Sea Salts, Part One: A review and a new study to determine their effects on reef aquarium inhabitants.
By: Eric Borneman
The first artificial sea salts designed to maintain marine ornamentals and concomitantly act as a surrogate for natural seawater were commercially developed in the 1960s. Over the many years since that first salt appeared in the aquarium trade, we now have over a dozen brands of artificial sea salt mixes from which to choose. Some of them, Instant Ocean®, for example, have been mainstays of the public and private aquarium sector for decades. Others, such as Oceanic®, are relative newcomers to the trade. Some are made in the USA; others are imported. Some are found in nearly every store that sells aquarium products, and others can be very difficult to locate, depending on the location.
Which brand of artificial salt mix is “the best” has been a longstanding debate among aquarium hobbyists, with advocates touting every brand. Opinions and reasons vary, and are not usually based on evidence. Unlike many other essential supplies or equipment, for which different tanks have different needs, a quality artificial salt mix is the very essence of marine aquaria to anyone who does not have access to, or utilize, clean natural seawater. It is the basis for the life sustained in our tanks and, at the very least, should provide a reasonable facsimile of natural seawater capable of fostering the sustenance, growth and reproduction of organisms living in it.
Older, opinionated views and early simple studies notwithstanding, the past few years have raised the artificial sea salt mix debate to new heights. A slew of “studies” have been produced and vehemently debated in terms of their results and their methods. This article focuses on a critical review of the recent studies and introduces a new long-term study that I am performing with the support of Kim Lowe and the members of the Marine and Reef Society of Houston (MARSH), sponsor of MACNA XVII in 2006. In terms of this study, it is important to recognize that at least some of the salts’ formulas have been modified over the years, and this may account for some of the variation between studies, although I am unaware of any manufacturer that has ever supplied independently verified information on its formulations or sources of ingredients, which would allow for an adequate discussion or explanation of any results of the studies listed below. Like so many other products in the aquarium trade, “proprietary secrets” seem to be the rule, although it would seem to me that a manufacturer that was sure of its salt’s quality would make this information public knowledge, and perhaps would even use it as a marketing tactic to attract customers. My view of all salts, therefore, is one of significant skepticism of all brands.
Review of Articles Concerning Artificial Salt Mixes
Bingman and Atkinson (1998, 1999)
Bingman and Atkinson analyzed various elements by spectrophotometry, ion chromatography and inductively coupled spectroscopy (ICP). They analyzed duplicate samples of salts at 35ppt salinity and 25oC, and averaged the values. Their results are summarized in tabular form here.
Their study tested eight artificial salt mixes, each represented by only one sample (N=1), with each sample then subsampled 10 times to obtain its average value. Because there is no replication, it is impossible to say if their data were applicable to the salt brands, in general, or if variation is significant within each brand.
Shimek (2001, 2002 a-e, 2003)
Perhaps no writings in the history of aquarium publications have produced more controversy than those of Shimek. It also could be said that perhaps no works were more misinterpreted. A look at the results of the numerous articles written since then, however, shows that many of the arguments made against Shimek’s articles can be discounted. This set of studies deserves careful analysis.
Shimek (2001) analyzed numerous foods and additives for their nutritional content and found that although some foods were virtually all water, others were highly concentrated and some also included high metal content. He viewed this as a matter of some concern and has repeated the idea that tanks are dynamic and that not all material that goes into an aquarium comes out again, even if export means are present. A year later, Shimek (2002a) had samples of aquarium water from 23 different systems and one water sample made with artificial sea salt (Instant Ocean) analyzed by a lab utilizing inductively coupled plasma emission spectroscopy and other EPA standard techniques, with each test run twice. The sample size for the treatments was, like in Bingman and Atkinson’s work, a limiting factor (N=1). It is important to consider that, despite the later controversies, Shimek was analyzing aquarium water samples subject to great variation in terms of imports and exports, as well as the use of potentially many types of artificial salt mixes. Only a single artificial salt mix sample was analyzed.
