The Data: Enrichment is an animal husbandry practice that that aim...

The Data:
Enrichment is an animal husbandry practice that that aims to improve a captive animal’s quality of care. There are a range of stimuli that can affect a captive animal’s psychological and physiological wellbeing. So far, there has been little investigation into the effect of the acoustic environment on the animal’s wellbeing.

You are asked to report on the results of an experiment conducted in a large bird-holding facility investigating the behavioural consequences for zebra finches (Taeniopygia guttata) kept under three different husbandry conditions. The data can be found in the file “acoustic.enrichment.data.csv”. A population of 100 birds was maintained in a common breeding facility. All birds were captive-reared and had never experienced background noise other than vocalizations of conspecifics and ventilation noise. For each trial, 10 birds of each sex (SEX) were randomly selected and released into a large aviary (i.e. 20 individuals in total), where they were exposed to one of three treatments (SOUND) for two days: 1) ambient sounds in the aviary (i.e. control); 2) human voices talking at 50dB; or 3) classical music (Vivaldi’s ‘Spring’) played at 50dB. At the end of the trial, birds were returned to their normal housing facility for at least two days, before the start of another trial. This was repeated 30 times in total (TRIAL), so there were 10 replicates for each sound treatment. The same individual bird [BIRD) was used in multiple trials; each bird was used for exactly six trials (i.e. on average once every five trials), but which group it was in or what treatment it experienced each time, was random.

On the second day, the behaviours of the 20 birds in each trial were recorded over a two-hour observation period. The observer recorded the frequency (% of observation time) that each individual spent displaying social behaviours (e.g. allopreening, interactions with conspecifics; [SOCIAL]) and fear behaviours (retreating to high perches, clumping together; [FEAR]). All birds within the aviary were individually marked with one coloured leg ring on each leg and a thread of corresponding colour hanging from each ring to ease quick identification; male and female zebra finches can very easily be distinguished from their markedly different plumage colouration. The birds were observed using an instantaneous sampling protocol: every 3 minutes a timer beeped and then the observer identified each bird in turn and recorded which of the target behaviours they performed at that time. Once all 20 birds had been located and their behaviours had been recorded, the observer waited for the next beep.

The key question is: Does acoustic enrichment affect the frequency of different behaviours shown by the zebra finches?   Each row of the data table contains the frequencies of behaviours over a two-hour period in individual zebra finches. Since the body mass of zebra finches (MASS, in g) or the sex of the bird (SEX) might influence behaviour, these data are also reported. So, for example, in rows 2-11 of the data table, the behaviours of 10 different female zebra finches that were played back the ambient sound in the first group that occupied the aviary (TRIAL= 1) are reported, rows 12-21 have the corresponding measurements for males from the same trial. Rows 22 to 601 contain the results of the other treatments and trials.

Guidelines:
It is highly recommended that you go back through your lecture notes and personal logbooks for the course, either before or as you are working on this assignment, to make sure that you are applying the strategies discussed in class for effective scientific communication. It is expected that you will have read all of the information provided. It is also a good idea to extrapolate feedback you received on R scripts and writing assignments The paper should be no more than 3000 words, excluding the abstract, figure legends, tables and references. The number of figures should not exceed that typically encountered in a Proceedings of the Royal Society B paper. The report should be double spaced, written in 12 pt font and include page numbers.

Title Page
Include an informative and interesting title, written on a separate title page that includes your name and contact details.

Abstract (300 word limit)
Follow the guidelines that you learned for writing abstracts. Make sure to set up the background to the problem, clearly state the objectives, briefly describe the methodological approach, outline the main results, and provide a concluding sentence.

Introduction and Literature Review
In the introduction you should adequately set up the problem of why this type of study is interesting and important, and discuss what has been done before and where gaps in knowledge might lie. Briefly introduce the overall problem, and discuss the relevance and scientific importance of this type of experiment. Make sure that you explicitly justify the motivation for each variable that is included in the experimental design (e.g. even if you find a nonsignificant result, you should include the motivation for testing that variable initially). End the introduction with a clear statement of objectives, stated in the form of questions that you will address using the experiment described. You should structure the rest of the paper accordingly.

Read and describe sufficient literature that you can adequately review the problem and identify any gaps in knowledge that can be addressed by this study but you don’t have to include as comprehensive a literature review as you would see in a published paper (where there would often be 50 or more references). You can include as many references as you think is appropriate but as a guideline you should aim for at least six and not more than twenty.

Include a statement on the ethical implications of using animals for this study. You should explain why this type of experiment might be important for managing captive populations and concentrate on how you would decide whether the welfare implications to the animals justify this type of study.

