Dams on Mekong tributaries as significant contributors of hydrological alterations to the Tonle Sap Floodplain in Cambodia

River tributaries have a key role in the biophysical functioning of the Mekong Basin. Of particular interest are the Sesan, Srepok, and Sekong (3S) rivers, which contribute nearly a quarter of the total Mekong discharge. Forty two dams are proposed in the 3S, and once completed they will exceed the active storage of China’s large dam cascade in the Upper Mekong. Given their proximity to the Lower Mekong floodplains, the 3S dams could alter the flood-pulse hydrology driving the productivity of downstream ecosystems. Therefore, the main objective of this study was to quantify how hydropower development in the 3S, together with definite future (DF) plans for infrastructure development through the basin, would alter the hydrology of the Tonle Sap’s Floodplain, the largest wetland in the Mekong and home to one of the most productive inland fisheries in the world. We coupled results from four numerical models representing the basin’s surface hydrology, water resources development, and floodplain hydrodynamics. The scale of alterations caused by hydropower in the 3S was compared with the basin’s DF scenario driven by the Upper Mekong dam cascade. The DF or the 3S development scenarios could independently increase Tonle Sap’s 30-day minimum water levels by 30± 5 cm and decrease annual water level fall rates by 0.30± 0.05 cm day. When analyzed together (DF+ 3S), these scenarios are likely to eliminate all baseline conditions (1986–2000) of extreme low water levels, a particularly important component of Tonle Sap’s environmental flows. Given the ongoing trends and large economic incentives in the hydropower business in the region, there is a high possibility that most of the 3S hydropower potential will be exploited and that dams will be built even in locations where there is a high risk of ecological disruption. Hence, retrofitting current designs and operations to promote sustainable hydropower practices that optimize multiple river services – rather than just maximize hydropower generation – appear to be the most feasible alternative to mitigate hydropower-related disruptions in the Mekong.


