The occurrence, causes, and consequences of estrogens in the aquatic environment


  • See Table S1 for the number of citations and rank of all the “Top 100” papers, which in this essay are references [Harries et al., Jobling et al., Allen et al., Routledge et al., 3–7, Huang et al., Lange et al., Spengler et al., Petrovic et al., Metcalfe et al., Gray et al., Arcand-Hoy et al.,9–15, Jurgens et al., 19, Arukwe et al., 20].


Purdom et al. 1 were the first to identify the problem that effluents of all sewage treatment plants (STP) investigated in the United Kingdom were estrogenic. They demonstrated this by placing caged fish in STP effluents and showing that, after a brief period of time, these fish had extremely high plasma vitellogenin concentrations. They correctly deduced that one or more estrogenic chemicals must have been present in the effluents. Although Purdom et al. 1 made their discovery in the late 1980 s, their results were not published until 1994.

If STP effluents induced estrogenic effects in caged fish, it seemed possible, even likely, that wild fish living downstream of STPs might also be affected. Jobling et al. 2 showed that this was indeed the case. In a very comprehensive, nationwide survey of a common species of freshwater fish, they reported that, indeed, wild fish living downstream of STPs were feminized to various degrees. Evidence of feminization ranged from inappropriate induction of vitellogenin synthesis to the presence of oocytes in the testes of male fish. Similar results have been reported from many countries in subsequent years. Two key questions arose from these findings: (1) What was the identity of the chemicals causing the feminization? (2)What were the consequences of being feminized to wild fish?


Purdom et al. 1 showed that effluents were estrogenic, but they did not investigate whether or not receiving waters (usually rivers) became estrogenic as a result of receiving the effluents. That step was achieved by Harries et al. 3, 4, who quite remarkably demonstrated that a significant length of one river in northern England (the River Aire) was strongly estrogenic. That particular river was known to contain very high concentrations of nonylphenol (NP) and related chemicals, which are degradation products of one group of nonionic surfactants. Nonylphenol had been shown to be a weak estrogen in mammals in the 1930 s. It also proved to be estrogenic to fish and to adversely affect the testes of male fish 5. Thus, NP soon became a prime suspect to account for the estrogenic activity of effluents. However, a completely different group of chemicals, the steroid estrogens, were also candidates because they were known to be potent in fish. In particular, Purdom et al. 1 had shown that one of the steroid estrogens, the pharmaceutical ethinyl estradiol (EE2), was extraordinarily active in fish: very low to sub-ng/L concentrations induced vitellogenin synthesis in a concentration-dependent manner. Perhaps of even greater surprise, some estuaries proved to be estrogenic 6, although the causative chemicals were not identified at the time and have not been since.

The only way to identify the main estrogenic chemical(s) in STP effluent, other than by informed guessing (i.e., systematic hypothesis testing), was to take an unbiased approach, of which the most appropriate seemed to be to conduct a toxicity identification and evaluation (TIE) exercise. To do a TIE, it is necessary to have a test available to detect the activity of interest. The test ideally needs to be able to cope with large numbers of fractions (samples) to provide results rapidly. No such test was available until a pharmaceutical company, GlaxoSmithKline (GSK), very kindly gave one of us (J.P. Sumpter) a strain of genetically modified yeast. This strain was used to develop the so–called yeast estrogen screen (YES), which proved to be an extremely efficient, effective, and useful tool 7. With this tool and access to only the second liquid chromatography–tandem mass spectrometry to become available in the United Kingdom, Desbrow et al. 8 identified steroid estrogens to be the main contributors to the estrogenic activity of STP effluents. Concentrations were extremely low, and close to or even below the detection limits available at the time, but nevertheless they were high enough to cause effects. Subsequent studies in other countries confirmed these results (e.g., Huang and Sedlak 9).

Despite the identification of steroid estrogens (both natural and synthetic) as major contributors to the overall estrogenic activity of STP effluents, it is still not clear which estrogenic chemicals contribute what proportion of that overall activity and if, as seems likely, those proportions vary from one effluent to another. Besides the steroid estrogens, some of which (especially EE2) induce effects at extremely low environmental concentrations 10, a range of xenobiotics known to be present in effluents 11, 12, including NP and bisphenol A, are estrogenic to fish 13. Furthermore, some of these xenoestrogens can feminize male fish, producing effects (e.g., ovotestis) that are reminiscent of those found in wild fish living downstream of STPs 2, 14. These laboratory studies, in which fish were exposed to (often) environmentally relevant concentrations of different estrogenic chemicals (e.g., Metcalfe et al. 13), demonstrated the environmental relevance of fish reproduction tests 15. It would be difficult to think of a more environmentally relevant parameter than reproduction when attempting to extrapolate from effects observed in laboratory studies to population-level effects on wild fish.