Later, (Shimek 2002b) reviewed and reanalyzed his work (Shimek 2002a) using more recent and accurate seawater values for metal content, and also attempted to determine correlations with numerous factors to explain his results. In the discussion, Shimek states, “There likely is no single cause for some of the effects, and it is just as likely that there is no defined cause at all for some of them. In these latter cases, the patterns would be the result of random factors or happenstance. The high metal concentrations may be due to build up in the tanks from foods, or from salt formulations, or from the ill-conceived use of poorly formulated additives. Other high concentrations, such as the fats and other metabolites, may be directly due to in-tank metabolism or be caused from food additions.” Despite this very clear statement, Shimek’s work was largely and wrongfully conceived in Internet forums as an assertion that artificial salt mixes were the cause of anomalously high metal levels.
Shimek (2002c) then assessed “food” inputs as a potential input of metals into aquarium water and concluded that food inputs could be a significant source of elements that could account for the high degree of variation in aquarium water samples compared to seawater. He also argued that many factors may account for between tank variation and the dynamics of aquarium water’s elemental composition, with the elements skewed from seawater having similarly variable but potentially negative effects, which were discussed in Shimek (2002d). Again, there is no implication that artificial salt mixes are the cause of high metal levels.
Continuing his investigations into the dynamics of tank water’s elemental composition beyond sample analyses and food/additive analyses, Shimek (2003) tested four artificial salt mixes and two aquarium water samples, using natural seawater that had been “filtered, (fiber, sand and charcoal) ozonated, protein skimmed and pH locked at 8.3 (Catalina Water Co.)” as a control, for their effects on sea urchin embryos, a common bioassay. Ten replicate tests were performed with each replicate containing 50-80 sea urchin embryos. Shimek’s results showed that the filtered seawater control had the highest survival rate, followed by two artificial salt mixes (Bio-Sea Marinemix® and Crystal Seas Bioassay formula®) and an aquarium water sample that allowed reasonable embryo survival, and then by poor embryo survival in two other artificial salt mixes (Coralife® and Instant Ocean®) and another aquarium water sample. Shimek suggested that the metal content in artificial salt mixes based on Bingman and Atkinson (1999) could explain the low survival in two of the salt mixes (Coralife® and Instant Ocean®), and questionably used manufacturer data for metal levels to potentially explain the higher survival in the other two artificial salt mixes (Bio-Sea Marinemix® and Crystal Seas Bioassay formula®) that had never been independently analyzed. It is also important to realize that although sea urchins are commonly utilized in bioassays, they do not indicate the test’s multiple potential effects on other organisms.
Harker (2003, 2004 a, b)
Harker (2003) offered a critique of Shimek (2002a, b) by indicating that the ICP analysis of seawater is not definitive, especially at the low levels present in seawater, and that other sampling and statistical errata may have occurred in Shimek’s works. He specifically argues that because the samples were not filtered, and because there was no handling control, contamination of the samples could account for the results. Harker (2004a) then used the very same ICP method he criticized in the earlier article (Harker 2003) to analyze an unknown number of seawater samples (presumably three, but not implicitly stated), and one sample from two of his own tanks, introducing sampling bias and ignoring randomness. Furthermore, Harker did not filter his water, did not control for handling, and allowed unfiltered seawater to stand for undetermined lengths of time in plastic water bottles before analysis, repeating the same potential flaws for which he criticized Shimek. The lack of methods provided, self-selection bias, publication bias, confirmation bias (existing even with no hypothesis to be tested), and the sample size of his work (N=1) offer little, but do add another set of data. Furthermore, the chemical analysis of seawater, including coral reef water, has already been performed in peer reviewed literature too voluminous to mention. Harker (2004b) is essentially an interpretation of his and Shimek’s results.
Holmes-Farley 2003
Holmes-Farley (2003) thoroughly reviewed the potential dynamics of metals in aquarium water, provides further interpretation of the results of Shimek (2002a, 2003), and also cites similarly elevated metal levels discovered by Sekha (unpublished). This work illustrates the highly dynamic nature of metals in saltwater solutions and the likelihood of each individual aquarium operating with potentially very different dynamics of import, export and speciation.
