Replumbing the Bengal Basin: A Shift in Recharge Driven by 1 Groundwater Irrigation Revealed by Stable Water Isotopes 2

17 Groundwater supports agriculture and provides domestic water for over 250 million 18 people in the Bengal Basin. Our analysis of stable water isotope ratios in rain, surface, and 19 groundwater shows that the proportion of groundwater recharge originating from stagnant surface 20 water bodies has increased by about 50% over the last seventy years while the relative contribution 21 from direct infiltration of rain has decreased. This regional shift in the source of groundwater 22 shows how the simultaneous expansion of irrigated rice, excavated ponds and groundwater 23 pumping has changed the hydrologic system by cycling evaporated standing water through the 24 subsurface. Analysis of water isotope data also reveals that most recharge from standing water 25 enters during the latter part of the dry season (February-April), while most rainwater recharge 26 occurs in the early months of the monsoon (June-August) before aquifers fill to capacity and reject 27 additional recharge of rainwater. 28


29
Groundwater is the main source of domestic and agricultural water for more than 250 30 million people living in the Bengal Basin of Bangladesh and India that lies in the downstream 31 floodplains and deltas of the Ganges, Brahmaputra and Meghna Rivers (1-3). Large-scale 32 abstraction of groundwater over the last six decades for dry-season agriculture has increased 33 recharge to groundwater (4-7), caused the water table to decline further during the dry season and, 34 in some parts of the basin, prevented the water table from rebounding as high during the monsoon 35 (1, 6). Pumping draws more shallow water to deeper aquifers increasing the risk of contaminating 36 relatively low-arsenic deep groundwater aquifers (8,9) and making coastal aquifers more 37 vulnerable to seawater intrusion (10). Understanding the ramifications of large-scale pumping on 38 groundwater dynamics is important for future sustainability of both the quantity and quality of 39 groundwater, the main source for household and irrigation water in the basin (11). 40 Groundwater in the Bengal Basin has been assumed to be recharged by direct monsoon 41 precipitation (9,(12)(13)(14)(15) largely based on the fact that rainfall exceeds potential evapotranspiration. 42 Several site-specific studies have challenged this notion by finding that shallow groundwater 43 aquifers today are recharged by rivers, local ponds and rice field water (16)(17)(18)(19)(20). Irrigation pumping 44 induces downward head gradients beneath these surface sources that increase as the water table  45 falls towards its dry-season minimum. Because recharge occurs while the water table is falling 46 annual recharge cannot be estimated by analysis of the rising portion of the groundwater 47 hydrograph alone. Groundwater hydrographs alone do not provide enough information to estimate 48 recharge. 49 The contribution of the different sources in the geomorphically active Bengal Basin can 50 vary spatially and temporally. Older groundwater may contain recharge from surface water bodies 51 that no longer exist because river meandering and avulsion moves channels across the landscape 52 over decades and centuries (21). For instance, the Brahmaputra flowed on the eastern side of the 53 Dhaka until it avulsed about 200 years ago and now flows about 80 km west of Dhaka (22,23). 54 Hydrograph analyses show that groundwater recharge has increased in the last three decades due 55 to intensive groundwater abstraction for irrigation. These studies argue that the increased recharge 56 is sourced from direct rain infiltration and river water (6, 24) during the monsoon. However, prior 57 regional scale studies have not (1) quantified the contributions of different recharge sources, (2) 58 accounted for recharge from ponds and rice fields (3) analyzed how recharge sources have changed 59 over decades and centuries. 60 Regional water and land use changes suggest that ponds and rice fields could contribute 61 significantly to groundwater recharge primarily during the dry season. In the last 35 years, the 62 number of irrigation wells has increased from < 100,000 to >1.7 million (25) primarily to support 63 dry-season irrigation that has also correspondingly increased from <1 million hectares to 5 million 64 hectares (~20% of the total land) (26,27). Groundwater abstraction in Bangladesh is estimated to 65 be >33 km 3 in 2010, 95% of which is used to support dry season farming (28); a volume that can 66 fill the 5 million hectares with a water depth equaling 650 mm. The volume of groundwater 67 abstraction in Bangladesh and West Bengal (India) has only increased in recent times (25). At 68 least 10 million hand-pumped wells have also been installed in the region, but they account for 69 only a minor fraction of total pumping (29). Man-made ponds cover an estimated 2.6% of the area 70 of Bangladesh, with more than 4 million households owning a pond (30); most of these ponds are 71 recent and have been constructed in the past 70 years (31), although the number of ponds may be 72 stabilizing post-2000 (32). 73 Evidence of rapid cycling of water through shallow aquifers, combined with the recent 74 declines in the water table elevation and the growth of standing water in rice fields and ponds, 75 motivated us to investigate how groundwater recharge may have changed over time. We postulate 76 that: 1) rice fields, pond water and rivers have now become a major contributor to groundwater 77 recharge across the basin and 2) the majority of the recharge from standing water occurs during 78 the latter part of the dry season (February to April) when groundwater levels reach their minima 79 so that large vertical head gradients draw more recharge from standing water into the available 80 aquifer storage. 81 In this work, we rely on radiogenic and stable water isotopes to answer our hypotheses. 82 We first establish the regional groundwater age profile by compiling existing radiocarbon and 83 tritium groundwater data from the region.

