«PREVIOUSLY PUBLISHED»

When we accepted “Developmental Stability: The Seeming Simplicity of the Methodology” by M. Kozlov for publication in our last issue, we hardly expected such repercussions.

In my profession, I have come across a few of cases or researchers using the method criticized by M. Kozlov. And I am sorry to say that their results made a very sorry impression on me. Their method seemed poorly investigated from a scientific point of view and lacking in methodological support.

On the other hand, in my twenty-five years in science I have also dealt with situations when a hasty and unskillful application compromised the most impeccable approaches and wonderful methods.

That is why the Bulletin considers it its duty to publish these two letters, the first from the authors of the method Kozlov criticized.

Dr. A. V. Scherbakov,
Deputy Editor

DEVELOPMENTAL STABILITY: ONCE AGAIN ON THE SEEMING
AND REAL SIMPLICITY OF THE METHODOLOGY

We were interested to learn of M. Kozlov’s the critical article “Developmental Stability: The Seeming Simplicity of the Methodology” (Bulletin No. 36) concerning our methodological manual “The Health of the Environment: Methodology of Assessment” (Zakharov, Baranov, etc., 2000). Criticism always promotes development. But, unfortunately, we did not find any constructive criticism of our approach in this article. Nevertheless, we believe it to be our duty to answer M. Kozlov’s claims since the Bulletin’s readership consists mainly of PNA specialists who have been developing a system of ecological monitoring and have used the developmental stability assessment method.

We proposed this method, as described in our manual, primarily to solve practical problems and for broad application in PNAs. That is why we gave a simplified version. However, the simplifications we made were absolutely correct.

Our “optimistic view of the universal correlation between unfavorable impact on an organism and the reduction of developmental stability which is shown in an increasingly fluctuating asymmetry” is based on our experience over many years in conducting ecological monitoring in different regions of Russia in populations subject both to chemical and radiation impacts. We used not only a morphological test, but also other tests that characterize the state of an organism: cytogenesic, immunological, biochemical and physiological approaches (Zakharov, Chubinishvili, etc., 2000). The results of our research showed that the indices change in correlation, and no wonder, since all of them reflect the state of an organism’s basic feature – that is, homeostasis.

Moreover, two thirds of the publications to which M. Kozlov refers confirm our “optimistic view” since the results of the researches described were also positive. Incidentally, M. Kozlov also arrives at positive results in his own work. The reasons for the “negative” results in the other third of the publications clearly vary. We’ll mention only those we consider most important.

The first reason is inaccurate initial data (mistakes made in taking measures or counting different morphological structures). Despite the seeming simplicity of our approach, it is very difficult to achieve simple initial results. If initial data were obtained inaccurately, no statistical processing would help. We know from our own experience that PNA staffers who do not have special training often fail to obtain reliable results. We could have written numerous articles casting doubt on the preciseness of the application of our method. Instead of doing that, we organize training workshops after which the “negative” results usually disappear. One more reason for “negative” results is that the researcher usually thinks he knows when there is stress, and considers the absence of changes of developmental stability as evidence of a weakness in the method. Sometimes it remains a mystery how the presence of stress was determined (not the impact, which can usually be determined by the results of chemical and physical analysis of the pollution, but the organism’s reciprocal reaction or feedback). On one hand, if the situation is clear, why do we need developmental stability analysis? On the other hand, if we do not see the answer we are searching for relating to developmental stability, what are the grounds for concluding that stress has occurred? Our approach is to obtain a simple answer concerning the presence of changes in the state of an organism in reaction to various impacts.

M. Kozlov is highly critical of the methods of statistical data processing. His recommendations are as follows:

1) check features for tendency to asymmetry and anti-symmetry;
2) do not summarize asymmetry using separate features;
3) use dispersion as an asymmetry index (not as an asymmetry average);
4) use Student’s t-criterion to compare samplings; use disperse analysis (ANOVA);
5) avoid pseudo-replications.

