The study was conducted on nine beaver-inhabited gravel-bed streams in the Outer Western Carpathians (Fig. 1). This region consists mainly of alternating layers of sandstones, conglomerates, mudstones and shales, which make up the sedimentary deposits of the Carpathian flysch (Ślączka et al., 2006; Łajczak et al., 2014). The relief morphology of the study area corresponds to the general underlying geological units. The lower-lying foothills in the north are composed largely of the shale-rich thin-bedded flysch of the Silesian Nappe, while the mountains in the south are composed of the Magura Nappe, dominated by thick-bedded sandstones (Rączkowska et al., 2012; Golonka et al., 2019; Święchowicz et al., 2021). In just over half of the studied catchments (sites 1, 2, 4–6), easily weathered, soluble shales comprise a large portion of the geological structure (Burtan et al., 1994; Jankowski, 1997; Kopciowski et al., 1997). In the southern part of the study area (sites 7–9), less weathered sandstones predominate (Kopciowski, 2000; Kováčik et al., 2011). The stream at site 3, despite its location in the northern foothills of the mountains, drains one of the highest foothill ranges, the Brzanka Range, which is largely composed of sandstone (Marciniec and Zimnal, 2009). Streams that drain areas underlain by shales and mudstones are typically characterised by higher concentrations of bicarbonate, calcium and magnesium ions due to increased weathering of carbonates and clay minerals, while streams in areas composed mainly of sandstone are usually more diluted and characterised by a lower level of total dissolved solids (e.g. Siwek, 2021; Solár and Tomaškovič, 2023).

The geomorphologically diverse terrain in the study area ranges from foothills to medium mountains (Gilewska et al., 1982; Borzuchowski and Olędzki, 2011). The highest parts of the Outer Western Carpathians are classified as medium mountains. These areas are characterised by steep slopes (with a gradient that often exceeds 30 %) and significant relative relief (ranging from 400–800 m). Low mountains can be found at the margins of these ranges and in isolated ridges. These feature steep slopes and smaller relative relief (ranging from 200–400 m). The foothill zone is primarily located in the northern part of the study area. It reaches elevations of approximately 500 m above sea level. This zone is mainly characterised by wide, gently sloping ridges that rise from 40–300 m above the flat valley floors (Starkel, 1990). The mean annual temperature in all the selected catchments ranges from 5.1–9.1 °C, and the average precipitation range from 770 to 1140 mm (based on the 2000–2010 period; CARPATCLIM, 2013; Szalai et al., 2013). Rivers in this area are mainly characterised by a nival–pluvial hydrological regime (Wrzesiński, 2017).

All of the catchments consisted of a mixture of forest and agricultural areas (i.e. arable lands and meadows). Despite the widespread abandonment of agricultural land and the generally low-to-moderate intensity of agricultural practices in the mountainous catchments (Pazúr et al., 2014; Kolecka et al., 2017), the impact of these activities may be locally significant, as evidenced by the increase in nitrate concentrations observed in the Ondava catchment due to the use of fertilisers (Pekárová et al., 2006; Balejčíková et al., 2020). Overall, agricultural practices in the Polish and Slovakian Carpathians are widespread, but characterised by low–input management and mowing and grazing on a small-scale (Pazúr et al., 2014; Kolecka et al., 2017; Affek et al., 2023). According to the national agricultural statistics (Central Statistical Office of Poland, 2025), the average consumption of mineral fertilisers (NPK) in the Małopolskie and Podkarpackie Voivodeships remains one of the lowest in the country (80–95 kgNPKha-1yr-1 compared to the national average of ∼125kgha-1) and has shown little change over the last two decades. A similar situation is observed in the northern part of Slovakia, where the average use of industrial fertilisers is less than 60 kgNPKha-1yr-1 (Siman and Velísková, 2018). Consequently, the direct inflow of nutrients from fertilisers into streams is minimal, although local enrichment may occur near farms or areas of intensive grazing. In contrast, domestic wastewater has been identified as the main source of stream water pollution in the rural areas of the Polish Carpathians (Siwek, 2021). The intensity of housing development across the study area ranged from negligible levels to artificial surfaces covering up to 4 % of the area and containing a few hundred buildings.

