Which statistics should tropical biologists learn ?

Tropical biologists study the richest and most endangered biodiversity in the planet, and in these times of climate change and mega-extinctions, the need for efficient, good quality research is more pressing than in the past. However, the statistical component in research published by tropical authors sometimes suffers from poor quality in data collection; mediocre or bad experimental design and a rigid and outdated view of data analysis. To suggest improvements in their statistical education, we listed all the statistical tests and other quantitative analyses used in two leading tropical journals, the Revista de Biología Tropical and Biotropica, during a year. The 12 most frequent tests in the articles were: Analysis of Variance (ANOVA), Chi-Square Test, Student’s T Test, Linear Regression, Pearson’s Correlation Coefficient, Mann-Whitney U Test, Kruskal-Wallis Test, Shannon’s Diversity Index, Tukey’s Test, Cluster Analysis, Spearman’s Rank Correlation Test and Principal Component Analysis. We conclude that statistical education for tropical biologists must abandon the old syllabus based on the mathematical side of statistics and concentrate on the correct selection of these and other procedures and tests, on their biological interpretation and on the use of reliable and friendly freeware. We think that their time will be better spent understanding and protecting tropical ecosystems than trying to learn the mathematical foundations of statistics: in most cases, a well designed one-semester course should be enough for their basic requirements. Rev. Biol. Trop. 59 (3): 983-992. Epub 2011 September 01.

Tropical biologists study the richest and often the most endangered biodiversity in the planet and in these times of climate change and mega-extinctions, the need for efficient, good quality research is more pressing than in the past.Previous authors have mentioned the need to improve the statistical training that tropical biologists receive, in order to reach those goals of efficiency and quality (Monge-Nájera 2002).
However, the statistical component in research published by tropical authors has been qualified as suffering from three problems: poor quality in data collection; mediocre or bad experimental design and a rigid and outdated view of data analysis (Fielding & Lauckner 1992).
In comparison with the scientific leaders, tropical statisticians are few, isolated and not always properly trained, having for example a poor grasp of the concept of variability and its usefulness in biology.Some scientific journals that publish in this field accept articles that lack appropriate descriptions of the experimental design and in some cases, 70 % of the articles have incorrect statistical analyses (Camacho 1997).
Some problems detected by the authors of Camacho's (1997) compilation are not so serious today, thanks to the Internet, which provides acceptable statistical software for free and helps statisticians from tropical countries to keep abreast of new developments and in SPECIAL ARTICLE contact with colleagues.However, the key to the problem lies in education and progress is slow in this field.The results of statistics courses taught to biology students are often unsatisfactory, and students too often end up hating statistics and being unable to select and apply appropriate test and procedures (Camacho 1997).
To summarize the problem, tropical biologists are not being taught the statistics they really need.The objective of this article is to identify the statistical tests that are most frequently used in the current literature of tropical biology, and to use that information to recommend an updated syllabus to all institutions that educate tropical biologists.

MATERIALS AND METHODS
We listed all the statistical tests and other quantitative analyses used in two leading tropical journals, the Revista de Biología Tropical and Biotropica.For the Revista we reviewed all of the articles from editions 56-4 (December 2008) to 57-4 (December 2009), not including the supplements.For Biotropica, we used only the articles included in the "Papers" section from editions 40-4 (July 2008) to 41-4 (July 2009).

RESULTS
We found that the articles in both journals used a large number of unusual statistics, but that each of these highly specialized procedures and tests was used very few times (Appendix). The

DISCUSSION
In tropical ecosystems, a few species are very common, while most species are rare; and in our sample, the same applies to statistical procedures.There is no justification to burden students with a knowledge of procedures that they will seldom, if ever, use; if the need arises, such procedures can be applied by professional statisticians.On the other hand, the main mathematical procedures were almost the same in both journals, even though Biotropica is more limited to terrestrial ecology, while the Revista publishes articles from a variety of fields in marine and terrestrial biology, as well as on the conservation of tropical ecosystems.
This gives us some confidence to recommend that statistical education of tropical biologists be based on learning when to use the dozen tests that they are more likely to need in their professional life (top of Table 1).The mathematical procedures involved in those tests, as well as any mathematical proof to justify those procedures, must be eliminated from the courses and left to professional statisticians.Highly useful tests such as Chi-Squared and the G test are simple, but more complex ones -such as multivariate tests-can also be understood from a practical perspective without need for the subjacent mathematics.
Descriptive statistics is also something everyone needs, but again only to understand what averages, standard deviation and the like mean.Learning the algorithms to calculate them is not necessary because computers already "know" how to do it for us.
A frequent weakness in most articles, not only in tropical biology but in all fields and nations, is the poor selection of graphics that has become omnipresent after the introduction of computer spreadsheets.But this problem is easy to identify (a simple guide to good statistical graphics is here: www.biologiatropical.ucr.ac.cr).
Students should also learn that the statistician must be consulted before collecting the data, and not afterwards, when it may be too late.They should not finish their statistical courses hating statistics, as we know is the case in some institutions, but rather feeling confident that they can select the correct procedure and apply it, as also found by previous authors (Garfield andBen-Zvi 2007, Metz 2008).
Software is another field in which current education clearly fails.Excellent free software is available to everyone in the Internet, yet most universities pay huge yearly amounts in licenses of professional software for students.Very good programs that are fit for tropical biologists are available in http://faculty.vassar.edu/lowry/VassarStats.html and in other addresses (e.g.http://statpages.org/,www.freestatistics.infoand www.macstats.org).Professional programs should only be bought for professional statisticians.
Students should also learn that statistics has space for humor, as shown by Gary C. Jokes (http://my.ilstu.edu/~gcramsey/Gallery. html), and that the motto of the guild is "Rubbish in, rubbish out".They must always put mathematics at the service of biology and not the opposite; they should understand that only controlled experiments can identify cause and effect relationships; if this requirement is not met, no statistical procedure will (James & McCulloch 1990).Additionally, it is of the greatest importance that:

Ramseyer's First Internet Gallery of Statistics
1. Statistics be taught within the career's subject courses (for example, as part of biology or medicine courses, instead of being a separate course) or at least, that all examples used in courses be from the career's field (Metz 2008).2. Related subjects, such as distribution, center and spread of data, be learned as a single concept (Garfield & Ben-Zvi 2007).
Currently, students can pass statistics courses with good marks even when they cannot understand statistics (Garfield & Ben-Zvi 2007, Metz 2008).Furthermore, they cannot apply statistical procedures outside the contexts in which they learned them and cannot understand the meaning of statistical tests and graphics (Garfield & Ben-Zvi 2007).
Every year unnecessary statistics textbooks continue to be written and published everywhere, despite the fact that all the information needed by students is freely available in Internet.Apart from that, almost all statistics textbooks available today are basically the same, a situation that would not be so bad if these they had the information that the students need, but a comparison of contents in books being published now and those from half a century ago will show little difference except for the addition of multivariate tests.Not surprisingly, student learning also is poor practically everywhere (Garfield & Ben-Zvi 2007, Metz 2008).
In conclusion, statistical education for tropical biologists must abandon the old syllabus based on the mathematical side of statistics and concentrate on the correct selection of procedures and tests, on their biological interpretation and on the use of reliable and friendly freeware.In most cases, a well designed one-semester course should be enough for them.It is a matter of common sense: their time will be better spent understanding and protecting tropical ecosystems than trying to learn the mathematical foundations of statistics.

TABLE 1 (
Continued)Statistical tests and other quantitative procedures that were used more than once during the studied year

TABLE 1 (
Continued)Statistical tests and other quantitative procedures that were used more than once during the studied year