It's only a matter of fashion. We like to think about science as a purely rational human culture, but in fact it's subject to all common human biases, such as fashion, prejudice, gossip, blood feuds etc. There is no doubt that, in any science, the descriptive, experimental and theoretical approaches are equally needed and feed each other, although people are forgetting it now. Another sad point is that, because most journals do not value descriptive studies any more, in some natural sciences, such as Ecology, many people are pushing it to make their studies look more theoretical or more hypothesis-driven. They advance pretty weak hypotheses and ethereal mathematical models, without a true biological basis. But those numerological studies have now a much higher chance of being published than true ecological studies oriented by theories and powered by field data.

Many ecologists, taxonomists etc have pondered this question. The reduction of outlets for such work is an increasing problem. From the viewpoint of scientists, environmental managers, conservation workers and decision-makers, getting access to this 'simple' descriptive information is extremely important, often providing valuable baseline data. Its value should be increasing in a changing world. However, journals are driven by different priorities, the most obvious of which are selling more journal subscriptions and increasing the 'apparent' status of their publications via enhanced impact factors. The implication of their decision-making is that descriptive work is not of value (to them at least).

In taxonomic work, some journals will no longer accept single-species descriptions. Another species, or multi-species revision, or addition of a cladogram, or molecular work, or a combination of these may be required to satisfy some editors! Our lack of knowledge about the organisms inhabiting our planet is well recognised: e.g., "So far, about 1.75 million species have been identified, mostly small creatures such as insects. Scientists reckon that there are actually about 13 million species, though estimates range from three to 100 million" (estimates still vary considerably since this was written). Governments eagerly sign up to the Convention for Biological Diversity (CBD), but obtaining the fundamental data for "Sustaining Life on Earth" does not appear to be prioritised. Wouldn't it be great if everyone (including scientific journals) played their part?

I agree about this problem. May be one reason is that investigation leading to descriptive results are camping at the basis of empirism and have low generalization value. Another limit of descriptive results is over inference without test and respect of the parcimony criteria. I said that but most of my research is descriptive :). But I know also that case study, descriptive works are seting futher research.

Sorry I forgot an important ref about this : the paper of M. Piugliucci 2002 : Are ecology and evolutionary biology "soft science" ?

You can get it at :

http://www.sekj.org/PDF/anz39-free/anz39-087s.pdf

Thanks Alex - a nice paper - I will read it in detail. We also have a lot of results that are published as "descriptive". Approach that is lately favoured in publishing policy is that higly ranged journal do not publish the data, but the syntesis only. The data are published in local journals. Sometime, this lead to lower confidence of the data. Am I right?

Yes, I agree, this is the case of Ecology letters, but a very good journal indeed. The problem is really important in Conservation. We get money for singular question and then the project is finished and we get new money for another topic. We have to stick to precise theory and to try to collect data, eventually according descriptive approach, but within the focus of this theory.... not easy.

Momir, I fully agree that observational data are neglected. To my opinion that´s a current fashion in ecology. I am sure that it will change again. Your question and also the positive response of the others show - that it will change (soon). A colleague of mine and me wrote a paper addressing that issue to a certain extend: http://www.researchgate.net/publication/221746208_Ecotoxicology_and_macroecology--time_for_integration

All the best, Matthias

Article Ecotoxicology and macroecology – Time for integration

Matthias, thanks for nice paper - I have read it briefly. I will reed it gingerly latter.

You are right - where to draw the line. Matthias (see previous discussion), for example, did a great job to do something in that regard. Actualy, we also tried to do something in that direction - we established journal at the beggining in 2011 (open acess) named Water Research and Management. Editorial policy goes in that dirrection to support publishing of real data. The things with journal goes slowly, but, we already published 7th issue.

Great topic this one. In particular for biodiversity descriptive data are still valuable for several journals, however as you have commented they are published mainly in local journas. The great exception is the good quality, high visibility and free online CheckList Journal, http://www.checklist.org.br/, that encourage this kind of data. Another journals as Zootaxa and Zookeys also could be considered (taxonomic) basically descriptive. Probably this effort could change the general trend discussed in this forum if as a community we use and cite such kind of papers.

Greetings!

Coincidently, notification of a new 'Biodiversity Data Journal' has just appeared. The Editor-in-Chief will be Dr Vince Smith from Natural History Museum, London.