Introduction
More than half of the world's greatest rivers have been altered by dams (Nilsson et al., 2005) and there is worldwide evidence showing that hydropower development causes significant hydrological and ecological disruptions to downstream freshwater ecosystems (Poff and Zimmermann, 2010).Understanding the cumulative impact of water resources infrastructure is important for sustainable development of river basins, and although hydrological alterations from dams have basin-wide implications, impact assessments typically concentrate on river segments directly upstream and downstream of single dam projects (Nilsson and Berggren, 2000).Impact assessments, however, become more challenging when critical ecosystems occur further downstream under the influence of multiple dams as well as other water infrastructure components (e.g., irrigation, water supply, and flood control).The situation becomes even more complex in large rivers where the interests of upstream Published by Copernicus Publications on behalf of the European Geosciences Union.stakeholders differ from those downstream.Such is the case of the Mekong, a transboundary basin with historically low levels of hydrological regulation (i.e., the fraction of the annual water discharge that can be stored in reservoirs) comparable to other large tropical basins such as the Amazon and the Congo (Lehner et al., 2011;Nilsson et al., 2005).Aggressive plans for multiple large hydropower schemes throughout the Mekong Basin for economic development are expected to bring significant disruptions to the hydrological regime (Lauri et al., 2012;Piman et al., 2013b), compromising the geomorphology (Kummu et al., 2010;Walling, 2009), fish ecology (Ziv et al., 2012), and productivity of the downstream floodplain ecosystems (Arias et al., 2014) that sustain the food security of millions of people.
The Mekong is the largest river and basin in Southeast Asia, covering 795 000 km 2 shared by six different countries: China, Myanmar (Burma), Thailand, Laos, Cambodia, and Vietnam (Fig. 1).Mean annual discharge in the Mekong at Kratie in Cambodia is 475 km 3 year −1 or 14 500 m 3 s −1 , varying from an average of less than 3000 m 3 s −1 dur-ing March-April, to nearly 40 000 m 3 s −1 during August-September (Adamson et al., 2009).The Sesan, Srepok, and Sekong river basins (collectively known as the 3S) cover an area of 78 650 km 2 distributed among Cambodia (33 %), Laos (29 %), and Vietnam (38 %).Due to its relatively high rainfall precipitation (1100-3800 mm year −1 ), the 3S provides the largest flow contribution among Mekong tributaries, with an average discharge of 510 m 3 s −1 during March-April and 6133 m 3 s −1 during September.In general, the 3S contributes 23 % of the annual Mekong discharge, compared to 16 % generated in the Upper Mekong in China (Adamson et al., 2009).
The Mekong meets the Tonle Sap system 300 km downstream from Stung Treng at the Cambodian capital, Phnom Penh.From October to May, water flows from the Tonle Sap to the Mekong at a maximum daily discharge rate of 8300 m 3 s −1 ; when the wet monsoon reaches the basin in May, the Mekong River rises to a higher level than the Tonle Sap, forcing the latter to reverse its flow towards its lake.This phenomenon creates a floodplain that extends over 15 000 km 2 and stores up to 76.1 km 3 of Mekong's annual flood-pulse (Kummu et al., 2014).Overall, 53.5 % of the water entering the Tonle Sap system comes from the Mekong, 34 % from 11 river tributaries, and 12.5 % directly from rainfall (Kummu et al., 2014).
Hydropower development in the Mekong is occurring in three distinct regions.The first is the Lancang Jiang cascade in the Upper Mekong River in China (Fig. 1), a series of 6 dams (5 already built) with downstream hydrological alterations expected as far down as Kratie (Räsänen et al., 2012).The second focus of development is a series of 11 dams along the mainstream channel in the Lower Mekong, only one of which is under construction, the Xayaburi Dam in Laos.The Lower Mekong mainstream dams have become very controversial due to their potential impacts on fisheries (Ziv et al., 2012) and their role in political affairs among the basin's countries (Grumbine et al., 2012;Grumbine and Xu, 2011;Stone, 2011).Of greater concern in terms of hydrological alterations is the third region of development occurring in the Mekong tributaries, in particular the 3S, where at least 42 dams are at some stage of development without much regional coordination or stakeholder consultation.Because of its proximity to the Tonle Sap and the rest of the Lower Mekong floodplains, flow regulation in the 3S will most likely affect the floodplain's hydrological seasonality.Should the Tonle Sap hydrology be altered, there could be serious consequences for the ecological productivity supported by its floodplain (Arias et al., 2014).
Thus far, existing dams are believed to have caused very little hydrological alteration in the Lower Mekong (Adamson et al., 2009).There have been alterations to the frequency of extreme events beginning in the mid-1970s, but this is probably linked to changes in the El Niño-Southern Oscillation (Delgado et al., 2012;Räsänen and Kummu, 2013).Several research efforts and modeling tools have been developed to evaluate ongoing and future hydrological alterations in the Mekong (Johnston and Kummu, 2011).The primary focus of these studies has been the cumulative impact of multiple water infrastructure development plans for the basin (Lauri et al., 2012;Piman et al., 2013b;WB, 2004).Other studies have scrutinized alterations in particular regions of development, such as the dam cascade in the Upper Mekong River (Räsänen et al., 2012) and the 3S (Piman et al., 2013a;Ty et al., 2011), but links between development in these regions and impacts on the Lower Mekong floodplains have not been assessed.Impact assessments of basin-wide alterations to the Tonle Sap, however, do exist and provide a good understanding of the general trends of future changes in the floodplain.Kummu and Sarkkula (2008) initially argued that the upstream development scenario from a World Bank report (WB 2004) could increase the Tonle Sap's dry season water levels by 15 cm and decrease wet season water levels by 36 cm, leading to a large reduction of seasonally inundated areas.Arias et al. (2012Arias et al. ( , 2013Arias et al. ( , 2014) ) demonstrated that hydropower-related alterations to the Tonle Sap's hydrology could cause major disruptions to existing floodplain habitats and their contribution to aquatic primary production.
Impacts of hydrological alterations in rivers and floodplains have been well documented for decades (Petts, 1980).Hundreds of studies provide evidence that hydrological alterations cause ecological disruptions in river and riparian systems (Poff and Zimmermann, 2010), but most of these studies have been carried out in single river reaches in North America and Europe, where more than three-quarters of river discharges are regulated (Dynesius and Nilsson, 1994), and where sufficient time series exist for statistical inference of pre-and post-dam alterations (FitzHugh, 2014;Poff et al., 2007).Studies in these regions have evaluated impacts of dam development based on the scale of alterations to the magnitude, frequency, duration, timing, and rate of change of natural flow regimes required for the integrity of river and floodplain ecosystems (Poff et al., 1997).Based on these properties, a method to assess the impacts of hydrological alternations (IHAs) to environmental flows was developed (Richter et al., 1996(Richter et al., , 1997)).This method defines 32 hydrological parameters and environmental flow components (EFCs) and assesses the magnitude and statistical significance of alterations caused by flow regulation.Recent developments have been proposed to the IHA method, including the analysis of multivariate components among indicators of alterations (Gao et al., 2009) and ranking of alteration levels for specific EFCs (FitzHugh, 2014).
Most of the current construction of hydropower projects is happening in the (sub-) tropics in South America, Africa and Asia (Kareiva, 2012), where hydrological and ecological monitoring has not been carried out to the required temporal span and resolution to comprehensively use the IHA method.(It typically requires time series with at least 20 years of daily measurements; The Nature Conservancy, 2009).Perhaps the only exceptions to this regional limitation are the Murray-Darling Basin in Australia (Kingsford, 2000) and the Paraná River in Brazil (Agostinho et al., 2009), where hydrological alterations and corresponding ecological disruptions have been well documented.Despite the obvious limitations, applying the IHA method to tropical rivers under development brings interesting challenges and benefits.First, IHA can be used as an a priori impact assessment tool applied to simulated scenarios of hydropower development in order to plan optimal and sustainable dam locations and operations.Furthermore, the tool can be used to compare the level of alterations between different projects and/or cascades, thus helping prioritize where sustainable hydropower and basin management strategies are most needed.Moreover, the IHA tool could be used to evaluate the cumulative impacts of dam cascades at critical downstream river reaches and high-value ecosystems, instead of just focusing on nearby downstream impacts of a single dam.With these particular applications in mind, an assessment of hydrological alterations in the Mekong would be an informative case study not only for researchers and managers in the basin but also to others managing (sub-)tropical rivers undergoing similar development and biophysical transitions.
The main objective of this study is to quantify how the proposed hydropower dams in the tributaries of the Lower Mekong, together with the definite future (DF) plans for infrastructure development throughout the basin, would alter the hydrology of the Tonle Sap's floodplain.This was carried out by first validating a 2-D hydrodynamic model of the Lower Mekong floodplains with historical water levels at the Tonle Sap.We then compared the expected hydrological alterations caused by scenarios of 3S hydropower development and the most likely development (i.e., DF) scenario for the rest of the Mekong Basin by 2015.Once these two scenarios were analyzed separately, their cumulative impact on hydrological parameters and environmental flows at the Tonle Sap's floodplain were estimated.We conclude with a discussion of the major implications of our findings as well as feasible alternatives to mitigate the expected hydrological alteration and consequent ecological disruptions.

