Ref. Ares(2018)3458435 - 29/06/2018
Ref. Ares(2019)2387732 - 04/04/2019
Cefas Document Control
Title:
Submitted to:
Welsh Government
Date submitted:
12/05/2015
Project Manager:
Report compiled by:
Quality control by:
Approved by &
date:
Version:
10
Version Control History
Author
Date
Comment
Version
2/2/15
First draft
1
4/2/15
Providing comments 2
and edits
17/4/15
Adding analytical
3
methods, survival
results
28/4/15
Results,
4-5
methodology,
formatting,
discussion
29/04/15
Accepted changes
6-7
v4, discussion and
figures, formatting
9/5/15
Adding photos,
8
conclusions, exec
1
summary,
references
11/5
Accepted changes
9
12/05/15
Providing comments 10
and edits
2
Estimating the discard survival rates of Common sole (
Solea solea)
and plaice (
Pleuronectes platessa) in the Bristol Channel trammel
net fishery and of plaice in the Bristol Channel otter trawl fishery
May 2015
3
Executive Summary
Discarding fish back to the sea that are caught during commercial fishing is often considered to be
wasteful. On 1st January 2014, the latest reform of the EU Common Fisheries Policy (CFP) came into
force and with it, under Article 15, a discard ban or landing obligation for regulated species (EU
2013). This discard ban is being phased in, and will cover all stocks of quota species in EU waters by
the end of 2019. The new policy includes a number of exemptions including for ’… species for which
scientific evidence demonstrates high survival rates, taking into account the characteristics of the
gear, of the fishing practices and of the ecosystem …’.
Research has shown that some discards survive and that in some cases, the proportion of discarded
fish that survive can be substantial. The principle of the new CFP is to motivate fishers to avoid
catching unwanted fish, whereby all fish are deducted from quota and fishers are obligated to land
all catches of quota species. When a quota is exhausted, fishing operations are to cease. However,
when avoiding unwanted catches is not possible, and the survival rate of discarded fish is shown to
be high, then the return of those fish to the sea is justifiable and allowable.
There are some published discard survival data but the results are highly variable and available for
only a few selected species and fisheries. Many factors, including biological attributes, environmental
conditions and technical elements of the capture process, can affect the survival rate of discarded
species. There is an immediate demand for scientific evidence on fishery specific discard survival
rates, which consider the specific characteristics of the gear and fishing practices.
To meet this requirement, this project aimed to generate discard survival estimates for key species in
Welsh fisheries. The specific objectives were to estimate discard survival rates of plaice (
Pleuronectes
platessa) and sole (
Solea solea) in the trammel net fishery and of plaice in the otter trawl fishery,
both fisheries operate off the south coast of Wales.
The structure of the project dictated the method that could be used, and this was developed within
the project and in parallel with the ICES’ Workshop on Methods to Estimate Discard Survival
(WKMEDS). The approach selected was to assess the health and vitality of fish at the point of
discarding during a representative range of conditions and combine this with survival rates of fish
held in captivity, also selected from the catch with a representative range of vitality conditions, and
combine these data to generate an overall weighted mean discard survival estimate.
This study demonstrated that after an observation period of 76-81h, the percentage of discarded
plaice surviving normal commercial fishing practice was 49%. For Dover sole, after this period,
discard survival was 21%. Model predicted final rates of discard survival were 3.6-39.1% for plaice
and 18.6-20.3% for sole. Using captive observation results from a similar otter trawl fishery in a
parallel study, combined with health assessment data in this study, produced inferred discard
survival estimates for plaice caught by an otter trawler in the Bristol Channel of 75-88%.
All estimates, included avian predation but excluded other marine predation. Furthermore, the
stressors exerted on the fish from the method applied, including temperature differences, handing,
confinement, proximity with other fish, dissolved oxygen depletion, were likely to have induced
some experimental mortality. Therefore, the results presented here should be interpreted as
minimum estimates of discard survival, excluding marine predation.
4
There were many factors identified with the potential to effect survival and the relatively low
number of replicates of the treatment made it difficult to identify the key influencing variables.
However, some initial analysis showed that lower survival was associated with poor weather
conditions. There was also an indication that higher survival was associated with monofilament nets
compared with multi-monofilament nets, suggesting that changing the net design could provide a
method to increase survival rates.
The survival estimates generated here are representative of the observed trips. Assumptions must be
made in order to extrapolate the data to vessel and fleet level. However, this evidence is considered
to provide scientifically robust estimates of discard survival and will inform fisheries managers of the
appropriateness and potential to develop proposals to gain exemption from the landing obligation
under the high survivability provision in European Regional Discard Plans.