The Slovak part of the Carpathians is characterised by large plots of agricultural land and small, dispersed settlements that are mainly concentrated in the valleys. This pattern is due to the impact of the post-1990 land reforms on land use and farm structure, both of which are characterised by large corporate farms and agricultural cooperatives (Lazikova et al., 2017). Despite the privatisation of land in Slovakia, land continues to be managed by large holdings and production companies (Bański, 2017), and the average size of an arable land parcel is 21.0 ha (Lieskovský et al., 2015). In the Polish Western Carpathians, settlements are more dispersed throughout the slopes, and land reform has led to a progressive fragmentation of agricultural land (arable land and grassland) due to the constant division of land between successive owners (Kroczak et al., 2018). In the Polish Carpathians, the average agricultural parcel size is 0.5 ha (Kolecka and Kozak, 2019).

The primary criterion for the selection of streams for the study was the presence of beaver dams that spanned the entire riverbed. The study sites were distributed randomly across the North–South transect of the Outer Western Carpathians in Poland and Slovakia. In addition to their varying topography, the sites differed in terms of human activity, such as catchment land use and the presence of river control structures (Table 1). The height of the beaver dams studied did not exceed 1.5 m. In most dam sequences, the ponds were directly connected. The only exception was site 7, where short sections of free-flowing reaches were observed between the ponds. For further analysis, the study sites were classified based on the beaver pond characteristics. Two types were distinguished, based on the pond flooding extent: in-channel (dams within river banks) and overflowing (dams that extended beyond the river bank, with flooding that affected adjacent parts of the river valley). The area covered by the overflowing ponds was two to seven times larger than their respective river channels, which indicates a substantial proportion of flooded riparian areas. The age of the beaver ponds was assessed based on archived orthophotos, and they were classified as: young (≤3 years old), moderate (4–9 years old), or old (≥10 years old) (Table 1).

The research followed a three-dimensional sampling scheme: seasonal sampling, longitudinal sampling, and vertical profile sampling (Fig. 2). The following water parameters were measured: temperature (T), dissolved oxygen (DO), specific conductivity (SEC), pH (H+) and the ionic composition (Ca2+, Mg2+, Na+, K+, NO3-, NH4+, HCO3-, SO42-, Cl-).

Although both geological variability and human pressure affect the chemical composition of stream water, a sampling design based on comparisons along the longitudinal gradients helped to minimise the impact of differences between the catchments. Three samples were collected at each site in late spring (June), summer (July–September), autumn (October–November) and winter (February), during baseflow conditions (i.e. excluding periods of rainfall events or snowmelts) in 2022–2023 (Fig. 2a). The section “above” the beaver pond or pond sequence that reflected baseline conditions was defined as 50–100 m upstream from a backwater, and the section “below” was 50–100 m downstream from the last dam. The “pond” sample was taken from the main channel section of the pond (i.e. excluding channel margins and backwaters). It was collected approximately 1 m above the dam, from the middle of the water column. For pond sequences, the representative “pond” sample was collected from the biggest reservoir in the system. Samples were collected between late morning and afternoon, with the sampling at each site being completed within about an hour. A total of 102 samples were collected from nine sites during the seasonal sampling.

Two sites were selected for additional measurements to create the longitudinal stream profiles (Fig. 2b). This sampling was conducted on 10 September 2023 at site 3 (an overflowing over ten years old beaver pond sequence) and at site 6 (a single in-channel young beaver pond). Measurements and water samples in the “above” and “below” sections were taken every 30 m, starting at a distance of 130 m from the ponds. Moreover, 4 locations within the “ponds” were sampled, to represent the front, middle and periphery of the water body.

To analyse the variation of physico-chemical characteristics in a pond's vertical profile, water samples were collected on 10 July 2024 at two beaver pond sequences, sites 2 and 3, which represented the two different pond types (Fig. 2c). Only in-channel ponds were observed at site 2. By contrast, overflowing stream banks were common at site 3. Measurements and water samples were taken from near the bottom of the water column, from its surface, and from dam outlets. Additionally, on-site measurements alone were taken from the middle of the water column. At site 2, outlet samples were taken from water overflowing the dam, while at site 3 they were taken from leaks in the upper part of the dam and near its bottom. In addition to the previously mentioned parameters, this scheme also involved determining the total organic carbon (TOC) concentration (mgL-1).

