Causes of the disappearance of the aquatic plant Egeria densa and black-necked swans in a Ramsar sanctuary : comment on Mulsow & Grandjean ( 2006 )

In an attempt to demonstrate the ultimate causes of the massive mortality and emigration of blacknecked swans at the Ramsar site Carlos Anwandter Nature Sanctuary, Mulsow & Grandjean (2006, available at: www.int-res.com/articles/esep/2006/E66.pdf) used field data, microcosm experiments and basic chemical theory, to postulate that calcium bicarbonate is removed from water, due to the increase in sulphate by the effluent of the pulp mill of CELCO, Valdivia, Chile. This process is deemed to affect the photosynthetic rate of the plant Egeria densa, the main food of the black-necked swans, which in turn provokes the known mortality and emigration of these birds from this site. We critically analyzed the study of Mulsow & Grandjean (2006), and we found a serious lack of consistent arguments that precludes any fundamental statement about the effects of sulfate and bicarbonate on E. densa. Here we discuss our concerns.

In an attempt to demonstrate the ultimate causes of the massive mortality and emigration of blacknecked swans at the Ramsar site Carlos Anwandter Nature Sanctuary, Mulsow & Grandjean (2006, available at: www.int-res.com/articles/esep/2006/E66.pdf)used field data, microcosm experiments and basic chemical theory, to postulate that calcium bicarbonate is removed from water, due to the increase in sulphate by the effluent of the pulp mill of CELCO, Valdivia, Chile.This process is deemed to affect the photosynthetic rate of the plant Egeria densa, the main food of the black-necked swans, which in turn provokes the known mortality and emigration of these birds from this site.We critically analyzed the study of Mulsow & Grandjean (2006), and we found a serious lack of consistent arguments that precludes any fundamental statement about the effects of sulfate and bicarbonate on E. densa.Here we discuss our concerns.

FIELD DATA
Mulsow & Grandjean (2006) described that 6 replicated samples were collected from 11 stations along the Cruces river; 4 and 7 stations were considered as affected and not affected by the effluents of CELCO, respectively.Concentrations of SO 4 2and HCO 3 -were measured at each station.Since these replicates were subsamples of the sites or stations (i.e.within nested design), the most appropriate analysis is a nested ANOVA.However, this configuration was not fol-lowed, and the ANOVA used by Mulsow & Grandjean (2006) results in pseudoreplication when using subsamples as replicates (Hurlbert 1984).This problem increases erroneously the degrees of freedom in the error source of variation and induces an increase in the probability of Type I Error (rejection of the null hypothesis being true).When averaged across subsamples at each station (Mulsow & Grandjean 2006, their  2and HCO 3 -can be correctly compared using ANOVA. After repeating the analyses presented by Mulsow & Grandjean (2006), we found (as they reported) a statistically significant difference in the concentration of HCO 3 -at unaffected sites with respect to those in areas affected by the effluents of CELCO (ANOVA, p = 0.001, F = 22.68; MS = 326.8;error df = 9).However, in contrast to results reported by Mulsow & Grandjean (2006), concentrations of SO 4 2in affected and unaffected sites were not statistically significantly different (ANOVA, p = 0.084, F = 3.78; MS = 4.65; error df = 9).Significant differences in SO 4 2and HCO 3 -(p-values < 0.05) are found exclusively by the effects of pseudoreplication.
These statistical problems show that the field data of Mulsow & Grandjean (2006)  and HCO 3 -.However, the authors present a confused description of their results and associated analyses; they report a determination coefficient (r 2 ), which implies the use of linear regression instead of correlation analysis.Using data from Mulsow & Grandjean (2006), we ran a correlation analysis which indicated a negative relationship between SO 4 2and HCO 3 -(r-Pearson = -0.687,p = 0.001), but the regression analysis showed that only 48.6% of the variation in HCO 3 may be explained by variation in SO 4 2 -(r 2 = 48.6,p = 0.017, n = 11; and r 2 = 47.2, p = 0.001, n = 18, including all the averaged subsamples in the analysis).Regrettably, the authors did not include any information on the number of observations used in the analysis and, of more importance, they did not describe the exact statistical procedure that was applied.Thus, any attempt to interpret these results is inappropriate.
We conclude that the incorrect analyses of Mulsow & Grandjean (2006) and their confusing report of their field evidence weakly support the hypothesis being tested in their laboratory experiment: the effect of SO 4 2on Egeria densa.

