Документ взят из кэша поисковой машины. Адрес оригинального документа : http://ecology.genebee.msu.ru/3_SOTR/CV_Terekhin_publ/1997_Scientometrics.doc Дата изменения: Mon Mar 16 10:56:41 2009 Дата индексирования: Mon Oct 1 21:34:07 2012 Кодировка:

Jointly published by Elsevier Science Ltd, Oxford Scientometrics,
and Akademiai Kiad6, Budapest Vol. 39, No. 2 (1997) 147-157

PRINCIPAL TRENDS IN MODERN ECOLOGY AND ITS MATHEMATICAL TOOLS: AN
ANALYSIS OF PUBLICATIONS*

E. V. BUDILOVA, J. A. DROGALINA, A. T. TERIOK.HIN

Department of Biology, Moscow State University, Moscow 119899
(Russia) E-mail: lenl@ATeriokhin.home.bio.msu.ru

(Received March 12, 1997)

The paper deals with a scientometric analysis of publications from
the journals "Ecology" and "Ecologia" (Russia) based on the
frequencies of individual and cojoint encountering of ecological and
mathematical keywords in these publications. Two main research
approaches are revealed: population ecology and system ecology. The
first one is used primarily in studies of plant communities, while the
other in terrestrial animals and birds. Water communities are the
subject of both approaches. The most spread mathematical methods are
the methods of mathematical statistics which can be clustered into
four groups: standard ones, multivariate methods, in particular
multiple regression and multivariate analysis of variance,
nonparametric or allowing deviations from normality, and methods of
analysis of categorical data. Differential equations and stochastic
process are used much lesser. The intensities of using mathematical
methods are notably different in two journals.

1. Introduction

There is a variety of judgements concerning the situation in modern
ecology: some scholars consider it as intensively evolving, while other -
as enduring a conceptual deadlock. That very fact inspired an attempt of
unprejudiced analysis of publications in ecological journals using formal
quantitative methods. We started from the conception that publications are
the principal components of the scientific information process (Nalimov,
1969). More specifically, we used the data on frequencies of encountering
of keywords and their pairs in papers published in the journals "Ecology"
and "Ecologia (Russia)" in 1991-1992. These data were analyzed using the
methodology we applied earlier in (Budilova, 1992; Nalimov, 1992). It is
based on the use of multidimensional

This paper (originally published in Russian in The Journal of General
Biology) was inspired, like many others, by Professor V. V. Nalimov, and he
contributed much in fruitful discussion and advices.

0138-9130/97/US S 15.00
Copyrightї 1997 Akademiai Kiado, Budapest
AH rights reserved
Е. V. BUD1LOVA et al.: TRENDS IN MODERN ECOLOGY

scaling methods and computer data bases, the approach widely used in many
scientometrics works (e.g., Tijssen, \981;Leydesdorff, 1994).
The analysis was aimed mainly at forming a general idea of basic research
trends in ecology. For that, a method of multivariate scaling of data
consisting of cojoint frequencies of different keywords was applied,
resulted in the conclusion that ecological researches can be naturally
clustered according both to their subject and basic methodological
approach, either populational or ecosystemic. It is turned out in addition
that the both clusterings are interrelated. These results well agree with
the similar conclusions made by some other authors (Cherrett, 1988;
Hoecstra, 1991; Ghilyarov, 1992).
Specifically, in the research earned out by the British Ecological
Society in connection with its 75-th anniversary (Cherrett, 1988), the data
of ranging of 50 ecological terms by 645 experts were analysed using a
factor analysis method. The main factor by which the keywords were
differentiated by experts was the opposition between "theoretical
reductionism" and "practical holism" that is similar to our opposition of
populational and ecosystemic approaches. In the work of Hoekstra (Hoekstra,
1991) the analysis of data from the "Biosis Previews" computer database
allowed to reveal that some of the keywords related to large biological
taxons (mammals, birds, fishes, reptiles, rosaceae et al.) are more closely
correlated with some of the keywords refered to theoretical ecological
concepts (ecosystem, competition, niche, population, evolution et al.). Our
interpretation of the obtained results, in according to which the
difference between the populational and ecosystemic approaches is the most
essential, is close to the similar conclusion by Ghilyarov (Ghilyarov,
1992).
Additionally, an analysis was performed related to the mathematical
methods used in the same publications. It allowed to expose the main
clusters of those methods and evaluate comparative frequencies of their
application. No significant difference in the intensity of usage of
different groups of methods with respect to ecological content of the
papers was revealed. Yet it was observed an essential difference in the
overall intensities of applying mathematical methods in "Ecology" and
"Ecologia" this intensity being higher in "Ecology".