We would like to comment on these recommendations.

1. In developing the system of features to analyze developmental stability for all the objects recommended for study in our guidelines we did an analysis of features for tendency to asymmetry and to anti-symmetry. Our research proved that there is neither directed asymmetry nor directed anti-symmetry for such features. That is why there is no need to do an analysis of objects whose system of features has already been investigated.
   
   
2. It is not possible to characterize the developmental stability of an organism reliably by analyzing only one feature. When you work with only one feature, as M. Kozlov does in his research, the chances are great of making a mistake. To characterize an organism we require features sampling. The use of integral characteristics helps to soften differences between features’ sensitivity in different samplings. Since in the end we are interested in developmental stability not of an individual but of a population, we use an average value for population sampling.
   
   
3. Since the use of an integral index for a number of features is expedient for our objectives, it becomes clear why we recommend the use of an asymmetry average value, not the dispersion. In principle, when we take several features it is possible to calculate the generalized dispersion for a sampling. However, this approach is rather difficult, and we cannot recommend it for practical use (Zakharov et al., 1991). Dispersion as an indicator has another major disadvantage: great sensitivity to variants with significant deviations. One of the most important advantages of using an asymmetry average value is the possibility on the basis of the integral index of drawing a points-based scale to assess the degree of deviation of developmental stability from the standard. Such a scale allows one to present the results more visually. When we were only starting work on our methodology, we also used a dispersion to characterize the degree of asymmetry. In time, however, we stopped using this indicator in our work.
   
   
4. A comparison of average values can be done by using Student’s t-criterion as well as by using the ANOVA method, as M. Kozlov recommends. If one compares samplings, then both approaches are very similar and can be deduced one from the other. Mathematically, these two tests are equivalent (Sokal, Rolph, 1981). The main aim of dispersion analysis is to assess the influence of certain factors, which corresponds to the aims of M. Kozlov’s research, though it is better to use non-dimensional criteria. The aim of our manual is to ensure the collection of data in accordance with the general methodology with the use of asymmetry integral indicators and to estimate their deviations from conventional normal conditions. The use of Student’s t-criterion is adequate and correct in this situation.
   
   
5. As for pseudo-replications, here everything depends on the hypothesis we are to check. If we compare developmental stability in two points located at different distances from the source of the pollution, and having found it we state that developmental stability at these two points is different, then our statement is correct. S. Hurlbert talks about this in his article (Hurlbert, 1984), to which M. Kozlov refers. We followed this plan in our research. Such an approach is well founded, especially if the influence of a source of pollution often has a directed character (along the direction of dominating winds) and there is no point in searching for the impact effect in the opposite direction, as M. Kozlov recommends we do.

In addition to the above criticisms, M. Kozlov says that it is necessary to process the collected data with great accuracy. He recommends taking measures with 0.5 mm accuracy and measuring one and the same sample 2—3 times. This criticism refers only to the objects for which we used dimensional features, i.e. the plants or, to be more exact, the birch tree. It is possible to determine the required accuracy of measurements only empirically since it depends on the variability of the changing structure and measuring equipment. Taking into account measuring error, there is no sense in taking measures of a birch tree leaf with an accuracy of less than 1 mm. Heightened accuracy of measuring, in case it does not exceed the error limits, may lead to nothing really important – only to a certain distortion of the results. The first thing we do to avoid the influence of leaf size on the degree of asymmetry is to compare the leaves of one size class. Secondly, we do not use absolute differences between the left and the right sides of the leaf; instead we analyze the values that have been normalized in accordance with the size of the leaf. Moreover, we see to it that all the data on plants is processed by the same operator. This reduces the influence of measuring mistakes on the final results. We repeat that the question here is about practical guidelines, and it is unrealistic to take all measures 2—3 times if the volume of collected data is large.