100
Radioactive isotopes in groundwater indicate that in the upper 100 meters of the aquifer 101 system much of the groundwater was recharged contemporaneously with the onset of groundwater 102 irrigated rice and the expansion of man-made ponds (37). Tritium active water is pervasive up to 103 a depth of 100 m ( Figure 1A) indicating modern recharge (post-1953)  Stable water isotope data 118 We combined stable water isotope data for precipitation, groundwater and surface water 119 (rivers, ponds and puddled water in rice fields) for the Bengal Basin from 31 sources (Table S1)  120  that together provide 580 precipitation, 1918 groundwater and 487 surface water values.  121 Precipitation samples were collected at Barasat and Kolkata in India and at Savar,Barisal,Sylhet,122 Chittagong and Cox's Bazar in Bangladesh (15,[40][41][42][43]. The distribution of precipitation stations 123 provided an extensive spatial coverage of precipitation isotope ratios from the basin (Figure 2A). 124 Groundwater samples were evenly distributed across Bangladesh and the bordering region 125 between Bangladesh and West Bengal lying to the East of the Hooghly River (tributary of the 126 Ganges, Figure 2). Some locations had high densities of groundwater isotope data as they have 127 been studied extensively for groundwater arsenic (17,19,(44)(45)(46). The lack of isotopic data from 128 the districts west of the Hooghly River in West Bengal is largely due to the fact that they have 129 been investigated to a lesser extent for groundwater arsenic contamination. 130 For all of the investigated precipitation sampling stations across the Bengal Basin ( Figure  131 2) isotope values varied seasonally with the heaviest values during the dry season (January-April), 132 followed by decline isotope ratios at the onset of monsoon and the lightest values during late 133 monsoon (September/October, Figure 3A and 3B). This seasonal pattern in the precipitation 134 isotope ratios is consistent across the Bengal Basin (43, 47). Rainfall is highly seasonal, and the 135 majority of precipitation occurs during the monsoon months of May to October ( Figure 3C). The 136 stable isotope ratios of precipitation showed a wide range of values from -120 to 25‰ (δ 2 H) and -137 15 to 5‰ (δ 18 O). The amount-weighted mean δ 2 H and δ 18 O of annual precipitation is -46.5‰ and 138 -6.9‰ respectively. We define the local meteoric water line (LMWL, δ 2 H = 8.2 δ 18 O + 11.2, R 2 = 139 0.97, p <0.05) as the best-fit regression line through all the precipitation data. 140 We divided surface water samples in groups representing (a) large rivers (Ganges and 141 Brahmaputra) and (b) standing water (small rivers, ponds and rice fields). The Bengal Basin is a 142 deltaic system with an intricate network of streams and rivers. In this analysis, by large rivers we 143 refer to the main Ganges and Brahmaputra channel as well as the tributaries feeding from the main 144 channels -a large proportion of water in these rivers is derived from Himalayan snowmelt and 145 higher altitude precipitation (48). This water is isotopically lighter (49) than the amount weighted 146 mean precipitation isotope ratios of the Bengal Basin ( Figure 4A). We define standing water as 147 waterbodies that are recharged primarily from the local precipitation in the Bengal Basin and 148 subsequently undergo evaporative enrichment during the dry season. This includes: 1) Ponds that 149 are filled up during the monsoon and are depleted by human consumption and evaporation during 150 the dry season, 2) Small rivers that do not contain water from the large rivers and may become 151 stagnant during the dry season, 3) Standing water in irrigated rice fields. 152 Samples from large rivers (Ganges and Brahmaputra) were isotopically lighter than the 153 standing water samples ( Figure 4A) In contrast to large rivers, standing water samples showed a large range of isotope values 162 ( Figure 4A). Seasonal measurements suggests that standing water undergoes progressive 163 evaporative enrichment during the dry season (November to April) and the heaviest isotope values 164 were observed towards the end of the dry season in March/April (13,19). The evaporation slope 165 obtained by regressing through the standing water data was 5.6 (δ 2 H = 5.6 δ 18 O -7.6, R 2 = 0.90, p 166 <0.05, Fig 4A), very similar to the theoretically calculated evaporation slope (between 5 and 6) in 167 the Bengal Basin (50) indicating evaporative enrichment of standing water bodies. 168 The stable isotope ratios of groundwater showed a wide range of values from -64.0 to 5.3‰ 169 (δ 2 H) and -9.6 to 0.1‰ (δ 18 O). The mean groundwater δ 18 O in shallow aquifers (0-50 m deep) are 170 heavier (mean ± σ = -4.1 ± 1.3‰) than intermediate (50-100 m, -4.6 ± 1‰) and deep (>100 m, -171 4.7 ± 1‰) aquifers. While some shallow groundwater (0-50 m depth) samples cluster near the 172 LMWL, many samples also fall below this line indicating groundwater recharge from both 173 unevaporated and evaporated sources ( Figure 4B). Intermediate (50-100 m depth) and deep (>100 174 m depth) groundwater samples cluster near the LMWL suggesting recharge dominantly from 175 unevaporated water ( Figure 4C and 4D Groundwater isotope mixing models 180 Groundwater in the Bengal Basin can be recharged by multiple sources including direct 181 rainfall, large rivers (Ganges, Brahmaputra and their tributaries), bodies of standing water 182 (irrigated rice fields and ponds), small rivers, or a combination of these sources. The contribution 183 from the different sources can vary locally depending upon the precipitation, proximity of the 184 aquifer to rivers, the hydrological connectivity between the rivers and the aquifers, and the density 185 of standing water bodies (ponds and