Spatial variability of water chemistry in the Ayeyarwady River Basin, Myanmar

Study area and water sampling

Water was sampled at 50 sites across the ARB (33 sites in the Ayeyarwady River; 16 sites in the Chindwin River, which a suspected spawning tributary for anadromous species; and one site at the Indawgyi Lake). Most of these sampling sites were located on the mainstem while nine sites were located on tributaries (Fig. 1). These sampling sites encompassed three broad regions of the ARB: (1) Ayeyarwady upstream (11 sites, A24–A33 and L1); (2) Ayeyarwady downstream (23 sites, A1–A23); and (3) Chindwin River (16 sites, C1–C16). At each sampling location, global positioning satellite (GPS) co-ordinates were recorded using a GPS instrument (Garmin 64). Three replicate river water samples were collected from just below the water surface at each site and one ‘blank’ sample was prepared (using deionised water) to detect contamination during the sampling process in the field. Water was then immediately filtered using 0.2 μm syringe filters (Sarstedt, Polyethersulfone membranes) into acid washed bottles containing high-grade nitric acid. Each sample ultimately contained 98 mL of filtered water and 2 mL of 65% nitric acid to preserve samples until analysis. Several field trips were conducted for water sampling in the Lower Ayeyarwady and Ayeyarwady Delta (October 2019) and the Middle, Upper Ayeyarwady, and Chindwin (October 2019–January 2020), covering a single season. In addition, existing hydrological data in 2019 (water level, rainfall, water temperature) were collected from the Hydrological and Metrological Department at three gauging stations (Hinthada, Mandalay, and Homalin) in the ARB.

Fig. 1.

Map of sampling sites for water chemistry (a) and the geography (b) of the Ayeyarwady River Basin. A, Ayeyarwady River; C, Chindwin River. For more details, see Supplementary material 1. The geology map was adapted from Garzanti et al. (2016).

Elemental analysis

For trace elements (Sr, Ba, Mg, Ca, Mn, Fe, Zn), water samples were initially tested using Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES; 5110 SVDV ICP-OES, Agilent Technologies, Santa Clara, CA, USA). Except Zn these elements are often abundant. The concentration of Zn was low in the ARB, but it was still detected at all sampling sites. The limit of detection and limit of quantification of Zn were 0.0002 mg L⁻¹ and 0.0006 mg L⁻¹, respectively. Calibration was achieved using a 5% HNO₃ matrix multi element standard (IntelliQuant Multi-element Standard #1, Agilent Technologies) containing 48 elements, and calibration standards were prepared using 1% HNO₃ at 50 ppm, 10 ppm, 5 ppm, 1000 ppb, 100 ppb, 10 ppb, 1000 ppt, 100 ppt, and 10 ppt. To account for machine detection limits, a replicate of each site was used to make two dilutions (1/100 and 1/1000) using instrument grade water (18.20 μΩ) fixed with 1% nitric acid. Parent samples were run for all samples as well as 1/100 dilutions, which were used for samples that were predicted to have high salinity concentrations. To account for field contamination, a ‘blank’ sample was prepared using instrument grade water (18.20 μΩ) instead of river water.

For Sr isotope concentrations (⁸⁷Sr:⁸⁶Sr), a multi-collector inductively coupled plasma mass spectrometry (ICP MS, Thermo Scientific, High Resolution NEPTUNE, Bremen, Germany) was used. Purification of strontium for ⁸⁷Sr:⁸⁶Sr ratio analysis was conducted using custom columns packed with Eichrom Sr-spec resin (~300 μL column with ~1 mL reservoir). Samples were acidified to 2 M with triple-distilled HNO₃ and (1–2 mL) loaded on to the columns. All solutions (samples and elution acids) were loaded onto the column on 0.3 mL aliquots to prevent disturbing the resin bed and to keep the resin wetted throughout chromatography. A waste beaker was placed under the column and 3 mL of 2 M HNO₃ was then passed through the column to elute Rb. Following this, 4 mL of 2 M, 2 mL of 7 M, and then 0.3 mL of 2 M HNO₃ was used to elute Ba and the matrix from the column. Finally, a clean 5.5-mL vial was placed under the column to collect the Sr cut, which was eluted using 5.3 mL of 0.02 M HNO₃. The collected sample (5.3 mL) was then analysed directly by a multi-collector ICP MS. The ⁸⁷Sr:⁸⁶Sr ratio of the standard reference material SRM987 (NIST; www.nist.gov) with an absolute abundance ratio was 0.71034 ± 0.00026. To determine spatial patterns of water chemistry, sampling points were overlaid onto the map containing the Ayeyarwady and Chindwin Rivers. A Jenks Optimisation Method was used to classify and display the trace element and strontium ⁸⁷Sr:⁸⁶Sr values into colour-coded categories.