Sampling and in situ measurements were conducted without disturbing turbidity. The basic physico-chemical parameters – T (°C), DO (mgL-1), SEC (µScm-1) and pH – were determined on-site using an ELMETRON CX-401 multifunction meter with the GXZ-3tk multiparameter immersion head. All pH, SEC and DO probes were calibrated before each field study at the laboratory of the Institute of Geography and Spatial Organization Polish Academy of Sciences. Calibration was performed according to the manufacturer's guidelines using certified analytical standards: pH probe: three-point calibration (pH 4.0, 7.0, and 9.0 buffers); SEC probe: calibration with a 1.41 mScm-1 standard solution, selected according to anticipated conductivity of the sampled waters; DO probe: air-saturation calibration (100 %) followed by zero-oxygen verification using Na2SO3. In addition, 250 mL water samples were collected at each site for laboratory analysis.

The HCO3- concentration (mgL-1) was determined by titrating the water with a standard solution of hydrochloric acid. The ionic composition (mgL-1) was analysed in a Dionex ICS-3000 ion chromatograph with the analytical columns IonPac CS16 3mm×250mm and IonPac AS18 2mm×250mm. Each sample was first filtered through a Whatman GF/D glass microfiber filter. The calibration curve was based on the Dionex certified reference materials (CRM) Combined Six Cation Standard II and Seven Anion Standard II. The detection limits were 0.4 mgL-1 for Ca²⁺ and 0.1 mgL-1 for the other ions. To ensure the accuracy of the analytical procedure, a certified reference material (CRM: KEIJM-02, GAW/WMO) was analysed. The percent relative error was less than 5.0 % for all parameters, except for K+ (5.8 %) and NH4+ (7.0 %). For any ion concentrations below the detection limit, the values were replaced with 75 % of the detection limit, which is a common procedure in chemistry (Palarea-Albaladejo and Martín-Fernández, 2013). Concentrations below the detection limit were found only for the NO3- and NH4+ ions. Across all the analysed water samples (n=102), the NO3- and NH4+ concentrations were below the detection limit in 17 and 20 samples, respectively. According to Hites (2019), the inclusion of values below the detection limit in analyses is justified if they do not exceed 20 %. This applies in the case of both these ions. In addition, the Vario TOC Cube Analyzer (Elementar) was used to determine the total organic carbon (TOC) concentration (mgL-1) of the unfiltered water samples.

To compare the study sites, they were first classified according to the age and type of each beaver pond sequence (Table 1). It should be noted that there was some overlap between the two categorisations, as most of the youngest study sites were characterised by in-channel pond type and all the oldest ponds were overflowing. To assess the influence of beaver dams on physico-chemical water properties, Linear Mixed Model (LMM) was applied to individual parameters. Two separate approaches were employed: one based on raw data, and the other based on differences between the upstream and the pond or downstream sections, expressed in the original measurement units. In the models, “sampling location” (3 categories for raw data: upstream, pond, downstream, and 2 categories for the difference-based analysis: pond and downstream), “season” (4 categories: spring, summer, autumn, winter), “pond type” (2 categories: overflowing, in-channel), and “dam age” (3 categories: young (≤3 years), moderate (4–9 years), old (≥10 years)) were treated as fixed effects. Both “pond type” and “dam age” were tested in interaction with “location”, to account for potential differences in their effects among sampling sections. Stream identity was included as a random effect to account for repeated measurements collected from the same streams in successive seasons. Effects were considered significant at p<0.05. For factors showing a significant overall effect, least-squares means (LS means) were examined, and pairwise comparisons were performed to identify differences between factor levels.

The LMM was complemented by a supplementary analysis based on percentage change (Dc) in the values of the individual physico-chemical parameters of the water in ponds and streams below the beaver dams relative to their values above the ponds. The Dc was calculated using the following formula: 1Dc=(Bx-Ax)Ax⋅100%, where Bx is the value of the physico-chemical parameter x in the pond or stream below the dam during a particular sampling session, and Ax is the value of the physico-chemical parameter x in the stream above the dam during the same session.