rice fields) above the aquifer. Furthermore, these sources and 186 their relative contributions have likely varied over time due to changes in climate, land use, and 187 water use. 188 These sources can be divided into three isotopic endmembers. First, the isotopically light 189 water in Ganges and Brahmaputra and their tributaries ( Figure 4A). Second, precipitation isotope 190 ratios (δ 2 H and δ 18 O) that exhibit a large range, however, are strongly correlated and fall on or 191 close to the LMWL ( Figure 3A). Third, ponds, local rivers and rice fields isotope ratios that show 192 strong seasonal variation. Samples collected from ponds and local rivers during the monsoon 193 season lie close to the LMWL and undergo evaporative enrichment during the non-monsoon 194 season (November to April) which makes then isotopically heavier than local precipitation ( Figure  195 4A). Similarly, standing water in rice fields (pumped from the groundwater) undergo evaporative 196 enrichments during the dry season and gets evaporatively enriched over time. Thus, there is a clear 197 distinction between non-evaporated water sources (large rivers and precipitation) and evaporated 198 standing water sources (local rivers, ponds and rive fields, Figures 4 and S3). With this variation 199 in the isotopic signature (δ 2 H and δ 18 O) of the endmembers, we model the proportion of 200 groundwater isotope ratios as a mixture of precipitation, standing water and large rivers. 201 Several combinations of precipitation, river and standing water isotope values can explain 202 the observed isotope ratios of groundwater samples ( Figure S3B and 5). Therefore, we use a 203 Monte-Carlo approach to find non-parametric distributions that describe possible mixes of 204 recharge reflected by groundwater isotope values. We constructed three different mixing models, 205 each with uniquely different assumptions about the source of recharge, to test the sensitivity of the 206 results to these assumptions. Complete details are provided in Methods section. 207 In the first model, groundwater isotope ratios are modeled as a linear mixture of a 208 precipitation and standing water sample ( Figure 5A), without any input from large rivers (Ganges 209 and Brahmaputra). This assumes that groundwater is recharged primarily from precipitation and 210 local surface waterbodies. The logic behind this assumption is that for regions not lying in the 211 river floodplain it is unlikely that rivers contribute significantly to shallow groundwater recharge. 212 Additionally, for sites in proximity to the Ganges and Brahmaputra, groundwater flow is toward 213 the river, except during the latter part of the dry season when flow may be reversed as the 214 groundwater head falls below the river water level due to extensive pumping (51, 52). During the 215 early monsoon, some local recharge along the Brahmaputra may take place driven by a rising river 216 stage from snowmelt in the headwaters. Water level in Ganges rises later in July (6), hence it does 217 not contribute to early monsoon recharge. 218 In the second model, we considered a 3-endmember mixing model with large rivers, 219 precipitation, and standing water as the potential sources ( Figure 5B) -any combination of these 220 sources could recharge the groundwater. The third model is also a 3-endmember model with a 221 standing water endmember and two precipitation endmembers ( Figure 5C), but no river water. The 222 idea behind the third model is to account for the possibility that groundwater samples contain a 223 mix of two, rather than just one, precipitation sources with distinct isotopic compositions. Because 224 precipitation isotope ratios vary systematically during the monsoon and transition from heavier to 225 lighter δ 18 O and δ 2 H values from early to late monsoon, this model can quantify the relative 226 contributions of each, and in doing so, the timing of recharge by precipitation. This model is similar 227 to model 1 in regard to the source endmembers used in the model but generates an independent 228 measure of contributions from standing water. 229 230

231
We observed a large scatter in the δ 18 O and d-excess profile in shallow wells with values 232 ranging from -7 to 0‰ (δ 18 O) and -2 to 12‰ (d-excess, Figure 6C) The moving depth-average groundwater δ 18 O decreased from -3.8‰ to -5.1‰ between 7 240 m and 100 m deep ( Figure 6A). In the same depth interval, deuterium-excess increased from 4.6‰ 241 to 7.2‰. Between 100 m and 250 m, the moving depth-average groundwater δ 18 O increased from 242 -5.1‰ to -3.8‰. The corresponding d-excess values did not change appreciably, increasing 243 slightly from 7.2‰ to 7.9‰ ( Figure 6C). 244 All three models, despite with their different assumptions, result in similar estimates of 245 standing-water contribution. Model 2, which includes river input, shifts some of the estimated 246 rainwater contribution to the river as both are unevaporated sources (i.e., lying on the LMWL) but 247 still yields about the same contributions of the evaporated standing water source over depth. 248 Precipitation is the largest contributor to groundwater recharge in all the mixing models. Temporal changes in groundwater isotope ratios 271 The distinct trend in the depth-average groundwater isotope ratios (δ 18 O and δ 2 H, Figure  272 6) where it decreases from 7 m depth to 100 m depth and subsequently increases from 100 m depth 273 to 267 m depth suggests that either the relative contribution of the isotopically distinct sources 274 recharging groundwater has changed over time or the isotope ratios of the sources recharging 275 groundwater have changed. In the Bengal Basin, it appears that both factors have contributed to 276 variations in groundwater isotope ratios over time.

