blog

Research published in F1000Research highlights an important issue regarding animal experiments, especially behavioural ones. The results of these studies could be flawed because unforeseen environmental changes are often not considered, even though they could influence the experimental results.

We invited one of the authors of the article, Rupshi Mitra (RM), Nanyang Technological University, Singapore, and one of the reviewers, Jaroslav Flegr (JF), Charles University in Prague, Czech Republic, to discuss how widespread the problem is and why it is important to conduct these kinds of studies.

Why is it important to consider the wider environmental surroundings during animal experiments?

RM: This study is important because it focuses on animal researchers’ most important resource, i.e. the experimental animals and their surrounding environment. Every experiment completely relies on the experimental subject since they are the primary source of all data.

Animals and their behaviour are a direct product of their immediate environment. Experimenters often take an objective view of animals as just an experimental subject, but it is crucial to remember that animals behave as a response to their environment. In fact, rodents and other small animals are capable of sensing sound-waves and other olfactory and visual cues that are less evident to us.

So, it is very important to keep animal’s immediate, as well as wider environment, as unperturbed as possible. For example, next door construction or an animal-house next to a high-traffic road is vulnerable to several sound and other disturbance which might go easily unnoticed. However, this might lead to a shift in the baseline of behaviour of control animals, which might result in significant variation in ultimate effect size.

JF: Animals are extremely complex systems. They have many inputs for various signals, not just for the signal carried by a physical, chemical or biological factor, the effects of which we want to study in our experiment.

Many of these factors can quantitatively or even qualitatively influence the responses of the animals to the focal factor. Researchers always try to maximally standardise experimental conditions to make the results of their experiments easily reproducible. However, they never fully succeed. It is more or less possible to control the factors that we are aware of.  However, it is almost impossible to control the factors that we are not aware of.

What type of environmental modifications can influence behavioural changes? And what are the behavioural changes?

RM: The most important environmental modification that can influence behaviour is that of the housing cages. Through several published literature we showed that stress-behaviour can be significantly altered just by changing the housing environment of experimental animals.

In addition to housing/cage/immediate environment, the environment of the holding room (for example, animals in small standard laboratory cages and big complex housing cages put together) as well as general noise-level of the animal facility matters for behaviour.

The most typical behaviour changes we observe are stress-sensitivity and alertness, locomotion and movement (that changes immediately following cage changes from simple to complex housing), playfulness, social interaction and food-intake. All of these behaviours are central to the basic well-being of animals.

JF: An animal organism is buffered against many environmental changes. Some parameters are kept relatively constant, regardless of any dramatic fluctuation in the environmental conditions.  Some biochemical and even some physiological parameters are relatively robust, so they do not change too much in response to changes of external conditions. However, animal behaviour is rather sensitive even to slight changes of conditions. Practically, any environmental factor can influence any output of an ethological experiment.

We know that we have to standardise room temperature and moisture, light, noise, and the amount and quality of food. We can at least partly achieve this. However, there are also other factors that are far more difficult to standardise, such as the vibration of a building; the composition of air; the mood or phase of a menstrual cycle of a researcher or laboratory technician. All these factors could have strong impacts on the behaviour of laboratory animals.

How can this influence the results of the experiment?

RM:  Major changes in daily behaviours, such as movement, eating and alertness, directly influence animals’ health, metabolism and overall fitness, and can influence almost any behaviour that need to be tested in animals. For example, extra-alert, anxious animals will eat less and be more aggressive/defensive to the experimenter’s handling. That itself will shift the behavioural paradigm that needs be tested.

We are also aware that an anxious rat will tend to run faster on a maze and can directly influence anxiety behaviour readout. This is in addition to stress-sensitive hormonal influence from the internal environment of the rat’s body. The influence of environment on animal behaviour becomes even more critical during animal breeding and pregnancy. We know now from several studies that mother’s environment (pre and postnatal) directly influence pup’s behaviour until adulthood. Thus, it is crucial to maintain minimally perturbed and a relatively stress-free environment for Dams and pups.

JF: The effects of an unknown factor can sometimes be dramatic. If we are lucky enough they just change the size of the effects of the focal factor on the output variable. However, we must be prepared for the possibility that the influence of an unknown factor could sometimes even be qualitative, causing a change from, e.g. positive to negative effect.

For example, male students infected with Toxoplasma show worse reaction times than the noninfected controls when the test is administrated by a male PhD student and better reaction times when it was administrated by a female PhD student. Such results are rather difficult to publish.

How frequently does such phenomena occur?