The impacts of the research summarized extremely briefly above were very much greater than the authors of this paper anticipated at the time, and we remain very surprised at the continuing impact of the research. Of course, the fact that the research dealt with a “sexy” topic (the feminization of male fish by chemicals in the environment) increased interest in the research not only from scientists but also from the media. Articles about the research in newspapers, in magazines, and on the radio and television took the research to a very broad audience. An accidental, but fortuitous, link with human health also significantly raised the profile of the research. In the early 1990s, a very vigorous debate took place about whether sperm counts in men were declining. The proponents of this argument suggested that estrogens were involved in falling sperm counts and disorders of the male reproductive tract 16. Thus, at the time, estrogenic chemicals were being implicated in feminization of everything from fish to men.

In addition to the high scientific impact of the research, there was substantial regulatory interest in the work; in fact, many of the key studies conducted in the United Kingdom 1, 4, 8 were supported and funded by regulatory and policy organizations rather than organizations that supported basic scientific research. If estrogens in the aquatic environment were adversely affecting fish health, then regulatory organizations had a legal obligation to address and manage the issue. But a variety of issues confused the picture and very substantially delayed any regulatory decisions (as they still do). Because it was unclear which chemical, or mixture of chemicals, was responsible for the feminization of fish, it was not possible to target one specific chemical for regulatory action. Instead, any regulation would need to be based on the total estrogenic activity (i.e., on estradiol equivalents). This caused major complications, such as the requirement to measure total estrogenic activity rather than quantify concentrations of individual estrogenic chemicals and frame regulations around one (or more) of those chemicals. There was also much discussion about whether the effects were adverse to the ability of the fish to reproduce and whether this was relevant as far as regulatory activity was concerned. Perhaps male fish with abnormally high plasma vitellogenin concentrations (as a consequence of exposure to estrogens) could still perform perfectly normally as males; that is, they could reproduce. Approximately 15 years after these issues were hotly debated in the United Kingdom Environment Agency, Harris et al. 17 showed that although intersex fish could breed, the intersex condition reduced reproductive performance by up to 76% for the most feminized individuals, demonstrating a significant adverse effect of intersexuality on reproductive performance. The weight of evidence from both laboratory and field studies now seemed to point to potential threats of fisheries from environmentally relevant concentrations of estrogenic chemicals, albeit the latter has not been proven.

Despite the lack of any current formal regulation, water companies operating STPs (and hence putting the estrogenic chemicals into the rivers) came under considerable pressure to clean up their act—in other words, improve the treatment of wastewater so that effluents contained lower concentrations of estrogenic chemicals. Of course, that pressure (which is ongoing) was not only due to the estrogens present in effluents. Nevertheless, the issue of estrogens in the aquatic environment undoubtedly highlighted the problem of inadequate treatment of wastewater and the pollution of rivers by chemicals present in the effluents that were discharged. In general, the authors' experience was that water companies (which are private companies in the United Kingdom) were reluctant to engage with the science and scientists and, at least initially, were hostile to it. Attitudes in the water industry have changed with time, and the industry recently has spent very considerable sums of money (tens of millions of pounds in the United Kingdom alone) investigating the abilities of various existing and new technologies to improve the treatment of wastewater to better remove estrogenic chemicals. Furthermore, some water companies (in both the United Kingdom and the United States, for example) have very actively engaged with academic scientists to better understand the problem and find solutions to it.

To date, NP is the only estrogenic chemical to be regulated via legislation. The European Union took the lead and a few years ago passed legislation that severely restricted the use of nonionic surfactants based on NP. Although this action was primarily based on the acute toxicity of NP to many aquatic organisms (including fish), the estrogenic effects of NP helped strengthen the case for the legislation. Very recently, in 2012, the U.S. Environmental Protection Agency (U.S. EPA) has recommended alternatives to NP-based surfactants 18.