This additional analysis was introduced to help reveal direction and magnitude of variation in analysed parameters and to identify consistent patterns that may have remained undetected in the LMM framework because of the limited sample size. For the purpose of calculating the percentage change, temperature values were converted from °C to K, to account for potential disruptive impacts of values close to 0 °C on the percentage change in winter. Seasonal and site-to-site variations in Dc were assessed for T and other physico-chemical water parameters if their mean absolute difference |Dc|‾ (for all sampling sites combined) exceeded 15 % (DO, H+, K+, SO42-, Cl-, NO3-, and NH4+; Table A1). The calculation of |Dc|‾ used the following formula (where N is the total number of sampling sessions): 2|Dc|‾=∑Bx-AxAxN⋅100%. This procedure was used to account for the impact of percent relative errors of the measurements on the Dc calculations. Thus, only the chemical parameters for which this impact was generally smaller than the measured changes were considered for further analysis. For each sample, the lowest (Dcmin) and highest (Dcmax) possible Dc values of a given physico-chemical parameter were calculated, with maximum percent relative errors assumed for each measurement (Fig. A1a). For most of the selected chemical parameters, such as DO, Cl-, K+, SO42-, and NO3-, the majority of the maximum possible differences in Dc (Dcmax-Dcmin) were less than 10 %–15 % (Fig. A1b). These differences were higher for H+ and NH4+, though still many times smaller than the mean absolute difference |Dc|‾ values (Fig. A1b and Table A1). This allowed for their inclusion in the subsequent analysis.

In addition, correlations between the upstream values of the physico-chemical parameters, catchment area, land cover, channel slope and channel width were analysed to identify potential causes of the initial differences between study sites. Moreover, the correlations between the upstream values and Dc were analysed, as this could indicate potential unreliability of the obtained percentage changes. Results characterised by |r|>0.6 (and p<0.05) were considered as demonstrating a strong correlation. All analyses were performed using STATISTICA 12.0.

The physico-chemical properties of the water upstream of beaver ponds varied between the studied catchments (Table 2). Higher SEC and major ion concentrations were recorded in streams draining catchments underlain by shale-rich bedrock (sites 1, 2, 4, and 6), whereas lower values occurred in streams draining catchments dominated by sandstones (sites 3 and 7–9). Concentrations of Cl-, Na+ and K+ showed strong correlations with the proportion of artificial surfaces in the studied catchments (r=0.85, 0.78 and 0.61, respectively). Additionally, concentration of Cl- was correlated with the number of buildings in the catchments (r=0.61). No strong correlations were found between the parameters of the stream water above the beaver ponds and their percentage changes (Dc) in the ponds and the streams below them (Table A2). This allows for the further comparative analyses of the effect of beaver dams on the physico-chemical parameters of stream water despite their initial variation across the study sites.

The large differences in the physico-chemical water parameters (|Dc|‾>15%) between the beaver-influenced river sections (pond and downstream) compared to upstream values allowed for the consideration of potential impact of beaver dams on H+ (pH), NO3-, NH4+, DO, K+ and SO42- (Table A1). Overall, greater differences in the physico-chemical water parameters were observed in the ponds than in the streams below the ponds. However, the directions of the observed changes were not strongly consistent across seasons or study sites (Figs. 3 and A2).

The LMM analysis revealed relationships between certain physico-chemical parameters and seasonality, the age of ponds and sampling locations (Tables 3 and 4). According to the analysis based on the raw values, season significantly affected all the parameters except pH, Ca2+ and NH4+, whereas sampling location (i.e. upstream, within the pond, downstream) significantly influenced DO, pH and T. In general, DO and pH values were lower in the ponds than in the streams above and below them, while T was higher in the ponds than in the streams above them. Beaver dam age significantly affected DO and the pH, with generally higher values in young ponds (≤3 years) than in ponds of moderate age (4–9 years). Pond type did not significantly affect any parameter in either the raw data or difference-based analyses.

