by Brian Tomasik
First published: 17 May 2017; last update: 18 May 2017


Water withdrawn from surface-water sources like rivers, lakes, and reservoirs contains zooplankton, which are often killed at some point during the processing or use of that water. This piece estimates that, very roughly, the typical American using public-supply water kills around ~2000 crustacean zooplankton per day for domestic water use. The average American also kills ~8000 crustacean zooplankton per day via water used for irrigation. These numbers are conservative because

  1. they don't include rotifers, which are often more numerous in freshwater than crustacean zooplankton
  2. they don't include nematodes or other aquatic invertebrates
  3. they don't include water withdrawals besides those for home use and irrigation
  4. they assume no zooplankton are killed by groundwater withdrawals.

The aquatic invertebrates that a typical American kills each year via domestic water use and consuming irrigated foods collectively have as many neurons as perhaps ~20 chickens, maybe several times more than this.

While the overall impact of water use on wild-invertebrate suffering complex, I prima facie encourage readers to reduce their own water consumption, especially if they use public-supply water. Reducing watering of lawns and houseplants is also good for another reason: namely, that attenuating plant growth decreases the number of terrestrial invertebrates who will be forced to endure suffering by being born and then usually dying not long thereafter.

In addition to reducing home water use, you could consider the source of your water supply as a factor when deciding where to live. Self-supplied water from wells and public-supply water from deep underground probably contain many fewer zooplankton than surface water. This section of the present article discusses how to identify whether a given residential area in the USA uses surface water or groundwater.

Copepods found in drinking water

In 2004, rabbis discovered copepods in the drinking water of New York City (NYC). This led to a flurry of debate among Orthodox Jews about whether such water was kosher, given that copepods are crustaceans like shrimp and lobsters. The New York Times explained:

the discovery has changed the daily lives of tens of thousands of Orthodox Jews across the city. Plumbers in Brooklyn, Manhattan and Queens have been summoned to install water filters -- some costing more than $1,000 -- and dozens of restaurants have posted signs in their windows trumpeting that they filter their water.

While water filters may make NYC tap water more kosher, they don't address the animal suffering that's involved in killing these copepods at whatever stages of the processes of water treatment and consumption cause them to die.

In NYC: "The copepods that come out of the tap are dead, in almost all cases. It is believed that the prolonged exposure to chlorine, along with the journey through the delivery system, are the causes. The lack of movement contributes to the difficulty in finding the copepods."

This post says: "The NYC water supply, like those of most cities, is treated at some point with chlorine to kill those living things that are in the water. However, although dead, the whole animal can be seen in the consumer’s tap water as visible white dots which can be identified sometimes as a whole sheretz [unkosher animal] (one can even see the antenna), sometimes they are broken and sometimes a magnification instrument is required to help identify it."

Cyclops thomasiOrthodox Union says of NYC water that "The primary species is Diacyclops thomasi". An individual from this species is shown in the picture on the right.

Cities that don't need to filter surface water

Copepods are not always present in the tap water of large cities. According to The New York Times: "The tumult is confined largely to New York because it is one of the few cities that is exempt from federal filtering requirements. Boston and Seattle are also exempt, but they have nothing like the city's numbers of Orthodox." This page adds that Farrell Sklerov of the NYC Department of Environmental Protection (DEP) "explained that many cities filter their water, but if the water quality exceeds federal standards -- which New York City tap water does -- it doesn't require filtering, a process that would remove the copepods. Among other cities that don't filter their water are Boston, San Francisco, Seattle and Portland, Sklerov said." Hence, many cities do filter zooplankton like copepods out of drinking water. But that doesn't mean those cities don't also kill zooplankton.

This page explains:

There are 7,400 surface water systems in the United States, of which 7,310 have filtration plants. Municipal filtration, if properly maintained, will successfully remove copepods and related organisms from the water, to a level acceptable in halachah. Of the 90 systems that do not filter their water, most serve a very small population. Aside from [the small cities,] New York City, Syracuse and Seattle are examples of large cities whose water supplies are (at least in part) not filtered.

Wikipedia echoes that: "Boston, New York City, San Francisco, Denver, and Portland, Oregon are among the large cities in the U.S. that do not need to treat their surface water sources beyond disinfection, because their water sources are located in the upper reaches of protected watersheds and thus are naturally very pure."

Evins (2004), p. 104: "Where water from upland reservoirs of good microbiological quality with a low content of suspended solids receives only limited treatment, planktonic species may enter the distribution system in appreciable numbers. However, they do not usually thrive there."

Where tap water comes from

NYC's copepods came from reservoirs: "In 1994 and 1995 the Department of Environmental Protection made studies of the copepod population of the reservoirs. These studies found copepods in all of the reservoirs." This page echoes that copepods "can be found in most freshwater habitats, including the reservoirs that supply public drinking water to cities like New York."

Lakes, reservoirs, and to some extent rivers are teeming with zooplankton, and many large population areas get their tap water from lakes, reservoirs, and rivers. For example:

  • NYC's water comes from the New Croton Reservoir, "two reservoirs in the eastern Catskill Mountains", and "tributaries of the Delaware River in the western Catskill Mountains". Also, water gets stored in reservoirs before reaching its destination: "Water from both [the Catskill and Delaware] aqueducts is stored first in the large Kensico Reservoir and subsequently in the much smaller Hillview Reservoir closer to the city."
  • The San Francisco Bay Area gets water from a variety of sources, including many reservoirs that were created by damming rivers, such as Hetch Hetchy Reservoir. Some water also comes "from local streams and rivers, fed by rainfall or [...] from underground aquifers."
  • Chicago's tap water comes from Lake Michigan.
  • Boulder, Colorado's water "comes from snowmelt, which flows through rivers, streams, lakes, ponds, and reservoirs".
  • Seattle Public Utilities's "Drinking Water Quality Report 2015" says (p. 6): "Two surface water sources provide the majority of water for our system. In 2015, about 61% was provided by the Cedar River. Approximately 37% came from the South Fork Tolt River. In addition to these two protected Cascade Mountain watersheds, the system has access to wells located in Burien that are only used to meet peak summer demand."
  • Chester, a small city in Pennsylvania, has its water treated by the Chester Water Authority (CWA). The CWA Octoraro Treatment Plant gets water from "the Octoraro Reservoir on the Octoraro Creek, and the Conowingo Pool of the Susquehanna River."
  • London's water mostly comes from the Thames and Lea rivers. River water may be stored in reservoirs, like the Wraysbury Reservoir and Queen Mother Reservoir.

This thesis says "Reservoirs provide much of the drinking water supplied to Georgia" (p. 4).

However, some cities get tap water from groundwater. And many rural areas use wells for drinking water. In the US, 40 million people get water from private wells, and "For over 150 million Americans, drinking water is supplied from ground water" in some form.

Zooplankton in water treatment


This page explains:

Zooplankton that pass into the clarification process of treatment works in elevated numbers will block filters, this can reduce the run times and efficiency of the works.

The organisms’ biomass in the process may also soak up Iron or Aluminium compounds from the flocculation stage, which will affect the required dosage levels.

Zooplankton that passes through into distribution can be unsightly, and will take up organic matter from the mains pipework to present a discolouration problem[. ...]

When surface water is treated, the vast majority of these animals and plants are removed. However there are some occasions when animals and algae can pass through water filters and enter the water distribution system, which conveys water to homes and workplaces.


Orthodox Union explains:

  • Water fleas (Cladocera) are also planktonic, but many species are present in large numbers for only several weeks a year. They are also much more delicate than copepods, and are less likely to survive the trip to the tap. They have appeared at the tap in large numbers only sporadically.
  • Nematodes, amphipods, and ostracodes all live at the bottom of reservoirs and lakes, rather than in open waters. They are thus less likely to be sucked into the effluent of the reservoirs. Isolated discoveries of all these organisms have been found in NYC tap water, but in frequencies low enough to be halachically unimportant.
  • Rotifers are common, but they are extremely small (~0.1mm) and are generally considered microscopic. They cannot be recognized with the unaided eye.