5
link to page 4 link to page 8 link to page 9 link to page 11 link to page 11 link to page 11 link to page 11 link to page 12 link to page 12 link to page 15 link to page 15 link to page 15 link to page 15 link to page 15 link to page 16 link to page 23 link to page 23 link to page 23 link to page 26 link to page 29 link to page 29 link to page 29 link to page 32 link to page 32 link to page 32 link to page 32 link to page 32 link to page 33 link to page 33 link to page 34 link to page 35 link to page 35 link to page 35
Contents
Executive Summary ................................................................................................................................. 4
Background .............................................................................................................................................. 8
Introduction ............................................................................................................................................. 9
Materials & Methods ............................................................................................................................ 11
Methodological approach ................................................................................................................. 11
What is survival?............................................................................................................................ 11
What influences survival? ............................................................................................................. 11
How do you estimate discard survival? ......................................................................................... 12
The limitations and assumptions of the selected approach ......................................................... 12
Specific study methods...................................................................................................................... 15
1) Static net study .......................................................................................................................... 15
Vessel & port of operation ............................................................................................................ 15
At sea ............................................................................................................................................. 15
Fishing activity ............................................................................................................................... 15
Data collection ............................................................................................................................... 16
On board tanks .............................................................................................................................. 23
Avian predation ............................................................................................................................. 23
Transit from sea to shore .............................................................................................................. 23
On shore ........................................................................................................................................ 26
2) Otter trawler study .................................................................................................................... 29
Vessel and port of operation ......................................................................................................... 29
Fishing activity and data collection ............................................................................................... 29
Analytical methods ............................................................................................................................ 32
Summary data from each study .................................................................................................... 32
Survival methods ........................................................................................................................... 32
Kaplan-Meier plots ........................................................................................................................ 32
Survival models.............................................................................................................................. 32
Applying survival rates to vitality data .......................................................................................... 33
Identifying factors that influence survival ..................................................................................... 33
The effect of reflex impairment and injury on survival ................................................................. 34
Results ................................................................................................................................................... 35
Overview of results ............................................................................................................................ 35
Study 1 – Trammel net, plaice and sole ............................................................................................ 35
6
link to page 39 link to page 40 link to page 40 link to page 41 link to page 41 link to page 41 link to page 42 link to page 44 link to page 44 link to page 45 link to page 45 link to page 46 link to page 46 link to page 47 link to page 47 link to page 47 link to page 48 link to page 48 link to page 49 link to page 54 link to page 60 link to page 61 link to page 62 link to page 63 link to page 63 link to page 64
Table 6 Data summary ....................................................................................................................... 39
Figure 13 Study 1 Length frequencies of plaice and sole in trammel net catches and held for
observation .................................................................................................................................... 40
Figure 14 Study 1 Semi-quantitative vigour vitality score for plaice and sole trammel net catches
....................................................................................................................................................... 41
Table 7 Study 1 Avian predation observations .............................................................................. 41
Figure 15 Study 1 Kaplan-Meier estimates of survival probability ............................................... 42
Table 8 Study 1 Survival of captive fish during observation time period and modelled for
extended period ............................................................................................................................ 44
Table 9 Study 1 Estimated discard survival for all plaice and sole on observed trips using vitality
as a proxy ....................................................................................................................................... 45
Study 2 – Otter trawl, plaice .............................................................................................................. 46
Figure 16 Study 2 Length frequency of plaice caught by a Bristol Channel otter trawler ............ 46
Figure 17 Study 2 semi-quantitative vigour vitality score for plaice catches in a Bristol Channel
otter trawler .................................................................................................................................. 47
Table 10 Study 2 Avian predation observations ............................................................................ 47
Table 11 Study 2 inferring discard survival for plaice on observed trips on-board a Bristol
Channel otter trawler using vitality as a proxy ............................................................................. 48
Factors influencing discard survival .................................................................................................. 49
Discussion .............................................................................................................................................. 54
Conclusions ............................................................................................................................................ 60
Acknowledgments ................................................................................................................................. 61
References ............................................................................................................................................. 62
Annexes ................................................................................................................................................. 63
Annex 1 STECF EWG 14-11 Table of survival estimates .................................................................... 63
Annex 2 Ad hoc physical environmental measurements .................................................................. 64
7
Background
Discarding fish back to the sea that are caught during commercial fishing is often considered to be
wasteful by fishers, conservationists and fisheries managers alike. On 1st January 2014, the latest
reform of the EU Common Fisheries Policy (CFP) came into force and with it, under Article 15, a
discard ban or landing obligation for regulated species (EU 2013). This discard ban is being phased in,
beginning with pelagic fisheries from 1st January 2015 and will cover all stocks of quota species in EU
waters (and those with a Minimum Landing Size in the Mediterranean) by the end of 2019. The final
text agreed by the European Council and European Parliament includes a number of exemptions and
flexibility tools. In paragraph 2(b), an exemption from the landing obligation is described for “
species
for which scientific evidence demonstrates high survival rates, taking into account the characteristics
of the gear, of the fishing practices and of the ecosystem”.
The discarding process can be defined by three phases: i) capture by fishing gear, ii) handling at the
surface, and iii) release back to the sea. Research has shown that some discards survive the process.
In some cases, the proportion of discarded fish that survive may be substantial, depending on the
species, the characteristics of the vessels and other operational, biological and environmental
factors. The principle of the new CFP is to motivate fishers to avoid catching unwanted fish, whereby
all fish are deducted from quota and fishers are obligated to land all catches of quota species. When
a quota is exhausted fishing operations are to stop. However, when avoiding unwanted catches is not
possible, and the survival rate of discarded fish is high, then the return of those fish to the sea is
justifiable and allowable.
The European Commission's Scientific, Technical, Economic Committee for Fisheries (STECF)
concluded that selection of a value which constitutes “high" survival is subjective and likely to be
species- and fishery-specific. The value will be based on “trade-offs” between the stock benefits of
continued discarding and the potential removal of incentives to change exploitation pattern and how
this contributes to the minimisation of waste and the elimination of discards (STECF 2014). Central to
any proposal for an exemption for selected species or fisheries, is the requirement for clear,
defensible, scientific evidence on discard survival rates.
Details of exemptions will be included in regionally formulated Discard Plans and Multi-Annual Plans,
and these will be based on scientific studies that have been independently reviewed before the plans
are assessed by the EU Commission. There are some published discard survival data but the results
are highly variable and available for only few species and fisheries. Many factors, including biological
attributes, environmental conditions and technical elements of the capture process, can affect the
survival rate of discarded species. Article 15 notes that consideration must be given to the specific
characteristics of the gear, fishing practices and of the ecosystem. Therefore, there is an immediate
demand for scientific evidence on fishery specific discard survival rates.
8
Introduction
In March 2014, Cefas was contracted by the Welsh Government to conduct a series of meetings with
the Welsh fishing industry to consider the impact of the landing obligation on the catching sector and
to see what specific actions or operational studies could alleviate this. Based on these meetings, the
main ‘choke species’ (those most likely to stop fishing activities) in Welsh inshore fisheries (using
both static and towed gear) is anticipated to be plaice, but Dover sole and rays could also have the
potential to limit fishing opportunities under the discard ban. There is a perception that certain areas
of the sea bed off the coast of South Wales have a high abundance of smaller sized plaice and sole.
These areas can be avoided to some degree, but in a mixed fishery of mostly quota species, where
some quotas are low, the discard rates can be high despite efforts to avoid these fish. In such
circumstances, even relatively low catches can risk a premature end to the fishing season.
There was considerable support from vessel operators in attendance at these meetings for a study of
the survivability of unwanted fish that cannot be avoided, particularly plaice, in Welsh inshore
fisheries. In recognition of this feedback from skippers, the Welsh Government agreed to fund a
research study to estimate the survival of discarded fish, with a focus on plaice. This work is expected
to complement other studies being undertaken in England and other Member States and the outputs
are expected to guide Welsh fisheries managers on whether exemptions from the Landing Obligation
should be applied for.