Season significantly affected the differences between upstream values of T, Mg2+, K+ and SO42-, and their respective values in the ponds or downstream of the ponds. For example, there was an increase in T and a decrease in SO42- in the beaver-influenced stream sections (pond and downstream) compared to upstream values in the warmer part of the year (spring and summer). In contrast, the values for both parameters remained relatively unchanged in autumn and winter. Moreover, a greater increase in K+ was observed in summer than in spring and winter. The differences in DO within the ponds were affected by dam age, with the highest decreases observed in the old ponds (≥10 years). The decreases in DO and pH were greater in the ponds than in the downstream sections. Although the LMM also identified the seasonal effect on Mg2+, this ion was excluded from further discussion because its |Dc|‾ value was low (less than 15 %) and did not exceed the potential relative measurement error.

The limited number of study sites (n=9) may have prevented the detection of weaker patterns. Analysing the positional measures of percentage changes (Dc) in physico-chemical water parameters in ponds and the streams below them, grouped by the factors indicated by LMM: season (spring/summer versus autumn/winter), pond age and sampling location, reveals additional effects that may occur during particular part of the year. For example, the greatest differences in SO42- were found in the old ponds, followed by the moderate ones and then the young ones during spring and summer (Figs. 4 and A2). Given the generally low upstream NO3- concentrations, the scale of NO3- decrease was constrained; nevertheless, the consistent decreases were observed within the ponds in most cases. There were greater differences in the old ponds than in the young ponds in spring and summer. The extremely high outlier value for Dc of NO3- disrupts this pattern for moderate ponds (Figs. 4 and A2). The concentration of NH4+ increased in most studied ponds compared to the streams above them during spring and summer.

As demonstrated by the longitudinal profiles (Figs. 5 and B1), the older overflowing beaver pond sequence (site 3) had a greater influence on the physico-chemical parameters of water than the young in-channel single pond (site 6). The greatest differences in the physico-chemical parameters at site 3 were seen between the free-flowing upper sections and the inside of the ponds; these changes tended to be partially neutralised in the stream below the beaver dam. The largest differences observed at site 3 were a decrease in NO3- within the ponds (-95.4%) and a decrease in SO42- both within (-64.5%) and downstream (-71.5%) of the ponds.

There were greater vertical variations in physico-chemical parameters in the old overflowing beaver pond (site 3) than in the younger in-channel beaver pond (site 2). DO, pH and T were found to decrease with pond depth in both types of beaver pond; however, the range of variation was greater at site 3 than at site 2 (Figs. 6 and B2). The pH value and DO concentrations were lower and the TOC concentrations were higher at the overflowing, old pond (site 3) than at the in-channel, younger pond (site 6). The pH value and the concentrations of DO and TOC of the outflow measured near the base of the dam were notably lower than those measured at the top of the dam at site 3. The TOC concentration in the beaver pond at site 3 was found to be more than 13 mgL-1, which was almost three times higher than in the stream flowing into the pond.

Beaver dams form pond environments that alter biogeochemical processes and thus the chemical composition of stream water (Larsen et al., 2021). Our study shows a significant decrease in DO concentrations within the beaver ponds compared to the upstream conditions. This suggests that DO is consumed during the oxidation of organic matter in the ponds. A beaver-driven decrease in stream DO concentrations was also found by Błędzki et al. (2011) and Stevenson et al. (2022). High TOC concentrations in beaver ponds we studied, compared to the stream water above the ponds, indicate an enrichment of organic matter in the pond water. This matter may be produced locally due to vegetation growth in the beaver ponds. According to Pollock et al. (2003) and Green and Westbrook (2009), the reduced stream flow velocity associated with beaver ponds promotes the deposition of particulate organic matter. Briggs et al. (2012, 2013) found that the accumulation of organic matter in beaver ponds generates specific streambed morphologies, which enhance the exchange of stream water and groundwater within the hyporheic zone behind the beaver dams. Furthermore, the high water pressure at the bottom of the ponds results in the downward movement of pond water, which promotes the transport of organic carbon and DO from the water column to the hyporheic zone, fuelling biogeochemical processes.