Lin et al. (2014), p. 2846:

In recent years there have been reports of the propagation of invertebrates in biological activated carbon (BAC) filters, thereby endangering the quality of the product water. Due to increasing eutrophication and over-fishing in China, large populations of grazing zooplankton have been found to occur in drinking water sources, which raises the possibility of zooplankton being present in waterworks.[1,2] The dissolved oxygen (DO) and particles in BAC filters provide a suitable living environment and a rich food source, for the breeding of copepods zooplankton.[3] Copepods exhibit great vitality and transfer potential and can easily penetrate BAC filters. They may contaminate the product water and even enter the municipal distribution network,[4,5] which poses a threat to human health.[6,7] Zooplankton can be seen by the naked eye, which would cause consumers to feel that the water is not sanitary. In addition to their negative effect on the aesthetic values of drinking water, the presence of zooplankton – which may carry microbes such as bacteria – increases the threat to human health.[8,9] Zooplankton is present in a waterworks in Shijiazhuang in northern China [10] and has been recorded in the water supply to households in Jilin province.[11]

Zooplankton losses during water treatment

General information

Christensen (2011), p. 33: "Sand filters retain some invertebrates and prevent them from entering drinking water distribution systems."

van Lieverloo et al. (2002), p. 1724:

Presently, in most surface water treatment plants, protozoan and invertebrate removal is part of algae removal and is usually considered of minor importance. [...] small motile invertebrates such as nematoda (59,60,62,64,65) and Rotifera (38,105) are hardly removed by some surface water treatment plants using coagulation and sedimentation followed by rapid gravity filtration, with or without chlorination. [...] Planktonic species however are better removed in treatment than benthic species (107). Furthermore, slow sand filters are known to remove invertebrates better than rapid gravity filters (27,107,108), and the backwash rate of rapid gravity filters controls invertebrate removal (109).

Liu et al. (2007) mention in their "Introduction" (pp. 1826-27):

Eutrophication caused by water pollution results in excessive propagation of zooplankton cyclops in waterbodies, which are hard to be removed by the conventional disinfection processes like chlorination due to its strong resistance to oxidation. In addition, the motility of cyclops enables it to easily penetrate from sand filter into the clear water tank in waterworks, even municipal distribution network. [...]

The occurrence of cyclops which penetrates the filter tank in drinking water showed that it cannot be thoroughly removed from water by the conventional water treatment process of flocculation, sedimentation and filtration (Liu et al., 2004a; 2004b; 2005; 2006). Therefore the removal of cyclops with conventional water treatment process, and then, the feasibility to remove it by chemical pre-oxidation in cooperation with the conventional process are put forward in this paper. The following steps are considered: thorough inactivation of cyclops utilizing oxidants, inhibiting its activity by pre-oxidation and then removing it thoroughly by subsequent clarification process (Kanio and Kimata, 2000; Reckhow et al., 1990; Fiessinger, 1991).

To inactivate or weaken cyclops with oxidants was believed to be the key to removing cyclops completely from water treatment system (Ruffell et al., 2000; Gunter and Pinkernell, 2000; Driedger et al., 2001).

I'm uncertain how common copepod-inactivating pre-oxidation is in actual water treatment.

Demir and Atay (2002)

Demir and Atay (2002) examined the Ivedik Water Treatment Plant, which is "Turkey’s biggest water treatment plant" and "is in the 10 biggest treatment plants in Europe". The treatment steps in the plant are illustrated by Demir and Atay (2002) as follows:

Water samples were taken at the locations marked with "X"s in the diagram. The following table shows various measurements, including zooplankton densities, at those locations.

The "Balancing room" is the last step in the figure before the water is distributed to end users. As you can see, only about 1 zooplankter per liter remains by this point, even though we started with 81 per liter in the "Mixing room" (the room "where the water of two reservoirs are mixed", p. 230). The zooplankton are gradually eliminated through the treatment process.

It's possible that the number of zooplankton killed here is higher than usual, given that "Both reservoirs were classified as eutrophic according to phytoplankton indices (13,22)" (p. 233). That said, the 81 zooplankton per liter in the intake water is a lower abundance than, e.g., the 139.7 per liter found by Snow (1974). On the other hand, the water taken for water treatment came from depths of 12 to 20 meters (Demir and Atay, 2002, p. 232), and "In these depths, the abundance of plankton decreased significantly", so presumably surface zooplankton were even more abundant than 81 individuals per liter?

Lin et al. (2010)

Lin et al. (2010) examined zooplankton in a water-treatment system in China. The authors explain by way of introduction (pp. 512-13):

In recent years, granular activated carbon (GAC) filters have been used as a method of advanced treatment for drinking water purification in some waterworks in southern China. The GAC filtration is usually applied following sand filtration and prior to final chlorination. However, some reports have indicated that the propagation of invertebrates can occur in GAC filters, thereby endangering the quality of the final product (Schreiber et al. 1997; Castaldelli et al. 2005; Weeks et al. 2007). Indeed, Copepods such as Cyclopoida and Harpacticoida are commonly found in GAC filters in waterworks in southern China (Li et al. 2007). Zooplankton has strong resistance against oxidation and is not effectively removed by conventional disinfection methods such as chlorination (Lin et al. 2004; Liu et al. 2007). In addition, the motility of zooplankton enables them to easily penetrate the GAC filter and enter the clear water reservoir of waterworks, and may even enter the municipal distribution network. Zooplankton can be seen by the naked eye, which would cause consumers to feel that the water is not sanitary. [...]

It is difficult to kill Copepod using liquid chlorine at dosages commonly used by drinking water utilities. Additionally, disinfection byproducts may be created when higher chlorine dosage is used to completely inactivate Copepod. As a substitute or a supplemental disinfectant for chlorine, chloramines produce fewer disinfection byproducts and have a longer exposure time than chlorine (Vasquez et al. 2006; Hua & Reckhow 2007). In addition, it has been reported that chloramines are more effective at inactivating Copepod than chlorine (Farrell et al. 2001). In this study, experiments were conducted to investigate the reasons for the propagation of Copepod in the GAC filters and to compare their disinfection by chloramines to disinfection with liquid chlorine.

In order to culture copepods (p. 513): "mature Copepods were initially collected from GAC effluent in a municipal waterworks in southern China (Jiangsu Province) and then artificially cultivated in the laboratory."

The following figure shows copepod abundances in the water source (p. 514). You can see that more eutrophic water had more copepods.

This table shows how the percentages of different kinds of zooplankton and protozoa changed over the course of water treatment, from raw water, through sand filtration, through ozone, and through GAC filtering.

You can see that in the raw water, copepods constitute only 5.3% of the measured organisms, but most of the non-copepods get removed by sand filtration.

However, I presume that most of those filtered-out organisms get killed in the sand filter, perhaps by being trapped there? Or maybe the organisms get removed by backwashing? Wikipedia explains: "Sand filters become clogged with floc after a period in use and they are then backwashed or pressure washed to remove the floc. This backwash water is run into settling tanks so that the floc can settle out and it is then disposed of as waste material. The supernatant water is then run back into the treatment process or disposed of as a waste-water stream." I would guess that most of the organisms get killed in some stage of that processing.

Lin et al. (2010) explain that because the copepods were motile, some of them could get through the sand filter and enter "the subsequent ozone contact tank" (p. 514). In fact, the following graph (p. 515) shows how many copepods got through each stage. These are densities in effluent, so, for example, the bar for "Sand filter" means the numbers of copepods getting through sand filtration.

Lin et al. (2010) note that "The densities of Copepod in the GAC filter effluent were approximately three times higher than that of the sand filter (Figure 2), which indicates that the propagation of Copepod occurs in the GAC filter." Why did copepods grow in numbers in the GAC filter? Based on growing copepods under similar artificial conditions, the authors speculate that copepods thrived in the GAC filter due to high dissolved oxygen, abundant food, a protective environment, and no fish predators (p. 516).

Lin et al. (2010) then experimented with inactivation of copepods using various doses of chlorine or chloramines applied to GAC filter effluent (p. 519). In the following graph, "CT" is the "product of disinfectant concentration and contact time" (p. 517).

How about groundwater?

My impression is that groundwater is unlikely to contain large numbers of zooplankton. Since it's underground, it doesn't get sunlight that could power algae growth. Some zooplankton (including some cladocerans and rotifers) feed on bacteria and protozoa, and groundwater can contain some of those microorganisms: "According to Oregon State University and the Extension Toxicology Network, bacteria such as coliform bacteria[ ]including Escherichia coli, parasitic protozoa such as giardia lamblia and cryptosporidium, viruses such as hepatitis A and parasitic worms such as helminths are common waterborne microbiological contaminants that affect the quality of untreated well water." This page says: "Things like microorganisms[...] and more can all be found in traces in ground water." However, the same page says "Most groundwater is dubbed safe". BWB (2014) says regarding Berlin's water (p. 6): "Drinking water coming from groundwater reserves is considered free of germs and therefore does not need to be chlorinated."

