Information about activities. How to join
                  the TT. Discussion
                  list, for members and others! Species-specific Care sheets Articles
                  from our Newsletter. Links to
                  important sites. Links to
                  members' sites. Search our
                  site! What has
                  been updated, and when.

Influence of environmental humidity and dietary protein on pyramidal growth of carapaces in African Spurred tortoises (Geochelone sulcata) Weisner & Iben 2003 : A criticism of the research methodology

A. C. Highfield

This paper by Wiesner and Iben has been uncritically cited numerous times and used frequently as key source material in discussions and reference texts related to the problem of so-called “pyramid growth” in tortoises.  One of the most striking recent examples is this statement, by world famous chelonian expert Peter C. H. Pritchard in his chapter “Evolution and Structure of the Turtle Shell” which features prominently in “Biology of Turtles” (CRC Press, 2008) pages 71-72:

“It has long been believed that the condition resulted primarily from a diet too rich in protein, and an alternative cause was believed to be an unnatural ratio of calcium to phosphate (sic) in the diet ... secondary causes were attributed to overfeeding, dietary fiber, temperature, UV light and so on... however current belief is that the condition derives from excessively dry conditions and when hatchlings are raised on a substrate of wet sphagnum, shells will develop normally (Weisner & Iben, 2003).

My italics. Note how other factors are dismissed by use of the past tense, and “current belief” identifies “excessively dry conditions” as the sole cause and “wet sphagnum” as the sole solution. The inconvenient fact that “wet sphagnum” (or anything like it) is not typically present at all in semi-arid tortoise habitats is entirely overlooked... and no physiological mechanism of any kind is suggested that could account for such an effect.

Furthermore, the above quotation is not directed merely at African Spurred tortoises.  It is directed at all tortoises, without exception or limitation.

With such uncritical acceptance, and such widespread and influential citation it is extremely important that the claims made by that paper, and the methodology employed to arrive at them, are indeed examined most carefully and with a critical eye. It is also important to highlight and identify cases where the paper is misinterpreted and misquoted as this creates yet another layer of obfuscation and confusion.

Humidity measurement and control

The most striking criticism of the data cited in their paper Influence of environmental humidity and dietary protein on pyramidal growth of carapaces in African Spurred tortoises (Geochelone sulcata) is that the bands of experimental humidity employed are so profoundly overlapping and are  of such wide individual range. This glaring and fundamental defect that runs through the whole paper. The terminology used is also contradictory in the extreme.  For example, they define what they term “dry environmental conditions” as consisting of the ranges  “24.3-57.8 % and 30.6-74.8% relative humidity”. They then go on to define “humid conditions” as comprising “45-99% relative humidity”. It is difficult to know what to make of this, when a relative humidity of (for example) 45% is described as “humid” in one sentence and then classed as “dry” in another.

It is also the case that very few ecologists or geographers would normally accept relative humidity of 74.8% as in any way representing dry conditions, or  45% as representing humid conditions (such conditions are normally defined as those containing a high percentage of water vapor; noticeably moist).  The use of such inconsistent and erratic definitions applied to the measured relative humidity levels must therefore give rise to serious concern and further confuses the earlier claims made in respect of “dryness” or “humid” conditions allegedly experienced in natural habitats.

The criticism of the methodology employed goes far beyond mere terminology, however. The actual bands used in the experiment are cited as follows:

Group A: 24.3-57.8%
Group B: 24.7-55.5%
Group C: 30.6-75.8%
Group D: 47.9-99%
Group E:  45-99%

The authors state that these measurements represent  the “mean of eight weekly measured values of the maximum and minimum relative humidity” in the experimental enclosures employed which consisted of  a series of 100 X 80 X 80 cm glass terraria fitted with two sliding glass doors. Hiding places or “caves” made of brick measuring 30 X 65 X 10 cm were provided in each enclosure, and the substrate used was the same in all units, comprising 4-5 cm of “bark humus”. The humidity levels cited in the above table were “measured directly under the top pane”, which would mean at the very top of the inside of the unit close to the heat lamps. They took a second set of humidity measurements in the caves of the “three humid terraria” (no mention at all is made of any such measurements in the “dry” terraria) which it is claimed were “10-15% higher” (than what?). It is unfortunately entirely unclear what these actual measurements were as no actual figures are given. If indeed the measurements were “10-15% higher” however, and were calculated on the same basis as Table 1, that would mean that humidity levels of up to 114% were achieved. This is a scientific impossibility (the maximum theoretical level of RH is 100%), therefore it is likely that they really meant “10-15% higher than the average ambient” -  but they do not say this, and they provide no figures for what those averages might have been, so again, we are left without a clue as to what conditions these tortoises were really subjected to.