277
Enriched δ 18 O and δ 2 H values in shallow aquifers were associated with lower d-excess and 278 vice-versa (r = -0.6); the inverse relation between the stable isotopes (δ 18 O and δ 2 H) and d-excess 279 provides evidence that the pattern in moving depth-average groundwater isotope ratios between 7 280 to 100 m depth is primarily due to increase in contribution of unevaporated sources as we move to 281 deeper depths. In other words, more recently recharged water (i.e., shallower depths) have higher 282 contribution from evaporated sources (ponds, rice fields and local rivers during the dry season) 283 and deeper water have lower contribution from evaporated sources. It is very unlikely that these 284 patterns in the top 100 meters of groundwater represent changes in precipitation isotope ratios 285 because most of this water recharged in the last millennium, much more recently than the last 286 changes in rainfall isotope ratios that occurred about 10,000 years ago (45, 53). 287 Between 100 m and 267 meters depth, δ 18 O and δ 2 H reversed the trend observed between 288 7 and 100 m depth and become progressively enriched. However, the trend in d-excess did not 289 reverse; instead, it increased nominally by 0.8‰ ( Figure 6) suggesting that the changes in 290 groundwater recharge are not driven by increase in contribution of evaporated water. Groundwater 291 below 100 m is, on average older than 4000 years and below 200 m is, on average, older than 292 10,000 years ( Figure 1B). In the Bengal basin, precipitation isotopes in the Pleistocene were 293 isotopically heavier than in Holocene precipitation (53). Thus, the observed pattern in the depth-294 average groundwater isotope ratios is most likely due to changes in the precipitation endmember 295 values; the groundwater isotope ratios become progressively heavier with increasing depth as the 296 component of older Pleistocene precipitation water increases. 297 298 Changes in groundwater recharge sources over time 299 There is a broad consensus that large-scale pumping of groundwater has altered 300 groundwater dynamics in the Bengal Basin by lowering the water table below large cities and  301 intensively irrigated areas, drawing modern water into deeper aquifers and reversing stream-302 groundwater exchange during the dry-season (3-5, 8, 18, 51, 54). Although field studies have 303 suggested changes in recharge sources in response to groundwater pumping -mostly increased 304 contribution from rice fields and ponds (16-19, 55) -a basin-wide shift in the sources recharging 305 groundwater has not been reported. Previous and recent basin scale work has mostly focused on 306 diffused and focused recharge during monsoon (6, 24) and have failed to explicitly consider the 307 possible recharge from ponds and rice fields at regional level. 308 The trends in isotopic composition are mirrored by changes in source attribution between 309 shallow, recent groundwater and older, deeper groundwater and provide clear evidence that recent 310 human perturbations have affected groundwater recharge sources on a massive scale. In all the 311 models, we observed a sharp transition in the contribution of standing water at the depth of 75 m 312 (Figure 7). The standing water contribution decreases as we move downwards from 7 m to 75 m. 313 However, between 75-100 m depth, the fractional contribution of the different sources does not 314 change appreciably. These differences in contribution with depth suggest that the proportion of 315 recharge from the different sources recharging the groundwater has changed over time. 316 We interpret our results within the framework of the available groundwater dating (C14 317 and tritium concentration) to suggest that the relative contributions in recent times have shifted to 318 contain more standing water sources that have been subjected to evaporation. Between 7 and 100 319 m depth, where the contribution of evaporated water is greatest, groundwater is dominantly 320 modern, with more than 80% samples containing high tritium levels. The average tritium 321 concentration decreases with depth providing evidence that the more recent the recharge is (i.e., 322 shallower depths), the higher is the contribution of evaporated water. Recent work have suggested 323 that total recharge has increased in the Bengal Basin in response to widespread pumping, however 324 these studies could not partition the contribution from the different sources (24). Groundwater 325 isotopic data and our mixing model suggests that the increased groundwater abstraction have not 326 only increased the amount of recharge but have also drastically altered the sources recharging 327 shallow groundwater aquifer in the Bengal Basin; evaporated water recharge during the dry season 328 (discussed subsequently) have become a major recharge source of shallow groundwater. 329 Below 100 m depth, changes in δ 18 O and δ 2 H is not mirrored by corresponding changes in 330 the contribution from the respective sources. Water below 100 m is typically >4000 years old 331 suggesting recharge that predates human influence. Therefore, the proportional contribution of 332 evaporated and non-evaporated sources has remained consistent even though the source water 333 isotope ratios changed over time.