The original aim of this work was to obtain estimates of the survival of commercially caught and
discarded plaice, Dover sole and rays. This would add to the evidence base on survival rates for these
species summarised by STECF (Annex 1). It was evident early in the project that imitations in
resources and time meant that we had to focus our attention on plaice and sole, these were
prioritised as these had the most limited evidence base on survival whereas there is some evidence
on the survival of discarded elasmobranchs. We aimed to estimate the survival rates across the full
length range of the catch, under the assumption that fish at any length could be discarded and an
exemption, if awarded, would not apply to fish only within a specific size range.
The original scope was to conduct experiments with both trawl caught fish and fish caught in gill
nets. The experimental approach was to: i) conduct vitality assessments on board commercial
vessel(s) during a representative range of conditions to quantify the reflex responses and physical
damage of plaice and sole, after having been caught, handled and discarded; ii) conduct captive
observations of individuals representing the various vitality levels to determine survival rates; iii)
combine the vitality scores with the likelihood of survival for each vitality category to estimate a
survival rate for the fishery.
The method used was that described in the report of the ICES Workshop on Methods to Estimate
Discard Survival (ICES 2014). The report details the different approaches and the limitations of the
conclusions that can be drawn from them (Table 1). The resources and, more critically, the time
available in this project, dictated which of the approaches was used and would deliver the most
robust evidence on discard survival estimates. The approach selected was to use vitality assessments
on-board commercial vessels during a representative range of conditions and combining this with the
captive observation of individuals with a different vitality levels to generate an overall weighted
mean survival estimate. It was decided that added to this we would provide estimates of avian
9
predation. This approach would provide an estimated discard survival rate, excluding marine
predation, which is representative of the fishery.
10
Materials & Methods
Methodological approach
Research aimed at determining whether aquatic organisms survive, which have been caught and
subsequently returned to the water, has been conducted over many decades. Although there have
been reviews of the outputs from this work (Broadhurst et al. 2006, Revill et al. 2013), at the
commencement of this project there had been no assessment of the scientific methods and
approaches that can be used to meet this aim.
Around the same time as the start of this project, an ICES (International Council for the Exploration of
the Sea) group on Methods to Estimate Discard Survival (WKMEDS) was initiated. The co-chair of ICES
WKMEDS provided the scientific advice for this project. The ICES workshop was initiated to develop
and describe the methods of best practice to quantify the survival of aquatic organisms caught and
returned to the water. The catalyst for the creating the WKMEDS was the change in European Union
fisheries policy, generating a need for guidance on how to investigate levels of discard survival, which
was absent at the beginning of this project.
Therefore, during the course of this project, the methods of best practice to derive estimates of
discard survival have been developing. The outputs from ICES WKMEDS have been applied to this
project, moreover, the experiences from this project have been used to improve the guidance on
how best to conduct discard survival assessments as reported by WKMEDS.
What is survival?
Before discussing the most appropriate methods for measuring the survival of discards it is useful to
consider what we mean by “survival”. The opposite of survival is death, which is a more definitive
state to identify. So typically when we measure the “survival” of organisms, after they have
experienced a particular treatment, we in fact quantify the number of individuals that died, based on
a measurable definition of death. More precisely, we usually measure mortality rates, which is the
number of individuals that die over a defined period of time. The inverse of the mortality rate is the
survival rate.
Death is not normally an instantaneous process and some time will elapse between an initial
exposure to a fatal stressor and the eventual cessation of life. Conversely, if observed long enough,
any individual will die. Therefore, the timeframe over which observations are made will have an
important influence on the estimated survival rate. There is no standard time frame for conducting a
survival assessment, as it depends upon the species in question and the nature of the fatal effects, as
well as the logistical limitations of the investigation. It is recommended that survival estimates should
be presented with reference to the timeframe over which they were derived (e.g. “40% mortality,
equating to 60% survival; 6 days observation”).
What influences survival?
A fish or other animal will experience an array of different potentially injurious events, or stressors,
throughout each phase of the capture process:
i)
capture by the fishing gear;
ii)
handling at the surface;
iii)
release back to the water
11
In this context, an array of factors that could potentially influence discard mortality can be identified.
These can be classified into three broad categories: biological (e.g. species, size, age, physical
condition, occurrence of injuries), environmental (e.g. changes in: temperature, depth, light
conditions) and technical (e.g. fishing method, catch size and composition, handling practices on
deck, air exposure). Each stressor and the additive effects of multiple stressors will influence the
survival of an individual.
How do you estimate discard survival?
There are three main approaches to conducting a discard survival assessment with the aim to
estimate discard survival (ICES 2014):
(1) Vitality Assessment: where the health status of the subject to be discarded is scored relative to
any array of indicators (e.g. activity, reflex responses and injuries) that can be combined to
produce a vitality score. Where these scores have been correlated with a likelihood of survival
they can be used as a proxy for survival likelihood;
(2) Captive Observation: where the discarded subject is observed in captivity, to determine whether
it lives or dies; and
(3) Tagging and Biotelemetry: where the subject to be discarded is tagged and released, and either
its behaviour/physiological status is remotely monitored (via biotelemetry) to determine its post-
release fate, or survival estimates are derived from the number of returned tags.
In isolation, each method has limitations which can restrict the usefulness of the survival estimates
they produce. However, when two or more of these methods are combined there is clear potential
for considerable benefits. The benefits from this integrated approach include: reducing resource
requirements, increasing the scope of the investigation, as well as improving the ac-curacy, precision
and application of the survival estimates.
A synthesis of the approaches recommended to meet specific objectives to estimate discard survival
is provided in Table 1 (ICES 2014). This table can be viewed either as means to identify a single
approach to meet a specific objective or as a stepwise process, from 1 to 6. In general, the
approaches taken from first to last increase in the level of resources and time required to achieve the
stated goal. The outputs from each approach, range from providing estimates of the proportion of
discards that appear dead or impaired at the point of discarding (referred to as “survival potential”)
(1), to generating a discard survival rate for a population that is representative of a fishery (6).