Modeling approach
This study integrates the results of four different sets of numerical models (Fig. 2).Basin hydrology and daily runoff flows were simulated in a daily time step using the Soil and Water Assessment Tool (SWAT) as described by Piman et al. (2013a).This SWAT model was calibrated for 28 different gauges upstream of Kratie.Sub-basin runoff flows were then used as inputs for two different models of development impacts on water resources.The first set of results came from simulations using the Integrated Quantity and Quality Model (IQQM) that Piman et al. (2013b) applied to assess the im-  1).Water movement from this location down through the Lower Mekong floodplains (including the Tonle Sap) was simulated with the two dimensional Environmental Impact Assessment Model (2-D EIA), a hydrodynamic model that numerically solves the simplified Navier-Stokes and continuity equations using a finite difference method (Koponen et al., 2010).The 2-D EIA Lower Mekong application covers an area of 430 km by 570 km from Kratie to the Mekong Delta at a grid resolution of 1 km 2 .An earlier version of this application was presented by Västilä et al. (2010).Daily water levels from the 2-D EIA model were extracted and validated at Kampong Luong, where the main water gauge on the Tonle Sap's lake is located (see Fig. 1).Simulated water levels were validated against historical measurements for the entire simulation period (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000).Validation results were evaluated according to the linear correlation coefficient (r) between observed and simulated results, as well as the Nash-Sutcliffe efficiency coefficient (Nash and Sutcliffe, 1970).