Also, no temperature measurements of any kind are provided anywhere in the paper. It is unclear why the authors considered this critical information unimportant. Information on temperature is absolutely essential when considering relative humidity and the dew point, for example. Temperature information is also extremely important when evaluating any method of tortoise maintenance.

Only eight weekly measurements per group with just the mean of the maximum and minimum values were cited. The range is so huge, and the method of measurement so unreliable and imprecise we have no idea at all what the true average relative humidity was for each group. None whatsoever. How can you possibly reach any valid conclusion on the effects of a particular level of relative humidity, when you are subjecting the tortoises to anything from 45% to 99% for unspecified periods as they did with ‘Group E’ for example, or 30.6% to 74.8% in ‘Group C’? Most people would regard 30% as quite dry and 75% as quite humid......the range is so enormous for each group that the results are completely meaningless. You cannot possibly infer anything from those results.

We simply have no idea at all of how long these animals were subjected to these various levels of humidity. Take ‘Group C’ again. In any 24 hour period, were they subjected to 30.6% for 6 hours and 74.8% for 18 hours? Or the reverse? Or some other period? We can only guess.

The method of taking a once weekly measurement from the very top of the tank is also incredibly misleading and inaccurate. Of far more relevance would be a set of continuously levels at just above the substrate taken at various points (near the animals) - but they did not even measure these and we have no idea what they were. They would certainly not be the same as what was recorded at the top of the tanks where the heat sources were situated. Higher? Lower? We simply have no idea (again). It is also usual in trials like this to specify what equipment was used to obtain the measurements. This is important, as there is likely to be a big difference between the results obtained from say, a greenhouse hygrometer and a piece of properly calibrated precision laboratory equipment. The authors were aware of this, because they do cite the make and model of the atomic absorption spectrophotometer used. They fail to say what they used for the most important measurements of all, however... the hygrometers. This is a very disturbing omission. Did they use the same instrument for all the measurements? Was it calibrated? We just don’t know.

The method of humidity control employed is said to have consisted of 40 X 40 X 15 cm plastic bowls filled with demineralized water and “atomisers to produce fog”. The use of a “fogger” type atomizer to produce humidity is also problematic. These do not directly produce humidity. They produce water droplets that are then vaporized by temperature. We have no idea what these temperatures were, as already pointed out. They are very prone to producing a ‘cold fog’ on the bottom of the tank while creating very high humidity at the top... amphibian keepers like them for this very reason. They also produce a profound evaporative, localized chilling effect toward the bottom of the terraria. Anyone who has ever used one will recognize this characteristic. We again have no idea what the effects of this really were in this trial, because insufficient data is provided. We can be absolutely certain, however, that what was happening at “ground level” next to the animals would profoundly differ from what was recorded on those mere eight occasions at the very top of the tanks.

Unfortunately, from an experimental point of view, other factors that would substantially influence the allegedly “controlled” humidity within these enclosures were also present: in each, a 13 cm diameter dish with drinking water, and the provision of soaked pellet food and “endive salad”. These would both evaporate adding to humidity within the units in an uncontrolled manner under the heat lamps. As so little environmental data is provided, it is impossible to guess what impact this may have had. It is disconcerting that this possibility does not even seem to have been considered by those conducting the experiment. It would have been preferable if feeding and drinking could have been conducted in a separate, isolated area within the “controlled” environment to eliminate this variable.

General maintenance

The authors state that while the animals were provided with “light for activity” for eleven hours a day, the UVB tube and heating light was only turned on for three hours in the morning (9-12 am) and 90 minutes each afternoon (3-4.30 pm). This is a quite unusual regime and is certainly not typical of what most captive tortoises experience. It would certainly have had some quite profound effects upon feeding, activity and upon digestive efficiency. It is yet another strange and unquantifiable variable present in this study. We cannot know what effects (if any) it had on the results or the behaviour of the tortoises.

One effect we can predict with confidence is that if the heat lamps were turned on for three hours in the morning, and again for 90 minutes in the afternoon, this would have had an immediate and massive impact on the RH experienced within the units, producing either a considerable rise, or a severe drying effect depending upon the amount of water evaporated within the unit. It is quite impossible it had no effect at all.  This  very considerable variable is not even mentioned by the authors and absolutely no data as to the levels that resulted is presented. The net result is that these tortoises experienced wildly fluctuating, unspecified levels of  RH on a daily basis. It is hard to reconcile this with the claims subsequently made as to the effects of any one particular level or levels of relative humidity.


There are a number of other serious methodological criticisms of this study that can be made, including the failure to adequate monitor and regulate actual calcium and phosphorus intake for the duration of the experiment, which would appear to have been highly variable and subject to individual preference (loose cuttlefish bone was provided in each enclosure and a tortoise could consume as much – or as little – of this as it preferred). Given that calcium intake and overal calcium to phosphorus balance is such a critical factor in bone development, the lack of control here is very worrying from the point of overal experimental validity. The authors conclude, however, that even on what they call a “higher protein” diet, the tortoises raised in what they call the “high humidity”  (45-99% RH!) environment of Group E had less “pyramiding” than those raised on a lower protein diet in Groups A and B.