Of particular interest currently is the situation with EE2. This is a human pharmaceutical and one of the two active ingredients of most hormonal contraceptives; hence, it is probably not a chemical that society would want to ban. Tens of millions of women currently choose the contraceptive “pill” as their preferred method of controlling their fertility. However, once in the aquatic environment, EE2 is extremely potent, causing effects on fish at sub-ng/L concentrations and preventing them from reproducing at concentrations of only a few ng/L 10. The European Union has very recently placed EE2 on a draft list of “priority pollutants,” chemicals thought to adversely affect the environment and/or human health, and hence chemicals whose concentrations in the environment ideally should be as low as possible. If EE2 remains on the new list of priority substances once this is finalized, then national monitoring programs would have to be initiated to determine the concentrations of EE2 in the environment and to assess their potential effects. Doing so would create many challenges, not the least of which would be the accurate and reliable measurement of pg/L concentrations of EE2. An environment quality standard of around 0.02 ng/L is possible.

Finally, one interesting and unexpected scientific impact occurred. The issues involved in addressing the problem of estrogens in the environment necessitated scientists from different disciplines working together. Biologists needed to work with chemists, and both of these groups had to work with the regulators if their results were ever to help improve the condition of the aquatic environment. The interdisciplinary nature of the research created challenges but was also very intellectually rewarding.


As always in science, research seems to pose more questions than it answers. It is possible to delve deeper and deeper into any issue, and doing so is very tempting. However, wise scientists try their best to answer the question “Do we need to know more?” before continuing their research program on any particular topic. There are undoubtedly very many unanswered questions relating to the issue of estrogens in the environment. For example, very little is known about the importance of degradation (by whatever means) in controlling the concentrations of estrogenic chemicals in the aquatic environment 19; hence, predicting environmental concentrations of the chemicals of concern is still difficult. Similarly, many, and perhaps most, estrogenic chemical have a range of effects on aquatic organisms. Knowing what these are and ranking them in importance (i.e., which is the most adverse mechanism-of-action) is currently very difficult. It was 15 years ago that NP was shown to affect biotransformation enzymes (such as the P450 enzymes) in fish 20, which could be an important finding because it could affect the metabolism of many chemicals. However, the possible consequences of those findings are still unexplored. In our opinion, most, and perhaps all, of the important questions relating to estrogens in the environment have been addressed and answered to a satisfactory degree. This implies that the field probably now receives too much attention, and too may resources, for the amount of new, useful knowledge being discovered.


It could be argued (and was, and still is by some) that estrogens in the environment was not, and currently is not, an environmental disaster. Nevertheless, a great deal has been learned from 20 years of research on this issue. Perhaps the most important contribution of all the research was that it added greatly to the topic now known as endocrine disruption. We have learned that many chemicals have endocrine activities of various sorts and that these activities might be important in the toxicities of these chemicals. Very many endocrine active chemicals are present in the environment, in biota, and in us. We do not know the consequences of that situation. How do we assess the effects of these very complex mixtures? It is relatively obvious that the toxicity of a mixture of similarly-acting chemicals (e.g., five estrogenic chemicals) can be predicted based on additivity, but what about a mixture of dissimilarly-acting chemicals (e.g., two estrogens, an anti-estrogen, and two anti-androgens)? It is not at all obvious how to predict or test the toxicity of this mixture, which is much more representative of the “real world” than one estrogenic chemical, or a mixture of a few estrogenic chemicals, tested in a laboratory

It should also be kept in mind that almost all the ecotoxicological research on estrogenic chemicals has focused on assessing responses in fish. Yet many other groups of animals live in the aquatic environment. They may merit more study than they have received to date

The development of the YES (and yeast androgen screen, or YAS) proved to be very useful not only for our own research but also for the research of many other scientists 7. The YES has been given, free of charge and with no conditions, to more than 200 laboratories worldwide. Those labs have used it successfully in a remarkable range of studies, to the considerable benefit of environmental science.

Finally, a word of caution—never dismiss a chemical because it is present in the environment at concentrations “too low to cause effects.” Concentrations of EE2 below 1 ng/L cause effects on fish. Perhaps other chemicals present in the aquatic environment might be active and cause adverse effects at exquisitely low concentrations, whereas still other chemicals may be present at comparatively high concentrations yet be of little, if any, toxicological concern. Judging which chemicals are of concern should be based on sound science, not on whether concentrations are considered high or low. Both potency and concentration define risk.


Table S1 (49 KB PDF).