Modeling scenarios
A total of four scenarios were considered for this study (Table 1).A baseline scenario (BL) represented recent historical conditions (1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) before major hydropower projects were built in the Upper Mekong and the 3S.We were limited to this 15-year time series because no continuous and reliable water level data exist for the Tonle Sap before this period, and because building of large dams began after 2000.Two of the wettest and five of the driest years in the past seven centuries occurred during this baseline period (Räsänen et al., 2013), and it was therefore considered a good representation of the range of historical hydrological conditions and variability in the basin.Although 17 dams were already operational by the end of this period (including the Manwan dam in China built in 1993), they were generally small and only accounted for 9.1 km 3 of active storage, in contrast to the approximately 38 km 3 that have been built since the year 2000 (MRC, 2009).The first scenario of water resource development that was analyzed resembles the definite future (DF) scenario proposed by the Mekong River Commission (MRC; Piman et al., 2013b), which represents existing and on-going water resources infrastructure development up to 2015.The DF scenario is primarily driven by the six dams in the Lancang Jiang dam cascade in the Upper Mekong in China, which provide an additional 23.2 km 3 of active storage from baseline (Räsänen et al., 2012).The DF scenario in our study does not consider any proposed dams in the 3S.The second scenario of water resource development was based on the simulations of dams operations in the 3S presented by Piman et al. (2013a).This scenario represents a total of 42 dams at different development stages (existing, under construction, and proposed) in the 3S tributaries and sub-tributaries with 26.3 km 3 of active storage.The last scenario analyzed represents the cumulative impact of both DF and 3S (DF + 3S) with an additional 49.5 km 3 of active storage from the baseline.All simulations were carried out in daily time steps for a period of 15 years from 1 January 1986 to 31 December 2000.

Data analysis
Simulated water levels were used to calculate 30 hydrological parameters and corresponding alterations using the IHA Tool (The Nature Conservancy 2009).This tool computes hydrologic parameters that are relevant to ecosystem processes and it calculates the level of alteration between the baseline and post-alteration periods.Analyses were carried out by combining the BL scenario time series with each of the water development scenarios so that the first 15 years defined the pre-alteration period and the second 15 years represented the post-development period, as if all dams were built at once on January 1, 2001.Three different sets of analyses were carried out: the DF, 3S, and combined DF + 3S scenarios.All analyses were carried out using non-parametric statistics.Data were analyzed according to calendar years (1 January to 31 December).Environmental flow components were set according to commonly used parameters.The water level 75th percentile for each year was defined as the threshold between periods of low flow and high flow pulses.Small floods were defined as those with a peak above the 2-year return period flood, whereas large flood events were  defined as those with a peak above the 10-year flood.Extremely low flows were defined as those with an initial low flow below the 10th percentile of the daily records for each period.
Annual summary statistics were used to compare the magnitude of alterations between scenarios.All hydrologic parameters were analyzed with the Kruskal-Wallis test (Kruskal and Wallis, 1952) to determine if differences among the BL, DF, and 3S scenarios were significant at the 95th percentile level.Once individual scenarios were compared, hydrological alterations were calculated for the DF and for the DF + 3S scenarios.Environmental flow components were estimated, exceedance probability charts plotted, and hydrological alteration factors computed for all parameters according to the range of variability approach (RVA; Richter et al. 1997).This approach consists of dividing the data into three different categories (bounded by the 33rd and 67th percentiles), estimating the frequency at which values are expected to occur within each category, and then estimating the percentage difference between the expected frequency and the simulated frequency for the impact scenarios.
In addition to the IHA analysis, changes in spatial flooding patterns were analyzed.Raster images representing cumulative flood duration were generated from the 2-D EIA model throughout the geographical extent of the Tonle Sap's floodplain (15 000 km 2 approximately).These were transformed into flood frequency rasters by normalizing flood duration according to the simulation's total length.Outputs from the impact scenarios were overlaid on the baseline raster in order to calculate and visualize spatial changes in flood regime.
The linear correlation coefficient between the observed and simulated daily water levels was 0.97 and the Nash-Sutcliffe efficiency coefficient was 0.91.

