provide a basis for defining conservation units and guiding management strategies from an evolutionary perspective (Williams et al. 2019; Hoban et al. 2020). The PIC values obtained indicated moderately to highly informative markers. However, despite regrouping the populations into watersheds and watershed systems, genetic variability estimates for each watershed remained similar to those previously reported for the corresponding A. quebracho-blanco populations (Almirón et al. 2022). Since overall genetic variability values changed little, the new conservation and restoration recommendations proposed here, relative to previous findings (Almirón et al. 2022), are based primarily on the observed differences in genetic structure.

Genetic structure analysis revealed a moderate genetic differentiation among watersheds (Rho_ST = 0.15), similar to previous estimates obtained at the population level (Rho_ST = 0.20). In contrast, lower levels of genetic differentiation were detected when populations were grouped into broader spatial units, such as watershed systems (Rho_ST = 0.07), ecoregions (Rho_ST = 0.03), or Bayesian genetic groups (Rho_ST = 0.10; Almirón et al. 2022). Although these values indicate low to moderate divergence and are statistically significant these groupings are less meaningful. In contrast, these findings suggest that watersheds may represent ecologically meaningful units and therefore be considered in conservation and restoration planning for the Gran Chaco forest. The relatively high Nm values may indicate the existence of historical gene flow among watersheds. However, Rho_ST values indicated moderate genetic differentiation, while the Mantel test, Bayesian clustering, and horizontal interpolation analyses supported an isolation-by-distance pattern. This apparent discrepancy may reflect the assumptions underlying Nm estimation and its indirect relationship with genetic differentiation. Furthermore, migration estimates are temporally limited and may fail to capture rare but biologically important events (Whitlock and McCauley 1999), including anthropogenic disturbances that have affected the Argentinian Gran Chaco during the last century or recent climatic changes. Therefore, conservation recommendations were based primarily on indirect estimates of genetic structuring.

Spatial genetic structure is shaped not only by the geographical distribution and distance among populations, but also by landscape features that act as barriers or corridors to gene flow (Trizio et al. 2005). Elements such as roads, urban areas, climatic conditions, vegetation cover, mountain ranges, and rivers can either impede or facilitate the movement of pollen and seeds, thereby influencing the extent and spatial pattern of genetic differentiation (Guillot et al. 2009). In this sense, studies conducted in Argentina, even when focused on herbaceous species, have shown that geomorphological and historical landscape features are key drivers of genetic structure, highlighting the pervasive role of geomorphology in shaping genetic variation across different plant life forms (Cosacov et al. 2010; Moreno et al. 2018; Baranzelli et al. 2022). In this study, the low but significant positive correlation between genetic and geographic distances found in the Mantel test indicates a pattern of isolation by distance, consistent with earlier studies on A. quebracho-blanco (Torres Basso 2014; Almirón and Solís Neffa 2024). However, at a given spatial scale, when migration rates are extremely low, only a very limited fraction of total genetic variation is expected to be spatially autocorrelated, as populations become genetically differentiated regardless of the geographic distance separating them. This limited spatial autocorrelation typically results in low Mantel’s r values, even when gene flow is restricted (Quilodrán et al. 2025).

Additional ecological and demographic factors may further modulate gene flow in A. quebracho-blanco. Population density and flowering phenology, for instance, can influence pollinator movement and consequently affect pollen dispersal distances (Ghazoul 2005). Moreover, the species’ prolonged fruit maturation period—lasting nearly one year—may be sensitive to water and temperature stress, potentially reducing seed viability under adverse climatic conditions (Ghazoul 2005; Alzugaray et al. 2007). Successful seedling establishment often depends on the presence of nurse plants that mitigate extreme environmental conditions, although interspecific competition during drought periods may inhibit recruitment (Barchuk et al. 2005).

In the genus Aspidosperma, diaspore dispersal is generally limited: although some species may reach distances greater than 100 m in continuous forests, most seeds do not travel beyond 30 m from the mother plant (Augspurger 1986; Menezes et al. 2024), the average dispersal distances recorded for other species in the genus, remain relatively low. Therefore, gene movement via seeds is expected to decline rapidly with distance, promoting spatial genetic structuring.

Likewise, although A. quebracho-blanco relies on mobile pollinators—mainly nocturnal moths (Noctuidae, Pyraloidea), which can travel long distances (up to 24 km in 1 h; Jacobson 1965; Silva Barros et al. 2020; Ge et al. 2021)—, the actual effectiveness of pollen-mediated gene flow is constrained by the availability and stability of these interactions. In fragmented landscapes, pollinators are particularly vulnerable (Siegel et al. 2024), which can decrease their abundance and diversity and, consequently, reduce effective pollen transfer among populations (Quesada et al. 2004). Collectively, these biological and environmental constraints may act in concert to limit pollen and seed dispersal, ultimately reducing effective gene flow across the landscape.