Wikipedia says: "Deep ground water is generally of very high bacteriological quality (i.e., pathogenic bacteria or the pathogenic protozoa are typically absent)". That said, this study about tube wells in Bangladesh reports:

Recent studies have demonstrated that underground water systems are increasingly vulnerable to both microbiological and heavy metal contamination, especially by arsenic, in Bangladesh. Such problems arise even in developed countries. For example, in 1994, an outbreak of cryptosporidiosis occurred in a rural community in Washington State, where water was supplied by two deep, unchlorinated wells (4).

Besides chemical contaminants, eukaryotic microorganisms (protists) are also a significant component of microbial communities inhabiting groundwater aquifers. This is not unexpected, considering that many protists feed heterotrophically, via either phagotrophy (bacterivory) or osmotrophy (able to grow in the dark on dissolved organic carbon). Protistan numbers in water are usually low [...] in pristine, uncontaminated aquifers but may increase by several orders of magnitude in polluted aquifers with a high organic content. Small flagellates (typically 2 to 3 μm in size in situ) are by far the dominant [protists] in aquifers, although amoebae and occasionally ciliates may also be present in relatively lower numbers (17).

In Sep. 2016 and again on 3 Mar. 2017, I briefly checked for invertebrates in the tap water of a house in a rural area near Albany, NY, USA. I used a microscope camera that allows me to notice moving organisms as small as about ~0.2 mm in diameter. I didn't find any invertebrates on either occasion when I looked for them, though I only examined a few milliliters of water and so wouldn't have noticed very low densities of organisms. The person who built this house tells me that its tap water comes from a deep well (several hundred feet down). The water is not treated at all.

According to van Lieverloo et al. (2002) (p. 1723), groundwater sometimes does contain invertebrates, though my guess based on this passage is that densities are higher in shallow groundwater than in deep groundwater:

Although less conspicuous than in surface water, groundwater biocoenoses may harbor an abundance of invertebrates that are visible to the naked eye. [...] Coarse sediments and cracks in rocky bottoms will facilitate the passage of surface water into groundwater supplies, especially via shallow wells. Soils with finer grains and coarse deep layers that are covered by fine-grained layers only harbor invertebrate species that are confined to groundwater and caves.

Evins (2004), p. 106:

Various observations and studies have shown a link between the type of water source, particularly its organic content, and the extent of animal populations in the water mains. Water from deep underground sources generally supports lower numbers of animals than water from surface sources, probably because water from underground has a low organic content.

While Evins (2004) is discussing invertebrates in water-distribution systems, rather than invertebrates in the original water sources from which water was initially withdrawn, the principle that "less organic matter implies fewer invertebrates" presumably holds in either case. See also this statement from van Lieverloo et al. (2002), p. 1724:

In a survey of 17 finished water reservoirs in Germany, sediments in reservoirs of surface water supplies were found to harbor high numbers and a wide variety of planktonic species next to some benthic species, whereas the sediments in reservoirs of groundwater supplies contained only low numbers of Rotifera and nematoda.

van Lieverloo et al. (2002) also list "benchmark values" for invertebrates in finished water from treatment plants in the Netherlands (Table 1, p. 1730). From that table, I see that the thresholds for declaring "Very high numbers" of organisms in "Finished Water From Treatment Plants" are often lower for groundwater than for surface water. For example, the so-called "Sample ≥" threshold value in numbers per 1000 L for Rotifera is 10,000 for surface water but only 500 for groundwater. For Cladocera, the comparison is 30 vs. 10, and for Copepoda, the comparison is 1000 vs. 50. van Lieverloo et al. (2002) also write (p. 1721):

In the Netherlands, many water utilities monitor finished water of surface water treatment plants for nematoda, among other invertebrates (68). Numbers of more than 1,000 per m3 are considered very high for finished water of surface water treatment plants. For groundwater treatment plants, numbers of 100 per m3 or more are very high (45).

Presumably, greater numbers of invertebrates in finished water correlate with greater numbers in the raw input water? If so, then the higher threshold values for invertebrates in finished surface water suggest that raw surface water generally contains more invertebrates than raw groundwater.

Finally, here's a question I'm curious about. When surface water seeps into the ground, it's naturally filtered (to a greater or lesser degree) by the sand and other materials in the ground.a Does this process kill zooplankton in a similar way as manmade sand filters do? Or are the zooplankton in the surface water able to swim up away from the sand and stay in the surface water? Even if zooplankton are killed by this "natural" filtration process, we might assume that this filtration would happen anyway, so that human use of groundwater doesn't increase zooplankton killing relative to the baseline. But is that true? Or might withdrawing groundwater lead more surface water to filter downward to replace what was lost? I would guess that if the groundwater is directly connected to the surface water near the Earth's surface, then withdrawing groundwater might indeed pull down more surface water? In contrast, if the groundwater is deep underground and surface water seeps down to that depth only slowly, then withdrawing groundwater wouldn't cause more surface water to filter downward?

An example waterworks system: Berlin

To illustrate some complexities that can be involved in the sourcing of water for a city, I'll discuss where Berlin gets its water from.

Directly, Berlin's water is all groundwater:

  • Grohmann and Petersohn (2000)b: "Berlin's water supplies originate exclusively from groundwater."
  • BWB (2014) is an informational report by Berliner Wasserbetriebe, the water company for Berlin. The report explains (p. 5): "We extract groundwater from the soil, which is then aerated and filtered in the waterworks. The Berlin drinking water is ready. Without any added chemicals." The report adds (p. 12): "Groundwater is pumped from over 650 wells between 30 m and 170 m deep to the waterworks, where it is further processed and stored in clean water tanks."
  • nuBerlin (2015): "Technically Berlin’s tap water is taken from the ground – not directly from the river Spree or from any of the many lakes. [...] Ground water is [...] pumped up by the local water supplier and then cleaned and processed at the local water supply works."

How does this groundwater get replenished? BWB (2014), p. 10: "Drinking water is extracted from groundwater. This is constantly renewed by means of bank filtration of stormwater and surface water from rivers and lakes. [...] Stormwater and surface water slowly and evenly pass down through layers of sand and gravel into the groundwater." In terms of zooplankton killing, I would guess that bank filtration is more similar to groundwater withdrawal than surface-water withdrawal because I assume that the river water filters through soil at its own pace, and zooplankton can perhaps stay in the river and avoid being forced into the soil of the river bank?? In contrast, in an artificial sand filter, zooplankton have no river to swim away into; they either get stuck on the filter or pass through it?

So, does this mean that Berlin's tap water kills very few zooplankton because it all comes from groundwater? Unfortunately, no. The reason is that some surface water is treated/filtered and used to increase groundwater supplies. BWB (2014), p. 16:

Sufficient amounts of groundwater do not always form naturally. Therefore, in order to extract the required quantities, we at Berliner Wasserbetriebe replenish groundwater with treated surface water. [...]

Surface water can be pre-treated before it is used for filtration purposes. In two treatment plants for surface water, the substances that can be filtered out, as well as the phosphates and nitrates, are reduced using the flocculation-sedimentation-filtering principle. To this end, flocculating agents are added to the water and then filtered. The Beelitzhof surface water treatment plant purifies water from Wannsee Lake, which is then discharged into the Grunewald lakes.

Grohmann and Petersohn (2000): "For sustainable water management, river water is treated by flocculation and filtration and used either for artificial groundwater recharge (rivers Spree and Havel) or for bank filtration (Nordgraben and Lake Tegel)."

I wonder what this filtration looks like. Is it a filtering process that kills most zooplankton?

How much of Berlin's groundwater is replenished with treated surface water rather than being replenished by natural surface water? BWB (2014) shows (p. 4) a map of Berlin's water system that indicates three "surface water treatment plant" locations (all on the west side of Berlin).

So Berlin water may not fully avoid zooplankton killing, though I would guess that water use in Berlin kills fewer zooplankton than water use in many other cities does.

Invertebrates in water-distribution systems

So far I've focused on the killing of zooplankton that were in the source water, such as lakes and rivers. However, small animals may also enter water-distribution systems later on. Christensen (2011) has an excellent illustration of possible invertebrate entry points into the water system (p. 18):

Evins (2004) likewise explains (p. 104) that small animals may be found in water-distribution systems for several possible reasons:

  • They might enter from the source water, passing through the water-treatment plant or colonizing the treatment plant
  • They might come through entry points in the distribution system, "such as badly screened service reservoirs"
  • They might breed within the distribution system.