EXPERIMENTAL DESIGN
The laboratory experiment of Mulsow & Grandjean (2006) consisted of a 'bottle experiment' (Diamond 1986) where plants of Egeria densa were cultivated in replicated compartments and oxygen production was measured.After these trials, some plants remained as controls; the rest were randomly assigned to the treatments, consisting of several doses of SO 4 2 -(K 2 SO 4 ).The unsolvable problem here is that the amounts of SO 4 2used were several orders of magnitude higher than concentrations ever observed in the field (up to nearly 2200).There is nothing to conclude from such experiments regarding what occurs in the field, especially in an experiment designed to describe the 'phenomenon that must have occurred in the Sanctuary after the CELCO-ARAUCO pulp mill began operations' (Mulsow & Grandjean 2006, p. 9).To explain why O 2 -production experiments were performed with up to 9800 mg l -1 K 2 SO 4 , when the maximum SO 4 2concentration reported in the river was 4.52 mg l -1 , Mulsow & Grandjean (2006) argued that these high concentrations were equivalent to the SO 4 2concentration accumulated during 15 to 60 d.However, there is absolutely no guarantee that this was the case in the Cruces river, and no data are presented to support this.Also, Mulsow & Grandjean ( 2006) described these concentrations as 'equivalent doses', while this term represents a radioactive definition, not a concentration of a formal chemical compound.It is highly unlikely that any plant of Egeria densa has ever been exposed to such high concentrations of SO 4 2 -, and therefore all conclusions based on the results of these experiments are fairly unrealistic.
Mulsow & Grandjean ( 2006) tested for differences among treatments using ANCOVA, comparing the slope of the relationship between O 2 and time (continuous variables) in different experimental treatments (discrete variable).This analysis also suffers from serious uncertainties both in presentation and interpretation.For instance, Mulsow & Grandjean (2006) stated, 'the rates of photosynthetic production of oxygen… were not statistically different', and the reported p-value suggested the contrary (i.e.significant differences, p < 0.05).The authors also state, 'when the same plants were exposed to a dose of SO 4 … photosynthetic rates decreased… and the differences were statistically significant', but there is no indication of the analysis applied (ANCOVA as well?) and its statistics (F-and p-values).Confusion reaches a maximum when, in the last sentence of the abstract, the authors reported the use of an ANOVA and p < 0.05, while in the text they indicated the use of an ANCOVA.
In summary, the conditions of the laboratory experiment have a reduced power to conclude the causes of disappearance of Egeria densa in the Sanctuary, and the inconsistent statistical analyses and interpretations raise serious concerns about the experimental results.

PLANT PHYSIOLOGICAL CONSIDERATIONS
Mulsow & Grandjean (2006) used volumetric measurements of O 2 release to quantify the photosynthetic rate of Egeria densa.This technique is somewhat obsolete (actually used only for teaching purposes) because it is not reliable, does not include other important parameters and only provides a rough estimation of photosynthetic activity.Thus, it cannot be used for comparison between treatments and never in a peer-reviewed research.The most useful and up-to-date system is the Clark electrode for O 2 evolution, which, in the liquid phase, is used to measure simultaneously temperature, light and pH within the leaf chamber (Mommer et al. 2005).
More doubts about the experimental setup of Mulsow & Grandjean (2006) arise from their Fig.1: photosynthetic rates in pre-treatment plants showed large differences between the respective groups (from 0.5 to nearly 4.0 ml O 2 g -1 ).A continuous increase in photosynthetic rate versus time appears in Figs. 1 &  by the pulp mill of CELCO, that resulted in precipitation of calcium bicarbonate, causing a HCO 3 -deficit which induced a decrease in photosynthesis in E. densa.These authors found low photosynthesis, with an increase in photorespiration and resultant loss of turgidity and a change in colour (plants turning brownish).It is well known in plant physiology, however, that even when photosynthesis is equal to zero (with an increase in photorespiration or respiration), most plants usually remain alive (Larcher 1995).Moreover, aquatic species are able to use the CO 2 released from photorespiration or respiration (Leegood 2002).Considering the large concentrations of SO 4 2used in the experiments, we could easily conclude that plants were probably subjected to osmotic stress (high sulphate salt concentration in the solution) or to the toxic effect of sulphate salt, which could induce an imbalance in the nutrient solution (or in the membrane transport process).