To conduct captive observation experiments to cover the full variability of conditions displayed by a
fishery and species is practically difficult and expensive. Instead, the vitality of discarded individuals
can be derived with relative ease from multiple fishing operations. In addition, estimates of survival
for the different vitality levels can be derived from captive observation. The proportion of survivors
at each vitality can produce a proxy estimate of survival that is representative of conditions in the
fishery (excluding predation) by applying it to the vitality data. This technique also gives the relative
influence on discard survival of selected variables.
The limitations and assumptions of the selected approach
1) The captive observation approach excludes predation and therefore may overestimate survival.
The inclusion of estimates of avian predation in this project meant that it is only marine
12
predation that is not accounted for, but the levels of this are unknown. To account for marine
predation requires the use of data storage or acoustic tagging techniques but these could not be
delivered within the time and cost structure of this project.
2) When using captive observation, the period of observation will dictate the context of the survival
estimates (e.g. 60% survival after 6 days). Ideally monitoring should continue until mortalities
cease or at least slow down. However, in practice, the duration of monitoring has to be a trade-
off between ideal scientific needs, the available resources (sea time, budgets and available tank
time) and occurrence of confounding mortality not associated with the process of discarding.
Therefore, if the observation period is too short, the survival estimates might be overestimated.
Models to project forward from a survival probability curve were used to inform whether a
longer observation period would have generated lower survival estimates (see Analytical
methods section).
3) For survival estimates to be representative of the fishery, vitality data should be generated for
fish discarded during all conditions of a fishery. However, because conditions are constantly
changing, without a continuous vitality monitoring programme, the survival estimates may be
representative only for the trips from which vitality data have been collected. To extrapolate the
results to a fishery, it must be assumed that the combination and strength of stressors on the
discarded fish are the same on all trips as those from which vitality data were collected.
4) It must be assumed that retaining fish in holding tanks does not have a recuperative effect and
artificially increase survival. This was considered unlikely in this project - see below (5).
5) Holding wild animals in captivity can induce stress, which can potentially increase mortality in
addition to the treatment effect. Moreover, physical damage from being held in tanks on-board a
moving vessel, changes in salinity, light, pressure and temperature, and being held in close
proximity with other fish, all exert stress on fish. When these stressors occur, they will likely have
additive effects to the treatment stressors and reduce observed survival rates.
6) To be able to use the assessments of fish vitality as a proxy for survival when combined with
captive observation results, two assumptions have to made:
a) Scientific fieldworkers need to be able to assess the vitality of fish consistently, in time, in
different conditions and between different workers. All the fieldworkers collecting data in
this project underwent training in handling live fish and performing vitality assessments. One
scientist oversaw all of the fieldwork.
b) Most importantly, to be able to use vitality assessments as a proxy for survival, there must be
a significant relationship between survival and vitality score. Therefore, the protocol used to
generate vitality scores must deliver scores that can consistently predict survival likelihood.
The results from the captive observation will determine whether assessed vitality is a good
predictor of survival.
13
Table 1 - An overview of possible objectives for a survival assessment and the recommended approaches
Objective (for the selected species,
Suggested approach
Resource Implications
variables & management unit)
1. To estimate discard survival
Vitality assessment on-board commercial vessel(s), with targeted
Personnel: Trained observers & fishers
potential for particular
observations of the factors that affect mortality.
Specialist equipment: None
conditions
Time frame: hours to days for field trials
2. To estimate discard survival
Vitality assessments on-board commercial vessels during
Personnel: Trained observers & fishers
potential that is representative
representative range of conditions
Specialist equipment: None
of the management unit
Time frame: hours to days for field trials
3. To estimate discard survival
Captive observation of individuals under particular conditions
Personnel: Experienced researchers & fishers
rate, excluding predation, for
Specialist equipment: Containment facilities (e.g.
particular conditions
aquaria & sea-cages)
Time frame: days to weeks for monitoring period
4. To estimate discard survival
Vitality assessments on-board commercial vessel(s) during a
Personnel: Trained observers, Experienced
rate, excluding predation,
representative range of conditions combined with
captive
researchers & fishers.
representative of the
observation of individuals representing the various vitality levels to
Specialist equipment: Containment facilities
management unit
generate an overall weighted-mean survival estimate
Time frame: days to weeks for monitoring period
5. To estimate discard survival
Tagging/biotelemetry on-board commercial vessel(s) under particular
Personnel: Experienced researchers & fishers.
rate, including predation
conditions
Specialist equipment: Tags
effects, for particular conditions
Time frame: days to months/years for monitoring
6. To estimate discard survival
Option 1:
Vitality assessment on-board commercial vessel(s) during
Personnel: Trained observers, Experienced
rate, including predation
representative range of conditions combined with
researchers & fishers.
effects, representative of the
tagging/biotelemetry of individuals representing the various vitality
Specialist equipment: Tags
management unit
levels on-board commercial vessel(s) to generate an indirect survival
Time frame: days - months/years for monitoring
estimate
Option 2:
Vitality assessment on-board commercial vessel(s) during
Personnel: Trained observers, Experienced
representative range of conditions combined with
captive
researchers & fishers.
observation (to estimate short term mortality) and
Specialist equipment: Tags,
tagging/biotelemetry (to estimate conditional long-term mortality) of
Containment facilities (e.g. aquaria & sea-cages)
individuals representing the various vitality levels on-board
Time frame: days to months/years for monitoring
commercial vessel(s) to generate an indirect survival estimate
14
Specific study methods
Two studies were completed as part of this project. The first was on the survival of Dover sole and
plaice in a
1) Static net study
Vessel & port of operation
The advantage of conducting field studies on board commercial fishing vessels during representative
fishing operations is that the fish under study have been exposed to realistic and combined stressors
associated with the capture and discarding process. The participation of vessels for this work was
sought through an open tendering process in accordance with government procurement
procedures. Although the invitation to tender was well publicised and a reasonable amount of time
was provided for tender submission, the number of applicants was low. The selection of vessels was
based on the willingness of the skippers to cooperate, the space on board and the safety of the
vessel to accommodate observers and necessary equipment and the track record of fishing in the
defined area. Upon evaluation of the received tenders it was concluded that a static netter, targeting
Dover sole, fulfilled the required criteria for the field trials. It was clear that extra effort was required
to locate a suitable trawler, but despite further publicity and lengthy negotiations we were unable to
source a willing and suitable otter trawler for this work based in Wales.