Christensen (2011), p. 1: "Reports of invertebrates in drinking water are global, and the World Health Organization (WHO) concludes that few if any drinking water systems worldwide are free of animals." Evins (2004), pp. 102-03:

the animal populations of water distribution systems were studied in the United Kingdom in the 1960s and 1970s; about 50 systems were sampled, and animals were found in all of them [...]. A systematic survey in the 1990s of water distribution systems supplied by 36 treatment works in the Netherlands also found animals in all of them [...].

For simplicity, in this piece, I've focused on animals in the source water. I conjecture that most of the invertebrates killed by water use are those in the source water, since it presumably has such high invertebrate populations? Note that the total number of invertebrates killed by water use overall is not the same as the total number of invertebrates in distribution pipes or that show up in tap water, because most invertebrates in source water are removed by water treatment. Moreover, not all the invertebrates in the water-distribution system are killed by being there. Evins (2004), p. 105:

it is not necessarily the species that pass treatment in the greatest numbers that are most common in the mains. A survey (Evins & Greaves, 1979) of treatment works and their associated distribution systems showed that, for most species, it is success of reproduction within the main that determines the size of the population. Thus, the species that are common in the distribution system are not necessarily those that appear most frequently at consumers’ taps (van Lieverloo, 1997). This is because the species that thrive in the pipework may resist dislodgement and suspension in the conveyed water, whereas those that are present in the incoming supply may pass directly to consumers’ taps.

That said, humans may intentionally kill invertebrates in water pipes periodically by physical and/or chemical methods (Evins 2004, sec. 6.4).

Evins (2004), pp. 117-18:

Planktonic species, which predominate in stored waters, are relatively easily removed by treatment and do not thrive in the distribution system. Benthic species, which account for a greater proportion of the raw water community in river waters, are more likely to pass treatment, and in turn are more likely to thrive in the distribution system.

Bichai et al. (2008), p. 510:

Although for the most part, invertebrates are intercepted and eliminated during sedimentation, some can reproduce inside the [treatment] plant and liberate eggs and larvae into the distribution system (Levy et al. 1986). Rotifers and nematodes abundantly colonize granular and biological filters, which constitute ideal media for the proliferation of benthic invertebrates (Lupi et al. 1994; Schreiber et al. 1997; Castaldelli et al. 2005). They are often released into the filter effluents (Matsumoto et al. 2002) and into the distribution systems. Investigations conducted on several drinking water distribution systems have confirmed the abundance of invertebrates (Chang et al. 1960b; Van Lieverloo et al. 1998), while amoebae are known to proliferate in the biofilms.

What are the densities of invertebrates entering and leaving the water-distribution system? van Lieverloo et al. (2002), p. 1728 (Fig. 6), give the following figures, where the circles, stars, and diamonds represent estimates from different sources:

Since 1 m3 = 1000 L, all of these densities are less than 1 individual per liter (within a given taxon). In comparison, as the next section explains, densities of zooplankton in surface-water sources are often at least 1 to 100 per liter. This reaffirms my supposition that most invertebrates are killed during treating or filtering source water, rather than within water mains.

Christensen (2011) presents (p. 16) the following table:

Where concentrations are given in the above table, the average is usually less than ~1000 per m3, or 1 per L, although adding these up across taxa will yield a somewhat bigger number. I haven't checked whether these concentrations are numbers of organisms that statically occupy a given section of pipes/etc. or whether these numbers are the organisms that flow with a given section of water.c

Zooplankton densities in surface water

In this piece, I've chosen to focus on crustacean zooplankton (such as copepods and cladocerans). I'm not counting rotifers because even though they're multicellular animals with nervous systems, they can be simpler than crustacean zooplankton, which casts their sentience more into doubt. I also haven't counted other planktonic or benthic invertebrates, which makes my calculations conservative.

This page reviews some estimates of zooplankton densities in rivers, and this page does the same for lakes and reservoirs. Numbers vary, but densities of crustacean zooplankton typically range between 1 and 100 per liter. In this piece, I'll use a point estimate of 10 crustacean zooplankton killed per liter of surface water withdrawn. This may be somewhat conservative as an estimate based on the densities discussed in the above links, since, for example, even if half of locations have crustacean zooplankton densities of 1 per liter and half have densities of 100 per liter, the average is 50.5 per liter, not 10 per liter. On the other hand, I want to keep the estimate low because some of the zooplankton density numbers were for water near the top of rivers or lakes, and we might expect zooplankton densities to be higher near the top of a water column rather than much further down??d That said, some of the studies in the above links (e.g., Ahmed (2010) for reservoirs and Orsi and Mecum (1986) for rivers) reported zooplankton densities in the whole water column (i.e., using a vertical plankton tow), and these numbers are still consistent with an average crustacean density on the order of ~10 per liter. (Of course, it's theoretically possible that a vertical plankton sample could have a high average density of zooplankton even if all the zooplankton were in the top part of the water column and none were at lower depths.)

This post says regarding NYC water: "Plankton [tows] of water in the reservoir system have captured as many as 350 copepods per liter with an average value of 50 copepods per liter. It is interesting that so few actually reach the consumers’ water taps."

This page reports: "The Department of Environmental Protection has confirmed that copepods are present in significant quantities throughout the water of New York City. The level of infestation varies from place to place. At one testing of 18 samples taken by the Department of Environmental Protection, the copepod count was, on average, 9 organisms per liter of water." I can't tell from that quote alone if this copepod density was in raw source water or in consumers' tap water. And keep in mind that this is just copepods, not cladocerans or other crustaceans.

Other aquatic invertebrates

This textbook reports that plankton may be less common in rivers than in lakes: "True plankton are common only in deep, slow-moving stretches of rivers or behind dams. Unlike [in] lakes, benthic invertebrates, especially insect larvae, constitute the bulk of the invertebrate fauna".

In addition to killing zooplankton, perhaps extracting river water can also kill benthic animals like insect larvae? Or maybe they're less likely to be sucked up? Maybe some would get captured as tychoplankton, i.e., as "organisms, such as free-living or attached benthic organisms and other non-planktonic organisms, that are carried into the plankton through a disturbance of their benthic habitat, or by winds and currents"?

This page reports: "According to spokesperson Edward Timbers at the DEP, midge fly larvae have also been found in NYC drinking water at times."

Evins (2004), p. 105: "larvae of many species of chironomid may be present in the distribution system in appreciable numbers. Larvae are often present in large numbers in rivers and reservoirs, and may penetrate treatment works."

Bichai et al. (2008) summarize (p. 519) densities of both metazoan invertebrates and protozoa at different stages of water treatment (raw water, water-treatment filters, and post-filtration water-distribution systems):

At the source [water], for instance, concentrations between 2 and 3000 amoebae/L and between 200 and 90 000 amoebae in river water were found during a 3 year investigation in Germany in untreated reservoir water and in rivers used as water supply sources, respectively (Hoffmann and Michel 2001). Densities of 200–300 rotifers/L are common in natural freshwaters and can occasionally reach 1000/L (Wetzel 2001). In terms of filter samples, important densities of nematodes were measured in sand samples taken near the surface of a slow sand filter bed (approx. 570 nematodes in a 30 g sand sample), as well as other types of zooplankton organisms, such as amoebae, rotifers, and copepods (approx. 140 amoebae, a similar quantity of rotifers, and about 60 copepods in a 30 g sand sample) (Hijnen et al. 2007). Invertebrate concentrations, mainly nematodes or rotifers, in the order of several thousands individuals per litre have been reported in the effluent of granular filters (Schreiber et al. 1997; Castaldelli et al. 2005), and concentrations of up to 400 amoebae/L were measured in filtered water from drinking water purification plants (Hoffmann and Michel 2001). As for distribution systems, protozoa are thought to be present in most systems in concentrations between 5 x 104 and 7 x 105/L (Sibille et al. 1998), whereas cladocerans and copepods have been found in concentrations between 600 and 750 organisms/m3 [i.e., 0.6 to 0.75 organisms per liter] in samples taken at hydrants (Van Lieverloo et al. 1998).

Chang et al. (1960) report the following results (p. 614), where "Raw Water" is before treatment and "Finished Water" is after treatment:


Zooplankton densities in surface water vary significantly depending on location, physical and biological parameters, time of year, etc. For example, Fig. 3 from Pace et al. (1992), shown on this page, reveals huge swings in zooplankton abundance over the seasons of the year. "For example, the B. longirostris population increased dramatically from 0.01 to 0.1 animals·L-1 during the spring flood to >100 animals·L-1 in early June. The timing and magnitude of this population bloom were repeated each year, although a lower maximum density was observed in 1989" (Pace et al. 1992, p. 1063).