Curiously, the three lateral views of example tortoises from Groups B, C and D (no images whatever from groups A and E were published) all show growth abnormalities to some degree. There is an apparent slightly reduced level of deformity on the example tortoise from Group D compared to that of Group B.  However, due to the lack of control over calcium intake and Ca:P balance (cuttlefish bone is rich in phosphorus) we cannot be sure how much of this effect might be due to any difference in humidity, or how much might be the result of different amounts of cuttlefish bone consumption or of food intake. It appears no attempt whatever was made to measure or record this variable. A very wide range of blood calcium, phosphorus values were recorded in each group throughout the study, but interpreting it is difficult when such unquantified variables are present. We also have no idea how truly representative the very few animals depicted really are.

Finally - but certainly not least -  the values of dietary protein provided in the diet are disturbing. By just about any normally accepted chelonian nutritional standards, all of these tortoises without exception were fed a high protein, highly digestible (low fiber, 10.9 to 13.4%) diet. Not one group was tested using a diet that could be defined as “low protein” or “high fiber”. The lowest protein level provided was 13.7% and the highest, a rather staggering 30.7% (both DM basis). It would have been far more realistic (and useful) if a genuine “low” protein, high fiber option had been included. This could be defined as something offering circa 7.5% protein (DM) and at least 30% fiber content, which more closely represents a typical ‘wild’ dietary profile.  There is no question whatever that the dietary regime used would have promoted artificially high rates of growth in all of the groups. This fact is admitted by the authors.

Misquotes and distortions
In stark contrast to the claims of what I might term “vocal supporters” of this particular paper (who consistently misinterpret it and misquote it), at no time do the authors themselves actually seek to deny the role of protein intake in the development of ‘Pyramidal Growth Syndrome’ (PGS), indeed, in their summary they clearly state “it is understandable that the abnormal combination of a dry environment and a persistent high level of protein may lead to health disorders, such as PGS. Such management conditions are very common both in private and institutional situations, which probably explains the high prevalence of PGS in tortoises in captivity” and go on to state “some influence of dietary protein is probable”. 

They also openly admit that “the combination of dry environmental conditions and comparatively high growth rates induced by a nutritionally dense diet led to pyramidal growth in the African spurred tortoises of this study”, thereby also acknowledging the key role of high growth rates and diet - another link which “supporters” of this paper have consistently misrepresented. Do not forget the fact that all of the tortoises in this study - without exception - developed “pyramiding” to some degree or other. Not a single really smooth, deformity-free animal was produced out of the entire fifty.

While I decry the methodology of this particular study, and many of the conclusions reached, I do actually agree with them on those particular points. High growth rates, promoted by nutritionally “dense” diets are also a factor and must be considered along with environmental influences.  Each of these factors is involved and we need to consider them all. We should not concentrate on just one aspect to the exclusion of the others.

There is one other widespread distortion and massive misunderstanding of what this paper proposes. These authors absolutely did not advocate keeping tortoises at constant levels of very high humidity. Nowhere do they even suggest that. What they actually say is very different. They simply say that “areas with a relative humidity of nearly 100% for hiding should be provided to the tortoises at all times”. This is not at all the same thing as forcing animals to endure constant levels of high humidity. They are specifically referring to the provision of a “humid hide” where a tortoise can retreat - voluntarily - on a temporary basis - not recommending an enforced 24/7 “sauna” type environment with no means of escape. This latter is, unfortunately, an apt description of the conditions adopted by some keepers who refer to this paper as their inspiration.... where constant round-the-clock 80-90% humidity at high temperatures (26C+) is the only environment ever available to them. Such conditions deprive those animals of thermoregulation opportunities, subject them to constant levels of humidity that would only be experienced on an infrequent basis in nature, and which could have unknown metabolic and physical consequences over the long term.

It is a strange and somewhat ironic fact that the authors of this paper were really very close to identifying the true causes of “Pyramidal Growth Syndrome”. Their experimental methodology may have been defective, but the link between humidity and this type of growth is real. Had they persisted further, and inquired in far more detail into the one material that is extremely affected by humidity and that almost completely surrounds a chelonian - keratin - they might have arrived at a genuinely viable answer. Instead, they overlooked keratin entirely (the word is not mentioned once in the entire paper), and they sought refuge in vague, unsubstantiated and unviable theories involving “intracellular and intercellular” pressure of osseous (bone) tissues caused by dehydration. Had they instead looked at what effect external hydration and dehydration has on keratin, and also at the two key  modes of cell proliferation in chelonian keratin, a great deal of confusion would have been avoided.

(c) 2014 A. C. Highfield / Tortoise Trust