Sea trials were carried out in Swansea Bay (ICES rectangle 31E6), off the coast of South Wales, using
the fishing vessel Seapie (NT28), a fibre-glass hulled netter of 9.88m overall length with a 90kw
engine (Figure 1). MFV Seapie operates from Swansea Marina, at the mouth of the River Tawe, with
access to and from Swansea Bay through the Tawe Barrage Lock.
The fishing activity during the study was representative of normal practice. All fishing was carried
out during neap tides in August and September 2014, on typical fishing grounds for this vessel at this
time of year (Figure 2). Sole was the main target species. The vessel was operated by the skipper
only.
At sea
Fishing activity
Sole trammel nets were shot from a net pound, hauled with a hydraulic hauler and cleared as per
normal commercial fishing practice; the nets were boarded on deck and were cleared once the final
anchor was retrieved and stowed. It is normal practice for the skipper to pick out/un-mesh sole and
plaice as a priority, where possible, leaving other species such as Starry smoothound (
Mustelus
asterias) and Nursehound (
Scyliorhinus stellaris) in the nets until all accessible sole and plaice have
been un-meshed. This routine was adhered to throughout the trials. Sole and plaice were handed to
the observer at the time that these fish would normally have been
retained in a fish box or discarded
back to the sea.
Occasionally, when the weather conditions were considered to be too uncomfortable or dangerous
to clear the nets at sea, the nets were boarded and cleared once the vessel returned to port. In
these instances fish were assessed only when the nets were cleared by the skipper at the port.
15
Figure 1 The static netter MFV Seapie (NT28) in Swansea port and fishing in Swansea Bay
Data collection
All sole and plaice caught were recorded by length (to the nearest cm below) and all other species
were recorded as numbers of individuals. The catch composition from each tier was recorded
separately, alongside the positional (lat/long; depth) and environmental information (air
temperature; sea surface temperature; light level) specific to that particular tier. Light levels were
measured using a Reed Instruments’ ST-1301 digital light meter, placed at deck level. The
specification of the fishing gear used in each individual tier was recorded (Table 2) and the times
were logged when tiers were shot, hauled and the subsequent catch sorting process began and
ended.
Once the sole and plaice had been un-meshed and handed to the observer, each individual was
measured and scored using a predefined assessment protocol. This assessment protocol was
developed using methods described in the ICES WKMEDS 2014 report and refined in the Cefas
laboratory using aquarium kept (unstressed) plaice. A series of behavioural reflex tests was
identified that consistently produced unimpaired responses in both free swimming and restrained
16
fish, and could be scored rapidly in a replicable manner. Injury types specific to the fishery of interest
were also defined.
Figure 2 Map of Swansea Bay & positions of gear
Table 2 Gear description
Gear Length
Hanging
Twine
Twine
Twine
Twine
Mesh
Mesh
Mesh
Mesh
of Fleet
Ratio
Type
Type
Diameter
Diameter
Size
Drop
Size
Drop
(m)
Inner
Outer
Inner
Outer
Inner
Inner
Outer
Outer
(mm)
(mm)
A
300
0.5
Mono
Multi
0.35
8x3
114
40
610
4.5
Mono
B
300
0.5
Multi
Multi
1.5x4
8x3
114
30
610
2.5
Mono
Mono
Vitality was assessed using a semi-quantitative assessment of activity (SQA) and a quantitative reflex
and injury scoring method. The SQA framework used was based on four ordinal vitality classes that
are defined, at one extreme as characterising very lively and responsive fish (excellent) and at the
other extreme unresponsive (dead) individuals (Table 3). The sole and plaice that showed no visible
response (body or opercular movement) to touching, prodding or immersion in water were classified
as dead and were simply measured and recorded. Sole and plaice that were assessed, using the SQA
17
scoring, to have excellent, good or poor health states were then scored by the presence or absence
of specific behavioural reflexes and injuries. Behavioural reflex tests were performed both in and out
of water (Table 4). A circular observation container was filled with approximately 50 litres of sea
water for the in-water reflex tests. A reflex action was scored as unimpaired (0) when it was strong
or easily observed, or impaired (1) when it was not present or if there was doubt about its presence.
An injury (including barotraumas) was scored as absent (0) when it was not present or there was
doubt about its presence, and present (1) when clearly observed (Figures 3 to 6).
Table 3 Vigour vitality assessment category definitions
Vitality
Description
Excellent
Vigorous body movement; lively
Good
Fair body movement; responds to touching/prodding
Poor
Weak or no body movement; fish can move operculum
Dead
No body or opercular movements (no response to touching or prodding)
Figure 3 Examination of a plaice; left, splitter used to divert water to the onshore tanks for
continuous water flow
18
Table 4 Vitality reflex and injury assessment protocol developed and applied to both species and in
both fisheries. * Injury specific to Experiment 2 (otter trawl)
Fish Reflex Actions
Description
Scored as 0 (unimpaired) / 1 (impaired)
Body Flex
Tested by holding the fish out of water, with both hands under the fish, and
rotating to get a 'ventral bend' (head and tail move together). Any fish
showing a 'ventral bend' or attempting to struggle free was scored
Operculum Closure
Tested by holding the fish out of water and lifting operculum with a blunt
object (pencil) to get a 'clamp'. Any fish showing an active 'clamp' reaction
was scored
Startle Touch
Tested in water by gentle grabbing of the tail of the fish to get an escape
reaction. Any fish that responded to the grab with a startled escape was
scored
Orientation Right
Tested in water by holding the fish upside down, just below the surface, to
get a 'righting' movement. Any fish actively righting itself within 5 seconds
was scored
Fish Injury
Description
Scored as 0 (absent) / 1 (present)
Exophthalmia
Eyes distended outwards from the head
Corneal Gas Bubbles
Air bubbles visibly present in the eye or the membrane covering the eye
Subcutaneous Gas Bubbles
Air bubbles visibly present under skin
Bleeding
Visible bleeding from any part of the body
Abrasion
Haemorrhaging red area from abrasion
Mucus Loss
Visible area of mucus loss
Scale Loss
Visible area of scale loss
Wounding
Shallow cuts on the body
Deep Wounding
Deep cuts or gashes on the body
Fin Fraying
Fins damaged
Predatory Damage
Bite marks or area of the body eaten or lice actively present
Prolapsed Internal Organs
Intestine protruding out of the anus
Net Marks
Visible line marks caused by the net
Bruises *
Red/purple bruising visible on the body
Scratches *
Scratch marks visible on the body
The measurements and vitality assessments were carried out by the same individual throughout the
experiment to eliminate potential observer effect. After the vitality assessment some of the fish
were then selected for retention in on board tanks. The selection of fish for the on board tanks was
based on the need to identify them throughout the experiment; only fish of differing total lengths,
by species, were placed in the numbered on board tanks. In order to minimise additional captivity
stress and to remove potential interspecific interactions, the stocking density of the on board tanks
was set at a maximum of four individuals and the two species were kept in separate tanks
throughout the experiment. The tank number was then recorded against the data for each individual
fish (haul number; species; length; SQA and reflex and injury scores) to ensure that each fish stored
in the on board tanks was uniquely identifiable. The temperature, salinity and dO2 concentration
19
(dO2) were monitored using an Oxyguard Handy Polaris 2 dissolved oxygen meter and an Aquamarin
refractometer. Fish that were not selected for the on board tanks (non-unique species/length
combinations or dead fish) were either retained by the vessel for sale or discarded back to the sea
after being measured and assessed for vitality.