Comparison between BL, DF, and 3S scenarios
Overall, similar scales and alteration trends between the DF and the 3S scenarios were found.Of the 30 hydrological parameters analyzed, 9 appeared to be significantly different (p ≤ 0.05) in either the DF or the 3S scenario when com- pared to the BL scenario (Table 2): April and May monthly water levels, water fall rate (that is, the difference between the annual minimum and maximum water levels divided by the duration between them), base flow index (that is, the 7day minimum over the mean annual water level), and 1, 3, 7, 30, and 90-day minima.None of the parameters appeared to be significantly different between the DF and the 3S scenarios.Box plots of some of the most representative parameters were prepared in order to demonstrate the general trends countered in this comparison (Fig. 4).For instance, the 30day minimum water level median was 1.52 m (with a range of 1.22-2.18m) for the BL scenario, which is significantly different from 1.84 m (1.51-2.48m) and 1.80 m (1.50-2.46m) for the DF and 3S scenarios, respectively (Fig. 4a).Water level fall rate for the BL (median of 3.2 cm day −1 , range 3.0-3.6cm day −1 ) was also significantly different from DF (median of 2.8 cm day −1 , range 2.7-3.4 cm day −1 ) and 3S (median of 2.9 cm day −1 , range 2.7-3.4 cm day −1 ; Fig. 4c).In contrast, the maximum annual water level of the BL scenario (median of 8.58 m, range of 7.42-9.67m) was not found to be significantly different from the other development scenarios (Fig. 4b).

Cumulative hydrological alteration from the combined DF and 3S scenario
The results of the simulations with the cumulative effects from the DF + 3S scenarios suggest that there could be a significant impact on the overall Tonle Sap flood regime.In terms of environmental flows, the cumulative impact of the DF + 3S scenario virtually eliminates all baseline extreme low flow conditions (Fig. 5): the frequency of these events is reduced from 11 to just one event in 15 years.Moreover, the BL scenario shows that high flow pulses and floods occur every single year, but that the frequency of these events decreases to two every three years.
Changes in the flood regime of the Tonle Sap are also reflected in the probability of water level exceedance (Fig. 6).Greatest deviations occur at exceedance levels above 70 %: for instance, 2.36 m corresponds to the 80 % exceedance level in BL, but this increases to 2.62 and 2.80 m for the DF and the DF + 3S scenarios, respectively.Mild declines occur at the 20 % exceedance level, but much milder changes were found for the greatest amplitude (and less frequent) events.
Hydropower development through the Mekong and its tributaries would alter multiple seasonal and annual hydrological parameters.The greatest alteration factors are expected to occur during the dry season months, with large alteration factors for monthly water levels during April and May, as well as other parameters including the 1-day, 3-day, 7-day, 30-day, and 90-day minima (Table 3).The DF scenario decreases the frequency of occurrence of the baseline dry season parameters by 40-60%), but the addition of the 3S hydropower network (DF + 3S) results in alteration factors of −100 % for all of these parameters (meaning that they are expected to be altered every year).Factors of alteration in annual rates of water rise by −33 % and fall by −20 % for the DF scenario, but the magnitude of alteration factors increases to −83 and −60 % for the DF + 3S scenario (Table 4).

Changes in flood duration
Both the DF and DF + 3S scenarios could bring changes to the long-term spatial patterns of inundation throughout 51-60 % of the Tonle Sap's floodplain (Fig. 7).In general, areas that are marginally inundated and areas that are permanently inundated are likely to expand, whereas areas that are seasonally inundated are likely to decrease.For instance, areas a CD is the coefficient of dispersion, i.e., the (75th percentile − 25th percentile)/50th percentile; b RVA is the range of variability approach; c HAF is the hydrologic alteration factor: the percentage difference between the expected baseline frequency and the simulated frequency for the impact scenarios.in the outermost class (inundated between 0.5-10 % of the time) expand by 177 km 2 (10.1 %) and 283 km 2 (16.1 %) as a result of the DF and the DF + 3S scenarios, respectively (Table 5).Moreover, the largest area shifts occur in areas inundated 90-100 % of the time, which expand by 279 km 2 (5.7 %) and 424 km 2 (8.6 %) as a result of the DF and the DF + 3S scenarios, respectively.On the contrary, classes inundated 20-90 % shrink by 600 km 2 and 994 km 2 as a result of the DF and the DF + 3S scenarios, respectively.

