Blue Flower

List of past and oncoming meetings.

BIGFish meetings usually take place in the Marine Institute, randomly alternating between talks and reading group discussions.

The exact time and meeting room are announced through an emailing list before the meeting.

Remote access via video is available on request.

DATE SPEAKER & TOPIC

26 February 2018

Hans Gerritsen, extracting catch data from Intercatch, filling in gaps in discard and sampling data, checking for outliers etc...

I will present some code I developed for extracting intercatch data for the anglerfish benchmark. The code takes the data submitted to intercatch and does the allocations of missing discard volume and missing sample data (age or length compositions) in R, rather than in intercatch itself. It was written to deal with a number of years of data but will also work for a single year. The talk is aimed at stock coordinators but may also be of some interest to data submitters.

08 February 2018

Colm Loardan & Claire Moore, Workshop on stock assessment and management strategy evaluation in a4a 

  1. Brief introduction to a4a (a summary can be found at A summary of a4a can be found at  http://www.flr-project.org/FLa4a/articles/sca.pdf ).
  2. Demonstration of how they have applied a4a to a number of demersal Celtic Sea stocks and run a management strategy evaluation. 
  3. Practical session, where participants can apply this framework on a training dataset. 
  4. Application of the method to your own stock. All you need to bring is an FLstock object for whichever stock you would like to look at.

24 January 2018

Julia Calderwood, Does catch composition influence fishing behaviour and can changes in the response to poor catch composition extend fishing opportunities under the Landing Obligation?

Further exploration of the data collected during Challenge Trials back in 2015 indicated that fishing behaviour had been altered to avoid unwanted catch with some success. Both depth fished and distance travelled between consecutive hauls were found to have an influence on catches of choke species such as cod and haddock. Although such tactical changes could result in the avoidance of unwanted catch there is little evidence to suggest that fishermen always change their behaviour following a poor catch. An update of this work and an overview of results from some data exploration will be presented.

19 July 2017

Hans Gerritsen, selecting fishing trips to sample through unequal probability sampling

This talk will outline some of the issues we have with trying to get a representative sample from the catches of a large number of demersal stocks, landed by a diverse fleet, mostly in mixed fisheries. We treat fishing trips as primary sampling units because that is how the catch is accessible to us for sampling. However, the distribution of catches from fishing trips are strongly skewed in a number of ways. If we sample trips at random we would spend most of our sampling effort on small vessels that contribute little to the overall catches. Hans and Sara-Jane have been working on a sampling design that is aimed at making maximum use of each fishing trip that is sampled through unequal probability sampling – but does it make sense? Have your say…

15 March 2017

Colm Lordan, estimating fisheries reference points theory and practice

This talk covered the following points

  • What reference points are and why they are so important.
  • The new ICES guidelines on estimating PA and MSY reference points.
  • How you estimate the reference points using FLBRP and msy.
  • A few examples of the process to estimate reference points.

lordan.2017.fisheries.reference.points.pdf

msy.whg7a.pdf

25 January 2017

Yves Reecht, course report : ICES training course on advance stock assessment.

Yves summarized the course about advanced fish stock assessment he attended in Copenhagen in November 2016. A bit of emphasised maximum likelihood estimates theory and the TMB package/software with a  few fisheries examples.

https://github.com/bigfishMI/reecht_2017_ices_advanced_stock.git

8 December 2016

Hans Gerritsen, Regional coordination in commercial catch sampling programmes – outcomes from the fishPi project.

Sampling programmes will be increasingly coordinated at the regional, rather than national, scale. This should ensure better coverage of the population and improved cost-effectiveness of data collection. The fishPi project trialled the way such sampling designs would be developed through four case studies. The project developed data formats and software for data sharing, checking and analysis, and for the simulation and testing of sampling designs. The main findings from the case studies was that considerable improvements can be made by adopting regional designs, using probability‐based selection methods and associated regional estimation methods. However, national sampling requirements also need to be taken into consideration.
Ireland was one of the project partners and contributed to the case study for northern hake. This bigfish talk will mainly focus on the methods and findings used in the hake case study but also address the overall conclusions of the project.

gerritsen.2016.fishpi.pdf

17 November 2016

Julia Calderwood, Can IGFS data be used to predict discarding hotspots in the Celtic Sea?