Another relevant factor to be considered is landscape influence. In this study, we identified eight putative barriers to gene flow, most of which correspond to major river watersheds or mountainous terrain. The Rho_ST values between watersheds further support the role of the region’s main rivers as significant barriers to gene flow. In the Humid Chaco, the Santa Lucía (Barrier A) and Paraná (Barrier B) rivers act as barriers to dispersal. Our findings support the role of the Paraná River and past hydrological changes in the Santa Lucía River that may have contributed to current genetic discontinuities in this species, consistent with patterns observed in other plant taxa (Paredes 2015; Morales et al. 2020; Moreno et al. 2020; Reutemann et al. 2025). Barriers C and D isolate population 4 from other northern populations, this corresponds with the different Bayesian genetic groups identified by Almirón et al. (2022).

In the Dry Chaco, the genetic discontinuities observed between population 4 and the northern Dry Chaco populations (ID 5–9)—corresponding to Barrier D—can be partially explained by differences in soil types, hydrological regimes, and the presence of alluvial fans. Seasonal river dynamics also influence local plant communities, potentially contributing to the genetic differentiation observed between populations 5, 6, and 7 from populations 8 and 9. In this area, two distinct watercourses—the Teuquito Stream to the north and the Bermejo River to the south—delineate the natural boundaries of the Protected Area where population 8 is located.

The Bermejo River is a dynamic watercourse with an irregular discharge regime shaped by highly seasonal rainfall patterns. Despite this variability, it maintains a permanent channel, which fosters the formation of deep pools and meandering paths. These geomorphologic processes give rise to shifting sandbanks and create highly dynamic riparian environments, where successional vegetation gradually establishes over time. In contrast, the Teuquito Stream is intermittent and typically dries out during the dry season. Its banks host tall, humid riparian forests sustained by well-drained soil and temporary water availability. These riparian formations differ markedly from the surrounding xeric forests—quebrachales and algarrobales—typical of the region (Administración de Parques Nacionales 2018).

Barrier E is associated with Dulce River, which separates the eastern lowland populations of A. quebracho-blanco located within the watersheds of allochthonous rivers (Paraná and Paraguay systems) from those situated in endorheic basins (Mar Chiquita and Serrano systems) that extend into higher altitudes in mountain regions. In this context, our results indicate that, although not physically separated by a clear barrier, A. quebracho-blanco populations belonging to Bayesian group I—characterized by the lowest levels of genetic variation—are associated with the Juramento River watershed (Paraná River system); whereas populations of group III, which exhibit higher genetic variation, are associated with the Bermejo River (Paraguay River system).

In contrast to Pilcomayo or Bermejo rivers, the Juramento River’s current diversion south of the Otumpa Hills—an ancient NNE-oriented ridge system reaching elevations of up to 210 m a.s.l.—alters local drainage patterns and creates pronounced topographic boundaries (Rossello and Bordarampé 2005; Peri 2012; Contreras 2019). As a result, vegetation cover varies markedly, with denser forests dominating in the northern Humid Chaco under the influence of the Pilcomayo and Bermejo rivers, while forest density decreases towards the south where the Juramento’s hydrological influence is limited (Contreras et al. 2015; Contreras 2019). This geomorphological complexity contributes to population differentiation. For instance, populations of Neltuma alba (Griseb.) C.E.Hughes & G.P.Lewis (formerly Prosopis alba Griseb.) from the northeastern Argentinian Gran Chaco differ morphologically from those in the central region (Verga et al. 2009; Cisneros and Moglia 2017). The Otumpa Hills may also contribute to the genetic differentiation of A. quebracho-blanco populations. In this sense, results of the Mantel test, PCoA, and interpolation analysis suggest that despite the genetic similarity observed between the Chaco National Park population and nearby populations, additional sampling in Paraná River sub-watersheds located between these localities is needed to determine whether Otumpa Hills function as barrier to gene flow.

In the western Dry Chaco, genetic discontinuities among populations located in the Dulce River watersheds (Mar Chiquita system) and those of the Serrano system (Barrier F), suggest that the Sierras Pampeanas—including the Ancasti Hills and Sierras Chicas—along with the intermountain depressions, act as natural barriers shaping regional patterns of genetic variation and structure in A. quebracho-blanco. The Sierras Pampeanas—a system of ancient faulted peneplains—form a rugged terrain characterized by north–south mountain ranges. The intermountain depressions are relatively flat and filled with sediments, creating endorheic basins (Cabido et al. 2003). Currently, these basins are partially occupied by the Salinas Grandes and Ambargasta salt flats. The Salinas Grandes acts as a barrier (Barrier G) to gene flow between populations on the Eastern hillside of the Ambato watershed and those in the Salinas Grandes watershed, both within the Serrano system. Finally, a barrier corresponding to the Sierras Chicas (Barrier H) separates populations belonging to Bayesian group II. The spatial discontinuities and the progressive uplift of the Pampean ranges associated with Andean orogeny (Gregory-Wodzicki 2000) likely constitute significant barriers that promote local adaptation and subsequent population differentiation, as suggested for other taxa (Solís Neffa 2010).