Data analysis

Data was grouped into the three regions for statistical analyses: (1) Ayeyarwady downstream (A1–A23); (2) Ayeyarwady upstream (A24–A33 and L1); and (3) Chindwin (C1–C16). Data from the coastal site (A1) was included purely for comparison, and not statistically analysed. This classification provides applications for future otolith studies to examine fish movements between different regions of the ARB. Differences in elemental concentrations and element:Ca ratios among regions were examined using univariate permutational analysis of variance (PERMANOVA) in Primer + Permanova ver. 7.0 (Anderson et al. 2008). Prior to analyses, data was log₁₀(x + 1) transformed to reduce the influence of outliers and Euclidean distances were calculated among samples for all PERMANOVA tests. Pairwise comparisons were further tested wherever significant differences (P < 0.05) were found in the main tests. Variability in elemental concentrations and element:Ca ratios were additionally assessed by calculating coefficient of variation percentages for each region. We also investigated whether there were any correlations between elemental concentrations and element:Ca ratios, and distance from the coast, by running Spearman Rank correlations tests in SigmaPlot for Windows ver. 15.

Variation in elemental concentrations

Most trace element concentrations differed between fresh and marine waters. For example, Sr, Ca, and Mg showed clear differences along the mainstem from the coast to upstream. Concentration of Sr at the coast (3.17 mg L⁻¹ at A1) were 63-times higher than the concentration further upstream (A33), which was similar to the 83-fold difference in Sr concentration observed between marine and fresh waters of the Mekong River (Vu et al. 2021), and is also similar to that observed for other river systems (He and Xu 2016; Tran et al. 2019; Stoot et al. 2024). In freshwater, concentrations of most trace elements such as Ba, Fe, Mg, and Zn were significantly different between regions of the ARB while concentrations of other trace elements (Sr, ⁸⁷Sr:⁸⁶Sr, Ca, and Mn) were not significantly different between regions (P > 0.05). For Fe and Zn, concentrations of these elements were higher in the Ayeyarwady upstream (Fe, 0.38 ± 0.37 mg L⁻¹; Zn, 0.0039 ± 0.0025 mg L⁻¹) and Chindwin (Fe, 0.36 ± 0.23 mg L⁻¹; Zn, 0.0039 ± 0.0036 mg L⁻¹) while these concentrations were low in the Ayeyarwady downstream (Fe, 0.04 ± 0.04 mg L⁻¹; Zn, 0.0007 ± 0.0006 mg L⁻¹). These trace elements and other heavy metals are likely discharged from mining operations in the upstream. Indeed, mineral resources such as gold, zinc, lead, and jade were mined in the Ayeyarwady upstream (Ketelsen et al. 2017), and these activities could contribute to higher concentrations of some heavy elements (e.g. Fe and Zn). Therefore, these mining activities raised a concern about pollution in the region (Krittasudthacheewa et al. 2021).

Myanmar was formed through the merging of various terrains at different geological periods. There are four major accreted terrains in Myanmar (Than et al. 2017). Additionally, the ARB is categorised into five hydro-ecological zones according to hydrology, geomorphology and ecology (Ketelsen et al. 2017). However, ratios of ⁸⁷Sr:⁸⁶Sr in water were not significantly different between regions of the ARB. This should reflect similar bedrock geology in the region. Ratios of ⁸⁷Sr:⁸⁶Sr in the lower Chindwin River (C1–C7) were lower (0.70926 ± 0.00027) in our study and close to that of the global ocean signature (0.70918 ± 0.00006). Another study sampling waters at Monywa (Chindwin River) also confirmed that ⁸⁷Sr:⁸⁶Sr ratios were lower (0.70903 ± 0.00029) in the area (Chapman et al. 2015). Overall, the Ayeyarwady River has the lowest value of ⁸⁷Sr:⁸⁶Sr ratios among the Himalayan–Tibetan region (Chapman et al. 2015). These authors also found seasonal variation in ⁸⁷Sr:⁸⁶Sr ratios in the ARB: peak at the onset of the flood season (June).