Figure 4 Examples of some injuries sustained by plaice; above, net marks; below abrasion
20
Figure 5 Example of fin fraying in plaice (above) and conducting the reflex assessments in
assessment container (below)
21
Figure 6 Example of body flex in plaice (above) and sole (below)
22
On board tanks
A vertical stack of six numbered grey polypropylene holding tanks was positioned on board the deck
of the vessel, roughly amidships on the port side, and secured to the vessel’s superstructure (Figure
7). A constant supply of sea water was supplied to this stack, in a flow to waste circuit, from the
vessel’s deck wash system. Sea water was pumped through the seacock valve in the hull of the
vessel by a Jabsco electric-clutch pump, and supplied to deck level using a reinforced PVC hose. This
deck wash hose was then connected to a ball valve on deck that was used to split the water supply
to feed the stack of tanks (Figure 7). The flow of sea water to the tanks was adjusted using the ball
valve to maintain a constant flow rate of 2-4l/min. Changes in engine revs during the fishing activity
changed the water flow rate, so regular adjustments and monitoring were necessary. The sea water
supply entered the stack through an inlet pipe in the top tank. The water then flowed through the
vertical stack by gravity-fed drainage, through interconnecting overflow pipes and exited the stack
through an overflow pipe in the bottom tank (Figure 7). This flow-through of fresh sea water was
initiated on the steam to the fishing grounds, after the vessel was clear of the brackish water
surrounding the mouth of the river and with sufficient time for the circulation of fresh sea water to
all six tanks, prior to hauling the fishing gear.
Avian predation
To evidence avian predation of discarded fish, individuals of known species, size and vitality scores
were released back to the sea, in a manner consistent with normal discarding during commercial
fishing on this vessel. These fish were then tracked visually by two observers and the presence or
absence of sea birds and the subsequent fate of the fish were recorded.
Transit from sea to shore
The vessel returned to port with the selected fish in the on board tanks. The pump supplying the
stack of tanks with sea water was turned off when the vessel reached an appropriate distance from
the port entrance to avoid subjecting the fish to substantial changes in salinity. The outflow of the
River Tawe meant that water salinity reduced when approaching Swansea port. This distance, and
position, was determined by taking repeated measurements of salinity and identifying the minimum
distance from the port at which the sea surface salinity was no less than on the fishing grounds. The
same distance, and position, was used throughout the trials and the pump was turned off at this
point; the fish remained in their tanks until the vessel was in port. Immediately prior to turning off
the pump, the observation container that was used previously for the reflex tests was filled with
seawater in preparation for the transportation of fish to the shore tanks.
The vessel entered Swansea Marina through the Tawe Barrage Lock. The time taken to pass through
the lock gates varied daily and was highly dependent on the number and behaviour of other lock
users. The amount of time that the fish were held in the on board tanks with no water flow varied
accordingly.
As quickly as possible after docking in port the fish in the six numbered on board tanks were
transferred to six identically numbered buckets (38L purple Tubtrugs® flexible) for transportation to
the shore tanks (Figure 8). Each numbered bucket contained a clear polythene bag that was partially
filled with sea water from the observation container before the fish (along with some of the water
from the on board tank) were carefully poured into the bag to minimise handling. The total volume
of water in each bucket when containing the fish was 16 litres. The six numbered buckets were then
23
transported by vehicle a distance of 2.5 miles to the shore tanks. In the absence of temperature
control apparatus, the water in the buckets was susceptible to the heat inside the vehicle.
Figure 7 On-board tank system
24
Figure 8 Tubs and process to transport fish from on-board tanks to onshore tanks.
25
On shore
A purpose built shore unit containing twelve separate holding tanks was used for the shore based
captive observations (Figure 9). Following build completion, the shore based tank unit was tested in
a control situation at the Cefas laboratory in Lowestoft. The tanks were supplied with sea water in a
flow to waste circuit, pumped from the underground sea water tanks that typically feed the
laboratory aquarium. Aquarium acclimatised plaice were assessed for activity, reflex and injury, prior
to their introduction to four of the individual holding tanks in the shore unit. Stocking densities of 3,
4, 5 and 6 plaice were observed. Water temperature and salinity were consistent with the typical
aquarium levels. The plaice were checked for vitality, using a gentle tail grab, at 24hr intervals for a
period of 72hrs. The dissolved oxygen concentration of the water was also checked. At the end of
the observation period, the plaice were assessed again for vitality, reflex and injury, before being
returned to the aquarium facility.