Dr Smith writes "We are establishing a next generation platform for publishing biodiversity science and data. The Biodiversity Data Journal (http://www.pensoft.net/journals/bdj) is a comprehensive online platform designed to accelerate publishing, dissemination and sharing of biodiversity-related data of any kind. Contributions will be accepted on a wide range of biodiversity subjects and will be published in association with major data repositories (e.g. GBIF, Dryad, GenBank) using a number of Biological Code-compliant templates. These templates make it easy to write manuscripts, review text and share data.

In preparation for launching the journal we are seeking editorial specialists with expertise in subjects and data types (e.g. taxon treatments, checklists, genomic-, ecological- and environmental-datasets, analytical methods and software) who can be involved in the innovative reviewing process. Potential editors should be interested in new ideas, methods and approaches to publishing, sharing and using biodiversity information. Established specialists, post-doctoral researchers and PhD students with relevant experience should not hesitate to apply using the short form linked below":

http://www.pensoft.net/editor_form.html

This is wonderful news

I attach the flyer

Hi everione,

Thanks for the encouraging info about new journals aimed to take into the consideration "descriptive" works, as well. I am particularly happy hearing that such initiatives (to provide platform for publishing important data without sophisticated additional analyses, which are often not so important for good presentation) are so widespread.

I am just afraid that changes in editorial policies could happened at stage when journals became more popular. It has happen in several cases (somebody mentioned Zootaxa).

Regards,

Momir

Hi everyone,

Thanks Momir for raising this interesting issue. I do and I don’t agree with the idea that “descriptive” approach is lately neglected. I think most of the time depends on the question whoever wants to answer and the quality of the data one has to answer the question. For example, in biogeography, climate change or evolutionary biology a good data set to address a question is un-neglectable. In evolutionary ecology, excellent data sets to explore heritability of a given trait or traits using animal models are largely accepted if the data set is good enough. Most of these new studies are published in great journals, including Nature or Science. Exploring microevolutionary process by selection analyses are commonly non-experimental and it is highly unlikely that they are neglected because of the descriptive nature of the data. The question is that if we wanted to support or reject an idea, or disentangle a mechanism. I think in there, it is needed experimental support. It is quite common that we (I include myself) try to oversell the paper pushing it a little going a bit further to explain whatever even knowing that we do not have the appropriate data. In summary, I think that good data sets, linked to good ideas, framed and discussed properly are highly unlikely to be rejected regardless of the descriptive nature of the data we are playing with.

Thanks everyone for the discussion!!

J

There are also two recently launched open acces journals (Dataset Papers in Ecology and Dataset Papers in Biology) that publish descriptive information and datasets (see http://www.hindawi.com/dpis/)

It seems that descriptive research span of a continuum of useful/essential to idiosyncratic and non general. A phylogenetic tree of a major taxa or list of their life histories is dt is essential (and highly valued!) descriptive research needed to address many ecological and evolutionary questions. The publication of the human genome was a great triumph in descriptive research. These types of research help us generalize about large scale processes and search for general mechanisms. But I think ecology can be easily swamped by the minutiae of local descriptions - the population dynamics of aquatic insects in a particular pond, for instance. Is this useful information to those not studying that specific question? If local data is used to address a general mechanism, with implications outside that particular system, then the answer is yes. Otherwise, its probably not terribly important. My sense is that high impact journals have the same perspective. Complicated statistics are often necessary to characterize descriptions and test mechanistic questions. Otherwise, there's the danger that ecologist become "stamp collectors", and the field ends up lacking a core body of knowledge.

A good middle road is to link local information to broader datasets (e.g. biodiversity, soil nutrient profiles, competition outcomes, etc), following a common protocol. This way, instead of flooding the field with local information, a good descriptive paper can contribute to a synthetic body of knowledge that others can make use of.

It's only a matter of fashion. We like to think about science as a purely rational human culture, but in fact it's subject to all common human biases, such as fashion, prejudice, gossip, blood feuds etc. There is no doubt that, in any science, the descriptive, experimental and theoretical approaches are equally needed and feed each other, although people are forgetting it now. Another sad point is that, because most journals do not value descriptive studies any more, in some natural sciences, such as Ecology, many people are pushing it to make their studies look more theoretical or more hypothesis-driven. They advance pretty weak hypotheses and ethereal mathematical models, without a true biological basis. But those numerological studies have now a much higher chance of being published than true ecological studies oriented by theories and powered by field data.