Discussion
This study presents an important contribution to the assessment of water resource management and development in the Mekong Basin.We have combined multiple hydrological modeling tools -all of which have been previously validated for the basin -and simulated the specific and combined impact of water resource development in two regions of great hydrological contribution to the whole basin.Piman et al. (2013a) had already showed that the scale of hydropower development in the 3S was as large as the Lancang Jiang dam cascade.In this study, we have taken a step further and shown that the corresponding hydrological alterations from the 3S hydropower projects are as large; perhaps more importantly, we have demonstrated that the cumulative effect of development in the Upper Mekong and the 3S will cause significant disruption to the inundation patterns of the Lower Mekong floodplains, in particular through an increase in dry season water levels as well as a reduction in water level rise and fall rates.
Our study (intentionally) assumed that there were no changes in rainfall-runoff from one simulation to the other in order to solely explore the issue of water regulation in tributary dams.This assumption, however, is not a complete representation of changes to the basin's hydrologi-cal cycle, as there are other key factors such as climate change (Kingston et al., 2011;Lauri et al., 2012), new irrigation schemes (Piman et al., 2013b), and land use/land cover changes (Costa-Cabral et al., 2007;Ishidaira et al., 2008) that alter rainfall-runoff characteristics and thus simultaneously affect the role of the 3S on the Tonle Sap hydrology.As Ty et al. (2012) demonstrated for one of the 3S rivers (Srepok), these other factors could also cause alterations, particularly a decrease in water availability during the dry season.Räsänen et al. (2014) showed that for the Sesan dam cascade in Vietnam irrigation water use during the dry season was relatively small compared to the increase in water flow caused by hydropower dams.In short, there is a great need for detailed modeling studies that take into account all of these major drivers of hydrological alterations.
This study demonstrated the use of IHA tools to assess the impact of future scenarios of water resources development.Although this tool has been previously used for simulated scenarios by Gao et al. (2009), their scenarios represented hypothetical reservoirs and dam operations, whereas our study represented existing and proposed projects based on actual design characteristics.IHA tools have been used in the Mekong by Ty et al. (2011) and Thompson et al. (2013), but their applications focused on climate change and excluded the Tonle Sap flooding characteristics.Our study has made a first attempt at quantifying environmental flows for the Tonle Sap using the simulations of baseline conditions, and our estimates could help guide environmental flow criteria based on the specific biological needs of this system.As the validation results showed, however, our model scheme had a slight tendency to overestimate historical dry season water levels; for that reason, the reported magnitude of water levels defining extreme low flows should be read with caution as they might be marginally higher than historical observations.We recommend that a closer analysis using long term observed water level records is carried out in order to more a CD is the coefficient of dispersion ([75th percentile -25th percentile]/50th percentile); b RVA is the range of variability approach; c HAF is the percent difference between the expected baseline frequency and the simulated frequency for the impact scenarios.
accurately define environmental flows and monitor ongoing alterations to these parameters.
Previous studies (Arias et al., 2012(Arias et al., , 2014) ) also assessed the impacts of water resources development on water levels and flood duration at the Tonle Sap.These previous studies used three representative hydrological years (dry, average, and wet) in order to characterize multiyear variability.In general, it was found that hydrological alterations increased from wet to dry years.While the results from this study still support this trend in representative years, we found that over a longer time series only alterations to dry season water levels are expected to recur.Furthermore, our estimates of dry season water level alterations for the DF scenario are consistent with values previously reported (Arias et al., 2012;ICEM: 2010), whereas our estimates for the DF + 3S scenario (+47 and +61 cm for April and May, respectively) are considerably larger than any of the MRC future development scenarios previously reported (maximum of +33 cm in April and +39 cm in May; MRC, 2010).This difference highlights the significance of tributary dams to the hydrology of the entire basin and the importance of modeling their dimensions and operations in detail.Differences between the DF + 3S scenario and previous estimates could also be partially attributed to water abstraction for irrigation during the dry season not considered in this study.A previous comparison of alterations from hydropower dams versus cumulative alterations of hydropower with irrigation did not show any major differences in the Lower Mekong (Piman et al., 2013b).In order to more comprehensively address this issue, further modeling studies in the Mekong should compare the effects of hydropower with irrigation development.
Significant hydrological alterations are expected in the Tonle Sap and the rest of the Lower Mekong floodplains if proposed hydropower development plans are acted upon.Ongoing trends and large economic incentives in the hydropower business tend to suggest that most of the hydropower potential will actually be exploited and dams will be built even in locations where there is a high risk of disruption to environmental flows.For instance, some of already operating dams in the 3S, such as the Yali Falls Dam (commissioned in 1994) on the Sesan River, were built without much consideration of transboundary environmental impacts and have in fact caused much damage in downstream Cambodia (Wyatt and Baird, 2007).Clear evidence of more recent trends are the Xayaburi Dam in the Mekong mainstream in Laos and the Lower Sesan 2 Dam at the confluence of the 3S tributaries in Cambodia.Both of these dams have been already commissioned despite not only being highlighted as having potentially large ecological impacts in the scientific literature (e.g., ICEM, 2010;Ziv et al., 2012) but also after causing much controversy in the international media.Under a likely, "development as usual" scenario, the most feasible alternative to mitigate the disruptions in the Mekong of both existing and proposed dams consist of retrofitting controls to current design and operational practices in or- der to optimize river services rather than just maximize hydropower generation.In other words, seasonal and diurnal operation rules should also aim at minimizing hydrological alterations downstream in addition to meeting electricity demands.From a hydrological point of view, run-of-the-river designs or operations in which power is gained primarily from flow volume and not elevation head would result in reduced environmental flow alterations.In addition to hydrological considerations, there are other aspects, such as sediment release and fish passage, that need to be implemented.These factors have not been widely considered in assessment studies in the Mekong (with the exception perhaps of Kummu et al., 2010;Ziv et al., 2012) and should therefore be the subject of further research.