This article reports:

"A lake doesn't go to sleep when it's covered with a blanket of ice and snow," said Liz Blood, program director in the National Science Foundation's Division of Environmental Biology, which funded the research. "While winter's lower temperatures and light levels may force lake life into a slower mode, algae and zooplankton are still abundant."

Smithsonian Institution (1999): "Not many insects are active in the winter, but the nymphs of dragonflies, mayflies and stoneflies live in waters of ponds and streams, often beneath ice. They feed actively and grow all winter to emerge as adults in early spring."

Mitcham et al. (1983)

Mitcham et al. (1983), discussing a treatment works in Langford, England (pp. 197-98):

The predominant zooplankton in Langford raw water are crustaceans, particularly Bosmina and Cyclops. [...]

During the growth period from June to August, the populations of either of these organisms may reach 20-50/L, accompanied by lower numbers of Daphnia, chironomid larvae, Nais, and nematodes. It is important to kill as many of these organisms as possible before filtration.

For this reason, a comparison of the ability of free and combined chlorine to kill the crustaceans was carried out in November 1979 and from April to August 1981 [...].

From the following table, we see that prechlorination killed most of the crustacean zooplankton present in the raw water.

The below table adds up the counts of all three types of crustacean zooplankton by month:

Month Crustacean zooplankton in raw water (numbers/L)
Nov. 1.1
Apr. 0.7
May 2.9
June 28.3
July 10.1
Aug. 10.4
Average over the months 8.9

We also see that these zooplankton seem to be generally more abundant in summer than in non-summer months, though I don't know how robust or generalizable these trends are.

Covariance of zooplankton and water use?

In this piece, I assume for simplicity that zooplankton densities are constant throughout the year. However, if densities are generally higher in summer, and if people use more water per day during summer (for watering lawns, filling swimming pools, irrigating crops, etc.), then the actual number of zooplankton killed by water use should be higher than (yearly average zooplankton density)*(yearly water use) because higher-than-average zooplankton densities and higher-than-average water use coincide.

As an oversimplified example, imagine that zooplankton densities are 0 per L in winter and 20 per L in summer, while water use for irrigation is 0 L in winter and 10 L in summer. The product (yearly average zooplankton density)*(yearly water use) is (10 per L)*(10 L) = 100 zooplankton killed, but the actual number of zooplankton killed is (0 per L)*(0 L) + (20 per L)*(10 L) = 200.

How many zooplankton are killed per person per day?

Maupin et al. (2014) is a report by the US Geological Survey (USGS) titled "Estimated Use of Water in the United States in 2010". This report (p. 18) estimates that "Approximately 42,000 Mgal/d [...] of water were withdrawn for public supply in 2010", where "Public supply refers to water withdrawn by public and private water suppliers that provide water to at least 25 people or have a minimum of 15 connections."e In addition, "An estimated 268 million people relied on public-supply water for their household use in 2010. This number represents about 86 percent of the total U.S. population." So, public-supply water per capita was roughly (42,000 million gallons per day) / (268 million people) = 160 gallons per person per day.f (Note: I assume that all "gallons" figures are US gallons, not imperial gallons.)

Maupin et al. (2014) add (p. 18):

Most of the public-supply withdrawals are delivered to customers for domestic, commercial, and industrial needs. Part of the total is used for public services, such as public pools, parks, firefighting, water and wastewater treatment, and municipal buildings, and some is unaccounted for because of leaks, flushing, tower maintenance, and other system losses. Domestic deliveries represent the largest single component of public-supply withdrawals, averaging 57 percent of the total nationally.

So public-supply water used for domestic purposes represented roughly 160 * 0.57 = 90 gallons per person per day (not counting leaks and other system losses for which domestic water use is partly responsible). That estimate matches with this source, which says regarding home water use that "Estimates vary, but each person uses about 80-100 gallons of water per day."

Maupin et al. (2014) say (p. 18): "Sixty-three percent of water withdrawn for public supply in 2010 was from surface sources, such as lakes and streams; the other 37 percent was from groundwater."g So on average, surface-water use per person per day for those on public supply was roughly 160 * 0.63 = 100 gallons. And just counting surface water for domestic use: 160 * 0.63 * 0.57 = 57 gallons (assuming the plausible premise that whether a given gallon of public-supply water is used for domestic or non-domestic purposes is roughly independent of whether it comes from surface water or groundwater).

57 gallons = ~220 liters of water per day. As discussed above, I'm assuming that ~10 crustacean zooplankton are killed per liter of surface water withdrawn. Hence, per-capita public-supply surface-water withdrawals for domestic use kill roughly ~10 * ~220 = ~2000 crustacean zooplankton per day.

Based on this source, here are some estimates of crustacean zooplankton killed by various uses of water, if all your water comes from surface-water sources.

Activity Liters of water used Approximate crustacean zooplankton killed
The water you drink per day 4 40
One toilet flush 11 110
Dishwasher 15 to 38 150 to 380
A load of wash 57 570
10-minute shower 76 760

This calculator lets you estimate your own personal water use. Mine is only 36 gallons = 136 liters per day, but that's partly because my house has low-flow water fixtures, and I don't do anything expensive like watering the lawn. (Watering the lawn is bad because, in addition to using lots of water, it increases plant growth and thus land-invertebrate abundances.)

This page says: "According to a 1999 study, on average all over the U.S. 58% of domestic water use is outdoors for gardening, swimming pools etc. and 42% is used indoors."

Filling a swimming pool requires 22,000 gallons of water (possibly killing almost a million crustacean zooplankton), "and if you don't cover it, hundreds of gallons of water per month can be lost due to evaporation." Swimming pools also trap and kill lots of bugs and possibly frogs. When I was a kid, I went to a friend's swimming pool almost every day during the summer. I would guess that the pool killed maybe ~1 to ~100 insects per day and 1 frog every few weeks. (That pool was in a rural area near a pond, so maybe animal mortality was higher than what's typically the case.)

NYC's water-supply system "provides over 1.2 billion US gallons [...] per day of drinking water to more than eight million city residents, another one million users in four upstate counties bordering on the water supply system, and visitors to the region." I'll ignore visitors, on the assumption that the number of people visiting NYC is roughly equal to the number of New Yorkers visiting other places at any given time.h So, assuming about 8 + 1 = 9 million people served, that implies ~130 gallons per person per day, for household, business, etc. uses.

Mitcham et al. (1983), p. 196: "The Essex Water Company is one of the largest water purveyors in the United Kingdom, supplying more than 1.3 million people and associated industry with a current daily average output of 340 ML [...]." That's a per-capita water use of 340/1.3 = 260 L = 69 US gallons.

BWB (2014), p. 12: "Each of Berlin's residents uses an average of 110 litres [i.e., 29 US gallons] of water per day."

Finding a water supply that kills fewer zooplankton

In order to kill fewer zooplankton via water use, when you move to a new location, you can look for locations where your water supply would come mainly from groundwater, especially deep groundwater. This will often be true for homes in rural areas that are supplied by wells. It may also be true for a few more populous areas, though my impression is that most large cities are supplied mainly by surface water.

This page mentions a few "Cities supplied primarily by groundwater" in the USA. But for a comprehensive list, see United States Environmental Protection Agency's database of Consumer Confidence Report information. Below is a screenshot of the top rows of this database when focusing on the state of California and sorting in descending order of population served.

As you can see, the big water systems tend to use surface water. If you sort in ascending order of population, you'll see lots of groundwater systems.

Some invertebrates may still be killed when using groundwater if invertebrates enter and breed in the water-distribution system, as we saw in the "Invertebrates in water-distribution systems" section of this piece. But I think this phenomenon usually involves many fewer organisms than are killed from raw surface water.

Another consideration is that moving to a place where water comes from groundwater slightly increases housing prices in that area, which will, on average and depending on price elasticities, cause some fraction of another person to live in a different area instead, plausibly one that uses surface water. (This ignores the possibility that your moving to a city makes other people more likely to move there, not less likely, perhaps because your friends and family want to join you.)

Also, increased groundwater use might hasten the day when the region begins using surface water, especially in cases where the region's groundwater use is unsustainable.