Overall, it appears that in the last 70 years the contribution of 334 mean contribution of standing water has steadily been increasing with time. For water recharged 335 before large-scale human perturbation, during late Holocene to Pleistocene, the respective 336 contribution of the sources has not changed appreciably. 337 This shift to increased fractions of standing water about 70 years ago coincides with the 338 shift to groundwater-irrigated dry-season rice farming and the growth of pond excavation (31), 339 both of which reflects the broader economic and population changes in the region. To our 340 knowledge this is the first study to show that large-scale groundwater pumping has altered the 341 source contributions to groundwater recharge in the Bengal Basin, with standing water in rice 342 fields, ponds and local rivers are now acting as major recharge sources. 343 The large decline (~1.6 m) in the annual minimum water table depth across much of Bengal  344 Basin (6) caused by groundwater extraction likely result in large vertical hydraulic gradient that 345 draws recharge in from the surface water bodies. Dry-season irrigation in Bangladesh increased 346 from <1 million hectares to 5 million hectares between 1975 and 2010 (26). Irrigated rice fields 347 now cover 21% of the land area and standing water in these field recharges groundwater as 348 irrigation return flow -groundwater pumped for irrigation that returns back to aquifers. Neumann 349 et al. (27) showed that return flow recharges aquifers by passing through macropores beneath 350 bunds (the raised berms that bound rice fields). This recharge circumvents the low-permeability 351 plough pan, the near-surface layer that restricts flow directly through rice fields. They estimated 352 the water budget for rice fields using an average ratio of field perimeter (bund length) to field area 353 and found that rice fields recharge in excess of 100 cm of water per year. Other field level analysis 354 also suggests significant water loss from rice fields from seepage and percolation. Working in the 355 Barind tract (northwest Bangladesh) Qureshi et al. (58) reported that during the dry season only 356 55% of the total applied water to rice fields was lost to evapotranspiration and the remaining 45% 357 was lost to seepage and percolation (56). On a seasonal basis, the authors estimated that net amount 358 of seepage and percolation were 34 and 60 cm for the dry and monsoon season respectively (56). 359 Similarly man-made ponds (31, 57) can recharge shallow groundwater. Stahl et al.,(19) 360 have shown that the presence of terrestrial crab burrows in pond-beds can short-circuit low-361 permeability surface sediments and provide widespread conduits for groundwater recharge; they 362 estimated the recharge flux to be 223 cm/year at their field site. 363 Although on an annual time scale there is a net discharge from aquifers to the rivers (58), 364 local rivers can also recharge shallow aquifers in response to lowering of groundwater level (24, 365 51) especially in the dry season when groundwater heads fall below the river head leading to flow 366 reversal that can recharge the aquifer or during the early monsoon when river and stream stage rise 367 precedes the rise in groundwater levels. Recharge from riverbeds is more complex than recharge 368 from ponds and rice fields. There are two modes of recharge from small rivers: (1) recharge during 369 the dry season when groundwater levels fall; (2) recharge in the early monsoon when river levels 370 rise faster than groundwater levels. During the dry season small river behave like other standing 371 water sources and have an isotopic signature indicating evaporation. This type of recharge is 372 clearly identified as standing water from its isotope values. During the monsoon, however, small 373 rivers act as conduits for recharge of precipitation and thus the isotopic signature matches that of 374 early monsoon precipitation. This type of recharge, although passing through riverbeds, is 375 identified as coming from precipitation in our model. 376 377

378
Groundwater isotope ratios in temperate regions are isotopically similar to the precipitation 379 amount-weighted isotope ratios (59-61). In the tropics, groundwater isotope ratios have been 380 shown to be isotopically lighter than the amount-weighted precipitation isotope ratios, which has 381 been interpreted as a recharge bias toward large rainfall events (62). However, unlike the 382 commonly observed pattern in the tropics (62), a majority of shallow and intermediate 383 groundwater isotope ratios (Figure 3) in the Bengal Basin are heavier than the modern amount-384 weighted mean precipitation isotope ratios (-46.5‰ and -6.9‰ for δ 2 H and δ 18 O respectively). 385 Our mixing models suggests that the combination of precipitation to recharge is skewed 386 towards heavier values along the meteoric mixing line (Figure 8). For all of the models, 70% of 387 the precipitation isotope ratios estimated to contribute to recharge were heavier than -6.9‰ (for 388 δ 18 O) -the amount weighted mean precipitation δ 18 O value. Therefore, our modeling suggest that 389 groundwater is not recharged evenly across the monsoon season and is seasonally biased toward 390 early monsoon rainfall, which is typically heavier than late monsoon precipitation (Figures 3A and  391  3B).  392 For all the models, the contribution from isotopically-light standing water sources (δ 18 O <-393 6‰) is low (<10%) and the proportion of isotopically enriched (δ 18 O >-2‰) standing water values 394 is > 50% (Figures 8B1-B3) suggesting that standing water recharges groundwater mostly during 395 the latter part of the dry season when the standing water bodies have become isotopically enriched 396 due to evaporation. Groundwater levels are lowest during the latter part of the dry season, which 397 leads to an increase in the difference between surface water and groundwater levels. The increased 398 downward head gradient during this period draws more recharge from standing water into 399 available aquifer storage. This process of induced recharge from ponds and canals is widely 400 observed in many regions with intensive groundwater pumping including the High Plains and the 401 Mississippi Alluvial aquifer systems of the US as well as the Indus Basin (63-65). 402 Monthly groundwater hydrograph data from 1230 groundwater wells in Bangladesh from 403 2000-2013 supports the isotopic derived inference on the timing of groundwater recharge. 