We cannot help but agree with M. Kozlov’s advice “to select the place of data collection with particular attention”. As a rule, we analyze samplings from places not more than 20 km apart. It is doubtful that points located within such a small distance could be influenced by different climate conditions.

We do not understand M. Kozlov’s unsubstantiated statement that by using our methodology he can prove that any source of pollution produces no negative impact on the environment. On receiving data on developmental stability we deliberately and successfully searched for confirmations in the change of indicators of other organism condition assessment approaches. Moreover, the research conducted in nature reserves in 2000 shows that in the presence of impact the users of our methodology manage to assess changes in the state of organisms (Zakharov, Chubinishvili, 2001). On what grounds and to what purpose have all these data been called into question? We leave that to M. Kozlov’s conscience.

The statement that the articles that contain data processed in accordance with our methodology cannot be accepted for publication in international journals has nothing to do with reality. Articles by our Lab specialists containing data obtained with the help of our method have been published several times in international journals; two international journals (Acta Zoologica Fennica, 1992; Acta Theriologica, 1997) devoted whole issues to these results.

One should note that the asymmetry indicator that we recommend is not our invention. It was first used in R. Leary’s works and the works of his co-authors (Leary et al., 1983) and has been used around the world since then. In a review article by R. Palmer and K. Strobeck (Palmer, Strobeck, 1986), to which, by the way, M. Kozlov refers, we came across the names of at least 14 foreign researchers who make use of a similar indicator.

By providing a list of literature at the end of our methodological manual, our aim is not to show how well we know the foreign literature, as M. Kozlov suggests, but to refer the reader to the specialized information available. Moreover, the data presented in the listed articles was collected and processed using our method. We are grateful to M. Kozlov for his remark that “he cannot suspect the authors of not knowing the English-language literature.” We, in turn, cannot suspect him of not knowing the Russian language, in which we continue to publish. M. Kozlov could have found answers to most of his questions in our publications, beginning with V.M. Zakharov’s monograph (1987) and including a special edition of the journal “Ontogenesis” in 2001(v. 32, ¹ 6).

The style of M. Kozlov’s critical review is rather surprising. He repeatedly expresses regret over the backwardness of Russian science. As a representative of Finnish science, he feels justified in using the patronizing tone of an older colleague. For instance, he says: “I consider such simplification very dangerous: it could result in a number of new publications in Russian editions which the world scientific community would see as informational turmoil.” Or: “This method, of which Russian biologists are so fond, is no longer used in the West, where comparisons of FA sampling are done with a disperse analysis.” Or: “Unfortunately, the problem of pseudo-replications in ecological research, with which Western scientists have coped successfully, remains unknown to Russian ecologists.” In fact, Russian science is not in such a bad way. If there are no more serious arguments than “it’s good because that’s how they do it in the West” then perhaps that postulate will have to do; but we find it unconvincing. In our opinion, the use of any method is not about fashion or trying to join western scientists by doing as they do; it is a deliberate choice with respect to a given task.

To return to the matter of the method’s simplicity, we would say that of course this simplicity would be imaginary if someone, instead of doing serious research, were to pick up a few leaves, attach a ruler to them and, judging from the incidence of asymmetry, claim to have determined the state of the environment and its health. But this method is realistic given that its every stage is strictly adjusted and represents only the tip of an iceberg that rests on the substantiation provided by a large group of various scientists over the course of more than 20 years. Simplification of the method to a certain degree is indispensable if we want it to be widely used; as well, it is evidence of the maturity of the approach.

In conclusion, we would like to thank M. Kozlov, whose critical article prompted us to review our previously published materials and reconsider the problems that readers of our manual could have.

Literature

The authors of the practical manual for nature reserves
“The Health of the Environment: Methods of Assessment”.
Russia, 119991, GSP-1,
Moscow, 26 Vavilova Street, CEPR.
E-mail: ecopolicy@ecopolicy.ru

<< | contents | top | >>