Other water uses

The following graph from this page shows freshwater use in the USA in 2005, where I think "Domestic" means "self-supplied domestic" (mostly well wateri), not domestic use of public-supply water.

As you can see, irrigation and thermoelectric water withdrawals are volumetrically bigger than public-supply withdrawals. I won't discuss thermoelectric water in this article, though I hope to write about it more in a later piece. Most thermoelectric water withdrawals are returned to where they came from:

Most thermoelectric power plants in this country use what is known as once-through cooling, a process that pulls in cold river, lake or coastal water to cool the steam that turns a power plant's turbines and then releases the heated water back into the environment. In 1995 it was estimated that less than three percent of water withdrawn for thermoelectric power was consumed (lost to evaporation).

I'm currently uncertain how many zooplankton are killed and injured by this process. My cursory impression from skimming some studies is that entrainment in cooling water sometimes kills zooplankton but doesn't always. To read more about this topic, search {zooplankton entrainment}.

In contrast, I assume that most zooplankton in irrigation water are killed eventually? Irrigation is discussed next.


Zooplankton deaths?

This page says (p. 1): "Irrigation ditches are manmade channels that deliver water to homes, farms, industries and other human uses. Most ditches divert water from natural creeks and rivers and bring it to other areas."

Zooplankton might die before reaching crop fields. For example, irrigated water may be filtered. This page says that the company's "filters are used to control zebra mussels and quagga mussels by stopping shells, adults, veligers, larvae and eggs from passing beyond the point of filtration."

If zooplankton survive the journey to crop fields, I would guess they still die fairly soon due to drying out, being eaten, etc.? (I couldn't find any information on this.)

Maybe zooplankton would survive the longest in flood-irrigation systems, where large collections of water remain on fields for some time. However, my impression is that even with flood irrigation, fields dry up relatively quickly?

  • In a section on "Flood (Levee) Irrigation", this page says "It is recommended that water not be allowed to stand on any area for longer than two days."
  • This page says that "High numbers of mosquitoes can develop in standing water as a result of flood irrigation." This suggests that flood-irrigated fields sometimes sustain standing water for many days in a row. That same page advises to "Minimize standing water in fields to less than four days by improving drainage channels."
  • I asked about this topic on Quora. One answer said: "A flood irrigation expert would have to supply a target standing time, but I would guess no more than an hour and probably less."

Stevens et al. (1985) describe irrigation near the Sacramento River (p. 26): "In general, water is diverted from the river or from reservoirs through irrigation canals, fields are flooded, pesticides are applied, and eventually the water is drained into sloughs and subsequently flows back into the river." I'm uncertain if the authors are referring specifically to rice irrigation or to all kinds of irrigation. I'm also uncertain whether some zooplankton survive the trip from the river, through crop fields, and back to the river. If so, and if this kind of round-trip path of irrigation water is common (is it?), then my estimates in this piece of zooplankton deaths due to irrigation may be a bit inflated.

If zooplankton do survive round-trip flows of irrigation water from a river back to the river, I would guess this mainly only happens in cases of flood irrigation?? And only a minority of all irrigation in the USA is flood irrigation. Maupin et al. (2014) say (p. 25) regarding the USA: "About 31,600 thousand acres (51 percent) were irrigated with sprinkler systems, 26,200 thousand acres with surface (flood), and 4,610 thousand acres with microirrigation systems."

Number of zooplankton killed

USGS figures

Maupin et al. (2014) say (p. 25): "For 2010, total irrigation withdrawals were 115,000 Mgal/d[...]. Withdrawals from surface-water sources were 65,900 Mgal/d, which accounted for 57 percent of the total irrigation withdrawals".j In 2010, the US population was 309 million, which implies an average of (65,900 million gallons surface water per day) * (3.785 liters per gallon) / (309 million people) = 807 liters of surface water used per person per day for irrigation. Assuming ~10 crustacean zooplankton killed per liter of surface water and no zooplankton in groundwater, this implies ~8000 crustacean zooplankton killed per day by the typical American via demand for irrigated food and fiber.

Actually, the above naive calculation isn't quite right, because of imports and exports. Americans consume some agricultural products grown in other countries, and some agricultural products grown in the US are exported.

This page claims: "The United States has about 5% of the world's population, yet it uses almost as much water as India (~1/5 of world) or China (1/5 of world) because substantial amounts of water are used to grow food exported to the rest of the world."

This page explains:

over 70 percent of the volume of U.S. production of tree nuts (largely almonds) and cotton were exported in 2011-13, as was more than 50 percent of rice and wheat production. Overall, the export share of U.S. agricultural production averaged 20 percent from 2011 to 2013 based on volume, the same average annual share since 2000.

On the flip side:

Over 95 percent of coffee/cocoa/spices [...] consumed in the United States are imported, as are about half of fresh fruits and fruit juices and almost a third of wine and sugar.

This graph shows that aggregate US imports and exports are fairly close in dollar terms. Assuming that surface-water withdrawals to produce a food product are roughly proportional on average to the dollar value of that product, then given that exports and imports are nearly equal in dollar-value terms, we may be able to roughly ignore agricultural trade for purposes of computing zooplankton killed per American, at least for a back-of-the-envelope calculation.

Mekonnen and Hoekstra (2010)

For alternate figures, we can consult Table 6 (p. 29) of Mekonnen and Hoekstra (2010), which mentions estimates of water footprints for various foods. For our purposes here, so-called "blue water" footprints are most relevantk, since blue water represents water withdrawn from lakes, rivers, groundwater, etc., mostly for irrigation. This page reports that the average American consumes 2,031 Calories daily (not counting food waste). Based on Mekonnen and Hoekstra (2010)'s Table 6, I constructed the following table showing the blue-water footprint of consuming 2,031 Calories = 2,031 kcal of each of the given foods. The Mekonnen and Hoekstra (2010) figures are global-average water footprints (p. 28), but assuming they roughly apply to the USA (do they?), then Americans' average per-capita daily blue-water footprint due to irrigation should be somewhere in the middle of the numbers in the below table.

2,031 kcal of this food... ...has this blue-water footprint (liters)
Sugar crops 370
Vegetables 360
Starchy roots 39
Fruits 650
Cereals 140
Oil crops 150
Pulses 84
Nuts 1100
Milk 310
Eggs 350
Chicken meat 440
Butter 120
Pig meat 340
Sheep/goat meat 450
Bovine meat 740

Since, according to the Maupin et al. (2014) numbers in the previous subsection, surface-water use for irrigation is only 57% of all irrigation water used in the USA, the numbers in this table should actually be multiplied by 0.57. An average of those numbers would be somewhat below the "807 liters of surface water used per person per day for irrigation" calculated based on Maupin et al. (2014). Adding on food waste and fiber products (including water-hungry cotton) would help a bit to bring up the total here. Imports/exports and differences between USA vs. global water footprints may also account for some discrepancies vis-a-vis the Maupin et al. (2014) number. Finally, I think the Mekonnen and Hoekstra (2010) figures count only blue-water consumption (p. 9), which doesn't include that portion of irrigation water that returns to local groundwater/rivers/etc.? In contrast, the Maupin et al. (2014) figures measure withdrawals, i.e., both consumed and non-consumed irrigation water.

Increased plant growth

Apart from its impact on aquatic invertebrates, irrigation seems fairly likely to increase wild-animal suffering for another reason: it plausibly increases terrestrial net primary productivity (on cropland, lawns, etc.), which creates more food for heterotrophs and thereby brings more animals into existence. Irrigation has additional environmental effects that are important to analyze as well. However, my intuition is that the harm caused by increasing plant productivity is the most significant consideration.

The following rough calculation compares aquatic-invertebrate suffering due to irrigation water use with terrestrial-invertebrate suffering due to irrigation increasing plant productivity. Maupin et al. (2014) explain (p. 25) that in 2010 in the USA, 129,000,000 acre-ft of irrigation water were applied to roughly 62,400,000 irrigated acres, implying that "The national average application rate for 2010 was 2.07 acre-feet per acre", i.e., irrigated crops got an average of 2.07 feet = 0.63 m of water per year over the soil surface. Over 1 m2 of cropland, that's 0.63 m3 = 630 liters of water used. The report also says that "Withdrawals from surface-water sources [...] accounted for 57 percent of the total irrigation withdrawals", so roughly 630 * 0.57 = ~360 liters of surface water were used per square meter of irrigated cropland. Assuming 10 crustacean zooplankton per liter of surface water, that's ~3600 crustacean zooplankton killed.