404 Typically, groundwater levels are the lowest in April-May and highest in September-October. 405 With these groundwater hydrographs, we considered rates of net discharge during the dry season 406 (November to April) and rate of net recharge during the monsoon (April to October). During the 407 dry season, the groundwater hydraulic head decreases rapidly from October to January (>50% of 408 the total decline in groundwater level) for most of the wells in response to natural discharge and 409 groundwater abstraction ( Figure S6A). Even though a large proportion of groundwater abstraction 410 for dry season farming occurs between December and April (28), the decrease in head between 411 February and April is smaller compared to the period from October to January (Figures S6B, S6C  412 and S6D), suggesting that a portion of pumped water might be infiltrating back -a phenomena 413 suggested by field experiments (27,56). It is important to note that a falling water table does not 414 imply that recharge is not occurring, but rather indicates that discharge is in excess of recharge at 415 the time. Thus, during periods of high levels of discharge (e.g., during the irrigation season) total 416 recharge may actually increase, while groundwater levels will nonetheless fall if discharge exceeds 417 recharge. In fact, in the case of induced recharge from standing bodies of water, a falling water 418 table will increase the downward head gradient and lead to greater recharge from standing water. 419 During the monsoon, the data suggests that most of the well experience their greatest 420 hydraulic head increase between June and early August ( Figure S7A and S7B) suggesting that a 421 large proportion of net recharge happens during the early part of monsoon -thus, verifying the 422 results obtained from the isotope mixing models. The head change in September is negligible 423 ( Figure S7C) across the country except in some northwestern parts which are experiencing long-424 term decline in groundwater level due to massive abstraction and reduction in precipitation (66, 425 67). The lack of rise in groundwater level in September suggests little recharge from precipitation. 426 Similarly, in October, heads do not change for several wells suggesting even less recharge from 427 precipitation as compared to the earlier months; for wells lying in the floodplains of Ganges and 428 Brahmaputra the head in fact goes down as groundwater starts discharging to the river ( Figure  429 S7D). 430 431

432
Our analysis of the combined isotope data base supports a conceptual model of modern 433 groundwater recharge in the Bengal Basin that divides groundwater recharge into three phases: (1) 434 dry season (November-April), (2) early monsoon (May-August) and (3) late monsoon . These phases of the seasonal hydrologic cycle are similar to the results of Harvey et al.,436 small-scale study (18) based on measured heads and water levels in one village. However, Harvey 437 et al did not distinguish between recharge from precipitation and recharge from standing water, an 438 important focus of this regional study.

439
(1) During the dry season, the groundwater head falls because of pumping and discharge 440 to rivers (6, 68, 69) and the resulting increase in vertical head gradients draws more recharge from 441 standing water into aquifers even as the water stored in aquifers decreases ( Figure 9A). The 442 proportion of recharge from different sources depends on several factors including water table  443 drawdown, distance from rivers, patterns of hydraulic conductivity, and availability of standing 444 water sources. 445 The depth of the dry season decline in the water table elevation has increased over recent 446 decades because of groundwater extraction. For rice field irrigation, which makes up more than 447 85% of groundwater pumping (70), the amount of dry-season decline in the water table due to 448 irrigation pumping is limited by the potential evaporation of rice fields. Groundwater pumped for 449 irrigation circulates through rice fields, and what is not evapotranspired recharges the aquifer as 450 return flow. Although drawdown is limited by the maximum rate of evapotranspiration, the rate of 451 recharge from irrigated fields that returns to aquifers is not constrained and will increase with 452 increasing rates of irrigation. 453 This circulation process can be formulated with a simple water balance: Q=RF+ET, where 454 Q is the extraction rate, RF is return flow, and ET is evapotranspiration, all in units of volume flux 455 of water per unit area of rice field. The decline of the water table is dH/dt = (Q -FR)/Sy = ET/ Sy 456 where Sy is the specific yield. This formulation assumes that irrigated fields are not simultaneously 457 drained to rivers. Since rice is grown in standing water, and rice farming techniques used in the 458 Bengal Basin are not the most efficient (71), a surplus of water (i.e. RF) recharges back to the 459 aquifer. An important implication for groundwater isotopes is that enrichment may continue as 460 groundwater is circulated. Over decades, groundwater may become progressively even more 461 enriched in heavy isotopes as it is pumped to the surface, subjected to evaporation, and returned 462 back to aquifers. 463 (2) During the early monsoon the unsaturated zone above the unconfined water table is  464 filled by monsoon precipitation ( Figure 9B). During this phase, maximum recharge by 465 precipitation takes place, as evidenced from the distribution of precipitation isotope ratios 466 estimated to be in groundwater by our model (Figure 8) and the rise in groundwater hydrographs 467 (Figure 9). Independent analysis of groundwater hydrographs also suggests a rapid rise in 468 groundwater levels during this period (6, 25). Recharge occurs across the entire basin and in some 469 regions (such as northeast, Figure 9) maximum groundwater levels are observed during this time 470 (6). Estimates of the mass of total water storage from GRACE and groundwater storage from wells 471 during 2002-2010 (72, 73) indicate that on average 85% of the water is recharged by the end of 472 July. 473 (3) In the late monsoon, groundwater recharge varies regionally, but is generally small. In 474 eastern parts of Bangladesh (light green region in Figure S7) where shallow groundwater aquifers 475 have already been recharged to a large extent during phase 2 the remaining space -if any-in the 476 shallow aquifers is recharged in the early part of phase 3 ( Figure 9C). Groundwater levels in the 477 shallow aquifers in these regions are highest in August or early September (6). Precipitation and 478 flooding events during this time do not recharge the aquifers and these waters are rejected as the 479 shallow aquifers have already been filled. This region not only receives more precipitation than 480 the western parts ( Figure S1) of the basin, but also has a lower proportion of irrigated area that 481 relies on groundwater (25). Other regions, mostly the western parts of Bangladesh and West 482 Bengal, receive lower rainfall ( Figure S1), are less prone to flooding, and are intensively farmed 483 and pumped for irrigation (25). In these regions (shown in light gray in Figures S7C and S7D), 484 where annual abstraction exceeds the groundwater recharge (6,25,74), the aquifer continues to be 485 recharged even during the late monsoon season. 