Methods
The methods section will be briefer than a published paper because you won’t have all of the intricacies of where samples were collected from but describe any details about the sampling design that you can from the write-up above and from the data provided. Most of the methods should focus on the statistical analyses that you conduct. This should be the same level of detail as in a typical journal article in Proceedings of the Royal Society B. Make sure to describe the response variables used (hint: there are two in this dataset) and what factors and interactions you are considering (either as fixed effects, random effects, or covariates; this will depend on how you set up your objectives). You should analyse the data according to what you learn during the Advanced Linear Models course.

You should use R to analyse your data and make graphs [Recommendation: as you learned in the Beckerman book, it is always best to plot your data before you do the statistics, so you can visualize whether the results are meaningful (i.e. that you have used appropriate statistics); this is particularly true for detecting when you might expect to see significant interaction terms].

Since there are factors other than the treatment that might influence the variance in responses, you should take the GLMM approach that you have learned in the Advanced Linear Models class. You will need to decide whether each possible factor is one that you are interested in in its own right (i.e. a fixed main effect or a covariate) or whether you just want to be able to account for the variation that it causes (i.e. a random effect). You should also take a model selection approach to assessing significance of the main effects and all interactions, as you will have learned in your statistics class. Although posthoc tests might be used to determine significance of any differences found for variables with more than 2 levels (or for interactions), you are not expected to implement these for this assignment, since you haven’t been taught this and it is not straightforward for GLMMs. You should describe what is driving any differences found based on the model coefficients and appropriate plots. If you want to make sure that the control (“Ambient”) is treated as the reference level of the variable SOUND, you can use the relevel command in r.

You should also use what you’ve learned about power analysis in the experimental design lecture. Specifically, you are asked to retrospectively determine the sensitivity of your experimental design with regard to the main fixed effects (i.e. each variable separately, using a power analysis method appropriate for a two-sample t-test). What effect size could have been detected given the sample size used and assuming a significance level of 0.05 and a power of 0.8? Use this to interpret the results you found and describe whether you think that the sample size could have been changed to either reduce use of animals unnecessarily or to increase the sensitivity for detecting a difference. While more sophisticated power analyses are possible, you are not expected to go beyond what you covered in the experimental design lecture. This is in fact realistic: power analyses usually assume a very simplified version of the real analysis. You should think about these findings in the context of the three R’s (i.e. reduce, replace, refine) and discuss the implications in the discussion section of your report (but describe the relevant methods and results in the appropriate sections).

Results
The results of the analyses that you conduct should be clearly described, using figures and tables, but also in words to highlight the results that you want the reader to notice.   The results are the most important part of the paper but won’t necessarily be the longest section. Make sure to include comprehensive captions for both figures and tables. Make sure to describe your final model in the results section [based on the model selection results], before explaining what the model predicts.

Discussion
The discussion should interpret the results in the context of other studies that have been performed and in terms of what else might be done to take the research further (including recommendations for how the experiment could have been revised based on your power analysis). You should start with a summary of the main results before describing any caveats or limitations. The discussion will be shorter than for most papers as you are not expected to comprehensively review the literature. You will want to focus on the ethical implications of the results, how they relate to previous studies and provide a critical analysis of the experimental design.

It is good to balance the Introduction and Discussion so that you can go back to the gaps in knowledge that you raised in the introduction and evaluate whether the data have filled these gaps or whether more work needs to be done. When you are setting your results in the context of other studies, make sure to include a critical analysis of both your experiment and those of others (i.e. what the results contributed to existing body of knowledge, along with limitations of the experimental design and statistical analyses). End with a discussion of what could be done next to advance the field of knowledge.

Bibliography
Include a bibliography and make sure to cite in-text references in the correct format. Use the style rules for the Animal Welfare Journal (i.e. the one published by the UFAW) as a guide.   Make sure to critically evaluate the references that you use, both in the introduction and in the discussion. All references cited should be listed in the bibliography and all references listed in the bibliography should be cited in the text.


Appendix
Include all of the R code that you use in an appendix; make sure to annotate it sufficiently that you will remember what you did but also to explain to someone reading it what each step is for and how to interpret the results. The appendix should be the raw R code rather than an HTML report but you should also submit an HTML report to MOODLE.

Timelines:
The assignments will be due before the semester break (Dec. 15th). You can start on looking up and reading literature during the statistics course but at least some of the references you choose are likely to change based on the results that you find. It is fine (encouraged) to discuss the approaches used with others but you should write the paper independently and write your own R code (including annotations). There will be two help sessions (Dec. 6 and 12); it is a good idea to try the analyses before the first help session so that you know what to ask about.

Evaluation:
Marks will be based on incorporation of skills that you have learned across the first term, so you will be evaluated on: Quality of Writing; Paper Structure; Use of Primary Literature; Data Presentation; Power Analysis; Statistical Analyses; Use of R (you should use this for your statistics and to make your figures); Critique of Experimental Design; Scientific Interpretation; and Critical Analysis abilities