This project is about exploring the advantages and disadvantages of different plant rooting structures to achieve different outcomes for the plant.
All the background is provided in the four papers available. You should definitely have a good look at those papers before you begin, especially the first three, which are pretty short.
Your task is to address one of the following questions:
1. What is the best plant rooting strategy for finding wet cracks in underlying rocks within a single wet season to provide moisture through subsequent drought periods?
2. What is the best plant rooting strategy for building maximum shoot biomass over a single wet season? (Shoot biomass is total above-ground biomass.)
The simulation model is provided as an R script, called “SCIE5500 sim.R”, and you will use another script called “rooting strategies SCIE5500.R” to run and interact with the simulation. You don’t need to even open “SCIE5500 sim.R” (though you are welcome to look at it if you want) and for this project you definitely shouldn’t change anything in that script. But you do need to make sure that both of these scripts are in one folder, and that you change the ‘working directory’ of R to be that folder. You then open the “rooting strategies SCIE5500.R” script and run it. You should see an animation of a plant’s roots growing in a shallow soil over a single wet season. When the simulation ends (at the end of the wet season), the last output in the R console window is the final shoot biomass achieved, and whether the roots discovered a crack in the underlying rock (TRUE/FALSE). You should probably run the simulation in the standard R Gui rather than Rstudio for example, because animations don’t work very well in Rstudio. You’ll still get the same results in Rstudio, but you won’t see how the plant grows as well.
The model is stochastic; for a given genotype/strategy you will get different phenotypes (structures) and thus different biomasses and successes in finding cracks. Therefore, to properly evaluate a strategy, you will need to do a number of different simulation runs using that strategy (ie replicates). This is easy to automate in the script – just change the number in the ‘replicate’ function in the second last line. The final output will then be the final biomass achieved, and whether the roots discovered a crack in the underlying rock, for each of the simulation runs. (Be careful though, it may take quite a while to run a large number of simulations, especially if you are using a laptop, or an older computer.)
To change the rooting strategy, you just change the values of the parameters defined in the parameter list in “rooting strategies SCIE5500.R”. For example, to change branching angle from pi/2 to pi/3 you would change the value for the ‘branchang’ parameter from pi/2 to pi/3 in the script. You certainly don’t need to change all the parameter values – better to just choose two or three that you think will be most important.
The papers talk about using evolutionary optimisation algorithms to address these two problems. You do not need to do anything like that. Rather you can just try out a number of different strategies and work out which one(s) is/are the best for your question. You can do this ‘by hand’, running simulations for different strategies one by one and recording results, or you can maybe automate the procedure using code if you wish – whatever is most efficient for you. You could evaluate and compare a number of well-defined different strategies that you define in advance. Or you could look at two or three parameters you think will have an important effect on the outcome, and look at the effect of varying the value of these parameters over a range of values.
The number of replicates should be enough for you to be sure that differences are ‘real’ and not just due to stochastic chance – you’ll need to do some kind of statistical testing to be sure. This could be a t-test or ANOVA to test for differences in biomass between two or more strategies, or a proportions/chi-squared test to test for differences in crack finding between two or more strategies. You could also consider a regression to test whether results change significantly as you change one parameter over a number of values. At the very least, you should test whether your final strategy is better than the initial strategy that you start with. It is quite easy to do a large number of replicates by changing the number in the ‘replicate’ function and they will be saved automatically to csv for file for analysing/plotting later.
You will submit a report for this project, which should be no more than 1500 words (not including tables and figure captions). It should be concise, clear and accurate – if you can present a clear and accurate description of a sound study in less than 1500 words then please do! The report should have four sections: Introduction, Methods, Results and Discussion.
• The Introduction should be quite short – it should state which question you chose to address and clearly explain how you chose to address it – that is, which parameters or strategies you chose to experiment with and why you thought they would be important. You should finish with some clear hypotheses about how you predicted success (biomass/crack finding) would be affected by the parameters you varied or strategies you tested. Normally you would provide some background in an Introduction, but you don’t need to give any background in this report since everyone reading it will know what the project is about.
• Methods should clearly define exactly what strategies/parameters you tested. For example, if you keep all parameters at their original values, but just vary branching angle, then give a table of all original parameter values, and another showing the various values of branching angle that were tested. It should be clear how what you did tested your hypotheses. You should explain here how you tested whether differences were significant. Note that it should be possible for someone to do exactly what you did based on your Methods.
• Results should present the results using figures, tables and/or text, as appropriate. It doesn’t have to be long. For example, if you look at biomass, you might show a figure with a few example pictures of the root structures of the most successful strategy you found, compared with a less successful strategy, along with a boxplot showing the ranges of biomasses achieved for each tested strategy. You would then maybe just need a paragraph or two of text referring to the figures. You should also present the results of your statistical tests of difference clearly and concisely – a few p-values in the text or within a table or figure is probably enough, definitely do NOT include whole ANOVA tables or anything like that.
• Discussion should include your main conclusions, along with thoughts about what you thought was interesting or surprising or exactly as expected, possible implications, discussion about the limitations of what you did (or of the model itself), and what you would like to do next or would recommend others to do next if the study was to continue.
You are welcome to work with other people as much (or as little) as you like. But you should submit your own report, and you should ensure that you either address a different question or use a different method to the people you work with (ie address a different success measure (biomass vs crack finding), consider different parameter values or test different strategies). Cases where reports or tested strategies are very similar will be penalised.
When the report has been submitted, you will then be directed to read and evaluate three other reports, and then self-evaluate your own report. You will get quite specific directions and criteria for the evaluations, based very closely on the directions above. Doing a thoughtful, fair and reasonable job on the evaluation will provide a significant proportion of the marks for the mini-project. The ultimate decision on the mark awarded for the project will be mine, but I will take peer- and self-evaluations into account when deciding. The final due date for peer assessment will be 14th November, so please plan some time for that as well.
There is a new forum for questions and discussion about the mini-project. Please use this in preference to direct email if you think your question may have any general relevance, so that everyone has access to questions and replies.
The assessment guidelines to be used for the mini-project are those stated in the unit outline, and repeated below. I will apply these based very closely on the specific directions above.
Assessment Guidelines: In general a mark of 80% should be awarded when an assessment or part of an assessment clearly meets all requirements. Extra marks can be awarded for extra insight or originality, up to a possible 100% for outstanding work. Marks can be lost for flaws, errors, omissions, lack of clarity etc. So for example:
less than 50% – doesn’t meet requirements
50% – barely meets requirements, with some fairly major flaws, omissions, lack of clarity
60% – meets the basic requirements, but with minor flaws and lack of clarity, or with some more major flaws
70% – meets requirements but with minor flaws or lack of clarity
80% – clearly meets all requirements
90% – clearly meets all requirements, with some extra insight or originality
100% – very clearly meets all requirements, with significant extra insight and originality