For the purposes of this experiment, the shore based was installed at Swansea University’s Centre
for Sustainable Aquatic Research facility and was supplied with sea water, originating from Swansea
Bay. The sea water was pumped from Swansea Bay via sub-sand filters and was then treated with
ozone to remove incoming pathogens, passed through carbon filters to remove residual ozone, and
into a re-circulating automated water treatment system made up of a mechanical sand filter, protein
skimmer, biological filter, UV lamps, temperature control, pH dosing and oxygen control. A supply of
water from this re-circulating system was plumbed into our twelve tank shore unit. The flow of
water to each of the twelve separate holding tanks was independent and could be individually
controlled using integral flow meters; the flow rate was set and monitored at a constant rate of
2l/min. A thin layer of aquarium silica sand was placed on the bottom of each holding tank to
provide a familiar substratum for the fish and minimise captive stress.
On arrival at the shore based unit, the six numbered buckets containing fish were topped up with
sea water taken from the pumped supply feeding the shore holding tanks; this process was carried
out in an attempt to acclimatise the fish prior to their placement in the shore holding tanks and to
stabilise any differences in the temperature, salinity and dO2 concentration of the holding water.
After a 5-10 min acclimatisation period the fish in the numbered buckets were transferred to the
numbered shore holding tanks by hand and the tank number was recorded; sole and plaice were
stored in separate holding tanks. At the point of transfer any fish that had died in transit were
declared dead, measured, identified, recorded and removed from the experiment.
A series of captive observations was then performed for a period of 72 hours, in agreement with The
Home Office. At 12-hourly intervals (from the point at which the fish arrived at the shore holding
tanks) the survival of the fish in the holding tanks was determined using a gentle tail grab. Fish that
responded to the tail grab by undulation of their fins were declared alive and fish that produced no
response movement were lifted to the surface and their health status was investigated further. Fish
that showed no visible response (body or opercular movement) to touching, prodding or immersion
in water were classified as dead. At the point of these 12-hourly inspections any fish that were
assessed to be dead were removed from the tank, measured, identified and recorded. After a
captive observation period of 72 hours all fish were individually removed from the holding tanks,
measured, identified and their vitality was assessed and recorded using the SQA and reflex and
injury scoring systems. The experiment for these fish was then terminated and they were disposed
of.
26
The process described in 2, 3 and 4 above was carried out for 12 consecutive fishing days over the
period 18th August to 6th of September 2014.
27
Figure 9 The onshore captive observation tanks
28
2) Otter trawler study
Vessel and port of operation
As a result of being unable to find a willing and suitable trawler based in Wales, the decision was
taken to approach the English North Devon trawler fleet and a vessel was selected for this work.
Further sea trials were carried out in Bideford Bay (ICES rectangle 31E5), off the coast of North
Devon, using the vessel Ann Louise (BD22), a fibre-glass hulled trawler of 9.95m overall length with a
148kw engine (photos). Ann Louise operates from the port of Bideford, on the estuary of the River
Torridge.
The fishing activity during this study was representative of normal commercial practice and was
considered to be comparable to that of the South Wales trawler fleet, with the exception that this
vessel towed two trawls in a twin-rig arrangement as opposed to the single trawl typically operated
by the Swansea vessels. All fishing was carried out during March 2015, on typical fishing grounds for
this vessel at this time of year. Rays were the main target species.
Fishing activity and data collection
The trawl gear was deployed, towed, and hauled as per normal commercial fishing practice. The cod
ends were emptied into the aft pounds and the nets were fully re-deployed prior to catch sorting.
The crew sorted the catch by hand, as they normally would, and any small, unwanted, rays present
in the catch were thrown back to the sea immediately. The unwanted plaice and other unwanted
species were left in the pound and, at the point of normal discarding, were collected from the deck
by the observer and placed into a 5-stone fish basket. A circular container was then filled with
approximately 50 litres of sea water, using the vessel’s deck wash system, and the basket containing
the plaice was submerged into it. A second circular container (38L Tubtrugs flexible) was filled with
seawater, using the vessel’s deck wash system, and was used for the in-water reflex tests.
Each plaice was measured and recorded by length (to the nearest cm below), then assessed for
vitality using the identical scoring protocol from study 1, with the addition of two gear-specific injury
types (reflex table). The measurements and vitality assessments were carried out by the same
individual throughout the experiment and that of study 1, to eliminate potential observer effect.
After the vitality assessment the fish were then thrown back to the sea. Avian predation
observations were made for a proportion of the plaice caught and discarded.
29
Figure 10 The otter trawler MFV Ann Louise (BD22) in Bideford port and fishing the in Bristol Channel
30
Figure 11 Example of bleeding injury (above) and bruising (below) seen on plaice from otter trawler
only
31
Analytical methods
As with the fieldwork methods, at the commencement of the project there were no accepted
analytical methods to apply to survival assessments. The statistical methods have been developed
from previous studies and within the work of the ICES WKMEDS.
Summary data from each study
Descriptive and summary data are presented, including the period of study, the number of fishing
days, the mean length of fish assessed for vitality, the mean length of fish under captive observation
and the length of observation time. The proportion of fish in the total catch at each vitality from the
vigour assessment and details of the reflex and injury assessment are presented. The summary table
also summarises the results from the captive observation trials and the survival estimates derived
from the different stages of the analysis for study 1.
Survival methods
The captive observation data provide the length of time that each fish was observed for following
capture and the state of the fish (dead or alive) when the final observation for that fish was made.
This type of data is called longitudinal data and is analysed using survival methods. These methods
provide estimates of the survivor function,
S(t), the probability of surviving for longer than time
t.
Survival methods account for a common propriety of survival data known as censoring. The data for
fish that were still alive at their final observation time are referred to as right censored. Here, we
know that a fish survived until at least that observation time but not how long it would have
survived if the observation period was extended.
Kaplan-Meier plots
The Kaplan-Meier (K-M) estimator generates the survivor function against time. K-M estimates with
95% confidence intervals were calculated for each category of fish vitality, using the R function
survfit. Confidence intervals were computed on the log-log scale as in Venables and Ripley (2002, pg
357).
The K-M method has the advantage of making few assumptions about the data, although it cannot
be used to predict outside the observed experimental period. K-M estimates can also be variable
towards the end of the experimental period when few fish remain observed. Therefore, a “plus-
group” time was defined and times greater than these assigned to the plus-group time when
calculating the K-M estimates. In this case the time was 73.03 hours.