I think descriptive information is still appreciated, but the "fashion" or "paradigm" has certainly placed greater value on "hypothesis testing" publications that have general application. It's not sufficient to describe a phenomenon in one lake, you have to demonstrate it across a gradient of lakes, etc. I think the value of this approach is clear: General principles can be applied widely without the need to re-test the hypothesis/theory in every new system. Sophisticated statistical presentation is not necessary, but is often a natural result of the desire to quantify an effect or statistical confidence that a theory is supported by the data. However, I wouldn't say the descritive information is "dead". In my own field, Limnology (study of inland waters), papers almost always have a descriptive section that *supports* the hypothesis-testing. I rarely see purely theoretical papers; even integrative reviews are quantitative and present data.

I don't know for sure, but I suspect (my "hypothesis" is :-) ) that this "fashion" for broadly applicable papers is an artifact of journal space being the limiting resource (at least historically). Some would have darker interpretations- that sinister vested interests try to block out rivals, etc., to increase the prestige of the most selective journals. With the revolution of digital and open-source publishing, that limit is removed and, whether sinister or not, the extra space is created for more descriptive presentations. I am looking forward to seeing more of this kind of thing.

Dear sir,

It is exactly what is happing in the wildlife research especially in the estimation of carnivore population. Researchers and scientific communities think and use complex statistics which no body could understand forgetting the basic sampling procedures and thinking statistics would bring the results. This complex statistics has lot of assumptions, this assumptions will not hold in greater number of cases, thus flawed methods results in poor estimation. But when simple statistical methods with adequate sample size is considered as poor .

All science is describing the world. Even the experiments. The question is in the quality of description, in attempts to generalize, explain the connections. Reading the articles presenting real data, you can decide something about the quality of the work.

Hi, Momir! Hi, everybody!

I agree with You, Momir. Even in the middle of 80ies the most "advanced" scientists of the r.i.p. USSR considered descriptive papers as "gray noise" in science. Though a variety of mathematical methods and dozens (hundreds!) of various indexes do not replace qualification, especially in our field ))

Hello,

thanks for launching this interesting topic.

We had a similar discussion some months ago during a meeting of a hundred vegetation ecologists in Switzerland.

Following the plenary discussion during the meeting, we investigated the methods published articles used in the field of vegetation ecology and came to the result that "observational or descriptive studies constitute a strong basis for further investigations", but that other methods (experiments, modelling) are recently chosen more often due to the "opportunistic behaviour of researchers who choose the approach that allows them to increase the probability of being rapidly published" (Spiegelberger et al,2012, Plant Ecology and Evolution).

Fashions are also important in science...

I completely agree with you Mello. I think there is just good and bad science, the rest are interests, fashion, prejudice, etc. Today most journals prefer to publish bad driven hypothesis that result in high impact bad results and conclusions (with important consequences in our understanding of the real world) than simple, less ambitious, but right and detailed description of nature. I understand in the past there were too many descriptive studies, but excluding descriptive data from most Journals is wrong, specially considering the bad quality science that is published in many distinguished and popular journals that only publish driven hypothesis studies.

A recent article in 'The Scientist' has just been brought to my attention by Kirk Fitzhugh and is of some relevance here.

Less Influence for High-Impact Journals

A new study reveals that more and more of the world’s most-cited articles are published outside of high prestige, high impact factor journals.

http://www.the-scientist.com/?articles.view/articleNo/33209/title/Less-Influence-for-High-Impact-Journals/

Redressing the influence of impact factors is, of course, one of the things that ResearchGate tries to do.

It is a good question. As a scientist I also suffered not dealing with descriptive studies. For example, one has more than 100 publications all are description/redescription of some parasites, has hard time understanding the value of hypothesis driven studies. Hypothesis driven studies are THE SCIENCE itself since they use Scientific Method. I have no interest to read a paper describing a gut flora of a fish, but would be very much interesting to see the change in the gut flora after application of a probiotic, for example.

In the latter hypothesis driven study, >you are already required to describe< normal or control flora in fish.

I mean that if we think/read/reasearch hard we develop an hypothesis and at the same time describe the situation. Hypothesis driven studies (using Scientific Method) will certainly speed up new developments in an area.

An attitude that is very human and unlikely to change. To agencies funding scientific research and institutions hiring scientists (such as university faculties), just describing organisms and species seems too technical and repetitive, not enough demanding on imagination and rational power, and unlikely to lead to exciting discovery. Is is considered an investment that is not conducive to promotion of the image of large institutions. In some groups such as insects ( I work with them) or other invertebrates, the perception is that there are too many 'species' to describe, and achieving close to complete description appears as a never ending task that is not worth the cost of it.