It is about some ongoing work being carried out as part of the Discardless project to assess whether survey data can be used to predict areas where non-target and low quota species are most (or least) likely to be caught. It will also show how fishing opportunities under a landing obligation could potentially be maximised if vessels used IGFS data to identify areas with low potential discards.

calderwood.2016.predicting.discards.pdf

27 October 2016

Sara-Jane Moore, course report : ICES training course on design and analysis of statistically sound catch sampling programmes

A summary of the course Sara-Jane attended

moore.2016.ices.course.survey.pdf

29 September 2016

Eoghan Kelly, course report : ICES training course on data limited stock assessment

Quick overview of the course content.

kelly.2016.dlm.course.pdf

15 September 2016

Debbi Pedreschi, spatial data analysis in ecology and agriculture using R, Chapter 7

pedreschi.2016.chapter.7.pdf

22 August 2016

John Phelan, the dispersal of Scallop larvae

This talk is about how to model the dispersal of Scallop larvae. Indeed, tracking the position of the larvae has important implications for the stability of the fishery. To do so he used Ichthyop which offers a platform to simulate the movement of larvae in the water column based VMS data in conjunction with ROM files. A discussion is provided about different ways of visualizing the outputs and some challenges that have been faced while developping thispeace of work.

phelan.2016.dispersal.scallop.larvae.pdf

19 August 2016

Macdara O'Cuaig, spatial data analysis in ecology and agriculture using R, Chapter 6

ocuaig.2016.chapter.6.pdf

10 August 2016

Cian Kelly, distribution and abundance of the Microsporidian parasite Spraguea lophii in both black and white monkfish

This thesis presents the first results on prevalence of the parasite Spraguea lophii in both of its hosts, the black anglerfish (Lophius budegassa) and the white anglerfish (Lophius piscatorius), in Irish Shelf waters of the North-East Atlantic. Anglerfish are a commercially important species off the Irish coast where they are found in shallow inshore waters down to deeper shelf waters. Spraguea lophii is a parasite in the Kingdom Fungi that attacks the nervous system of their hosts and produces numerous lesions. This study measured the prevalence of the parasite in both species and amongst different regions, sizes and for both sexes. Samples were collected from Irish Shelf waters during the Irish Groundfish Survey (IGFS) 2015 on board the RV Celtic Explorer. The observable pathology of the parasite was described as well as its ultrastructure, where xenoma cysts harboured numerous capsules and those capsules enclosed individual spores. PCR, sequencing and phylogenetic analysis then confirmed the parasite species as S. lophii as well as examining its phylogenetic position relative to previously studied parasites of similar SSU rDNA. Overall prevalence of S. lophii in Irish Shelf waters was detected as 51.27% (n=513). Prevalence in L. piscatorius was 59.88% (n=324), while in L. budegassa it was 36.51% (n=189). Chi-square tests confirmed significant associations between prevalence and species, and prevalence and size. L. piscatorius had a higher prevalence than L. budegassa, whilst larger sized individuals had a higher prevalence than smaller individuals.

kelly.c.2016.anglerfish.parasites.pdf

20 July 2016

Coilin Minto, including unclassified individuals in sex specific growth moodels

Estimating sex-specific growth models typically proceeds by fitting to individuals recorded as male or female. Yet, for many animal species sex may not be apparent until the onset of maturation. As a result, sex-specific growth models are often only fit to known-sex individuals that occupy a limited region of the fitting space. This results in biased parameter estimates. An alternative approach is presented here whereby the sex of the unclassified individuals is treated as a classification problem to be estimated simultaneously with the sex-specific growth models. An introduction to the Expectation Maximisation(EM) algorithm will also be provided.

https://github.com/mintoc/lhmixr

29 June 2016

Guillaume Bal, bayesian assessment and forecasting of covariates impact on fish growth

