by Brian Tomasik
First published: 8 May 2017; last update: 16 May 2017


Irrigation of crops, pastures, and lawns plausibly increases the world's net primary productivity (NPP) and, therefore, the amount of plant food available to feed heterotroph populations. This probably increases wild-animal suffering. However, some side effects of irrigation, like soil salinization and reductions in groundwater-dependent vegetation, may reduce NPP. Assuming that irrigation does increase wild-animal suffering on balance, it would be good to remove government subsidies for irrigation. Opposing irrigation subsidies may be a point of convergence between environmentalists and wild-animal-suffering reducers.

How irrigation affects global NPP

Impact on crop yields

Irrigation can substantially increase crop yields (and hence NPP), especially in arid regions. Pictures of center-pivot irrigation like the one here demonstrate this starkly. (Many more photos of this type can be found in "Center Pivot Irrigation: The Real Crop Circles".) Crop circles along the Columbia, Washington, USAThe green circles are where irrigation water reaches, and the browner exterior is not irrigated. (Of course, other factors like fertilizer use may also be relevant here.)

Irrigation along the Nile has substantially increased plant productivity. The Brazilian savanna is able to grow substantial grain crops "thanks to irrigation and soil correcting techniques".

One forum discussion anecdotally agrees with the idea that irrigation allows for more life:

Cooler weather & less rainfall help keep bug populations down. In general, states like California have less mosquitoes & other bugs due to no rainfall during summer & dry countryside. But if a person drives into places that are heavily irrigated for agriculture [like the Central Valley] the bugs are everywhere & splattered all over cars windows.

Muir (n.d., "Waterlogging"): "one could argue that the Central Valley of CA shouldn't be lush and green; it is one of the most agriculturally productive regions of the world only because of irrigation, which may not be sustainable there."

This study reports that in the US, "About 61.1 million acres were irrigated in 2005." It adds that "The majority of [water] withdrawals (85 percent) and irrigated acres (74 percent) were in the 17 conterminous Western States. The 17 Western States are located in areas where average annual precipitation typically is less than 20 inches and is insufficient to support crops without supplemental water." This suggests that crop cultivation in Western US states is worse than in Eastern US states because, it seems, irrigation has more impact in terms of increasing NPP in the West.

Lustgarten and Sadasivam (2015): "California and Arizona are able to produce more than twice as much cotton on each acre they plant as can cotton powerhouses like Texas and Georgia because they irrigate their fields more often."

Bradford et al. (2005) suggest that crop cultivation has increased NPP in the US Great Plains and postulate that irrigation and fertilizer were main driving forces:

irrigation may be accounting for much of the change in primary productivity as a consequence of cropping practices. In semiarid areas, water availability is crucially important for vegetation dynamics (Noy-Meir 1973), so it is not surprising that widespread irrigation elevates primary production. Although our analysis did not explicitly address the [e]ffects of fertilization, it is also reasonable to expect that fertilizer additions in cropped areas may contribute to our observations of increased net primary productivity.

This figure from p. 1869 of Bradford et al. (2005) shows the impact of irrigation:

Hydrological changes

Irrigation moves water around. This can result in increases of water in some places and decreases in other places. For example, if irrigation is fed by surface water, such as from rivers, irrigation will probably reduce river flow, and it may increase groundwater. Meanwhile, irrigation from groundwater probably does the opposite. This study says:

The results indicate that the impact of irrigation water withdrawals on the long-term annual mean global water resources is small. However, in some river basins, the impacts of irrigation are significant with two major opposing effects on surface water (SW)/groundwater (GW): SW depletion/GW accumulation in regions with irrigation primarily fed by SW, and SW accumulation/GW depletion in regions with irrigation fed primarily by groundwater.

However, Edwards and Hill (2012) claim that the idea that "irrigation from surface water would create increases in western groundwater[...] is no longer supported."

In some cases, surface water and shallow groundwater are effectively "one resource" because "Groundwater and surface water are connected in a variety of ways" (Moran et al. 2014). However, my impression is that in some cases, deep groundwater doesn't interact much with surface water. For example, two different residents of the area of upstate New York where I grew up told me that they think the groundwater in this region is quite deep and isn't accessible to plants or surface streams. Ponce (2006): "Shallow groundwater flow systems should be distinguished from deep groundwater flow systems; the former interact with surface water, while the latter do not."