Very interesting discussion. I often tell my students that the simple observation of nature or biodiversity can lead to fantastic discoveries, even in the XXIst century. Just think that scientists observing male penguins regurgitating fishes to feed their offspring before the return of females (the film March of Penguins, 2005) have discovered new micropeptides present in the stomach of these birds (Y. Le Maho and coll.). These micropeptides could become our antibiotics of tomorrow. There are other examples with insects (Nober Prize 2011 Jules Hoffman).

Gerard

As a marine biologist I feel that the 'descriptive' approach is fundamental to doing the science. This is part of the scientific method after all. There is no sense in doing experimental or theoretical work under one has described the system one is working on. Otherwise you risk asking the wrong questions or applying theory that is not relevant. To the extent that I have been successful it is because I let the natural history of my systems inform the questions I am asking. This is why I think it is important for biologists to gain expertise with a group of organisms before attempting to apply theory to system. After all, this worked with Darwin and his barnacles.

I think good descriptive work derives its value, to a large extent, by its relationship to theory. Characterizing the human genome was a descriptive scientific milestone, because we knew how important - from mechanistic theory - the genome was. Testing theory ultimately requires hypothesis based science. Even good, basic description implicitly relies on theory or hypotheses - there is a hunch that x is important to fitness or life histories, etc. Description that seems untethered from formal or intuitive theory/hypothesis may be deservedly of lower importance. If scientists can clarify those links in the introduction of a paper, I would think there'd be less resistance to its publication.

When scientists describe the phenomenon not explainable by recent paradigm, it may led to new discoveries. We try (using the coat of statistics) explain all things, to sell themselves, to be clever. The need to get money does not support DOUBT, so characteristic for every creative work. However, description in taxonomy, floristic and and faunistic works must also persist. Science needs consistency and revolutions friendly together.

Description is still valued in the sciences but has taken a back seat to the perceived activities of what is understood to be “real science”. First, description per se requires context for appropriate evaluation; numerical quantities such as five light years or thirty-seven centimeters do not, as their standards of comparison exist by agreement and are based on what is clearly measurable, e.g. the distance light can travel in one year. In turn, what is clearly measureable is also explicitly verifiable, a precondition of the reductionist approach that was so successful during the development of 20th Century science, particularly in the physical sciences. Again, description per se is not immediately verifiable, particularly in biology and biologically related sciences where evolving systems exhibit variability. The second reason then is simply a default condition of the nature of science conducted during the 20th Century and has had a limiting effect on what is considered to be “real science”, particularly among the public but also in science education.

Also associated with description is the process of comparison and the comparative sciences, e.g. much of ecology and evolutionary biology falls under the banner of comparative science. Very often these sciences among others are labeled as inexact sciences, largely because their predictive value does not meet the standards of physics, chemistry, et al, and also very often the initial findings are embedded in description. However, if one considers the problems that each domain of science addresses and understands that it is the problems that define the methods used for their potential solution and not the other way around, then the stain of the comparative sciences not appearing to be a true science disappears.

One further note is pertinent here and bears on a comment made by one of the doyens of population genetics, C. C. Li, in his seminal text, Population Genetics,

“In the employing of mathematical methods, however, biological facts must be reduced to a mere abstract of their real complexity. It is important to understand that this simplification is recognized for what it is: a working method. It does not mean to ignore the complexity and totality of biological relationships. These clearly remain the foundation upon which any mathematical model may be built.”

This quote turns the tables on those enamored with the precision of mathematical acuity by clearly recognizing that precision is achieved at some loss, that of the reality of a more complex world, the world that science strives to understand. Another way of considering the quote above is to recognize the limitations imposed by any model. Those limitations were set out clearly by Richard Levin in his book, Evolution in Changing Environments, and are as follows: All models are subject to any two and only two of the following constraints, generality, precision, and reality. Specifically, models can be general and precise, but if so, cannot satisfy the condition of reality. The Hardy-Weinberg Equilibrium is a model that fits these conditions. It is general, since it can be applied to any one gene-two allele case, and it is precise in that it makes specific predictions given the values of the allele frequencies. However, it is not a realistic model, since it only addresses the one gene-two allele case. It can be modified to fit more genes, but very quickly becomes too cumbersome to work with. Interestingly, description and the comparative method may (do) give a broader view of than that of mathematical modeling of the world that science attempts to understand.

As a marine biologist I have found that Journals are less interested in much of what I have been studying on account of 'descriptive biology' not conforming to editorial policy. I agree with David Garbury very much on this matter and it is because the aquatic environment is one where descriptive accounts are of great importance to continue to present to the wider scientific community. Perhaps we might be able to change the views of some Journals through this link, should they be interested!