Conclusions
This paper presented a study in which hydrological modeling and assessment tools were used to provide evidence of the expected hydrological alterations that hydropower development in the Lower Mekong tributaries could bring to the Tonle Sap.Hydrological alterations caused by dams in the 3S were of similar magnitude as the DF scenario, which resembles water infrastructure development up to 2015 particularly driven by China's Lancang Jiang dam cascade in the Upper Mekong.DF plans in combination with the full development of the 3S dam network will most likely cause significant and undocumented hydrological alterations to the Tonle Sap and the rest of the Lower Mekong floodplains.
The most significant alterations are in terms of water levels during the dry season (April and May) and rates of water level rise and fall; these hydrological parameters are crucial for biological factors such as tree seed germination and fish migration, and therefore major ecological disruptions are likely to follow.Although there could be a decrease in wet season water levels in years of low flow from the Mekong, wet season disruptions are not recurrent in years of larger floods.Given the importance of the 3S to the rest of the Lower Mekong, we recommend that more detailed studies of drivers of hydrological change in the 3S are carried out, including irrigation, land use/land cover conversion, and climate change.Moreover, optimization of hydropower operations considering both electricity generation and environmental flows should be sought as a feasible alternative, and further studied and implemented in existing and proposed dams in this critical tributary.

Figure 1 .
Figure 1.Map of the Mekong Basin highlighting its floodplains and dams in the definite future (black dots) and 3S development scenarios (violet triangles).The green triangle shows the Kampong Luong water level gauge location on the Tonle Sap.

Figure 2 .
Figure 2. Models used and their general features.DF is definite future.

Figure 4 .
Figure 4. Boxplots of (a) 30-day minimum water level, (b) 30-day maximum water level, and (c) water fall/drop rate for the baseline (BL), Definite Future (DF), and 3S hydropower development (3S) scenarios.30-day minimum and drop rate for both DF and 3S are significantly different from BL (p ≤ 0.05).Water fall/drop rate refers to the difference between the annual minimum and maximum water levels divided by the duration between them.There are no significant differences in 30-day maximum water level among scenarios.

Figure 5 .Figure 6 .
Figure 5.Comparison of daily water levels and environmental flow components between the baseline scenario (BL) and the combined effect of the definite future and the 3S hydropower scenarios (DF + 3S).

Figure 7 .
Figure 7. Maps representing the duration of flooding during the 15-year simulations: (a) shows the baseline (BL) flood duration map as percentage of total simulation time, while (b) shows the difference in flood duration between the combined definite future and 3S hydropower scenarios (DF + 3S) and the baseline.Areas with an expected increase in flood duration are shown in blue, those where a decrease in flood duration is expected in the marginally inundated areas are shown in red, green and orange.

Table 4 .
Summary of annual hydrological parameters.Largest alterations factors (HAFs) estimated for base flow index and water level fall rate.

Table 1 .
Description of water infrastructure development scenarios.

Table 2 .
Kruskal-Wallis test resultsfor comparison of annual parameters.Each column group represents a one-to-one comparison between the baseline (BL), definite future (DF) and 3S hydropower (3S) scenarios.χ 2 is the test statistic and p is the probability value of χ 2 .Significant p values (≤ 0.05) are highlighted in bold.

Table 3 .
Summary of monthly and minimum/maximum hydrological parameters.Largest alterations occur during the dry season (April, May, 1-day min, 3-day min, 30-day min, 90-day min).

Table 5 .
Changes in spatial patterns of flooding in the Tonle Sap.