CHEMICAL MECHANISMS
The title of the paper by Mulsow & Grandjean (2006) is misleading in stating, 'Incompatibility of sulphate compounds and soluble bicarbonate…': SO 4 2is compatible with HCO 3 -at the pH of the Cruces river (ca.6.0 to 7.4) (UACh 2005).The sentence 'Calcium bicarbonate in water can be easily disassociated by a stronger anion such as SO 4 2 -.' (Mulsow & Grandjean 2006, p. 6) is not correct because the authors refer to the following reaction: with an equilibrium constant of 4 × 10 -9 (Harris 1999).
In other words, the former reaction does not occur as written, because it goes back to the left (to reactant).Furthermore, the same authors declared that 'After losing the ionic equilibrium, addition of SO 4 2resulted in the loss of (HCO 3 ) -according to…' (see second equation of Mulsow & Grandjean 2006, p. 10), which is wrong because, as indicated above, there is no such loss.These authors also indicated that the addition of SO 4 2results in the loss of HCO 3 -and then present the equation with H 2 SO 4 instead of SO 4 2 -.Finally, they write, 'Thus, a strong negative correlation results between SO 4 and (HCO 3 ) -as demonstrated here.At near neutral pH, this relationship can be summarized as…:' (see third equation of Mulsow & Grandjean 2006. p. 10).This equation is incorrectly presented, since concentrations must be written in individual form (i.e.2[Ca 2+ + Mg 2+ ] must be 2[Ca 2+ ] + 2[Mg 2+ ]).It seems to us, that with this equation the authors are saying that when SO 4 2increases the concentration of HCO 3 decreases.However, it is not a traditional algebraic equation, but a charge balance.Since all solutions are electrically neutral, the addition of an extra amount of SO 4 2also introduces the same amount of positive charge (i.e.cations; in the study by Mulsow & Grandjean [2006] these were K + ions from their use of K 2 SO 4 ).The corresponding equation of the new charge balance of that dissolution is as follows: ].This shows that the concentration of HCO 3 -is not affected by the presence of SO 4 2 -.The only factors that could affect the equilibrium of HCO 3 -in the Cruces river would be water temperature and, especially, pH.However, pH values in the river fluctuate between ca.6.0 and ca.7.4 (see UACh 2005).This is far lower than the value needed to change HCO 3 -into CO 3 -(i.e.nearly 10 to 11).

CONCLUSIONS
Using data from Mulsow & Grandjean (2006) we have demonstrated here that there is not a significant spatial variation in sulphate concentrations across the Cruces river.Furthermore, in our comment on the chemistry we have shown that sulphate and bicarbonate are compatible and not negatively related as proposed by Mulsow & Grandjean (2006).Both observations allow us to conclude that the unrealistic and reduced repeatability of their experimental design performed in order to test the effect of SO 4 2on photosynthesis in Egeria densa has no empirical or theoretical basis.Thus, the question about the causes of the disappearance of this plant cannot be related to the interaction between sulphate and bicarbonate.The social, political and scientific confrontations derived from the Cruces river ecological disaster require precise and objective analyses to discuss its ultimate causes.Consequently, our best efforts are focused upon setting up a constructive dialogue among scientists, using scientific language and replicable methods.Also, as scientists, we are committed to sharing our results based on solid evidence with the public and the government, independently of which interests they support or threaten.This commitment includes the criticism of peer reviewed research, such as this paper.

Disclosure:
The authors have no personal, financial and/or other conflict of interest with the pulp mill of CELCO, other timber industries or governmental agencies linked to the study area.Our main motivation is to use scientific methods to produce high-quality information to understand the causes of environmental degradation occurring in the Cruces river wetland ecosystem.