Hydrological regimes and climate exhibited strong seasonal variation in the Ayeyarwady River, suggesting that they play key roles in driving the variation in water chemistry in water chemistry in the region. Seasonal variation of some elements (e.g. Sr, ⁸⁷Sr:⁸⁶Sr, and Ca) were found in the ARB (Chapman et al. 2015). Impacts of hydrology on trace element concentrations were also found in other tropical rivers such as the Mekong and Amazon (Castello et al. 2015; Vu et al. 2021). For example, seasonal variation of elements (e.g. Mg, Ca, and Na) was related to variation in hydrological regime due to salt intrusion in the Mekong Delta (Vu et al. 2021). Similarly, ratios of ⁸⁷Sr:⁸⁶Sr were higher during the wet seasons in the Daly–Katherine River in Australia (Crook et al. 2017). Our study found that ratios of ⁸⁷Sr:⁸⁶Sr were usually different between rivers. For example, in the Ayeyarwady upstream, ratios of ⁸⁷Sr:⁸⁶Sr were 0.7136 in the Ayeyarwady River (A31), 0.7090 in the Mali Kha River (A33), and 0.7185 in the N’Mai River (A32). In addition, Chapman et al. (2015) found that ⁸⁷Sr:⁸⁶Sr ratios were 0.7142 ± 0.0013 at the Myitkyina (Ayeyarwady River, about 40 km downstream from A32 and A33). Mixing water from two river branches (A32 and A33) would cause the differences in water chemistry.

Other studies have shown distinct patterns of ⁸⁷Sr:⁸⁶Sr ratios (low at river mouth and high in fresh water) in other tropical rivers (Crook et al. 2017; Tran et al. 2021; Höpker et al. 2022). The ratios tend to be globally stable in the ocean (0.70918 ± 0.00006); however, they are often highly variable in river systems due to different river bedrock types (McArthur and Howarth 2004). Although, ratios of ⁸⁷Sr:⁸⁶Sr showed clear longitudinal concentration profiles in the ARB (constantly lower in the Ayeyarwady downstream, but higher in the upstream), they were not significantly different among regions (P > 0.05). The mean water ⁸⁷Sr:⁸⁶Sr was 0.7109 ± 0.0027 (range, 0.7074–0.7206) in the ARB. The concentration of these ratios in the ARB is slightly higher than in the Mekong River at 0.7102 ± 0.0011 (range, 0.7089–0.7139) (Tran et al. 2021), but much lower than in the Daly River catchment (Northern Territory, Australia) at 0.7334 ± 0.0218 (range, 0.7093–0.7806) (Crook et al. 2017). Some trace element concentrations (e.g. Sr, Ba, Ca, and Mg) significantly differed between the mainstem and tributaries (P < 0.05): the concentrations in tributaries were higher than in the mainstem. Similarly, another study showed that these ratios were substantially different between the mainstem (0.7198 ± 0.0078) and tributaries (0.7597 ± 0.0150) in the Daly River catchment (Crook et al. 2017).

The concentrations of trace elements varied considerably between river arms at the mouth, except for Ba and Ca. For example, concentrations of Sr and Mg in one river branch (A3) were much higher than in other branches (almost two and three times higher for Sr and Mg, respectively). River mouths are dynamic environments due to effects of both sea tides and river discharge. The Ayeyarwady River is strongly influenced by sea tides. Monsoon rains usually start in June to October, and water levels peak in July or August, when a large amount of river water is pushed to the sea. Meanwhile the dry season from November to May (winter and summer) results in the salt wedge moving further upstream into the delta. The Ayeyarwady Delta exhibited semidiurnal tidal regimes (two high and two low tides daily), which led to highly dynamic variation in water chemistry in the Ayeyarwady estuary. Hydrological regimes and salt penetration strongly exhibit seasonal and dynamic variation. Mean tidal ranges were observed over 3 m at the shoreline. Such dynamic variation was also observed in the Mekong estuary (Vu et al. 2021). Brackish water often penetrates about 80 km inland in the Ayeyarwady Delta, particularly during the dry season (Kravtsova et al. 2009; Sakai et al. 2021). Similarly, tidal range in the Mekong estuary was up to 3.74 m over a tidal cycle, hence brackish water was frequently pushed up to 50 km further inland (Gugliotta et al. 2017; Vu et al. 2021).