486 487 Implications for groundwater quantity, groundwater quality and human 488 habitability 489 Changes in the contribution from different sources suggests that the flow path of water 490 entering the subsurface had changed. Large scale shifts in the flow paths would likely have major 491 effects on the geochemistry of groundwater. Regional increases in contributions from ponds, rice 492 fields, and rivers could threaten the regional quality of shallow groundwater. Circulating irrigation 493 return flow may contaminate groundwater with increasing levels of solutes from rice field. Input 494 of reactive organic carbon has been implicated in arsenic mobilization within aquifers (17, 18, 55, 495 75, 76), although the source of this reactive carbon remains an area of active inquiry. Small-scale 496 field studies have documented pollutants transported from ponds to drinking water wells (77-80), 497 there is a growing concern of nitrate pollution in surface and groundwater (81), and local 498 groundwater quality assessments have documented E. coli, major ions, trace elements and organic 499 compounds (82,83,92,(84)(85)(86)(87)(88)(89)(90)(91). 500 The Indian subcontinent accounts for more than 25% of the total global groundwater 501 withdrawal and several aquifers are experiencing rapid declines in groundwater levels or reduction 502 in groundwater quality (84,85). A variety of studies in the subcontinent such as (9,24,57,64, 97) has focused on understanding the rate of recharge and the chemical load carried in recharge, 504 regional groundwater depletion and sources recharging groundwater, however a systematic 505 analysis on the effect of large-scale pumping and dry-season farming on the timing and sources of 506 recharge across the Indus-Ganges-Brahmaputra Basin is missing. Understanding how intensive 507 groundwater pumping has affected, and will continue to affect, both the sources and timing of 508 groundwater recharge is essential to understanding and managing the regional hydrologic system. 509 Our isotopic analysis provides evidence that across much of the eastern and southern 510 Bengal Basin late monsoon precipitation does not recharge groundwater, likely because aquifers 511 are full by August so that precipitation in September-October is lost as runoff. This "rejected 512 potential recharge" was the focus of Roger Revelle's seminal paper (98) that promoted the idea of 513 reducing monsoon flooding by pumping groundwater to lower the water table during the dry 514 season so that more precipitation would infiltrate during the monsoon rather than contribute to 515 flood water. Local hydrological modeling and field studies have arrived at a similar conclusion (6, 516 24, 25, 99). Our isotope analysis suggest that increased pumping could be sustained in many parts 517 of the basin where precipitation is sufficient to return the water table to the land surface every 518 monsoon. 519 520

521
Model 1: Mixtures of precipitation and standing water 522 We model groundwater isotope ratios as a mixture of precipitation and standing waters 523 (i.e., ponds and rice fields): 524 where δ 18 Ogw and δ 2 Hgw are the isotope values of the groundwater, δ 18 Oprecip and δ 2 Hprecip 527 are the isotope values of precipitation and δ 18 Ostand and δ 2 Hstand are the isotope values of the 528 evaporated standing water bodies. fprecip and fstand are the fractional contribution of precipitation 529 and standing water to the groundwater and fprecip + fstand = 1. 530 We simulated all pairs of groundwater and standing water samples. We first obtain the 531 δ 18 Oprecip and δ 2 Hprecip endmember values defined as the intersection between the local meteoric 532 water line (LMWL) and the line joining the groundwater and the standing water sample in the dual 533 isotope space (see Figure 5A). Then, fprecip and fstand were calculated as: 534 535 After calculating the contribution of precipitation and standing water endmembers for each 538 pair ( Figure 5A), we calculated the mean to obtain the average contribution of precipitation and 539 standing water endmembers respectively. 540 541

542
In the 3-endmember second and the third mixing models, we applied a Monte Carlo method 543 (36,(100)(101)(102)(103) to estimate probability distributions of the contributions of different sources 544 recharging groundwater. This method entails more steps than model 1, that had only the two 545 endmembers. The Monte Carlo simulations each draw from the large data set of precipitation and 546 standing water isotope data to fully cover the probabilistic range of values as characterized by the 547 available data. 548 Below, we describe the isotope values of each endmember. For the river endmember, we 549 chose a fixed value (δ 18 Oriv = -7.4‰ and δ 2 Hriv = -50‰), as the isotopic variation in the river values 550 were negligible ( Figure 4A). We calculated amount-weighted mean precipitation isotope ratios by 551 weighting each sample by its proportional contribution to total annual precipitation. We assumed 552 that precipitation isotope values follow a bivariate normal distribution with values {m2-H , m18-0, 553 σ2-H , σ18-O , ρ} of {− 46.5, − 6.9, 31.2, 3.7, 0.9} where m2-H, m18-0, σ2-H, σ18-O are the amount-554 weighted mean and standard deviation of the precipitation isotope values and ρ is the correlation 555 between the O and H precipitation isotopes. For the standing water endmember, we assumed that 556 all the standing water samples included in this study (n = 447) are representative of the true 557 distribution of the standing water isotope values and we treated each standing water isotope value 558 as a potential endmember. 