2. Analysis of ecological keywords lists

This part of analysis is based on the lists of keywords from 412 papers
published in "Ecology" in 1991-1992 which were obtained using the computer
database Science Citation Index (SCI). These data were used for forming a
single common list of

148 Scientometrics 39 (1997)

Е. V. BUD1LOVA et al.: TRENDS IN MODERN ECOLOGY

keywords and computing the frequency of each of them. The derived amount of
the keywords from 412 papers counted 6775 with 3613 different ones that is
53% of the total number that testifies to the lack of commonly adopted
keywords thesaurus. About 70 most frequent keywords (encountered more than
10 times) constitute only 20% of the total amount that also confirms the
conclusion on non-established character of the ecological keywords
thesaurus. The list of keywords includes, in addition to properly
ecological terms, also names of taxons, parameters and processes under
investigation, methods used, geographical names.
In Table 1 the list of 69 keywords encountered not less than 10 times is
presented. The most frequent ecological terms are the following ones (the
number in brackets is the frequency of the keyword): "competition" (81),
"growth" (61), "predation" (56), "dynamics" (53). These four terms
(together with two associated ones - "population dynamics" and
"interspecific competition") are the most specific for modern ecology: they
appear in 200 of 412 papers.
The most frequent subjects of research are given by the following
keywords: "herbivory" (50), "plants" (34), "nitrogen" (29), "forest" (23),
"birds" (22). The most frequently investigated parameters are "size" (27),
"clutch size" (16), "body size" (15). Within the geographical names the
most frequent are "California" (18) and "Costa Rica" (11). Research and
data analysis methods are presented in Table 1 by only two keywords -
"field experiment" and "model".
In addition to computing the frequencies of individual keywords, the
frequencies of cojoint encountering of pairs compiled of 69 most frequent
keywords were also computed. These data allowed to present, using a method
of multidimensional scaling (e.g., Computer Biometrics, 1990), all the 69
keywords on the plane (Fig. 1) in accordance with their relative
similarities measured as the frequencies of their cojoint occurrences in
the same papers.
The analysis of Fig. 1 permits to mark out four groups of keywords
varying in according to the type of a community which they incline to: A -
terrestrial animals; В -birds; С - water communities; D - plants. But in
the same time each of these four groups is characterized also by terms
which could be named "methodological" and which can be devided into two
classes that could be considered as associated either with "population
dynamics" or "ecosystems".
Groups A (animals) and В (birds) are inclined to the population dynamics,
while group D (plants) to the ecosystems. Group С (water communities) is
closed to both methodological approaches. The fact that different
theoretical conceptions are, as a rule, closely associated with a specific
type of biological community was already

Scientometrics 39 (1997) 149

Е. V. BUDILOVA et al.: TRENDS IN MODERN ECOLOGY

revealed by Hoekstra who affirmed that "various ecological concepts are
associated with their own specific types of organisms" (Hoekstra, 1991).

Table 1
Frequencies of 69 most used ecological keywords from 412 papers
published in "Ecology"
in 1991-1992

