Practicum I Pau Barquin Definitiu

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Research Project Progress Report

Project Name: Tracking The Travels Of North Atlantic Blue Sharks


(Prionace Glauca): A Satellite Telemetry Study Of
Horizontal Movements And Potential Hotspot
Identification

Student Name: Pau Aleix Barquín Barrafón

Report Submission date: 01/07/2023

Summary

This progress report covers part of the practical and documentation work of a project that has the aim
of inferring the spatial distribution, horizontal movements and migration patterns of the North Atlantic
blue shark populations. This will be accomplished by using data that comes from satellite telemetry
tagging and will encompass the processing and analysis of this data. The procedures covered by this
report include bibliographic documentation and data analysis using R programming language.
- State of art

Satellite telemetry has reinvented the study of marine megavertebrates, including mammals, birds,
fishes, and turtles. The functionality of this tool involves automatic recording of remote location’s
position through satellite technology. This, has enabled researchers to track the movements and the
habitat preferences of these animals despite the remoteness and the wide distance that they cover,
providing valuable insights into the understanding of their ecology and life history. The first satellite
tracking studies were published in the 80s thanks to a pioneering development of instruments and
infrastructure. The posterior technological advances yielded to a huge expansion and established the
foundations for a very fruitful and operational research field (Hart and Hyrenbach 2009).
Nowadays, the advancements that makes possible the expansion of this field are mainly associated
with tag miniaturization, which enables tracking a larger diversity of animals, also with improvements
in battery technology, in software and in data recording and processing, which led to a big reduction
in data loss. As a result, we can now obtain high-resolution, long-term data at small spatiotemporal
scales, while also recording location data in real time and along with other ecological variables
(Hussey et al. 2015).
The understanding of all this knowledge enables us to perceive the wildlife world in ways never seen
before and is usually straight related with the willing to improve marine ecosystem conservation. This
research field has demonstrated to be successful in influencing the management and the
conservation policies of marine species. The translation of animal tracking data into conservation
protocols can be done by many ways such as the identification of critical habitats and migration
corridors, by supervising the management methods and also by working with a worldwide data
sharing among researchers (Hays et al. 2019).
The benefits of using this technology are more than demonstrate, although, these benefits usually
come with significant problems such as tagging failures, limited battery life, transmission issues or a
Research Project Progress Report

high economic cost, leading sometimes to weaker study design, reduced sample sizes and inferior
statistical inference (Hebblewhite et al. 2010). Also, there are some ethical considerations associated
with this type of research; Tissue damage, premature tag loss or decrease in the animal’s swimming
capacity are the most common problems found in the external tag attachment. Although, this highly
depends on the type of species and on the context. The studies need to be the least invasive possible,
in order to achieve this, some recommendations have been suggested like the realization of pilot
studies to quantify the possible effects of the tagging before starting the complete field work (Jepsen
et al. 2015).
The continued development and application of advanced tagging techniques will be key to improve
our comprehension of marine ecology and supporting conservation efforts to protect these vital
environments.
- Hypothesis and objectives

We started this project with the initial hypothesis that the tracked North Atlantic blue shark populations
might exhibit standardized common movement patterns that could be related to environmental factors
and with the animal’s life history as it is discussed in previous works.
These movements can be monitored by satellite tagging to see both the different routes established
and also other key parameters such as the speed and the tortuosity along the track. Move persistence
index analysis will provide us with crucial information about the migration paths and the travel patterns
of the sharks across the different Atlantic regions. The low move persistence movements will reveal
the encirclement and detour zones so we can identify the behavioural tendencies and the possible
habitat preferences.
This information can be used to demonstrate the second hypothesis, by which we suggest that these
common movements may result in high activity areas with a large shark congregation that are referred
to as hot spots. The process will be based on pooling and relating the previous results with existing
environment variables such as the presence of high productivity areas, current systems or
topographical features. The determined hotspots will mean the settlement of crucial marine
ecosystems where mating, breeding or feeding by the blue shark populations is carried out, also
meaning high activity zone for other organisms.
Our objectives are:
- Objective 1: Through the processing of satellite data and with computational tools we have the
aim to analyse the routes and the space-use patterns of these sharks as well as possible
congregation hotspots.
- Objective 2: By characterizing the horizontal movements we aim to establish blue shark
foraging, breeding or mating hotspots along the North Atlantic region.
- Objective 3: Identify the changes in spatial distribution along the long different seasons of the
year and reveal the migration routes.
Research Project Progress Report