559 For each standing water sample, we simulated 500 random draws of precipitation isotope 560 values to obtain precipitation endmember values. For each groundwater sample, we repeated the 561 above process of simulating random draws of precipitation isotope values for each of the 447 562 standing water samples. In total, our approach generated >200,000 (447 standing water values 563 multiplied by 500 precipitation values) possible "mixing triangles" for each groundwater sample. 564 For each combination of standing water, river and precipitation endmembers, where the 565 endmember values were solvable (0 <= {fprecip, friv, fstand} <=1 and fprecip + friv + fstand =1), we 566 obtained the fractional contribution of the endmembers ( Figure 5B). When a groundwater sample 567 fell outside the triangular domain ("mixing tringle") defined be the three end members ( Figure  568 5B), we discarded this combination and did not calculate fprecip, friv, fstand as they would not fulfill 569 the criteria: 0 <= {fprecip, friv, fstand} <=1. After calculating the contribution of precipitation, river 570 and standing water endmembers for each solvable triangle ( Figure 5), we calculated the mean to 571 obtain the average contribution of precipitation, river and standing water endmembers 572 respectively. 573 We only included groundwater samples that fell below the LMWL and were not highly 574 evaporatively enriched. For groundwater samples, falling on or above the LMWL (14% of the total 575 groundwater sample), the contribution of surface water is minimal, and we assumed that the 576 groundwater sample is completely recharged from precipitation i.e., fprecip = 1 and fstand = friv = 0. 577 Similarly, highly evaporated groundwater samples (deuterium excess <0‰, 5% of the total 578 groundwater sample) also typically fall outside the mixing triangle because the groundwater 579 isotope ratios will have heavier isotope values than most of the standing water endmember values; 580 this will result in only a small number of plausible mixing triangles. We assumed these evaporated 581 samples to be recharged solely from standing water i.e., fstand =1 and friv= fprecip = 0. These 582 assumptions are valid as it is very unlikely that samples falling above the LMWL are recharged by 583 standing water. Similarly, highly evaporated groundwater samples are likely to be recharged 584 primarily by standing water. 585 586 Model 2: Mixtures of one precipitation event, river and standing water 587 In the second model, we considered a 3-endmember mixing model with precipitation, large 588 rivers and standing water as the potential sources ( Figure 5B). Groundwater sample isotope ratios 589 (δ 18 Ogw and δ 2 Hgw) were modeled as a mixture of precipitation, river and standing waters 590 endmembers: 591 where δ 18 Ogw and δ 2 Hgw are the isotope values of the groundwater, δ 18 Oprecip and δ 2 Hprecip 594 are the isotope values of precipitation, δ 18 Oriv and δ 2 Hriv are the isotope values of the large rivers 595 and δ 18 Ostand and δ 2 Hstand are the isotope values of the evaporated standing water bodies. fprecip, friv 596 and fstand are the fractional contribution of precipitation, large rivers and standing water to the 597 groundwater and fprecip + friv + fstand = 1. After defining the isotope ratios of each endmember, we 598 solve the above equations to obtain the fractional contributions of the endmembers (friver, fprecip, and 599 fstand.). 600 601 Model 3: Mixtures of two precipitation events with standing water 602 The third model consists of two precipitation and a standing water endmember ( Figure 5C). 603 In this model, groundwater isotope ratios were modeled as: 604 where δ 18 Ogw and δ 2 Hgw are the isotope ratios of the groundwater, δ 18 Oprecip1 and δ 2 Hprecip1 and 607 δ 18 Oprecip2 and δ 2 Hprecip2 are the isotope ratios of precipitation endmembers respectively. fprecip1, 608 fprecip2 and fstand are the fractional contribution of precipitation and standing water to the 609 groundwater and fprecip1+, fprecip2 + fstand =1. In this model, for each standing water endmember, we 610 generate two distinct sets of 500 random precipitation isotope values, and from each set a random 611 precipitation value is selected to obtain a pair of precipitation isotope ratios ( Figure 5C). We then 612 solve the above equations to obtain the fractional contributions of the endmembers (friver, fprecip2 613 and fstand.  is modeled as a mixture of precipitation and standing water. The line joining the surface water 970 endmember and the groundwater (solid red) is extended to join the LMWL (dashed red). The point 971 of intersection is the precipitation endmember isotope ratio. In model 2 (B), groundwater is 972 modeled as a mixture of precipitation, large river and standing water endmembers. The small light 973 gray circles are the 500 randomly generated precipitation isotope ratios. The red triangle 974 connecting the large river (purple circle), standing water (blue rectangle) and precipitation (orange 975 circle) endmembers illustrates one of the 500 possible triangles for each standing water sample. In 976 the example shown here the groundwater sample (open red circle) falls within the red triangle and 977 hence the mixing model is solvable. In model 3 (C), groundwater is modeled as a mixture of 2 978 precipitation endmembers and a standing water endmember. For each standing water endmember, 979 we generate two random distribution of 500 precipitation isotope values and from each of those 980 distributions chose the precipitation endmember values (orange and green circles). The red triangle 981 connecting two orange circles (precipitation endmembers) and blue rectangle (standing water 982 endmember) results in a solvable mixing triangle i.e., groundwater isotope ratios can be modeled 983 as a mixture of these three endmembers. The green triangle connecting two green circles 984 (precipitation endmembers) and blue rectangle (standing water endmember) results in an 985 unsolvable mixing triangle i.e., linear mixing between these endmember values cannot explain the 986 observed groundwater isotope ratios. Figure S3 shows Distribution of modeled precipitation and standing water δ 18 O values in groundwater (A1 to B3). 1008 The red line in panels A is the amount weighted mean precipitation δ 18 O value. 1009