On the other hand, Principal Component Analysis revealed spatial clustering of environmental conditions that corresponded with the distribution of Bayesian genetic groups. According to Almirón et al. (2022), the species’ climatic optimum—represented by suitability values between 0.61 and 1.00—is concentrated in the Dry and western Humid Chaco, aligning with the distribution of the genetic groups I and II. Our results showed that populations within these groups, located in the Dry Chaco, exhibit lower genetic variability and occupy areas characterized by a higher mean diurnal range, greater elevations (140–1,067 m a.s.l.), and lower precipitation levels. In contrast, group III (Humid Chaco), which displays higher genetic variability, is restricted to lower elevations (0–183 m a.s.l.), with higher precipitations and lower diurnal range.

The topographic, climatic, and ecological contrasts among the watersheds in the Gran Chaco appear to influence dispersal dynamics and population connectivity, contributing to genetic divergence among populations located in different watersheds. Similar patterns of genetic differentiation have been reported in other Gran Chaco species (Pometti et al. 2012; Paredes 2015, Solís Neffa et al. 2022).

Overall, these findings highlight the combined influence of landscape complexity and environmental variation on the genetic structure of A. quebracho-blanco. In the context of increasing anthropogenic pressure and climate change, the integration of genetic, topographic, and climatic data into conservation and restoration planning is essential to ensure the long-term persistence of this keystone species in the Gran Chaco.

The levels of genetic diversity observed in each watershed are similar with those previously reported for A. quebracho-blanco populations (Almirón et al. 2022), suggesting that overall patterns of genetic variability are maintained across both population and watershed scales. In that sense, we support the areas previously proposed for in situ and/or ex situ conservation and restoration of A. quebracho-blanco but emphasize the importance of incorporating the landscape barriers identified in the present study. These barriers suggest that the management of each conservation area should not rely solely on the broad genetic clusters previously defined. Rather, populations within each Bayesian genetic group may already be undergoing divergence due to restricted gene flow. For areas proposed for in situ conservation, where populations i) harbour high genetic variability, ii) occur in areas of high and stable suitability under present and future conditions, iii) are separated within watershed by barriers associated landscape features such as mountain systems, and iv) previously showed significant positive isolation by distance among nearby populations, it is essential to preserve the independent genetic viability of each population. This situation applied to populations 18 and 19 (Genetic Group II), and populations 5, 6, 7, and 9 (Genetic Group III). Such an approach would allow natural differentiation processes to continue while maintaining the evolutionary potential of the species.

For ex situ conservation, involving populations 1 to 4, which i) harbour high genetic variability and ii) occur in areas with lower present and future climate suitability, and for restoration areas involving populations 8, 10, 11, and 12, which i) harbour moderated to low genetic variability and ii) occur in areas with high and stable climatic suitability for present and future but under strong anthropogenic pressure, we proposed, based on the results obtained here, that populations located in different drainage watersheds should be treated as distinct management units.

In the case of population 8, Almirón and Solís Neffa (2024) reported a moderate level of genetic structure and suggest that forest fragments may play a key role in conserving the genetic diversity of A. quebracho-blanco. Our results further revealed genetic discontinuity between nearby populations in this area, evidenced by horizontal genetic interpolation and the presence of a genetic barrier coinciding with the Teuquito stream. Consequently, population 9 is currently the best (but not the only) candidate for the restoration of population 8. Conversely, although previous and current findings suggest that populations 10, 11, and 12 could potentially be restored using genetic material from population 4 located in a protected area, additional sampling is needed to confirm the absence of a barrier to gene flow among these populations before proceeding with restoration actions. Such local studies would also help determine whether existing production systems could contribute to the conservation of A. quebracho-blanco genetic diversity, as has been observed in populations from the north of the Dry Chaco (IUCN-CMAP Working Group on OECMs 2019; Almirón and Solís Neffa 2024).

However, the success of any restoration strategy will also depend on demographic and ecological factors that remain poorly understood in the study area. Information on population density per hectare and regeneration rates of the species is still lacking. In the Dry Chaco, conserved environments may contain at least 340 saplings but only about 7 adult individuals per hectare (Nai Bregaglio et al. 2001), suggesting high mortality within the seedling bank, likely related to the strict requirements during early establishment stages (Barchuk et al. 2005). Furthermore, regeneration is negatively affected by regional disturbances such as fires and land clearing (Tálamo et al. 2013).