Although our study did not examine seasonal variation in trace element concentrations in the ARB, seasonal variation of water chemistry was found in the basin by other studies. For example, Chapman et al. (2015) showed that concentrations of some trace elements (e.g. Sr and Ca) in the dry season were relatively higher than the flood season, particularly at the Hinthada station (Ayeyarwady downstream) and Monywa station (Chindwin). This study also showed that elemental concentrations had a negative relationship with river discharge in the ARB. However, ratios of ⁸⁷Sr:⁸⁶Sr were relatively stable between seasons in the ARB, except the Ayeyarwady upstream (Chapman et al. 2015). Seasonal variation in water chemistry was also found in other river systems, particularly in tropical rivers (Fukushima et al. 2014; Crook et al. 2017; Vu et al. 2021). For example, there was a strong negative relationship between river discharge and Sr concentrations, while ratios of ⁸⁷Sr:⁸⁶Sr were positively related to river discharge (Crook et al. 2017; Peucker-Ehrenbrink and Fiske 2019). It is noted that large tropical rivers such as the Ayeyarwady, Mekong, Congo, and Amazon Rivers often exhibit strong seasonal variation in hydrology (Kravtsova et al. 2009; Castello et al. 2015; Laraque et al. 2020; Vu et al. 2021). Strong seasonal variation in hydrology is likely an important factor driving variation in water chemistry. Therefore, the concentrations of most trace elements (e.g. Sr, Ca, and Mg) in the dry season were higher than in the wet season, but ⁸⁷Sr:⁸⁶Sr ratios were higher in the wet season in the ARB (Chapman et al. 2015). Such trends were also observed in other tropical river systems (Crook et al. 2017; Vu et al. 2021).

Implications for fish migration studies

The ARB is home to over 600 fish species (Vidthayanon et al. 2005; Zöckler and Kottelat 2017), with many migratory species found in a wide range of environments (e.g. fresh, brackish, marine). Most knowledge of fish migration is anecdotal and based on local knowledge (Conallin et al. 2019). This is an important knowledge gap for fish migrations in the ARB. As a result, understanding fish migrations in the basin is crucial for better management and conservation. Otoliths and other calcified structures such as spines or eye lenses are widely used to reconstruct fish life histories, as these structures permanently incorporate trace elements from ambient water (Campana 1999; Elsdon and Gillanders 2006; Limburg et al. 2015). Therefore, it is suggested that variation in elements should be examined before such studies. Most common elements (Sr, including Sr:Ca; Ba, including Ba:Ca) and isotopes (⁸⁷Sr:⁸⁶Sr) as salinity proxies were used to trace fish movements and migrations (Walther and Limburg 2012). For example, life history strategies of many Mekong fishes were revealed by otolith chemistry, contributing to better management and conservation (Vu 2022).

Our study confirmed that Sr (including Sr:Ca) is the best indicator among elements to differentiate between marine and freshwater environments. For example, concentrations of Sr at a coastal site (A1) were 63-times higher than in a freshwater site further upstream (A33). Hence, these Sr concentrations show distinct longitudinal concentration profiles in the ARB. A recent study also showed that element:Ca ratios (e.g. Sr:Ca and Ba:Ca) were significantly different between spatial scales (up, middle, and downstream) or river orders (mainstem and tributary) in the upper Nu-Salween river (Tang et al. 2024). Ratios of ⁸⁷Sr:⁸⁶Sr could have more advantages over Sr and Ba for otolith chemistry studies (Crook et al. 2017), but we found that these ⁸⁷Sr:⁸⁶Sr ratios highly varied (0.7109 ± 0.0027; range, 0.7074–0.7206) in the ARB, lower in the Ayeyarwady downstream and Chindwin, and higher in the Ayeyarwady upstream. However, they were not significantly different among these regions (P > 0.05). Interestingly, we did not expect that concentrations of Ba at a coastal site (A1) would be over two times higher than in freshwater environments (A33). Although Walther and Thorrold (2008) suggested that using four markers such as Sr:Ca, Ba:Ca, ⁸⁷Sr:⁸⁶Sr, and d¹⁸O could geo-reference most fishes, we suggest that Sr (including Sr:Ca) alone or in combination with other elements as salinity proxies, should be used in otolith chemistry studies to understand fish connections in different salinity gradients in the ARB. The interpretation of fish movements using otolith chemistry data should be cautious because these suggested proxies varied seasonally to some extent (related to river discharge) (Chapman et al. 2015).