Restoring Drying Lakes with Treated Agricultural Euents under a Carbon-based Payments for Ecosystem Services Incentive

9 Wetlands store a significant proportion of terrestrial carbon, however, when degraded and dry, 10 they can become net carbon emitters. Climatic stressors, such as rising temperature and reduced 11 precipitation, further exacerbate carbon release risks. This study explores incentivizing adoption 12 of constructed wetlands (CW) on agricultural farms for treating effluents and releasing into drying 13 lakes. A payment for ecosystem services (PES) framework is developed to analyze land use 14 allocation decisions of farmers towards adopting CWs on their private farms. Release of treated 15 agricultural wastewater helps a drying lake remain wet under a warming climate preventing release 16 of carbon stored in its soils. Results indicate that PES payments equaling 10,000 to 20,000 INR 17 (150-300 US dollars) per mega litre (ML) can be effective in incentivizing adoption of CWs on 18 farms in India, and their benefits to drying lakes can be significant. Specifically, life of lakes can 19 be prolonged to more than 100 years under such PES based schemes besides resulting in substantial 20 carbon storage in soils. Such PES schemes can be a cost-effective way to not only protect and 21 conserve lakes for their biodiversity and livelihood benefits but also from a carbon mitigation 22 perspective. Results further show that when a social planner allocates land between farming and 23 CW, incorporating the carbon sequestration benefits of lakes and when facing a risk of abrupt and 24 permanent drying of lakes, their adoption rate is higher compared to that of the farmers. When 25 extrapolated, carbon benefits from such PES programs for the entire country could be nearly 15 26 trillion USD over the next 100 years.


30
Wetlands occupy about 12 million km 2 of global geographical area (Moomaw et al., 2018). They 31 also store the most amount of carbon in the biosphere at nearly 30% of the total 15 Pg soil carbon 32 (Nahlik and Fennessy, 2016). In the United States, there is about 10 times more carbon (at 9.6 33 x10 15 g C) stored in wetlands than in tidal saltwater sites (Braun et al., 2019). However, owing to 34 anthropogenic pressures, the planet has been losing wetlands at a rapid pace. Since 1700 AD, we 35 have lost 87% of the world's wetlands (Davidson, 2014). In the 20 th century, coastal wetlands 36 globally declined at a rate of 0.7-1.2% annually resulting in more than 60% loss to their total area. 37 The loss in coastal wetlands has been highest in Asia at 1.1%, whereas in Europe it has been 0.99%  In India, more than a third of existing wetlands have been lost due to encroachment, 41 urbanization, land use changes and warming temperatures. Wetlands in India occupy almost 5% 42 of the country's geographical area. Currently India is losing wetlands at a rate of 2-3% per year 43 (Prasher, 2018). Foote et al. (1996) identified 12 major causes of wetland loss in India, including 44 agricultural conservation, deforestation, hydrological alteration, water quality degradation, 45 wetland consolidation and groundwater extraction. More than 50,000 wetlands in India have been 46 rendered dead from water pollution alone (Prasad et al., 2002). Owing to urban expansion and 47 encroachment, in Mumbai, more than 70% of wetlands have been lost over the past four decades 48 (Chauhan, 2020). Similarly, lake Shambhar in Rajasthan faces extinction threats due to 49 encroachment and groundwater extraction (Rana, 2019). Whereas Karanji lake in Mysuru, India 50 has completely dried up (Milton, 2019). Warming temperatures are further projected to cause an increase in lake water evaporation 52 rate over the course of this century . Based on IPCC's A1B scenario, climate 53 change could cause significant drying of Montane wetlands in the US Pacific Northwest (Lee et 54 al., 2015). Similarly, warming temperatures and reduced precipitation are expected to decrease 55 wetlands in Canada by between 7 and 47% (Withey and van Kooten, 2011). Due to frequent 56 drought conditions, Chad and Urmia lakes in Africa have shrunk by 80% and 65% (Smith, 2012). 57 Warming temperatures also increase the risk of fires. The great Dismal Swamp sequesters 58 significant amounts of carbon, however catastrophic fires can cause a release of 6.5 million tonnes ). Therefore, it is important to prevent degradation of wetlands in order for society to 76 maximize benefits from carbon mitigation efforts.

77
There exists a significant body of work in the literature dealing with the determinants of 78 wetland loss as well as analyzing the use of existing and new policy tools for their protection.

79
Section 404 of the Clean Water Act in the United State mandates that for every wetland that is lost 80 due to draining, a new wetland must be created to offset the services provided by the lost wetland 81 (Ribaudo et al., 2009). As a result, there are currently more than 600 mitigation banks in the US 82 that provide credits for wetland restoration which can be sold to developers who destroy wetlands.