The thing with 'descriptive biology' is that it is not mechanical- it requires a certain talent with words as well as a keenness of eye to know what to describe and how to describe it. Secondly, there is much scope to obfuscate. Seeking clarity in 'rigorous' science, journals with limited space chose the path of statistics and mathematical models. However, what Antony Harper says above is perfectly true and instead of recognizing that mathematical models sacrifice complexity at the altar of clarity, the trend is to rely more and more on such models. Sometimes, basic facts are overlooked in the beauty of mathematics, and this has led to some indefensibly poor work being published in prestigious journals.

The current fashion is to encourage wide ranging conclusions at the cost of specific observations. However, the thrall of impact factors, citation indices and other approaches to 'monetizing' science and scientists will perhaps not last long, given that work published in non impact factor or low impact factor journals have higher citations, as noted by Andrew Mackie above.

Perhaps, in an ideal world, the 'anecdotal' and 'rigorous' will both find their place in the sun. Certainly, the anecdotal or descriptive will regain its respect in the long run, as overly rigorous science stumbles on the problem of lack of comprehension and inapplicability on the ground, since much of it takes into consideration unrealistic situations. I always point out to my audiences that most scientific theories, hypotheses and laws begin with the qualifier, "Other things remaining constant"; however, nothing remains constant in nature. Therefore, all these are hypothetical explanations are to help us understand the interaction of the factors being compared and not necessarily reflective of the situation in toto on the ground.

It is the situation on the ground that concerns policy makers and managers, so descriptive science will continue to be valued, while modeling will continue to provide the theoretical basis, regardless of the fashion prevalent among editors of prestigious journals.

Yes, I agree that descriptive science has taken a back seat. Descriptive science was at one time the very basis of any further questions. However man has moved on from appreciating natural phenomena as a whole to asking questions and wanting proofs for everything at the molecular level. The key lies in seeing the details in the larger picture. It will not be out of place to say that these days one often encounters students who know the greatest details on molecular level functioning of every aspect of science, however would not know where these functions fit into in physiology or evolution for instance.

Description of natural systems is a prerequisite for understanding them. Darwin and Margalef are two examples of people having a tremendous background in natural history, thus being able to understand ecology and evolution. The pressure and fashion to publish analytical papers is a gateway for bad science. Consider the recent claim that all seagrass meadows in the Mediterranean will die out in less than a century due to climate change. I looked at this amazing claim, and found nothing supporting it: poor experimental design, voodoo statistics, virtually no data, biased interpretation... all seasoned with impressive modelling and graphs. Fortunately, some journals now publish large datasets, or at least require deposition of all data. It's simple: no data, no science... perhaps academic promotion, but not science.

The descriptive approach has marked most of the last century, contributing greatly to the advancement of scientific knowledge. However, it has its limitations, as it does not investigate the functional relationships in terms of cause and effect, but merely advances a correlative approach. This is no longer sufficient to understand the phenomena we have to deal with. This does not mean that the descriptive approach should be abandoned, but rather geared to the needs of knowledge that emerge from the study of the processes in which we are immersed and, in many cases, we are drivers ourselves thereof.

“We say it is "explanation" but it is only in "description" that we are in advance of the older stages of knowledge and science. We describe better we explain just as little as our predecessors. We have discovered a manifold succession where the naive man and investigator of older cultures saw only two things "cause" and "effect " as it was said we have perfected the conception of becoming but have not got a knowledge of what is above and behind the conception. The series of "causes" stands before us much more complete in every case we conclude that this and that must first precede in order that that other may follow - but we have not grasped anything thereby. The peculiarity for example in every chemical process seems a "miracle " the same as before just like all locomotion nobody has "explained" impulse. How could we ever explain We operate only with things which do not exist with lines surfaces bodies atoms divisible times divisible spaces - how can explanation ever be possible when we first make everything a conception our conception It is sufficient to regard science as the exactest humanizing of things that is possible we always learn to describe ourselves more accurately by describing things and their successions. Cause and effect: there is probably never any such duality in fact there is a continuum before us from which we isolate a few portions - just as we always observe a motion as isolated points and therefore do not properly see it but infer it. The abruptness with which many effects take place leads us into error it is however only an abruptness for us. There is an infinite multitude of processes in that abrupt moment which escape us. An intellect which could see cause and effect as a continuum which could see the flux of events not according to our mode of perception as things arbitrarily separated and broken - would throw aside the conception of cause and effect and would deny all conditionality.”

Nietzsche, the gay science. [https://www.goodreads.com/work/quotes/2382792-die-fr-hliche-wissenschaft]

To make the research article more concise short description is given and more details about description can be included in as supporting information