RM: This is rather common phenomena. This could be one reason why outcome from behaviour experiments differ widely depending on location of animal holding area, air-conditioning and other facilities within holding room as well as different batches of animals likely exposed to different immediate environments. Another very important factor that counts in immediate environment is the cage-mate. However, the stress-sensitive environment of small laboratory cages and single/group housing of animals makes the most significant impact on these phenomena. This in turn makes animal-behavior even more vulnerable to small changes in wider environment.

JF: I think that in the field of ethology such phenomena are very common. However, their existence is infrequently recognized as ethologists only rarely repeat their experiments.  When they the experiments are repeated and different results obtained, they usually have difficulty publishing their results in scientific journals. Possibly, the current trend of uploading preprints into public preprint archives could at least partly solve this problem and could protect science against the serious problem of the so-called ‘drawer effect.’

To achieve this, we must somehow motivate researchers to make negative or controversial results of their studies publicly available, in the preprint archives. To succeed in this, we should probably at least partly shift the centre of gravity of the scientific evaluation from counting the number of papers a researcher has published to counting the number of times the work has been referenced.

Currently a serious problem with the publication of negative and controversial papers is that some important preprint biological archives have relatively strict and very subjective criteria regarding which preprints can and which cannot by uploaded to the archive.

What can be done to overcome this?

RM: I think the most important step is to spend the first few days daily observing the animals and critical assess them once they are habituated to an experimental environment. It is also important for the experimenter to familiarise themselves with the animals, through minimal handling. Before starting an actual experiment it is worthwhile to run a small batch of control animals to study and record their baseline behaviour to assess whether that matches published reports of similar behaviour.

These initial few steps will provide an important clue about the general wellbeing of animals, as well as readiness for the behaviour-testing. Another major improvement could be to change the housing environment to minimise animals stress-sensitivity to lower the possibility of stress-induced variation in behaviour.

This could be done at several levels, starting from immediate housing in cages, a holding area to animal facility and location of the animals-house. Thus, it is critical to keep in mind animal’s well being and sensitivity while constructing animal facilities.

Since animals’ wellbeing strongly influences their behaviour, it is likely mandatory to assess the initial behaviour of an experimental animal, for any research study, not only for behaviour experiments.

JF: We need to try to standardise the conditions of our experiments as much as we can, preparing for the large number of unknown factors that could both quantitatively and qualitatively affect the experimental result. We must always consider these phenomena during the interpretation of our results, as well as somebody else’s experiments.

Most importantly, when we try to answer any scientific question, we should not rely on the result of just one type of experiment, but test our hypothesis using several experimental approaches and in many independent ways. Only then can we carefully count and weigh the evidence for and against the hypothesis.

Please explain why your research article published on F1000Research is an important contribution to this field of animal research?

RM: This research article shows how change in the wider environment, i.e., next-door construction of a building affected behaviour of experimental animals, adult male Wistar rats, for a specific time-window. We noticed distinct anxiolysis (non-anxious behaviour) in stressed animals exposed to early-life maternal separation stress and enhanced innate aversion to predator odor in toxoplasma-infected animals. These results are surprising and the exact opposite to what is widely reported on stress-induced anxiogenesis (generation of anxiety) and toxoplasma-induced loss of predator-fear.

This distinctive reversal in the rats’ behaviour could be recovered soon after the building-construction was over. Thus, the robust reversal in behaviour coincided specifically with the phase of the building construction. Although an accidental finding, our study highlights the dominant effect of the bigger environment beyond the animal facility that can directly affect behaviour of experimental animals. This study also shows the sensitive nature of animal’s behaviour as well as animal’s general well-being, which is most often reflected through behavioural output.

This study is important because it focuses on animal researcher’s most important resource, i.e. the experimental animals and their surrounding environment. Every experiment completely relies on the experimental subject since they are the primary source of all data. So, if our animal subjects are doing well, and are not overtly sensitised to their environment, their behaviour will likely be reliable as well. This might also help reduce the wide variation in behaviour readouts across different experiments.

JF: This paper is a wonderful illustration of the discussed phenomenon, namely of the fact that an unknown factor can qualitatively affect the output of ethological experiments. The fatal attraction phenomenon; the change of natural fear of Toxoplasma-infected animals to the smell of a cat to become attracted to a cat’s smell, was widely studied in many laboratories in the past 18 years. Its existence has been demonstrated not only in mice and rats, but also in chimpanzees in 2016 and men in 2014.

In the present F1000Research paper the research team, which also included the current principal figure of the fatal attraction research, showed that the influence of an unknown factor could change the attraction back to the fear.