Activities and Progress

- Search of documentation
During the initial weeks of the project, the focus was on gaining a basic understanding of the satellite
telemetry research field as well as familiarizing oneself with the fundamental aspects of the R
programming language. To establish a strong basis for the future analysis and interpretation of
satellite telemetry data, a pre-selected bibliography was consulted.
This previous preparation mainly turned around a manual for the R package AniMotum (Jonsen et
al. 2023), which contained the necessary information for the correct understanding of the applications
and the possible uses of the package with real data. For a further and detailed understanding some
links from AniMotum GitHub were also checked, this page provided comprehensive information about
the functions of the package, as well as access to an example data to facilitate practical application
of the package's uses (https://ianjonsen.github.io/aniMotum/index.html). At the same time, I was
expanding my concepts and overcoming my limitations with the R programming language by
consulting general manuals for beginners such as the one written by Emanuel Paradis in 2005.
To gain a better conception of fundamental concepts in the field of satellite telemetry, various papers
were consulted. One such example is Weng et al.'s (2008) paper, which provided insight into the
common methods used, data analysis techniques, the current state of research, limitations, and other
relevant aspects of the field. Moreover, Nakano and Stevens in Biology and Ecology of Blue Shark
(2008) provided an extensive summary of the blue shark biological data (age and growth,
reproduction, diet and distribution patterns). This book was consulted in order to gain a basic
knowledge about the specie and about the current research frontiers and limitations.
- Procedures and work plan used

Sharks studied and tagging


Initially, we had satellite telemetry data from forty-six blue sharks. However, we only included a total
of thirty-seven shark tracks because some of them were corrupted and with atypical values. The data
comes from sharks captured over several years: 2006 (three), 2007 (one), 2008 (two), 2009 (three),
2011 (three), 2014 (ten), 2015 (five), 2017 (ten) and 2020 (nine) amongst the North Atlantic
populations. Sharks were captured using a hook and line method, after which, a satellite tag was
attached to each animal’s dorsal fin. The model of tags used were the smart position only-transmitter
(SPOT) by Wildlife Computers, included between the Argos-linked satellite transmitters (SAT). Data
was collected and processed by the Argos DCLS, offering information such as latitude and longitude
coordinates, as well as location quality classification ranging from best to worst, with values of 3, 2,
1, 0, A, B and Z.
Research Project Progress Report

- Data processing
The raw tracks of the different years were obtained from the Argos system. This data was transformed
into a standardized format making possible the computational multi-analysis.
The project involved processing satellite telemetry data with a new R package called AniMotum. The
package is designed to be easy to use for novice researchers, while also offering advanced users
additional control through optional arguments. Its purpose is to facilitate quality control and help draw
insights from animal satellite telemetry data, as described in Jonsen et al. (2023). Some of the key
aspects that the package includes are the capability to fit State-space models (SSMs), which can be
defined as a time-series models that predict the future state of a system from its previous states
probabilistically (Patterson et al. 2007). In addition, the package enables the analysis of changes in
animal movement behaviour along satellite tracks.
All of these features make it easy to analyse satellite telemetry data, and also enable the extraction
of valuable information for inferring animal behaviour, habitat preferences, and movement patterns
of tagged populations.

TABLE 1. Mainly functions of the Animotum R package with a short


overview and a use example. Each function has optional arguments that
allow more control over the data. Adapted from Jonsen et al. (2023).
Research Project Progress Report

We applied a standard workflow from our package to process the tracks. In the begging, we only
examined data from 11 sharks to ensure the correct application of the package and familiarize with
the procedures, but later we merged all the data for a comprehensive analysis.
According to Jonsen et al. (2023) the standard workflow consisted in:
1. Introducing and reading the data into R
2. Transforming data into the correct format for the package
3. Choosing a movement process model (RW, CRW and MP)
4. Fitting the model to the data
5. Validation the fit of the model
6. Visualizing the track routes into the map and also the move persistence estimates of the track
7. Simulate and/or re-route the tracks
This package works fitting the data into 3 different predefined models called State-space models
(SSMSs). After applying the model, the fit can be measured and validated by using a set of residuals
calculated from the x and y values. The prediction residuals can be visualized as time-series plot,
quantile-quantile plots or as autocorrelation functions. Also, in order to approach of the model to the
data it is also important to pay attention into the confidence ellipses shown around each point of the
track, large ellipses mean a possible bias in the data.
During this project we used, as in other studies, the Correlated-random walk model (CRW) for
mapping the movements of the sharks and the move persistence model (MP) for mapping the
movements of the sharks in relation to the move persistence index. This index is another key aspect
of the package, ranging from 0 to 1 it reflects alterations in movement patterns based on the speed
and the direction of the track, this way, we can obtain information about the parts of the track intended
to migrate or to foraging. Also, we applied control over some other arguments in order to obtain more
accurate data, like a speed filter, where only speeds under 2 m/s were considered, and a 12 hours’
time step.
Although, some tracks were dismissed because of the abnormal values, some others could be treated
with a speed analysis in order to find the reason of the corruption. We used the traipse package,
which is included in the AniMotum package, and it incorporates the ‘track speed’ function (Karney
2013). This function calculates the speed (m/s) based on geodesic distance using input vectors for
longitude, latitude, and date-time. We estimated the speed for each point of the track in order to
detect the segments that were altered with abnormal speeds. After identifying these changed parts,
we were able to apply the necessary correction. Also, some outlier values in many tracks were filtered
to obtain more accurate data.
All the analysis was realized with RStudio, R version 4.2.2. AniMotum package version 1.1-03. The
software used was Windows 10 Pro with a processor Intel® Core™ i5-4210H.
Research Project Progress Report