The oxidation of organic matter in ponds may also have caused the decrease in SO42- concentrations at most of the study sites during warm seasons (spring-summer). Under the anoxic conditions that prevail at the bottom of ponds (as measured at site 3 in July 2024; see Fig. 6), as well as in the hyporheic zone, SO42- ions are likely to be reduced to H2S. These reactions are carried out by microorganisms through the microbial sulphate reduction, where SO42- acts as an electron acceptor in the anaerobic oxidation of organic matter under low redox potential (Sutton-Grier et al., 2011). The decrease in SO42- concentrations in streams with beaver dams has been previously identified, and attributed to the expansion of anoxic conditions (Larsen et al., 2021). According to Naiman et al. (1986; 1994), large accumulations of detritus, in combination with decomposition in expanded wetted areas, result in the development of anaerobic biogeochemical processes.

The decomposition and mineralisation of organic matter in the process of ammonification, under both aerobic and anaerobic conditions, leads to the production of NH4+ ions in water (Stefanakis et al., 2014). In our study, the ammonification was indicated by the increase in NH4+ concentrations in most of the beaver ponds studied in spring and summer. The increase in NH4+ concentrations in ponds may also be partly explained by the production of this ion through the process of dissimilatory nitrate reduction to ammonium (DNRA), as indicated by a decrease in NO3- concentrations in the studied ponds. Numerous pathways exist for converting NO3- to NH4+ under anaerobic conditions within DNRA processes, but all of these lead to the production of NH4+ within the system (Wang et al., 2024). Denitrification, alongside DNRA, could also explain the decrease in NO3- concentrations in beaver ponds. Denitrification reduces NO3- to gaseous N2O and N2 under anoxic conditions in water (Drever, 1997). Decreased DO concentrations in the studied ponds could positively influence the denitrification processes in beaver impoundments, as suggested in previous studies by Lazar et al. (2015), Wegener et al. (2017) and Dewey et al. (2022). According to Lazar et al. (2015), denitrification in beaver ponds can remove 5 %–45 % of NO3- loading at the catchment scale.

The decomposition of organic matter results in the formation of organic acids, and this may have contributed to the significant decrease in water pH in the studied beaver ponds. According to David and Vance (1991), organic acids constitute most of the dissolved organic carbon in streams and lakes. Organic acids with high dissociation capacities significantly reduce water pH (Driscoll et al., 1989). A decrease in stream pH due to the presence of beaver dams is rarely reported in the literature. In fact, beaver dams most often cause its increase. For example, in the Appalachian Mountains, USA (Smith et al., 1991; Cirmo and Driscoll, 1993; Cirmo and Driscoll, 1996; Margolis et al., 2001) and the lowlands of Devon, England (Puttock et al., 2017), the pH of the stream water below beaver dams increased due to reduced concentrations of acidic ions such as NO3- and SO42-. In the streams we studied, pH decreased despite the decrease in acid ion concentrations (i.e. NO3- and SO42-). Furthermore, the streams and beaver ponds studied in the USA and England (Smith et al., 1991; Cirmo and Driscoll, 1993; Cirmo and Driscoll, 1996; Margolis et al., 2001; Puttock et al., 2017) had similar organic carbon concentrations to the streams and ponds in our study. However, the streams that fed the beaver ponds in the USA and England had lower pH (<7) than the streams we studied (>7). The difference in baseline water pH is likely a key factor in explaining the difference in results, because the intensity of organic matter decomposition increases with increasing pH, and the latter is also associated with better conditions for the growth of the bacteria involved in decomposition (Walse et al., 1998). Higher pH values in the streams that feed the studied beaver ponds likely facilitated the greater decomposition of organic matter, which would be in line with the increased production of organic acids in the ponds. Another process that can decrease the pH of pond water is the production of CO2 during the aerobic and anaerobic mineralisation of organic matter (Drever, 1997).