This talk presented an extended Von Bertalanffy growth model to assess, compare and forecast the impact of competitor densities and temperature on juvenile salmon growth in the wild. The approach used in that case can easily be tailored to the different species and covariates you may work one.

http://onlinelibrary.wiley.com/doi/10.1111/j.1095-8649.2011.02902.x/abstract

bal.2016.growth.pdf

31 May 2016

Cormac Nolan, morphometrics and discriminant analyses

This talk will deal with two subjects. The first, morphometrics, will focus on how to capture shape data and how to process it in R. It will mainly deal with otolith morphometry by introducing two useful R packages (Momocs and shapeR)  but will also touch on fish body morphometry. Morphometrics, the analysis of form, is a great tool for getting complex shape data into a format that can be used as a variable in further analyses, which brings us nicely to the second subject. Discriminant analysis uses pattern recognition to find a combination of features that characterizes or separates two or more classes of objects; it is classification by machine learning. The characterising features can be anything from environmental variables or species composition to the shape of a fish or its otolith. I will try to provide an introduction to some basic analyses and their requirements e.g. linear or quadratic discriminant analysis, the training data set, normality. Finally, I will share some resources that we found useful and that detail some more powerful classification methods (K nearest neighbour, support vector machines (!), etc. ).

Code available at https://github.com/bigfishMI/morph_discr

17 May 2016

Dave Reid, Real Time Incentives (RTI) : a different way to manage fisheries

RTI is a spatially and temporally explicit way of managing fisheries. Each part of the fishery area will have a different tariff based on fishery or ecosystem management objectives, e.g. reducing cod F, or avoiding catches of endangered elasmobranchs. Fishermen will have a set number of “credits” based on vessel size and gear. The presentation will show how the approach works, some of the simulations we have carried out to test the method, and what we are doing in the new SFI funded RTI project. 

reid.2016.RTI.pdf

13 May 2016

Debbi Pedrechi, spatial data analysis in ecology and agriculture using R, Chapter 5

pedreschi.2016.chapter.5.pdf

11 May 2016

Debbi Pedreschi, course report : ICES training course in the R environment

Quick overview of the ICES course Debbi went to, including a tour of the topics covered and an overview of the material

pedreschi.2016.r.training.pdf

29 April 2016

Julia Calderwood, spatial data analysis in ecology and agriculture using R, Chapter 4

calderwood.2016.chapter.4.summary.pdf

19 April 2016

Guillaume Bal, why time series decomposition matters : an example with river temperature modelling and forecasting

bal.et.al.2014.pdf

bal.2016.temperature.talk.pdf

12 April 2016

Julia Calderwood, spatial data analysis in ecology and agriculture using R, Chapter 3

calderwood.chapter.3.summary.pdf

6 April 2016

Hans Gerritsen, The Irish Anglerfish and Megrim Survey: why, when how and what now?

The 2016 Irish Anglerfish and Megrim Survey has just finished. I will give a brief overview of the objectives of the survey, the survey design and gear, some preliminary results and what to do next.

gerritsen.2016.survey.talk.pdf

29 March 2016

Yves Reecht, spatial data analysis in ecology and agriculture using R, Chapter 2

reecht.2016.chapter.2.summary.pdf

reecht.2016.chapter.2.r.code

15 March 2016

Guillaume Bal, spatial data analysis in ecology and agriculture using R, Chapter 1

chapter.1.summary.pdf

10 March 2016

Andrew Campbell, Development of an index for Western Horse Mackerel

The western horse mackerel is a widely dispersed pelagic stock of significant importance with annual catches averaging over 150kt. However, in part due to a lack of fishery independent data, the assessment model outputs are highly uncertain which presents a significant challenge for the management of the stock. Using data from the IBTS, this presentation will outline some recent work to develop an index of stock abundance.