This page says: "irrigation has immediate effects on the provision of moisture to the atmosphere, inducing atmospheric instabilities and increasing downwind rainfall, or in other cases modifies the atmospheric circulation, delivering rain to different downwind areas." Does this mean that irrigation increases rainfall beyond the crop fields, thereby increasing plant growth? If the irrigation water is withdrawn from deep underground, is this a one-time effect, because after the water reaches the ocean, it's effectively lost?

Once water reaches the oceansa, I assume it has little impact on NPP (apart from bringing nutrients downstreamb) because ocean NPP is limited by light/nutrients, not by the total volume of water in the oceans? I assume that small increases in ocean volume probably make little difference to the rate of water evaporation from the oceans?

This page says: "Aquifer drawdown or overdrafting and the pumping of fossil water increases the total amount of water within the hydrosphere subject to transpiration and evaporation processes". It's unclear to me how significant this effect is, and it seems like it might be a one-time effect, since once the water reaches the ocean, it's effectively lost, and evaporation from the oceans won't appreciably increase as a result of slightly more ocean water? But insofar as groundwater extraction does in fact increase total freshwater availability on land, this might increase total plant growth.

Likewise, if desalination becomes cost-effective in the future, it could increase the total amount of freshwater available, increasing the amount of water that can (unfortunately) help terrestrial plants grow. This page reports:

In 2005 over 2,000 desalination plants with a capacity of more than 100 m3/day had been installed or contracted in every state in the U.S. with a total capacity of more than 6 million m3/day. Only 7% of that capacity was for seawater desalination, while 51% used brackish water and 26% used river water as water source. The contracted capacity corresponds to 2.4% of total municipal and industrial water use in the country in 2000. The actual share of desalinated water is lower, because some of the contracted capacity was never built or never operated, was closed down or is not operated at full capacity.

This article mentions that "Rivers and lakes are the hardest hit habitats, with animals['] populations down by 81% since 1970, due to excessive water extraction [among other things]." But if rivers and lakes are shallower, then there's more land area on which terrestrial plants can grow. For example, Muir (2012) says: "The San Joaquin River in California is so permanently dewatered that trees are growing in its bed".

Groundwater-dependent ecosystems

In some cases, groundwater contributes to native plant growth, especially in groundwater-dependent ecosystems. This study gives examples of such ecosystems:

Ecosystems like wetlands can have very high NPP.

Ponce (2006): "Depletion of groundwater places at risk the survival of the well plants [aka phreatophytes]. In turn, the loss of vegetation can produce a domino effect on the rest of the landscape, with increased erosion, increase in surface albedo, decreased ambient moisture, and climate change."

This page reports that in response to removal of surface water and groundwater: "Owens Valley springs and seeps dried and disappeared, and groundwater-dependent vegetation began to die.[2][23]"

I'm uncertain what fraction of all plants are fed to a significant degree by groundwater. This page says:

Not all ecosystems need groundwater, of course. Some terrestrial ecosystems – for example, those of the open deserts and similar arid environments – exist on irregular rainfall and the moisture it delivers to the soil, supplemented by moisture in the air. While there are other terrestrial ecosystems in more hospitable environments where groundwater plays no central role, groundwater is in fact fundamental to many of the world’s major ecosystems. Water flows between groundwaters and surface waters. Most rivers, lakes, and wetlands are fed by, and (at other places or times) feed groundwater, to varying degrees. Groundwater feeds soil moisture through percolation, and many terrestrial vegetation communities depend directly on either groundwater or the percolated soil moisture above the aquifer for at least part of each year. Hyporheic zones (the mixing zone of streamwater and groundwater) and riparian zones are examples of ecotones largely or totally dependent on groundwater.

Minerotrophic vegetation gets its "water supply mainly from streams or springs." In contrast, ombrotrophic "vegetation receive[s] all of its water and nutrients from precipitation, rather than from streams or springs."

This thesis says:

With respect to the role of groundwater in transpiration, the groundwater subsidy (GWS) to a plant has been defined as the additional water available for root water uptake as the result of a shallow groundwater table [...] [Lowry and Loheide, 2010]. GWS acts as a source of water for plants during times with little to no precipitation when the surface soil moisture has been evaporated or transpired. [...] Previous studies of GWS have focused on systems with declining water tables and vegetation that are adapted to wet environments [Lowry and Loheide, 2010]. These systems are not affected by oxygen stress and therefore always experience benefits from a shallow water table. Notably, oxygen stress can cause a reduction in root water uptake during times of shallow depths to water table thus experiencing a groundwater taxation (GWT).