|Frequenc|Keyword |Frequency |Keyword |
|y | | | |
|81 |competition |16 |selection |
|61 |growth |16 |fish |
|56 |predation |16 |clutch size |
|53 |dynamics |16 |behavior |
|50 |herbivory |15 |coexistence |
|49 |communities |15 |body size |
|34 |plants |15 |algae |
|32 |patterns |14 |streams |
|31 |populations |14 |reproductive |
| | | |success |
|30 |population |14 |daphnia |
| |dynamics | | |
|29 |nitrogen |13 |stability |
|29 |community |13 |responses |
| |structure | | |
|27 |size |13 |lepidoptera |
|27 |field experiment |13 |habitat use |
|27 |disturbance |13 |habitat |
| | | |selection |
|26 |evolution |13 |consequences |
|26 |ecology |12 |recruitment |
|25 |reproduction |11 |trees |
|24 |succession |11 |prey |
|24 |ecosystems |11 |nutrients |
|23 |forest |11 |insects |
|23 |dispersal |11 |grassland |
|22 |birds |11 |foraging |
|21 |model |11 |fire |
|20 |life history |11 |density |
|20 |food webs |11 |Costa Rica |
|19 |photosynthesis |11 |allocation |
|19 |diversity |10 |temperature |
|19 |demography |10 |mortality |
|18 |interspecific |10 |food |
| |competition | | |
|18 |density dependence|10 |defoliation |
|18 |California |10 |chemical |
| | | |defenses |
|17 |vegetation |10 |biomass |
|16 |zooplankton |10 |biological |
| | | |control |
|16 |survival | | |
150
Scieniometrics 39 (1997)
Е. V. BUDILOVA et al.: TRENDS IN MODERN ECOLOGY


Fig. 1. Multidimensional scaling of the ecological keywords cojoint
encountering data ("Ecology", 1991-1992)

As to the frequencies of keywords of different groups, the most frequent
are the keywords of group A (76% of papers) and D (61%). Keywords of group
С are encountered in 17% and В in 11% of papers.



3. Analysis of mathematical keywords lists

An analysis similar to the above was made for the lists of mathematical
terms extracted from 182 papers published in "Ecology" in 1991 and 139
papers published in "Ecologia" in 199.1-1992. These lists were obtained by
a direct inspection of the papers content because the section "Keywords" of
the "Science Citation Index" database does not include factually
mathematical terms for "Ecology" and does not contain any information about
"Ecologia" at all. All the terms related to mathematical modelling or data
processing were classified as mathematical keywords. The obtained lists
were used to compute both the frequencies of all individual keywords and
the frequencies of pairs of most frequent of them.
The total amount of derived keywords for "Ecology" counted 1774 with 349
(20%) different ones. The most frequently encountered (5 times and more) 63
keywords


Scientometrics 39 (1997)
151
Е. V. BUDILOVA et al.: TRENDS IN MODERN ECOLOGY

constitute 79% of the total amount. They are presented in Table 2. We see
that the most frequent statistical characteristics are the following ones:
"mean" (138), "standard deviation" (66), "standard error" (62),
"probability level" (49). The most exercised methods of data transformation
are "log transformation" (46), "arcsine square root transformation" (15),
and "arcsine transformation" (14). The most used statistical methods are
"ANOVA" (107), "regression" (73), "correlation" (69), "t-test" (44), "F-
test" (33), "ANCOVA" (26). The most used packages of statistical programs
were "SAS" (42) and "SYSTAT" (11).

Table 2 Frequencies of 69 most used mathematical
keywords from 412 papers published in "Ecology" in 1991

|Frequen|Keyword |Frequency|Keyword |
|cy | | | |
|138 |mean |11 |SYSTAT |
|107 |ANOVA |11 |multiple |
| | | |regression |
|73 |regression |11 |Mann- Whitney U |
| | | |test |
|69 |correlation |11 |linear |
| | | |regression |
|66 |standard deviation|10 |square root |
| | | |transformation |
|62 |standard error |10 |MANOVA |
|49 |probability level |10 |factorial design|
|46 |log transformation|10 |contingency |
| | | |table |
|44 |t test |9 |quadratic |
| | | |regression |
|42 |SAS |9 |nonparametric |
| | | |test |
|41 |one factor |9 |median |
|39 |two factors |8 |nonlinear |
| | | |regression |
|33 |Ftest |8 |Kolmogorov-Smirn|
| | | |ov test |
|30 |histogram |8 |Bonferroni |
| | | |method |
|30 |chi-square test |7 |type iii sum of |
| | | |squares |
|29 |simple regression |7 |Student-Newman-K|
| | | |euls test |
|26 |ANCOVA |7 |sign test |
|24 |GLM |7 |paired |
| | | |comparisons |
|19 |three factors |7 |LSD |
|19 |repeated measures |6 |stepwise |
| | | |regression |
|17 |Wilcoxon test |6 |normality test |
|17 |indexes |6 |mixed model |
|16 |Spearman rank |6 |log-linear |
| |correlation | |analysis |
|16 |experimental |6 |Bartlett test |
| |design | | |
|15 |Tukey test |5 |transition |
| | | |matrix |
|15 |Kruskal-Wallis |5 |one-tailed test |
| |test | | |
|15 |arcsine square |5 |nested design |
| |root transform | | |
|14 |arcsine |5 |log-likelihood |
| |transformation | |ratio test |
|13 |normal |5 |Dunnett test |
| |distribution | | |
|13 |Gtest |5 |Duncan multiple |
| | | |range test |
|12 |In transformation |5 |differential |
| | | |equations |
|11 |two-tailed test | | |
152
Scientometrics 39 (1997)
Е. V. BUDILOVA et al.: TRENDS IN MODERN ECOLOGY