The water that flows out of the ponds takes on the physico-chemical properties of the pond water. This is evidenced by the similar directions of changes in physico-chemical parameters within and below the beaver ponds compared to the streams above the ponds. However, our studies show that the downstream effect of beaver impoundment depends on the dam structure and, thus, the type of outflow. Due to the vertical physico-chemical stratification of water within the ponds, water flowing out over the top of the dam may have different physico-chemical properties from water flowing out through the lower parts of the dam. For instance, the DO concentrations in the water flowing out from the bottom of the dam were lower than in the water flowing out through the top. Decreasing DO with beaver pond depth is consistent with the results reported by Stevenson et al. (2022). Nevertheless, the changes in the physico-chemical parameters in the water below the ponds, regardless of the type of outflows, are smaller than the changes in the ponds compared to the upstream sections. This likely reflects processes typical of flowing water, which tend to shift water properties back towards upstream conditions. For example, the deoxygenated water flowing out of the ponds is reoxygenated, aided by the turbulent water flow in the mountainous stream channels. Similar decreases in DO within beaver ponds and the subsequent reoxygenation of downstream waters have been observed in previous studies (Smith et al., 1991; Harthun, 2000; Błędzki et al., 2011; Ecke et al., 2017). The increased DO concentrations in the streams below ponds favour the aerobic oxidation of organic matter, including the oxidation of organic acids. This increases the pH of downstream water compared to pond water. The increase in pH may also be partly due to the release of CO2 from the water, facilitated by the turbulent flow of streams below the beaver dams (Horgby et al., 2019). The reoxygenation of the water in the streams below the beaver ponds involves a series of biogeochemical processes, such as the aerobic oxidation of organic matter and nitrification, which counterbalance the effects of the ponds. Our longitudinal study showed that the changes in the physico-chemical properties of the water were partially neutralised within a relatively short downstream distance from the ponds (<100m).

The greatest changes in most of the physico-chemical parameters within and below the beaver ponds compared to the streams above the ponds were observed during the warm season (spring and summer). This result can be explained by the greater intensity of the biogeochemical processes that modify the chemical composition of water at higher temperatures. In the winter, the water temperature in the streams and at the surface of the studied ponds was low, and did not exceed 1.5 °C. In the summer, water temperatures in the studied streams and the near-surface layer of the ponds generally exceeded 15 °C and sometimes rose above 20 °C. Biogeochemical processes in beaver pond water and the sediments in the hyporheic zone, such as organic matter decomposition, denitrification, DNRA, and microbial sulphate reduction, are very slow at low temperatures, compared with warm conditions (Qu et al., 2022). As a result, there are intensive physico-chemical transformations in the stream water flowing into the ponds in the warm season. The increase in NH4+ concentrations and the decrease in NO3- concentrations in the studied beaver ponds were particularly prominent in summer due to intensive ammonification (NH4+ production), denitrification and DNRA (NO3- depletion). This is consistent with the findings of Margolis et al. (2001) and Law et al. (2016): the greatest NO3- decrease downstream of beaver ponds occurs in the summer. The NO3- ion is the most effective electron acceptor (oxidant) in the anaerobic oxidation of organic matter; therefore, NO3- is the preferred oxidising agent for anaerobic microbes. The NO3- reduction under the anaerobic oxidation of organic matter occurs at a relatively high redox potential level (Anderson and Fidel, 2025). However, the decrease in NO3- concentrations in the beaver ponds in the summer may also be caused by NO3- uptake by aquatic vegetation and microorganisms (Naiman and Melillo, 1984; Maret et al., 1987; Songster-Alpin and Klotz, 1995). Furthermore, the decrease in SO42- concentrations in the studied ponds and downstream of the beaver dams was mainly observed during the warm period. This is also likely due to the high rate of anaerobic oxidation of organic matter during that time. According to Smith et al. (1991), Cirmo and Driscoll (1993), and Margolis et al. (2001), anaerobic oxidation of organic matter is facilitated by the low redox potential characteristic of summer low-flow conditions. During this period anaerobic microbes would have used even relatively inefficient oxidants, such as SO42-, in the anaerobic oxidation of organic matter, due to the low concentration of NO3- (a more effective oxidant) in the studied waters. The average concentration of NO3- in ponds and streams below them was very low during the summer, the average was 1.5 mgL-1.