Ponce (2006):

Another significant example of groundwater depletion is the Ojos Negros valley, in Baja California, Mexico, about 40 kilometers (25 miles) east of Ensenada. In the past 35 years, the water table in the valley has dropped up to 45 meters (150 feet). The extracted groundwater is primarily used to support intensive agriculture. The experience of the Ojos Negros valley is particularly sobering, because the valley derives its name (Spanish for "Black Eyes") from two oval-shaped swamps, or "cienagas" [...], that existed prior to the development of groundwater, and have since dried out and disappeared in the face of protracted aquifer depletion. Other wetlands near the valley's mouth have been reduced to a fraction of their original size (Ojos Negros Research Group, 2003).

Despite the above, I would conjecture that groundwater or surface water used for irrigation increases NPP on crop fields and neighboring regions more than it reduces NPP of groundwater-dependent vegetation, because irrigation water is targeted to maximize yields, whereas groundwater isn't always as accessible to plants. Some agricultural areas in the Western US, for example, would not have much plant growth at all without irrigation. That said, further quantification of the increases and decreases in NPP due to irrigation is needed.

Ponce (2006) mentions that some groundwater is lost to the oceans as "deep percolation", which doesn't contribute to surface fresh waters. I expect that tapping this "deep percolation" water for irrigation increases NPP. Ponce (2006) adds: "On an annual basis, deep percolation is approximately 1/6 of shallow percolation, or baseflow [...]. Thus, deep percolation amounts to 2% of precipitation (L'vovich, 1979)."

Soil salinization and waterlogging

Irrigation can damage the ability of land to grow crops via increasing soil salinity:

Salinity from irrigation can occur over time wherever irrigation occurs, since almost all water (even natural rainfall) contains some dissolved salts. When the plants use the water, the salts are left behind in the soil and eventually begin to accumulate. Since soil salinity makes it more difficult for plants to absorb soil moisture, these salts must be leached out of the plant root zone by applying additional water.

This page explains:

According to a study by UN University, about 62 million hectares (20%) of the world's irrigated lands are affected [by soil salinity], up from 45 million hectares in the early 1990s. In the Indo-Gangetic Plain, home to over 10% of the world's population, crop yield losses for wheat, rice, sugarcane and cotton grown on salt-affected lands could be 40%, 45%, 48%, and 63%, respectively.

According to this article:

improper cropping and irrigation practices have led to desertification in areas of Uzbekistan, where cotton is a major export. In the days of the Soviet Union, the Aral Sea was tapped for agricultural irrigation, largely of cotton, and now salination is widespread.

Rees (1992) says there are 1.5 million hectares per year of "soil salination from failed irrigation projects" (p. 128).

Muir (n.d., "Salinization"):

It is hard to find consistent estimates of how much of the world's irrigated acreage is affected by salinization -- you can find credible sources claiming anywhere between 8 and 32%. Ten to 20% seems to be mid-line for the estimates, globally. Salt concentrations are high enough in much of this irrigated acreage to decrease yields significantly. In extreme cases, land is abandoned because it is too salty to farm profitably.

This page says that irrigation can cause "waterlogging and drainage problems", which "can lead to reduced agricultural production." The "Case studies" section of that page gives many examples of waterlogging and salinization. Muir (n.d., "Waterlogging"): "Worldwide, as much as 10% of all irrigated land may suffer from water logging. This is an area about the size of Idaho. As a result, productivity has fallen in this area of cropland."

Other ecosystem changes due to irrigation

This page says:

Sedimentation is an essential part of the ecosystem that requires the natural flux of the river flow. This natural cycle of sediment dispersion replenishes the nutrients in the soil, that will in turn, determine the livelihood of the plants and animals that rely on the sediments carried downstream. The benefits of heavy deposits of sedimentation can be seen in large rivers like the Nile River. The sediment from the delta has built up to form a giant aquifer during flood season, and retains water in the wetlands. The wetlands that are created and sustained due to built up sediment at the basin of the river is a habitat for numerous species of birds.

Of course, if the sediments aren't washed downstream, do they contribute to life in other regions, such as on river banks, or on crop fields if sediment gets caught in irrigation water?

Damming a river to create a reservoir can also reduce downstream sediment: "As all dams result in reduced sediment load downstream, a dammed river is said to be 'hungry' for sediment. Because the rate of deposition of sediment is greatly reduced since there is less to deposit but the rate of erosion remains nearly constant, the water flow erodes the river shores and riverbed, threatening shoreline ecosystems, deepening the riverbed, and narrowing the river over time."