Sekha 2003
Sekha (2003) examined the potential toxicity of elevated levels of metals in aquarium water and provided data for copper and zinc levels in the skeletons of the same genet of four coral species from the wild, the Waikiki aquarium and two private aquaria. He concluded that the two elements, known toxins, might not have been toxic because of their variability, despite their high levels in some samples due to chemical speciation and bioavailability. However, there was no measure of what constituted toxicity and no data on their possible effects on the living corals. A cellular bioindicator, such as reproductive development, development of shock proteins or obvious visible changes in the corals’ normal physiological appearance and behavior, could help determine if toxicity were present. The fact that the corals were alive when sampled for skeletal analysis is not evidence that toxicity was not occurring.
Hovanec and Coshland (2004), Hovanec et al (2005)
Hovanec and Coshland (2004) expand on the analyses of artificial salt mixes by Shimek (2002a) and Atkinson and Bingman (1998, 1999). The authors concluded that their work was more valid than the previous works because they used a more sensitive method of inductively coupled plasma-mass spectrometry (ICP-MS) than the ICP emission spectrometry used in earlier works, and suggested that they had falsely inflated values due to interference. Furthermore, they claimed that the other studies were flawed because they did not utilize a seawater control. Unfortunately, these researchers utilized surf water near a heavily populated coast as their seawater control, and this cannot be considered a random or representative sample of seawater. They explain the high levels of lead and zinc found in their filtered seawater (Catalina Water Co.) by suggesting that fuel fumes in transit contaminated their sample, yet presumably the same company transport methods for Catalina water would have been used in the sample provided for the Shimek study. These anomalies may have been circumvented were this study not also compromised by its sample size (N=1). Furthermore, because one of the authors is the chief scientist of the company that makes several of the tested salts, numerous types of bias were introduced: self-selection bias, sample bias, publication bias and confirmation bias (despite the study’s lack of a hypothesis and its inability to be statistically analyzed with N=1). Finally, the results of their work also suggest that their sampled artificial salt mixes were not necessarily good approximations of seawater, with all samples having at least twice the levels of metals in seawater, and some much higher.
Hovanec et al (2004) then attempted to repeat the work of Shimek (2003). Their results differ in the mean survival of sea urchin embryos. While Shimek performed 10 replicates, Hovanec et al performed only four. Furthermore, they used Catalina water as a seawater control despite having found contamination in this water (lead and zinc) in their previous work (Hovanec and Coshman, 2003). They also found it to be toxic to sea urchin embryos in the present study in a second bioassay performed to analyze a new artificial salt mix (Oceanic®) not available at the time of the first bioassay. To compound this experimental problem, they used only two other artificial salt mixes, both produced by the company that employs the lead author. Thus, the smaller sample size (N=4), lack of a valid control, self-selection bias, untenable second bioassay, sample bias, publication bias and confirmation bias leave little value to this study.
Marulla and O’Toole (2005)
This recent article represents an oddity in that the work’s apparent objective is to re-evaluate Shimek (2003), yet of the nine salts tested, the authors opted not to test one of the four salts used by Shimek. The reference being questioned, however, is a bioassay utilizing sea urchin embryos, and this work is an analysis of artificial sea salt mixes’ elemental composition. In addition, they included several salts that are among the least available to aquarists. As no hypothesis was stated, I have assumed the work’s hypothesis to be the refutation of Shimek (2003) based on the presentation of results. Their methods states that “several” bags and “several” samples per bag were utilized in the study, but do not quantify the amounts used. The authors also used purified water as a negative control and natural seawater as a positive control, but do not discuss where the seawater was obtained, so it is impossible even to make an assumption about whether the positive or negative controls were representative of true control values; and the purified water contained unexplained high levels of zinc. The authors also failed to include the values of the negative control in their results.
As is the case in other studies, the sample size once again allows for no statistical analyses (N=1). The authors utilized an unidentified independent lab using ICP-MS. They also presented results where other tests were performed, but it is not clear what the authors were testing, what methods were used or what they found. Later in the methods, the authors state that they used a variety of reports and studies to determine natural seawater levels, yet no references are provided. It is unclear why a positive control was used in addition to the uncited seawater values, nor are the results of the positive control given in this article.