Our research suggests that the age of beaver dams plays an important role in altering the chemical composition of stream water (Fig. 7). Previously, the impact of beaver dams on various physico-chemical water parameters has typically been found to be inconsistent across different study sites. For example, different directions of change in pH were documented by Lizarralde et al. (1996), Harthun (2000) and Čiuldienė et al. (2020). The beaver pond characteristic proposed in our study may partly explain those site-to-site differences. For instance, the magnitudes of the decrease in pH and DO (based on LMM and the positional measures of the Dc) and SO42- (based on the positional measures of the Dc) in the studied beaver ponds were found to be positively related to the age of the beaver pond sequences. These water parameters are influenced mainly by the decomposition and aerobic/anaerobic oxidation of organic matter. According to Gurnell (1998), Meentemeyer and Butler (1999), Bason et al. (2017) and Puttock et al. (2018), older ponds have greater pools of organic matter deposited on their bottoms. Although the pond age does not cause the difference directly, it can be used as a proxy for the accumulation of organic-rich sediments over time and the enhancement of respective biogeochemical processes.

The wide spatial scope of the study and the high diversity of beaver-influenced streams introduced certain limitations. The spread of the study sites across a vast area restricted the number of samples collected to one sample per site each season, due to time and cost constraints. In addition, dam failures resulting from floods and their targeted elimination during the research period decreased the planned sample size by half. The remaining study sites may have been affected by continuous dam maintenance by beavers. In the most extreme cases, temporal breaches of the dam may have been overlooked after its reconstruction. The small number of study sites (nine) may have limited the identification of the existing patterns in the effect of beaver dams on the physico-chemical properties of stream water. Therefore, we used the additional analysis of the positional measures, which is suitable even for small statistical samples. Moreover, since the study focused on the baseflow conditions, early spring high flows have been omitted. This probably explains the similarities between the results from the late spring and the summer.

Although a Before-After-Control-Impact (BACI) model would be the most appropriate approach for studying the impact of beavers on the chemical composition of water, it was not possible to use this method in the area concerned, due to the lack of investigations prior to dam construction. Instead, this study employed a Control-Impact (CI) design based on spatial contrasts. In this framework, upstream sections served as control sites, representing the reference conditions, while sections within and below the pond sequences represented the impacted areas. This comparative “upstream-downstream” approach is a standard field method for assessing hydrochemical gradients generated by beaver dams when “before” data are unavailable (Green and Westbrook, 2009; Klotz, 2010; Fuller and Peckarsky, 2011; Puttock et al., 2017; Stevenson et al., 2022). It allows for the isolation of local effects caused by damming while minimizing the confounding influence of catchment heterogeneity. Although this approach lacks independent control sections, as in a full BACI model, it provides robust comparative evidence for evaluating the magnitude and spatial extent of changes induced by beaver activity under natural field conditions. Future research should be more detailed and extensive to enable the application of the BACI model.

As there was no identified relationship between the initial values and the percentage change (Dc) in the physico-chemical parameters, it was assumed that the observed changes would not differ significantly based on upstream conditions. This allowed various catchments with different land uses and levels of anthropogenic influence to be included in the study. The diversity of the studied streams was also partly managed by including the stream identity as a random factor in the LMM. However, it should be noted that the inclusion of the streams characterised by particularly high concentrations of certain components (e.g. highly anthropogenically modified) may have resulted in different observations.

Although the sample locations were standardised to a certain extent, they may not reflect typical stream or pond conditions. Stream samples are more likely to be well-mixed and representative due to turbulent flow conditions. This is also evidenced by the homogeneity of the longitudinal profiles upstream and downstream of beaver ponds for most studied parameters. However, the spatial distribution of the pond samples may be more biased, as it did not cover the full spectrum of hydrological units observed within the ponds (Majerova et al., 2020). Additionally, as the samples were taken from the middle of the water column, they do not represent the surface and bottom water layers, which, according to the vertical sampling scheme, could substantially affect observed physico-chemical parameters. One of the greatest limitations of the study was the absence of field measurements of redox potential in streams and vertical profiles of the beaver ponds. Such measurements would have enabled the direct identification of redox conditions in the studied waters, and thus avoided the need to rely on speculation based on the concentration of key redox reaction ions. Direct measurements of redox potential, microbial activity and vegetation cover in future studies would enable empirical assessment of the role of the discussed processes.

The age of beaver pond sequences appears to be partly related to their longitudinal development and flooding extent. This limits the opportunity to assess the impact of each characteristic separately. Moreover, the analysis would be greatly improved by the use of continuous values representing pond age, if they could be obtained.