It is to be noted the variety of regression methods (with prevailing of
simple regression): "simple" (29), "linear" (11), "quadratic" (9),
"nonlinear" (8), "stepwise" (6). The total number of different statistical
tests is over 60. The most frequently used are "t-test" (34), "F-test"
(33), "chi-square test" (30), "Wilcoxon test" (17), "Kruskal-Wallis test"
(15), "G-test" (12), "Mann-Whitney test" (11).
A two-dimensional representation of similarities between the keywords,
based on the frequencies of their cojoint encountering in the same papers,
is shown in Fig. 2. It is possible to cluster them into six groups:
I. - standard statistical methods;
II. - multivariate methods (multiple regression and multivariate analysis
of
variance);
III. - deviations from normality, nonparametric methods;
IV. - contingency tables and multiple comparisons;
V. - Markov processes;
VI. - differential equations.
VII.


[pic]Fig. 2. Multidimensional scaling of the mathematical keywords cojoint
encountering data ("Ecology", 1991)



Scientometrics 39 (1997)
153
Е. V. BUDILOVA et al: TRENDS IN MODERN ECOLOGY

The major amount of terms (61 of 63) refers to the statistical data
analysis and among them are the most frequent the basic statistical
methods. Groups II and III contains more advanced methods as compared with
group I: the methods of group III admit deviations from normality and those
of group II admit simultaneous treatment of many variables. Group IV is
characterized by a discontinuous nature of treated variables. At last,
group V and VI are aimed to dynamic modelling, either stochastic or
deterministic. The frequency of the keywords of these two groups appeared
to be unexpectedly small especially taking into account that such
ecological terms as "competition" or "dynamics" were among the most
frequent ones.
An attempt were made to compare relative frequencies of encountering
mathematical terms in papers of different ecological groups (Table 3) but
no significant differencies were not revealed.
A statistical analysis of occurrence of mathematical terms was also
carried out for the Russian-written journal "Ecologia". The total amount of
mathematical keywords encountered in 139 papers published in 1991-1992
counted 402 with 101 (25%) different. 21 of the most frequent ones
(encountered 3 and more times) covered 77% of the total amount.
Their frequencies are presented in Table 4. We see that practically all
keywords in Table 4 can be found in Table 1. As compared with "Ecology",
the intensity and diversity of mathematical methods in "Ecologia" is lower.
The analysis of frequencies of application of various mathematical methods
(Table 5) also shows that in "Ecologia" dominate the papers of narrative
and survey character lacking mathematical methods (24% as opposed to 4% in
"Ecology"). Out of six above mentioned groups of methods only two are
presented in "Ecologia": standard and multivariate methods. Partly all this
can be explained through a lesser amount of papers in "Ecologia" and its
specific character, yet evidently it exposes also the fact that
mathematical methods are not so much used in Russian ecological papers.
Similar conclusion are drawn by some other authors (Beschastny, 1990;
Orlov, 1990) who explored the usage of mathematical methods in some other
natural science researches.