Meanwhile, a square meter of soil can contain tens to hundreds of thousands of terrestrial invertebrates (such as springtails and mites), even ignoring nematodes (see, e.g., the Curl and Truelove (1986) graph here). Greater crop productivity presumably increases populations of soil invertebrates because of more plant-root biomass and more aboveground plant biomass that will later decompose. So even if irrigation only increases crop yields by a low-ish amount (say, 20%? I'm just making this up), it seems likely that irrigation affects at least as many invertebrates by increasing plant productivity (thousands to tens of thousands of invertebrate-years per square meter?) as by killing aquatic animals (thousands of invertebrates per square meter?), although both factors are nontrivial. It's also worth remembering that:

  • The zooplankton killed by irrigation would have died of other causes later, so the badness of killing them is lower than the badness of bringing into existence terrestrial invertebrates that wouldn't have suffered and died at all if plant productivity had been lower.
  • An increase in the soil-invertebrate population by 1 invertebrate over the year during which the crop grows may imply creation of several individual invertebrates when the average invertebrate lifespan is shorter than a year. For example, creating 1 extra springtail-year would imply creating 10 extra springtails if the average springtail lifespan is, say, 1/10 of a year.

How many neurons?

Most people agree that a single zooplankter warrants less ethical weight than a single mammal or bird. But how much less weight? One crude proxy for moral weight is number of neurons. This page very roughly estimates that a crustacean zooplankter may have on the order of ~103(?) neurons.

Assuming that typical per-capita public-supply water withdrawals for domestic use in the USA kill ~2000 crustacean zooplankton per day and irrigation kills ~8000 per day, the total number of crustacean zooplankton killed per person per year would be roughly (2000 + 8000 crustacean zooplankton per person per day) * (365 days per year) = ~4 million. And assuming ~103 neurons per crustacean zooplankter gives ~4 billion total neurons.

For comparison, a chicken has around 221 million neurons, so 4 billion neurons is about 20 chickens' worth of moral importance. Of course, the cumulative suffering over time that we cause to an individual chicken, relative to the organism's maximum possible level of suffering, is arguably greater than that we cause to an individual zooplankter because factory-farmed chickens have terrible lives in addition to painful deaths. Also, the zooplankton we kill would have later died in some other way, while if we don't buy chicken meat, some number of broiler chickens are never born to begin with.

Personally, I think smaller animals deserve more moral weight than their numbers of neurons alone would suggest (Tomasik 2013). For example, suppose we take moral importance to scale as (number of neurons)3/4. Then a chicken matters (221 million)3/4 / 10003/4 = 10,000 times as much as a crustacean zooplankter. A typical American kills 4 million crustacean zooplankton per year, and according to the 3/4 exponent on number of neurons, these zooplankton have ~400 chickens' worth of moral importance. In comparison, a typical non-vegetarian American eats ~23.7 chickens per year. Maybe this calculation places more weight on zooplankton than we're willing to accept. In general, giving zooplankton more weight than a linear count of neurons may cause them to dominate our calculations relative to vertebrates.

Also remember that I've ignored rotifers and other non-crustaceans from this calculation. Their neurons might collectively be of a comparable magnitude as those of crustaceans. For instance, rotifers have perhaps ~200 neurons each but can be several times more numerous than cladocerans + copepods combined in rivers and lakes.

Is water use net bad?

Killing zooplankton is directly bad because it causes them immense suffering, relative to whatever levels of sentience they may have. However, one might argue that the zooplankton would have died later anyway, and killing them prevents them from giving birth to more future zooplankton. Unfortunately, if a given set of zooplankton is killed, then whatever food they would have eaten will be available for other animals to eat, so it's not clear that killing actually reduces long-run animal populations, and it may just increase the total, long-run number of painful deaths. Full analysis of questions like these is complex and non-obvious (Tomasik 2017). However, I think it's a reasonable heuristic to suspect that killing animals in painful ways is at least somewhat bad, because doing so may increase the total number of deaths that will occur.

Impact on phytoplankton productivity?

Is it true that withdrawing surface water leaves constant the size of aquatic food supplies?

Withdrawing water extracts some phytoplankton as well as zooplankton. (Which organisms end up eating this withdrawn phytoplankton?) My intuition is that this consideration is fairly small, since except in the case of algae blooms, phytoplankton biomass per liter of water tends to be pretty low:

The pyramid of biomass may be "inverted". For example, in a pond ecosystem, the standing crop of phytoplankton, the major producers, at any given point will be lower than the mass of the heterotrophs, such as fish and insects.

Thus, the amount of phytoplankton destroyed by water withdrawals is probably not huge.

It's also possible that water withdrawal affects phytoplankton in other ways. Stevens et al. (1985), p. 23:

Jerry Turner of the Striped Bass Working Group observed that only two notable spring [phytoplankton] blooms have occurred in the western [Sacramento-San Joaquin] delta since 1976, and both immediately followed shutdowns of the [California State Water Project] SWP diversion pumps for repairs [...]. The first incident was in May 1981 when the first samples following the pump shutdown indicated that a significant phytoplankton bloom had suddenly developed. The second incident of this kind occurred early in June 1982 when the SWP pumps again were shut down for repair work and a major phytoplankton bloom followed.

These results suggest that the water project diversions are, in some as yet unexplained way, having a major effect on the phytoplankton population and basic productivity of the western delta.

Given that this effect was "as yet unexplained", it's not clear if this observation is a fluke or if it's genuine and generalizes to other water-diversion projects. Stevens et al. (1985) also mention another possible explanation of phytoplankton declines based on reduced organic-waste inputs to the water (p. 24). Also, I don't know over what volume of water these declines occurred and thus how much phytoplankton was affected.

Replacing land with water

There are many other factors to consider when assessing the net impact of water withdrawals. For example, if your water use contributes to a future decision to create a reservoir over an area that's currently land, then your water use slightly contributes to long-run conversion of a land ecosystem to an aquatic ecosystem. (This video shows time-lapse footage of a reservoir's construction.) It's not clear if this is good or bad, but this section presents some evidence that lake ecosystems may contain more suffering per unit area than terrestrial ecosystems. If so, then water use that leads to the creation of more reservoirs would be worse than we thought. That said, I haven't explored this issue in any depth yet.

There are other environmental impacts of reservoirs and of water use more broadly that need to be considered in a full analysis.

Example: Glades Reservoir

Here's a back-of-the-envelope calculation to get a sense of how the magnitude of the "land replaced by water" effect of water use compares with the magnitude of the "kill withdrawn zooplankton" effect. Schnabel (2011) is a letter describing a "Safe Yield Analysis" for a proposed new Glades Reservoir that would meet projected increases in water demand in Hall County, Georgia, USA (p. 1). "The Chattahoochee River and Glades Reservoir components, as configured, provide an additional safe yield of 72.5 mgd" (p. 2). Figure A-2 of Schnabel (2011) shows how much area and volume the Glades Reservoir would occupy depending on how high it would be filled. At an elevation of 1180 feet, the reservoir would cover 850.0 acres and have a volume of 11.714 billion gallons. These numbers are corroborated by this page: "The proposed dam would impound an approximately 850-acre reservoir at a normal pool elevation of 1180 feet mean sea level (msl) and provide 11.7 billion gallons of water storage capacity. [...] The safe yield for the proposed Glades Reservoir is estimated to be 72.5 million gallons per day (mgd) on an annual average daily basis." A yield of 72.5 million gallons per day from a reservoir with 11.7 billion gallons is a daily yield of 0.6%.

I mentioned in an earlier section when discussing Curl and Truelove (1986)'s numbers of invertebrates per square meter of soil that invertebrates (not counting nematodes) on land number ~104 to ~105 per m2. Meanwhile, we also saw above that invertebrates in fresh water often number ~10 to ~100 per liter, which is ~104 to ~105 per m3. Assuming these densities only extend a few meters into the water column, then total invertebrate populations in water would probably fall in the same range (~104 to ~105 per m2) as those on land. So it's not obvious which type of environment hosts more invertebrates, but these numbers give a sense of the orders of magnitude we're talking about.

Creating the Glades Reservoir would convert from land to water 850.0 acres = 3.440 million m2. Let's assume this would change invertebrate densities (either up or down) by something on the order of 104 per m2. So the whole reservoir would increase or decrease invertebrate numbers by maybe (~106 m2) * (~104 per m2) = ~1010. This is a change in population size, i.e., a change in the number of invertebrate-years that would exist per year. If we assume that the typical invertebrate (including those failing to reach maturity) lives, say, ~1/10 of a year, so that there are ~10 expected lifetimes per year, then we have ~1010 * ~10 = ~1011 invertebrate lives and deaths created or prevented per year by the reservoir.

