Embedded Files
Contributing your data
Contributing your data
(What you do)
1. Read the data management plan
2. Tell us about your dataset
3. Upload your data
What happens next
What happens next
(What we do)
4. We process your data
5. We integrate it into SWARMatlas databse
6. test hypothese of collective movement
What you do
What you do
1. The data management plan
1. The data management plan
For details on how data is stored and used read the data management plan (DMP) - https://drive.google.com/file/d/1sIHUBSIt9iJ3VZ4scOldUqBVRwt31FfD/view?usp=sharing
If the DMP is too long, the key points are:
SWARMatlas is a private data repository accessible by invite only.
all data-providers are given ‘put-only’ access meaning you can upload (but not download) data.
no Data will be used in publications without first contacting the data-provider to discuss how your contribution will be acknowledged.
data is stored in a SharePoint OneDrive hosted by Swansea University. For details please see the data formatting tab or click the link.
2. Tell us about your dataset
2. Tell us about your dataset
For each dataset, please complete a short submission form - Dataset Submission Form – Fill in form
This enables to process the data understanding the full context and gives you the chance to explain aspects of the data not covered in the associated readme/reference data files. This also makes integrating data into the SWARMatlas database easier and enables easy categorisation and exploration of the data. The form covers the basics like:
Tracking method
Sampling interval
Study location
etc.
Additional notes
Tracking data can be difficult to work with and the accuracy/precision of data may vary within a collection period. If there is any information that may be important to note when interpreting a dataset please include it in the readme file or submission form. For example:
“In the first 30 seconds of each dataset, the subjects are acclimatising to new conditions”
“Due to technical issues, sampling frequency is inconsistent from minutes 13-15”.
“At minute 3 of each dataset a predator cue is presented”
3. Upload your data
3. Upload your data
Before uploading, please review the tips below.
General
Please upload data in a folder with the same filename provided in the form above.
Raw data
There is no expectation that significant changes must be made before uploading. Any transformations necessary will be done by us during data processing
Any data uploaded will be kept in its original format and any necessary transformations will be saved in the same folder as “filename_TidyData” preserveing the original formatting.
Associated data
Each dataset should be accompanied by a readme or reference data file that explains:
corresponding publication and authors
list of files
brief description of data, specifically describing each column if necessary
For datasets with individual-level data (i.e. age, sex, weight etc. of each individual), if available, this ‘reference data’ should also be uploaded.
Follow the link below to upload. Thank you for your contribution!
What we do
What we do
4. Data processing
4. Data processing
Data processing
All data processing takes place in R. All code, raw and processed data, and outputs are kept on the SWARMatlas platform, so every result is secure and can be traced back to the original data.
Each dataset goes through four processing steps:
Tidying data (data validation)
Plotting tracks (visual validation)
Sensitivity analyses (defining collective movement events)
Calculating collective movement metrics (using swaRmverse)
After this dataset-level processing, the data enter database-level processing, which will be described at a later date.
1. Tidying data
Datasets may arrive in many different formats. To make the pipeline consistent, each one is first “tidied” into a standardised format. This includes:
renaming columns to a standard naming convention (e.g. x, y, id, t)
converting unsupported file types (e.g. .txt) into accepted formats
converting wide-format data (e.g. x1, y1, x2, y2…) into long format, with one row per individual per time step (x, y, id, t)
setting the parameters used throughout the pipeline, based on a preliminary look at the data, any associated documentation, and the submission form
2. Plotting tracks
We plot the tracks of every dataset. This is a visual check that catches problems automated checks may miss. This confirms that we have understood the data and that the calculation of collective movement metrics is accurate. In particular, we check that:
there are no unusually large jumps between consecutive points, which can indicate tracking error or inconsitent sampling frequency.
individual identities are consistent through time, with no swapped or merged IDs
separate events, such as different trials or recording sessions, have not been combined into a single track (or vice versa)
tracks fall within the expected area and have a plausible scale, which confirms that units and coordinates have been interpreted correctly
Track plots also provide a permanent visual record of the raw data. Every comparative result in SWARMatlas can later be linked back to the raw movement data it came from, which makes our analyses easier to check, interpret and present.
3. Sensitivity analyses
Not every movement of mulitple individuals constitutes a collective movement event. Animals may be resting, moving in different directions or at different speeds. To make appropriate definitions of what constitutes a collective movement event, certain thresholds (e.g. speed & polarisation) must be set. We run sensitivity analyses to choose suitable thresholds for each dataset.
4. Calculating collective movement metrics
For each collective movement event, we calculate a standard set of group-level metrics with the R package swaRmverse. These include for example, group speed, how aligned individuals are (polarisation) and how they are spaced. Because the same metrics are calculated in the same way for every dataset, collective movement can be compared directly across species, environments and tracking methods.
5. Integrating into SWARMatlas
5. Integrating into SWARMatlas
Now, each dataset takes its place in the wider SWARMatlas database. Its details, such as species, setting, tracking method and location, are added to the SWARMatlas catalogue, which:
makes the dataset findable, so researchers can discover it by species, method, setting or region
adds it to the summary statistics and maps on the 'Explore the Data' page
makes it ready for comparative analysis, because each dataset is categorised and processed in the same way, so we can select exactly the datasets a question needs (for example, only fish, only field data, or only GPS-tracked groups) and compare their collective movement metrics directly.
6. Testing collective movement hypotheses
6. Testing collective movement hypotheses
A large, standardised database lets us test hypotheses that no single study can address. These hypotheses fall into two broad groups:
Mechanistic hypotheses (i.e. how animals move together): the mechanisms that coordinate a group. For example, does the sensory information available to individuals determine where they position themselves within a moving group?
Functional hypotheses (i.e. why animals move together): the adaptive benefits that have shaped collective movement through evolution. For example, is avoiding predators a common driver of collective movement across species?
Additionally, by comparing many species, environments and tracking methods, SWARMatlas can reveal which patterns of collective movement are general, and which are specific to particular animals or settings.
Page updated
Google Sites
Report abuse