1/2 BEES2041 Data Analysis for Life and Earth Scientists Practical report 2 Independent data analysis Background In recent years, there has been a very strong focus on the need for science to be reproducible. This means that all the methods (in the laboratory and field) are presented in enough detail for another researcher to be able to repeat the work and check the conclusions. It also means that the data collected, and all analytical methods used are also presented in a public forum. This movement has resulted in scientific journals frequently requiring authors to publish their data and all code used to process, analyse and visualise their data on publicly available platforms. Support for open science comes from the desire to have publicly funded research able to be scrutinised by anyone, for data and methods to be shared freely to advance science more quickly, and from those involved in synthetic research (i.e., meta-analysists who combine many studies to gain an overall understanding of a problem) needing to assess the reliability of a single study. The rare cases of scientific fraud can also be more readily detected when data and methods are openly available. Further reading on reproducibility in environmental sciences • Powers et al. 2018 Ecological Applications link • Liu et al. 2019. Eos link Your task Open science requires the sharing of data sets and the code required to process, visualise and analyse the data. When groups of researchers are working on the same problem, they also need to share their work prior to the study’s completion and publication. To give you experience in producing a document that would allow open sharing of all analytical methods, you are required to prepare notes and code that would detail each of the steps required to address a single hypothesis Step 1) Choose one of the two environmental science scenarios below (Frog ID or Waterbird survey) Step 2) Create a document (R Studio Notebooks are ideal for this), that will include: a) Notes to introduce the question and detail the hypotheses being tested (1-2 paragraphs) b) Notes and code to load any R packages that you will require c) Notes and code to import the data set and extract relevant parts d) Notes and code to create a graph that can visualise the how the data addresses the question e) Notes to describe the analytical methods (1-2 paragraphs) f) Code to formally analyse the data to address the question g) Notes to interpret the results (1-2 paragraphs) The notes should include enough detail for a co-worker to be able to understand what a given piece of code will be doing (imagine they are just learning data analyses like you are). 2/2 1. Use of citizen science to monitor frogs across Australia FrogID is a successful citizen science program led by Dr Jodi Rowley (Australian Museum and School of Biological, Earth and Environmental Sciences UNSW). Members of the public record frog calls on a smartphone app which are then uploaded to contribute to a large, nation-wide data base of frog species distributions. Read more at www.frogid.net.au The effectiveness of citizen science programs to document flora and fauna depend on how well the public sampling represents the range of habitats and areas available. To test this, Callaghan et al. 2020 recently contrasted the number of frog species detected by FrogID (in each 30’ grid cells across Australia) to the number of frog species known from a data set collected by frog experts over many decades. Your question: • Does the data on species richness of frogs from FrogID accurately predict the known species richness from previous expert data sets? Data: • The data sets from Callaghan et al. 2020 are provided on Moodle. The number of frog species are provided in each 30’ grid cell. • Hint: you will need to join data sets to get all data in the same spreadsheet (see Combining data sets on Environmental Computing) Required reading: Callaghan, CT; JD Roberts, AGB Poore, RA Alford, H Cogger & JJL. Rowley. 2020. Citizen science data accurately predicts expert- derived species richness at a continental scale when sampling thresholds are met. Biodiversity and Conservation 29: 1323– 1337. link Rowley, JJ and CT Callaghan. 2020. The FrogID dataset: expert- validated occurrence records of Australia’s frogs collected by citizen scientists. ZooKeys, 912, p.139. link 3/2 2. Aerial Waterbird Surveys The Centre for Ecosystem Science at UNSW (led by Professor Richard Kingsford) has been monitoring waterbirds across Australia’s wetlands since1983. Read more at the Centre’s website. Your question: • Is there a decline in numbers of waterbirds observed across the years surveyed? • Address this question for a bird species of your choice in one of the wetlands that has many observations. • Wetlands with many observations include: Lindsay-Walpolla-Chowilla, Lower Lakes & Coorong, Lowbidgee, Kerang, Menindee, Gwydir, Booligal, Narran Lake, Lake Moondarra, Coolmunda Dam, Barmah- Millewa, Hattah Lakes, Cumbung, Darling River, Mullawoolka Basin, Macquarie Marshes, Murray River, Paroo Overflow Lakes, Chow_Lin_Wall, and Fivebough Swamp Data: • The survey data are freely available at https://aws.ecosystem.unsw.edu.au/ To explore this yourself, you need to register with a user name and password • While you can export parts of this data set, we have provided the entire dataset in a spreadsheet for you on Moodle (WaterbirdSearch_20200311-1247b.csv). Please use this data file as we have added a year variable that you will need for your analyses. • Hint: You will need to be familiar with the functions in dplyr for selecting rows. Help on Subsetting Data is on Environmental Computing. Required reading: Kingsford RT & JL Porter 2009. Monitoring waterbird populations with aerial surveys—what have we learnt? Wildlife Research 36: 29–40. link Kingsford RT, G Bino & JL Porter. 2017. Continental impacts of water development on waterbirds, contrasting two Australian river basins: Global implications for sustainable water use. Global Change Biology 23: 4958– 4969 link 4/2 Assessment Submit the notes and code as a R Studio notebook (i.e. .rmd file) or R script for those few students that had troubles with notebook on older Mac laptops Upload those files to the Practical Report 2 Assignment in Moodle (found in the Assessment section) Due date: Week 8, 11:59 pm, Monday 6th April Marking: a) Notes to introduce the question and detail the hypotheses being tested (10 marks) b) Notes and code to load any R packages that you will require (5 marks) c) Notes and code to import the data set and extract relevant parts (5 marks) d) Notes and code to create a graph that can visualise the how the data addresses the question (25 marks) e) Notes to describe the analytical methods (15 marks) f) Code to formally analyse the data to address the question (25 marks) g) Notes to interpret the results (15 marks)