Meanwhile, how about zooplankton killed by water withdrawals? Assume the safe yield of 72.5 million gallons per day = 26.5 billion gallons per year = 100. billion liters per year. Assuming ~10 crustacean zooplankton killed per liter, that comes out to ~1012 crustacean zooplankton killed per year by water withdrawals. Of course, these animals would have died anyway, so the badness of killing a single one of them is less than the badness of creating a new life and death that wouldn't have existed at all. If we assume that killing, say, 10 zooplankton only increases the total number of deaths that happen by, say, 1 (since the rest of the deaths would have happened anyway), then directly killing ~1012 crustacean zooplankton would only increase total deaths by ~1011 per year, which is the same order of magnitude as the number of zooplankton deaths created or prevented by the reservoir's existence.

The numbers here are highly uncertain and will vary from case to case, but at least we've seen that it's not obvious whether one or the other of the considerations at hand is more important. Of course, converting land to a reservoir might increase invertebrate populations, so even if that consideration is more important than killing withdrawn zooplankton, it might still recommend reducing water use.

Other considerations

Another factor to consider is that water taken from rivers may reduce the volume of the river downstream. However, except for water that you drink, use to water plants, or that evaporates, tap water that you use in your home will often be returned to streams later, after being treated as wastewater. So it's not clear there's much change in the net amount of water in streams. Irrigation does take a lot of net water out of streams.

How about using groundwater? As mentioned above, withdrawal of deep groundwater probably doesn't kill many zooplankton, so this analysis might be dominated by other considerations. In some instances, groundwater contributes to the growth of terrestrial plants: "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." Therefore, we might naively favor more groundwater use, especially if, after the water is extracted from the ground, it ends up in rivers that flow to the ocean, rendering the water not useable by terrestrial plants. Using groundwater for drinking may also decrease the amount available for farmers, which has the benefit of reducing irrigation. On the other hand, if you live in a rural area with a well, you probably also have a septic system, which I would guess returns the water you use to the ground near where it's withdrawn, in which case the impact on net groundwater supply might be small, unless water percolation to deeper groundwater is slow?l Groundwater depletion may also lead to greater pressure for surface-water use, greater imports of water-intensive products grown elsewhere, and other side effects.

In conclusion, based on what arguments I've considered so far, I suspect that home water use—in areas where public supply comes from surface water—is somewhat negative in expected value, though there's huge uncertainty in this analysis. Using water to irrigate lawns or farmland seems even more likely to be bad.


This piece has focused on zooplankton in freshwater. Non-fresh water is also sometimes withdrawn for various purposes, one of which is desalination, although desalination makes up a low share of total water supply in the USA. While I haven't looked into the issue in depth, it's plausible that desalinating water also kills zooplankton and other critters that it contains.

Scheer and Moss (2007):

"Ocean water is filled with living creatures, and most of them are lost in the process of desalination," says Sylvia Earle, one of the world's foremost marine biologists and a National Geographic Explorer-in-Residence. "Most are microbial, but intake pipes to desalination plants also take up the larvae of a cross section of life in the sea, as well as some fairly large organisms[:] part of the hidden cost of doing business," she says.

Other entrained animals

In addition to zooplankton, other small aquatic animals may be killed by water withdrawals. Stevens et al. (1985), p. 25:

Striped bass eggs, larvae, and juveniles are lost via entrainment in diversions of [Sacramento-San Joaquin] delta water by the [Central Valley Project] CVP, the [California State Water Project] SWP, delta agriculture (DA), and the Pacific Gas and Electric Company (PGE). Fish losses depend on the density of organisms at the pump intakes, the pumping rate, and (in the case of PGE) mortality occurring during passage through the power plants before the cooling water is discharged back into the delta.

I assume they say mortality rate only matters for PGE because for non-PGE water withdrawals, the entrained organisms aren't returned back to the source water (so that 100% of them are "lost")?

To estimate fish losses, some investigators used the same basic approach as I did to estimate zooplankton mortality: namely, multiplying densities of these organisms in the water by the amount of water withdrawn/diverted. Stevens et al. (1985), pp. 25-26:

Losses of striped bass have been estimated for power plants based on sampling within the cooling systems. Similar estimates of striped bass losses in CVP, SWP, or DA diversions are precluded by inadequate sampling. However, indirect estimates of these losses have been made by [various people]. These estimates were derived by multiplying estimates of striped bass egg and larva densities in the delta channels within the influence of the diversions by the amounts of water being diverted.

That said, "Except for the PGE power plant estimates, these various entrainment-loss estimates are only gross approximations" (p. 26).

Given that striped bass eat other animals, it's not obvious how their entrainment affects total animal suffering, although death by entrainment is presumably awful for the fish themselves.


  1. For example, nuBerlin (2015) says of Berlin's groundwater: "by seeping into the ground the water is already filtered by the different layers of sand and soil."

    Obermueller (2013) says that Berlin’s tap water "has, in effect, been doubly filtered: first, as it passes through different layers before settling in natural underground aquifiers; second, when this water is extracted by one of the BWB’s [i.e., the water utility's] nine water plants and purified in a three-step system."  (back)

  2. Note: I don't have access to the full text of this article and so am only quoting from its "Abstract".  (back)
  3. Why does this distinction matter? Here's an example. Imagine that we have a short pipe that contains 1 m3 of volume, and we have 1000 invertebrates in it, one of which gets carried away by the flowing water for every m3 of water that flows through the pipe. The static concentration of invertebrates is 1000 per m3, while the density of invertebrates flowing along in the water is only 1 per m3.

    The number of zooplankton killed by withdrawing surface water for water treatment is the latter kind of quantity—it's a number of invertebrates per unit of flowing water, not a static collection of invertebrates over which lots of water passes.  (back)

  4. Do water-treatment plants usually take in water from the surface or the middle or the bottom of a water body?  (back)
  5. The "Glossary" of Maupin et al. (2014) says (p. 50) regarding "public-supply water use": "Public suppliers provide water for a variety of uses, such as domestic, commercial, industrial, thermoelectric-power, and public water use." While my impression is that in many cases, thermoelectric-power water doesn't need to be treated(?), I assume that the portion of thermoelectric-power water that comes from public supply is treated(?), in which case that portion does kill zooplankton in the same way that other public-supply water uses do, and hence it should be counted.  (back)
  6. Technically, some people who aren't on public-supply water may be responsible for a portion of public-supply water withdrawals. For example, people in rural, non-public-supplied areas may buy products made by a company or sold in a store that uses public-supply water. However, for simplicity, I'm assuming that all public-supply water use is due to people who get public-supply tap water. Anyway, this distinction isn't very important because the fraction of the country that doesn't get public-supply water is so low.  (back)
  7. For comparison: "In England and Wales over 50% of drinking water is derived from surface water i.e. rivers and reservoirs. Surface water may contain small plants (algae) and animals, all of which are present in a healthy aquatic environment."

    van Lieverloo et al. (2002) report that "In the Netherlands, 65% of drinking water is produced from groundwater supplies" (p. 1729).

    Regarding water supply in Germany, this page says:

    The sources of public water supply are as follows:

    • 65% from groundwater
    • 9% from springs
    • 5% from bank filtration, i.e. from wells close to rivers and lakes, drawing essentially surface water
    • 20% from surface water[.]


  8. Because NYC is a tourist destination, perhaps there are more people visiting than New Yorkers leaving at any given time.  (back)
  9. This USGS page says: "At other homes, mainly is more rural areas, people provide water for themselves from sources such as a well, a cistern, a pond, or a stream. Largely groundwater wells provide water for these users, with almost 98% of water coming from wells."  (back)
  10. For global context, Muir (2012) says: "About 62 % of the water used in agriculture, globally, comes from surface sources (e.g., rivers) while about 38 % comes from ground water (underground aquifers)."  (back)
  11. "Green water" (i.e., rainwater used by plants) doesn't contain zooplankton. "Grey water" used to dilute pollutants might involve some zooplankton deaths from toxins in the water, but that issue is beyond the scope of this piece.  (back)
  12. Muir (2012): "You don't 'consume' water when you brush your teeth if you live in a town such as Corvallis[, Oregon] -- the water goes back to to the river via the sewage system -- or percolates back to ground water, in the case of septic systems."  (back)