Edwards and Hill (2012) mention "contamination of water systems by irrigation runoff." This page says: "Owing to drainage of surface and groundwater in the project area, which waters may be salinized and polluted by agricultural chemicals like biocides and fertilizers, the quality of the river water below the project area can deteriorate". Does this increase or decrease aquatic NPP? I don't know. I would speculate that runoff of herbicides and salts would reduce phytoplankton growth, while runoff of fertilizers would increase it. Stevens et al. (1985), p. 27: "From stream flow records, Striped Bass Working Group member Stephen Hansen estimated that the five major sources of return irrigation water contribute between 5% and 20% of the total Sacramento River flow at or near Sacramento during April-July [...]. Pesticides and herbicides used in rice culture [...] are applied extensively during these months."

This page notes: "The Manatali reservoir formed by the Manantali dam in Mali [...] destroyed 43000 ha of savannah[...]. Further, the reservoir destroyed 120 km2 of forest." In general, dams may replace land with bodies of freshwater. However, it's not clear that this reduces animal populations and in fact may increase them.

This page says "Aquifer drawdown or overdrafting and the pumping of fossil water may be a contributing factor to sea-level rise." Since the NPP per unit area of open ocean is less than that of most land types, this might somewhat reduce NPP?c

Policy recommendations

Irrigation water efficiency is probably bad

Muir (2012):

The potential to increase substantially the gross irrigated area of the world is limited. Gains from new capacity are expected to be largely offset by losses from [...] problems [such as] waterlogging and salinization, as well as retirement of areas being irrigated by pumping water in excess of rates of recharge.

In fact, most new water capacity is predicted to come from increasingly efficient use of existing supplies rather than harnessing of new supplies.

I would guess that more water-efficient irrigation doesn't appreciably reduce crop NPP per hectare (unless it reduces weed growth?). However, water efficiency may allow more total hectares of land to be irrigated because more water is available, thereby significantly increasing crop NPP.

Moreover, water-efficient irrigation might actually increase long-run crop NPP per hectare by reducing waterlogging and salinization. NSW (2013) explains that one way to reduce soil salinization is "Avoiding over-irrigation of crops by using techniques such as drip irrigation, soil moisture monitoring and accurate determination of water requirements." This page says: "The use of sprinkler irrigation and micro-irrigation systems decreases the risk of waterlogging and erosion."

Finally, federal subsidies for irrigation-efficiency improvements may increase the amount of irrigation that farmers do. Nixon (2013):

A study by researchers at the University of California, Davis, this year concluded that Kansas farmers who received payments under the conservation subsidy were using some of their water savings to expand irrigation or grow thirstier crops, not to reduce consumption.

Another study by researchers at New Mexico State University in 2008, which studied an area running from Colorado to New Mexico, came to the same conclusion.

Irrigation subsidies are probably bad

McMaken (2013):

Water shortages occur in the West not because too many people are flushing their toilets too often, but because agriculture, heavily subsidized through cheap water made possible by the federal government, continues to grow crops in places that would never support agriculture on a similar scale in a free market. Indeed, agriculture uses well over 80 percent of all the water used in Western states, and most of that water is stored, pumped, and diverted using dams, pumps, and aqueducts paid for by the U.S. taxpayer.

As a result, growers don’t have to face the real-life costs of transporting water to their farms. They only need consider the subsidized price, which is far below what it would be in a private market. Consequently, water usage for growers across the West is much greater than what it would be were there a functioning market for water in the region.

Edwards and Hill (2012):

Many cities and farm businesses in the West have become dependent on water from Bureau of Reclamation projects. However, the agency's policies have created economic distortions and environmental damage. Numerous Reclamation dams have not made any economic sense because the costs of the dams have outweighed the benefits gained from irrigation farming and other marginal uses of water. [...]

About four-fifths of Reclamation water is directed to agriculture.[...] Generally, the higher prices paid by urban water users and power customers subsidize the much lower water prices paid by irrigators. Prices vary widely in the West, but farmers often pay no more than 10 percent of the water's market value.

Reducing water subsidies would probably increase the price of crops grown with irrigation and thus favor crops grown with rainwater. This would presumably reduce the USA's total NPP because dry Western states would have less plant growth, and crop cultivation would displace more already vegetation-dense land elsewhere.

There are some counterarguments to this conclusion. For example, reducing water subsidies would also encourage more efficient irrigation methods, which, as mentioned above, might increase NPP per hectare on crop fields as well as total long-run water supplies—due to, e.g., slower depletion of fossil-water reservesd. My sense is that these effects are less important than the direct effect of reducing irrigation in dry regions, but I haven't run the numbers.