Results

- Presentation of the data and results obtained


For every track, the correlated random walk (CRW) demonstrated to be more accurate for the data
than the random walk model (RW) as it is shown by the prediction residual diagnostics and the Akaike
Information Criterion (AIC) (Figure 4). In comparison to the RW model, the CRW prediction residuals
have no temporal trend and little autocorrelation (Figure 1), implying the CRW process model is a
better fit to the data. Also, confidence ellipse around the fitted and predicted location must be
examined for every track. The result obtained in the 202920 shark (Figure 2) can be extrapolated to
the other tracks, where there is a small ellipse around the location points meaning that we have a
(a) (b)

(c)

(d) (e)

(f)

FIGURE 1. Prionace glauca. Example of the model validation for one of the tracked sharks
(202920). Validation of the fit to the RW (a- prediction residual time series; b- prediction
residual quantil-quantil; c- prediction residual autocorrelation) and CRW (d, e, f) state-space
models. All residual plots generated using plot.osar function.

good fit for the model and we can assume the insignificant error. The routes that showed a larger
confidence ellipse around the locations were treated removing outlier values and parts of the track
that could be altered, those which, despite treatment, continued to have large ellipses were excluded.
Research Project Progress Report

This exclusion was associated with routes which had a poor data recording or with issues during the
transmission.

FIGURE 2. State-space model fit to 202920 shark with the corresponding 95%
confidence ellipse.

The map of the different shark paths (Figure 3) shows a general view of the many routes taken by
the sharks across different years of recordings. Each point in the map represents one remote location
detected and processed by the Argos DCLS, these points were overlapped with the North Atlantic
map. Overall, a total of 8187 location points were introduced, with an average number of 293 locations
registered per shark. Blue shark movements were tracked for an estimate recording period of 86 days
per shark. Besides, an average of 12 locations were collected each day from all the shark.
Although each shark was captured in a non-identical time and place, we still can analyse the general
patterns of space-use by populations though the years in the North-Atlantic region. As we can
observe (Figure 3), while the blue shark routes can cover vast distances of direct and straight
movement, there are also some regions where these animals have undirected and encirclement
movements. Between our tracked sharks most part of the activity takes place in latitudes from 30ºN
to 45ºN with also activity detected in the Cabo Verde Islands and in the nearby, in the Portuguese
southern coast and in Southern England.
Research Project Progress Report

FIGURE 3. Prionace glauca. Blue shark paths (n=37), each route is distinguished by a
different colour, the legend provides the corresponding shark ID for each colour-coded route.

The tracked routes from the blue shark were further analysed by inferring the move persistence index
(γt), which indicates the regions where the animal spends more (low γt) or less (high γt) time. With the
proposal of analysing this and drawing conclusions about the shark behaviour, we created a map of
the tracks interpolated with the index.
The map was generated by fitting the Move Persistence model into the location data that was used
previously, providing the same spatial distribution as it was shown in the blue shark path map in
Figure 4. The results show that the major part of the paths is represented with high γt values (general
average of 0.87), meaning long-distance travels with directed and fast movements patterns.
However, some certain regions are distinguished by having low γt values, suggesting crucial zones
where sharks spend more time and have detour and slow movements patterns. Identifying these
zones is fundamental for the understanding of ecological and biological behaviour of these animals.
Research Project Progress Report

FIGURE 4. Prionace glauca. 2-D tracks for the inferred move persistence (γt) of each
shark (n=37). Locations associated with low γt (purple) are indicative of slow, undirected
movements, while high γt (yellow) indicate fast and directed movements.