83
Similarly, the possibility of using environmental impact bonds for restoring wetlands in Louisiana

106
Given that agriculture directly and indirectly poses a significant threat to wetlands, the 107 solution to their conservation must also be explored within that sector. For instance, farmers could 108 be incentivized to reduce effluent loadings into streams that end up in wetlands. Heberling et al.    In this study, a dynamic optimization model of CW adoption on agricultural farms is 141 developed keeping in mind the long run benefits of releasing treated effluent water into the lake. Water released into the lake from farms helps mitigate the effects of a warming climate, which is 143 projected to cause increasing evapotranspiration rates over the course of the century. Optimal rate 144 of CW adoption is compared from private farmers' and the social planner's perspective, where the 145 latter also incorporates the risk of carbon release from a sudden drying of the lake. Implications  CWs on individual farms are connected through a drainage canal to a main channel that empties 159 into the natural lake. The irrigation water used for farming is treated in the CWs before being 160 released into the canal. Assuming a water scarce world, farmers do not extract groundwater for 161 irrigating their crops as groundwater levels are too low. It is further assumed that the groundwater 162 aquifer is not linked to the wetland so that a reduction in the groundwater levels does not impact 163 on lake water storage. Under a PES mechanism, the farmer is paid for carbon sequestered due to 164 enhanced water availability in the wetland. Optimal CW investment is also derived from a social  The wastewater treated by CW plants is discharged into the lake through canals resulting in 196 augmentation of lake water volume as: where is fixed lake area, > 0 is the rate of evapotranspiration loss in each time t, > 199 0 is a parameter converting evaporation in mm per day per ha to ML per year, 1 > > 0 is 200 the fraction of lake water that is lost due to overflow and groundwater seepage and > 0 is a 201 parameter converting rain falling on lake into water volume (in ML).

202
The rate of evapotranspiration of the lake water is further given as: where >0 is a parameter which could take on a value higher than 1 under extreme dry conditions.

219
The farmer's total income comprises the sum of PES and crop incomes, given as: The optimal control problem for the framer involves maximizing: where is their rate of time preference. Next, consider that a social planner maximizes the income 224 of the farmer net of societal damages from carbon release in the state when the lake turns dry. The 225 optimization problem of the planner is to maximize: where ̇ is the instantaneous hazard rate of lake running dry, is the cumulative hazard, and is 228 the net present value of future damages to society from carbon released into the atmosphere in 229 each year the lake is dry. The expected valuation of the pre-and post -dry states is derived using 230 the methodology for exponential hazard functions as given in Reed and Heras (1992).

231
The lake will continue to release carbon until all stored carbon stock has been released.   Table 1.

307
The numerical example is solved using GAMS software and a time horizon of 150 years. Results is also assumed to increase at a rate of 0.2% each year (or 16% over the next 80 years), which 320 would further exacerbate water loss rate from the lake (fig 3). However, adoption of CWs begins 321 to show its effect after 7 years when sufficient capacity is accumulated in treating effluents. Lake In another scenario, where rainfall is 1.8mm/day, the lake survives for only 4 years. This suggests 357 that CW adoption needs to be promoted keeping in mind the shortfall in the long run rainfall 358 experienced by a particular region. 359 We next consider two scenarios from the farmer's land size and agricultural opportunity 360 costs perspective. When the farmer owns a lower land area (of 50 ha), adoption rate is delayed as 361 the opportunity cost of enrolling land is higher, but more importantly, because revenues generated 362 through farming fall short of the requirements for making faster investments into CW technology.

363
When farm size is lower, agricultural income is also lower, which means that the farmer has to 364 wait longer to reach the maximum CW capacity. This has an adverse impact on lake water volume 1000 ML, this risk increases significantly. We consider a base case risk scenario where the risk of 379 lake drying is low. Compared to the no-risk case, the presence of risk results in higher rate of CW 380 adoption by the planner (fig 6). This helps improve lake volume in the short term (fig 7). However, 381 in the long run there is decline similar to no-risk case, given the water supply constraints. When other studies have found that the impact of carbon storage may be poverty alleviating or 420 exacerbating depending upon various circumstances (Ferraro et al., 2015).

421
There are a few limitations of the modeling approach of this study which are worth mentioning.

422
The role of groundwater withdrawal on wetland water dynamics is not explicitly modeled.   Time paths of cumulative total lake water volume under no-risk scenarios Time paths of cumulative lake water volumes (ML) compared for risk scenarios Figure 8 The probability that the lake will not turn dry until year t, depicted for lake drying risk scenarios