This page says "Conservation advocates have urged removal of all [water] subsidies to force farmers to grow more water-efficient crops and adopt less wasteful irrigation techniques." Does growing more water-efficient crops reduce NPP? I don't know; maybe some plants have high NPP despite requiring little water?

Consuming non-irrigated food

One idea for reducing one's contribution to irrigation could be to buy foods from regions of the country that require less irrigation. If crops require less irrigation, it's likely that the crop fields are displacing native vegetation that was relatively productive (because the region isn't arid), which means that the change in NPP relative to that of native vegetation caused by crop cultivation may be small or even negative.

The following figure is from this document:

It seems difficult to trace down the state of origin of most foods and food ingredients. Still, we can know that, for example, some crops are grown primarily in California:

California produces a sizable majority of many American fruits, vegetables, and nuts: 99 percent of artichokes, 99 percent of walnuts, 97 percent of kiwis, 97 percent of plums, 95 percent of celery, 95 percent of garlic, 89 percent of cauliflower, 71 percent of spinach, and 69 percent of carrots (and the list goes on and on).

This document says:

About 80 percent of all land in orchards, berries, and vegetables is irrigated. Other crops with more than 25 percent of total acres irrigated in 2012 include rice (100 percent), cotton (41 percent), alfalfa hay (35 percent), peanuts (32 percent), sugar beets (32 percent), dry edible beans (29 percent), and barley (26 percent).

If a given food is grown both in irrigated and non-irrigated regions, then buying more of it from the non-irrigated region might lead to substitution in which other people buy more of it from the irrigated region instead. If items of a given food are considered interchangeable by most consumers, then if you buy one particular item of that food, it may be best to imagine that you're actually slightly increasing production of all the interchangeable versions of that food (both irrigated and non-irrigated). If a given food is mostly non-irrigated, then buying it will mostly increase production of non-irrigated versions of it.

We should also remember that since the goal is to reduce NPP and not just to reduce irrigation per se, one should also consider the intrinsic NPP of different crop types and the NPP of the native vegetation they displace. My informal guess is that fruit/vegetable crops often have lower NPP than grain crops, although this is just non-rigorous speculation based on looking at pictures of different crop fields. Lower NPP generally means less invertebrate suffering. Still, if California land would otherwise be arid in the absence of irrigated farmse, then maybe lower-NPP California fruit/vegetable crops actually cause a bigger increase in NPP relative to the counterfactual without farming than high-NPP Midwestern grain crops, which displace more productive prairie grasslands? A full analysis here is complex, and it would help to gather specific numbers for native NPP and crop NPP by crop type and region, rather than continuing to guess based on crude generalizations.


A friend introduced me to the topic of groundwater-dependent vegetation.


  1. Ponce (2006) explains that while most groundwater flows toward the closest ocean, "A smaller portion of groundwater accumulates in closed, endorheic drainage basins lying in the interior of continents". Wikipedia says: "Approximately 18 percent of the earth's land drains to endorheic lakes or seas, the largest of these land areas being the interior of Asia." Water in these basins is often salty. How much vegetation does such water support?

    In any case, even if endorheic lakes aren't technically oceans, they're still (often saline) regions covered by water to which river water and groundwater eventually flow. So water flowing to such lakes is still "lost" from being able to contribute to much plant growth on land.  (back)

  2. This textbook says:

    Since phosphate and most metals are generally transported as particles, flood transport is important to the phosporus budgets of unpolluted lakes [...]. Nitrogen, which is largely transported in dissolved forms, is affected more by the volume of water dischrged than by current [...].


  3. Another question: Does damming rivers to create reservoirs reduce sea levels by keeping more water on land? My intuition is that this effect is extremely small. For example, if a reservoir with volume V covers over land area A, I would guess that by keeping the volume V of water out of the oceans, sea levels decline and expose additional land area around the coasts of much less than A. I think this because when water covers land, every place that it goes covers what was formerly land. In contrast, when water enters the ocean, in order to cover a little bit more land around the coasts, you also need to raise the water level in the whole ocean as well.  (back)
  4. Are there subsidies for groundwater? Or are most subsidies for surface water? If most subsidies are for surface water, then maybe removal of those subsidies would increase substitution toward groundwater because surface water would become relatively more expensive. This would presumably accelerate aquifer depletion and thereby reduce long-run NPP by decreasing fossil-water supplies?  (back)
  5. This report says regarding rainfall in Western states:

    The majority of total U.S. irrigation withdrawals (83 percent) and irrigated acres (74 percent) were in the 17 conterminous Western States [...], which are typical of areas where average annual precipitation is less than 20 inches and generally insufficient to support crops without supplemental water.