154 Scientometrics 39 (1997)

Table 3 Frequency of application of six
groups of mathematical terms in four groups of papers differing by
their ecological content
|62 |mean |7 |least squares |
| | | |method |
|48 |standard error |6 |linear |
| | | |regression |
|34 |probability |6 |two factors |
| |level | | |
|32 |indexes |5 |one factor |
|22 |correlation |5 |nonlinear |
| | | |regression |
|16 |regression |5 |log |
| | | |transformation |
|12 |t test |4 |variance |
|12 |histogram |4 |chi-square test |
|8 |variation |3 |maximum |
| |coefficient | | |
|8 |ANOVA |3 |confidence |
| | | |interval |
|7 |standard | | |
| |deviation | | |


Table 5 Frequencies of application of six groups
of mathematical methods in "Ecology" and "Ecologia"




Mathematical methods
"Ecology"
"Ecologia"

|0|Narrative and review |4% |24% |
|.|articles | | |
| |without statistical | | |
| |data | | |
|1|Standard methods |77% |69% |
|.| | | |
|2|Multivariate methods |60% |13% |
|.| | | |
|3|Nonparametric methods |35% |0 |
|.| | | |
|4|Categorical data |15% |0 |
|.| | | |
|5|Markov processes |2% |0 |
|.| | | |
|6|Differential equations|2% |0 |
|.| | | |


4. Conclusion

The performed analysis allowed to clear up some problems dealing with the
modern situation of scientific research in ecology. First of all, there are
well observed two main trends: population ecology and ecosystem ecology. At
the same time there is well seen another classification of ecological
researches - by research subjects such as plants, birds, terrestrial
animals, water communities. Ecosystem approach is more typical in plant
communities research, while the populational one is more characteristic for
terrestrial animals and birds. As to water organisms, they equally are the
base for both approaches. Purely numerically in the examined stuff dominate
the papers of populational approach. The analysis of application of
mathematical methods in



156
Scientometrics 39 (1997)
Е. V. BUDILOVA et at: TRENDS IN MODERN ECOLOGY

ecological researches showed that statistical methods add up to the
majority and surprisingly small is the proportion of articles which
includes differential or stochastic dynamic models.

The authors would like to remember with gratitude Professors V. Nalimov
and A. Ghilyarov for benevolent attention and helpful comments. This work
was supported by RFBR Grant 95-04-13573.

References

1. BESCHASNY, A. A., NEMTSOV, A. V., State of the art of application of
mathematics in psychiatry, Journal of Neurology and Psychiatry (Russia),
(1990) No. 2, 144-146.
2. BUDILOVA, E. V., teriokhin, А. Т., A bibliographic data
base. "Neural networks and

neurocomputers", The RNNS/IEEE Symposium on Neuroinformatics and
Neurocomputers, Institute of

Electrical and Electronics Engineers: New York, 1992, Vol. 2, pp. 1125-
1126.
3. CHERRETT, J. M., Key concepts: The results of a survey of our members
opinions, In: Ecological

Concepts, J. M. CHERRETT (Ed.), Blackwell Scientific Publications:
Oxford, 1988, pp. 1-17.
4. Computer Biometrics, Moscow State Univ. Press: Moscow, (1990).
5. GHILYAROV, A. M., One hundred and twenty five years of E. Haeckel's
"ecology", Journal of General

Ecology (Russia), (1992) No. 1, 5-17.
6. HOEKSTRA, T. W., ALLEN, T. F. H., FLATTER, С. Н., Implicit scaling
in ecological research,

Bioscience, 41 (1991) No. 3, 148-154.
7. leydesdorff, L., The generation of aggregated journal-journal citation
maps on the basis of the CD-

ROM version of the Science Citation Index, Scientometrics, 31 (1994) 59-
84.
8. NAUMOV, V. V., MULCHENKO, Z. M., Scientometrics: Study of the Science as
an Information Process

(in Russian), Nauka: Moscow, 1969.
9. NALIMOV, V. V., drogalina, J. A., Transpersonal movement: emergence and
perspective, Psychology

Journal (Russia), 13 (1992) No. 3, 130-139.
10. ORLOV, A. I., About reorganization of statistical science and its
application, Statistics News (Russia),
(1990) No. 1,65-71. \\.Science Citation Index (Ed. by E. GARFIELD).
Institute of Scientific Information: Philadelphia,
1991-1992. 12. TIJSSEN, R.J.W., DE LEEUW, J., VAN RAAN, A. F. J.,
Quasi-correspondence analysis on square
scientometric transaction matrices, Scientometrics, 11 (1987) 351-366.

Scientometrics 39 (1997) 157