7 January 2016

Colm Lordan, Why everyone in FEAS should do fully reproducible research

In fisheries we tend to do a lot of repetitive data analysis.  We work up survey data into results in reports the same way.  We work up sampling data in the same way. We re-run stock assessments in the same way.  However there are always issues; data changes, data problems, new data, new tools, new models etc.  We have to do fixes, filters fill ins on the fly.  In fact very little of our scientific analysis and results could be independently reproduced even when detailed protocols exist.   This is a problem.  Especially when you can never find the original analysis! I will demonstrate a few typical examples of fully reproducible research using rMarkdown.  I will give some tips and tricks I have picked up over the last year or so of experiment with rMarkdown.  Then I would like to discuss a strategy for rolling out fully reproducible research across the FEAS and the MI.

lordan.reproducible.research.talk.pdf

28 October 2015

Hans Gerritsen, Mapping in R

There are many ways to produce maps in R. Hans will show some example code of working with shapefiles, spatial dataframes and rasters; colouring by numbers; geoprocessing; finding out which points lie inside which polygons; working with projections etc.

gerritsen.2015.spatial.packages.in.R.pdf

13 October 2015

Eoghan Kelly, Automatic reports for discard trips

After the data from a discard trip has been entered into the database, we automatically generate reports that are used for QC and to provide feedback to the skippers. The reports are generated using SQL, R, LaTeX and Shiny. Eoghan will briefly outline the technicalities of generating these reports and will also illustrate how they can make our lives easier and less prone to error.

AutomaticReportsForDiscardTrips.pdf

22 September 2015

 

Damian Smith, first steps with sharing and collaboration tools : git / github and jupyter

Damian will show us some of the things he has found  whilst learning how to use these popular tools and how we could start using them

GithubJuypyter_2015-09-22.pdf

8 September 2015

 

Jonathan White, Partial year salmon counts as predictor of full year counts

Over the previous eight years there has been a declining trend in the forecast and observed numbers of returning wild salmon to Irish rivers, and in recreational catches.  There is currently interest to see if counts of salmon made by fish counters during the year, can be used as indicators of the full run of salmon in a year, in order to use early counts as an indication of potential full year runs.  This study aims to investigate the potential use of three partial year salmon counts, made up to the end of May, to the end of June and to the end of July, made by fish counters to:

i. Compare partial year counts against previously observed partial counts, to assess – mid season – how the salmon run in a river is progressing.

ii. As a means of estimating what the size of the full run (with error margins) may be, based upon partial year count counts made earlier in the year.

 A method is outlined to compare the partial counts made in the current year againts previously seen partial counts.

white.2015.mid.year.counter.ppt.pdf

white.2015.mid.year.counter.word.pdf

25 August 2015

Sarah Davie, Making plots in R using the gplot package

Sarah will introduce you to the use of the ggplot package. Your graphs should be a piece of art after that.

So below we have the ggplot cheat sheet (tutorial not found), the presentation as a pdf, R code that makes the plots, and the bit of code for plotting with google earth we looked at in the meeting. Happy plotting all :) 

Using ggplot for plotting presentation

Using ggplot for plotting code

ggplot cheet sheet link

Ggmap and google

13 August 2015

Oliver Tully, estimations from Spatial data

Oliver will lead a chat about estimations from spatial data (eg fisheries survey data). The following points will be considered:

- choice of spatial interpolator,

- spatial autocorrelation,

- variance of the interpolated surface,

- estimating variance in point estimates (sampling locations), many sources possible,

- final estimates and total variance (variance due to interpolations, variances at sampling points...).

tully.2015.spatial.estimation.pdf

28 July 2015

 

Hans Gerritsen, introduction to smoothers and GAMs

Hans will give a summary of a workshop on spatial modelling methods he attended earlier this year. This workshop addressed the use of smoothers (for summarising spatial trends, but they can equally be applied to trends over time or any other explanatory variable). The quirks of some of these smoothers will be discussed. General Additive Models (GAMs) are typically used to model smooth terms and some examples of GAMs will be illustrated, highlighting the things to look out for (link functions, autocorrelation, criteria for model selection).

 gerritsen.2015.smoothers.gams.pdf