The authors measure the pH level of each salt solution by unknown means 24 hours after mixing, allowing for a potentially great change in the solutions due to their equilibration with environmental gases. The pH should have been measured immediately after complete dissolution, ideally without free gas exchange. In their presentation of results, the authors fail to adequately present their results except mainly as a comparison to the results of Shimek (2003). They also describe the elemental composition results as being “significantly” different from either Shimek’s (2003) values or from natural seawater, but because no statistical analyses are performed (or possible), there is no validity to the term “significant.” The authors present their ICP-MS results with no real logic, combining discussion with results and emphasizing the elemental values of some salts while not mentioning others with equal or higher values. Using aluminum as an example, high levels were found in five salt mixes, yet no mention is made of the salt mix with the highest value (Oceanic®) or a relatively similarly high value (Kent®), but they do emphasize the values that do not correlate well with Shimek (2003). This trend is continued in the analysis of all results. Other results are discussed without reference; for example, barium concentrations, despite being nearly a magnitude higher than stated seawater values, were described as being nontoxic despite the lack of any references to substantiate the statement, and strontium was described as essential for stony coral growth, yet no study is cited (or to my knowledge, even exists) that suggest that strontium is essential for the growth of stony corals and at least one: Wright, O. P. and A. T. Marshall. 1991. Calcium transport across the isolated oral epithelium of scleractinian corals. Coral Reefs. 10:37-40; indicates that it decreases calcium transport across the coral epithelia. For some potentially toxic metallic elements, all salts could be 2.5 to 5 orders of magnitude higher than seawater, yet the authors cursorily dismiss these results (e.g., lead), unless the results conflict with Shimek (2003). The authors also confirm Shimek (2003) with several elements, yet fail to acknowledge similarity of results (e.g., titanium). Finally, the authors even make the obvious mistake of capitalizing elements, obviously indicating their unfamiliarity with scientific nomenclature and procedure. The publication of articles such as this pseudo-scientific “study” is not only a waste of precious resources (> $15,000 by the authors’ own statements), but does a great disservice to a hobby attempting to gain insight into admittedly questionable data. Its many types of bias, lack of statistical significance and lack of hypothesis render this study meaningless except to provide another set of elemental data for artificial salt mixes.
This article will be continued in the next issue with the introduction of the MARSH salt study.
References
Atkinson MJ and C Bingman. 1998. Journal of Aquariculture and Aquatic Sciences. 8(2): 39-43.
Atkinson MJ and C Bingman. 1999. The composition of several synthetic seawater mixes. March, Aquarium Frontiers online.
Harker R. 2004b. Is it really in the water? A critical reexamination of toxic metals: Part 3. Advanced Aquarist online.
Harker R. 2004a. Is it really in the water? A critical reexamination of toxic metals: Part 2. Advanced Aquarist online.
Harker R. 2003. Is it Really in the Water? A Critical Reexamination of Toxic Metals in Reef Tanks. Advanced Aquarist online.
Holmes-Farley R. 2003. Reef aquaria with low soluble metals. Reefkeeping 2(3).
Hovanec TA, EL Toy, J Westerlund, JL Coshland. 2005. The Toxicity of Synthetic Sea Salts and Natural Seawater to the Development of White Sea Urchin (Lytichinus pictus) Larvae. Advanced Aquarist online.
Hovanec TA, JL Coshland. 2004. A chemical analysis of select trace elements in synthetic sea salts and natural seawater. Advanced Aquarist online.
Marulla M, T O’Toole. 2005. Inland Reef Aquaria Salt Study, Part I. Advanced Aquarist online.
Sekha H. 2003. Toxicity of trace elements: truth or myth. Advanced Aquarist 2(5).
Shimek, R. L. 2001. Necessary Nutrition, Foods and Supplements, A Preliminary Investigation. Aquarium Fish Magazine. 13: 42-53. Available online at: http://web.archive.org/web/20030419215821re_/www.animalnetwork.com/fish/data/foods.asp
Shimek, R.L. 2003. The toxicity of some freshly mixed artificial seawater; a bad beginning for a reef aquarium. Reefkeeping 2(2).
Shimek, RL. 2002a. It’s (in) the water. Reefkeeping 1(1).
Shimek RL. 2002b. It’s still in the water. Reefkeeping 1(2).
Shimek RL. 2002c. What we put in the water. Reefkeeping 1(3).
Shimek, RL. 2002d. Our coral reef aquaria – Our own personal experiments in the effects of trace element toxicity. Reefkeeping1(8).