- Conclusions and recommendations


The results presented here are a fundamental part of the project. During this process, the tracks were
pre-selected, corrected and examined in order to generate the basis of this research work. These
maps exhibit valuable information about the main spatial distribution presented by sharks as well as
the zones where slow and undirected movements were performed. This information will be crucial in
the following steps of the project, where we will represent our data into kernel density maps in order
to have a better understanding of the data. This will be the last step and we will be able of drawing
conclusions and discuss the results along with previous studies information.
Research Project Progress Report

- Bibliographic references
Hart KM, & Hyrenbach, KD. (2010). Satellite telemetry of marine megavertebrates: the coming of
age of an experimental science. Endangered Species Research, 10, 9–20.
https://doi.org/10.3354/esr00238

Hays, G. C., Bailey, H., Bograd, S. J., Bowen, W. D., Campagna, C., Carmichael, R. H., Casale, P.,
Chiaradia, A., Costa, D. P., Cuevas, E., Nico de Bruyn, P. J., Dias, M. P., Duarte, C. M.,
Dunn, D. C., Dutton, P. H., Esteban, N., Friedlaender, A., Goetz, K. T., Godley, B. J.,
… Sequeira, A. M. M. (2019). Translating Marine Animal Tracking Data into Conservation
Policy and Management. Trends in Ecology & Evolution (Amsterdam), 34(5), 459–473.
https://doi.org/10.1016/j.tree.2019.01.009

Hebblewhite, M., & Haydon, D. T. (2010). Distinguishing technology from biology: a critical
review of the use of GPS telemetry data in ecology. Philosophical Transactions of the
Royal Society B: Biological Sciences, 365(1550), 2303–2312.
https://doi.org/10.1098/rstb.2010.0087

Hussey, N. E., Kessel, S. T, Aarestrup, K., Cooke, S. J., Cowley, P. D., Fisk, A. T., Harcourt, R.
G., Holland, K. N., Iverson, S. J., Kocik, J. F., Flemming, J. E. M., & Whoriskey, F. G.
(2015). Aquatic animal telemetry: A panoramic window into the underwater world.
Science (American Association for the Advancement of Science), 348(6240), 1221–
1221. https://doi.org/10.1126/science.1255642

Jepsen, N., Thorstad, E. B., Havn, T., & Lucas, M. C. (2015). The use of external electronic
tags on fish: an evaluation of tag retention and tagging effects. Animal Biotelemetry,
3(1), 1-23.

Jonsen, I. D., Grecian, W. J., Phillips, L., Carroll, G., McMahon, C., Harcourt, R. G., ... & Patterson,
T. A. (2023). aniMotum, an R package for animal movement data: Rapid quality control,
behavioural estimation and simulation. Methods in Ecology and Evolution, 14(3), 806-816.

Karney, C.F.F. Algorithms for geodesics. J Geod 87, 43–55 (2013). https://doi.org/10.1007/s00190-
012-0578-z

Nakano, H., & Stevens, J. D. (2008). The biology and ecology of the blue shark, Prionace
glauca. Sharks of the open ocean: Biology, fisheries and conservation, 140-151.

Paradis, E. (2005) R for Beginners – The Comprehensive R Archive Network. Institut des Sciences
de l’Evolution, Université Montpellier.

Patterson, T., Thomas, L., Wilcox, C., Ovaskainen, O., & Matthiopoulos, J. (2008). State–space
models of individual animal movement. Trends in Ecology & Evolution, 23(2), 87–94.

Weng, K. C., Foley, D. G., Ganong, J. E., Perle, C., Shillinger, G. L., & Block, B. A. (2008). Migration
of an upper trophic level predator, the salmon shark Lamna ditropis, between distant
ecoregions. Marine Ecology Progress Series, 372, 253-264.
Research Project Progress Report

Auto evaluation

While doing this practicum I had the chance of approaching to a professional scientific group and to
have access to real data from one of the most important predators in our oceans. The AniMotum
package demonstrated to be an effective and efficient tool for analysing satellite telemetry data,
especially, for the beginners in the field like me. The large variety of functions and parameters made
me able to explore and process the data in a comprehensive way. The creation of the maps from real
data was something really satisfying and enriching for me, allowing me to reveal the paths and the
movement patterns of these specimens. Although learning a programming language can be a slow
process and occasionally frustrating, it is very rewarding to overcome every challenge presented. In
conclusion, this practical work, besides being an interesting way of approaching to the current
scientific work, it has given to me new skills and knowledge very useful for the future of my career.

You might also like