For each case study, the survivor curves from each vitality category (Excellent, Good, Poor) were
then compared using the log-rank test (R function
survdiff). First, an overall comparison of all curves
then comparisons between each pair of vitality categories.
Survival models
For discard survivability studies, a plausible description of the results is that the proportion of fish
surviving will gradually decrease and then flatten off with a proportion of fish surviving the capture,
handling and release process. To model this process and predict the long-term survival probability
requires an extension of standard survival analysis models as these assume that the discard-related
mortality must extend until survival is zero. The extended models required are referred to as mixture
cure models or mixture-distribution models.
32
Two such models were fitted to the case study results: (1) a semi-parametric proportional hazards
mixture cure model (PHMC) as implemented in R package
smcure (Cai et al. 2012); (2) a parametric
mixture distribution model (Benoit et al. 2012), fitted by maximizing the likelihood function for the
model within the R optimization function
optim. Fitting more than one model, using different
implementations, is valuable to provide evidence on the sensitivity of the estimates to the model
properties.
Model (1) fits a common baseline survivor curve across all SQA categories (fish quality), based on the
observed pattern of mortalities, and then scales the risk to reflect the survival within each SQA
category. Model (2) assumes that the survival pattern can be modelled by the Weibull statistical
distribution, this is a relatively flexible distribution that can represent a range of survival functions
commonly encountered in ecological data. Here, we fitted Model (2) to each SQA category
separately to remove any assumption of similarities in their survivor curves.
The estimate of survival probability from each model was extracted to apply to the vitality data.
Applying survival rates to vitality data
For each species, the survival rate for each of the categories in the vigour assessment (Excellent,
Good, Poor, Moribund) were applied to the proportion of fish assessed with that category from all
sampled catches. Data were raised where appropriate to give the proportions at each vitality
category pooled across all sampled trips.
Summing across the proportions of catch at each vitality, multiplied by the survival rate for that
category gave an overall estimated survival rate of the observed trips. Three survival rates are
presented, one in the context of the captive observation period, the other two using the predicted
final survival rates for each of the vitality categories from the extension models.
Identifying factors that influence survival
Potential links between the vigour assessment in the sampled catch and variables related to each
fishing haul were examined for plaice from the trammel net fishery. This study was selected as a
range of variables covering the sea conditions, environmental variables, catch processing and catch
composition were available to analyse within the time constraints of the project. Vigour assessment
in the sampled catch was used as the response (rather than survival at the end of on-shore
observation), as links between vigour assessment and survival had been observed, using the
sampled catch provided a greater sample size and allowed the focus to be on factors related to the
hauls. The number and proportion of fish in each vigour assessment category was calculated for
each haul, and then linked to the haul data using a unique combination of haul date and haul
number. As a visual analysis, the vigour category proportions were plotted against each potential
influencing variable. Where appropriate, smooth curves (loess smoother with span of 0.75) were
added to the plots to aid interpretation.
To assess each variable’s ability to describe patterns in vigour category proportions, multinomial
statistical models were fitted to the counts in each category using function
multinom in R package
MASS (Venables et al. 2002). A separate model was fitted for each potential influencing variable,
with categorical variables as factors and continuous variables as linear terms within each vigour
category. A model’s fit was measured using the likelihood ratio statistic from comparing the model
to a null model which had the same vitality category probabilities for every haul.
33
The effect of reflex impairment and injury on survival
A Generalized Linear Model (GLM) with the binomial family and a logit link was used to examine which
injuries and reflexes had a significant impact on proportion of dead (D) and alive (A) fish. For both
species in study 1 we fit a binomial GLM to the reflexes and injuries, separately. The models were
estimated using the software R 3.1.0.
34
Results
Overview of results
Data from both fishery studies are summarised in Table 6.
Study 1 – Trammel net, plaice and sole
In total, 44 hauls of commercial sole trammel nets were made during two neap tides between 18th
August and 6th September 2014 (Table 5). The nets were deployed on the sea bed at depths ranging
from 14m to 30m (mean 21m), for soak durations of between 19hr 20min and 28hr 52min (av. 23hr
52min). Once the nets had been hauled, the time taken for the catch to be sorted, and hence the
maximum amount of time fish were exposed to the air, ranged from 20min to 2hr 34min.
The catch composition was dominated by starry smooth hound (1055), with sole (455) and plaice
(409) featuring as the next most abundant species. All sole and plaice caught were recorded and the
length distribution is shown in Figure 13. The mean lengths of sole and plaice caught in 4 ½ inch
(inner mesh) trammel nets were 35.8cm and 29.8cm respectively; plaice caught in nets designed to
catch marketable sized sole were notably smaller than the sole.
A total of 409 plaice and 455 sole were caught and assessed for vitality. In total, 83 plaice and 189
sole were assessed as being dead/moribund at the point that they were unmeshed from the nets.
The remaining fish were scored as either Excellent, Good, or Poor (Figure 13), and a proportion of
fish at each of these vitality scores was selected (by length) for the on-board observation tanks).
The 107 plaice and 96 sole retained for captive observation had a length profile comparable to the
total catch (Table 6, Figure 13). The Kaplan-Meier plots (Figure 15) show clear separation between
the vitality (vigour) categories, with the amount of survival in the expected order – the best survival
with Excellent vitality. This finding is supported by the results of the log-rank tests comparing the
survivor curves. Overall, there are statistically significant differences in survivor curves between
vitality categories for both species between Excellent fish and Good and Poor fish. These results
demonstrate that the vitality assessment effectively distinguished the chances of survival of
Excellent fish from the other vitality categories.
Fish were held in captivity for 76-81 hrs; survival probability for plaice was 72.7% for Excellent fish,
36.4% and 42.1% for Good and Poor. When weighted to the proportion of fish in each vitality
category in the total catch, the estimated survival in the observation period was 49.3% (37.1-59.8%)
(Table 8). Two models were used to forecast forward from the KM survival plots; when combined
across all hauls, because the a small number of fish died at the end of the observation (Figure 15);
the forecast survival estimate varied between 3.6-39.1% owing to the different sensitivities of the
model (Table 9).
For sole, the survival probability was 50.0% for Excellent fish, 0.0% and 6.3% for Good and Poor.
When weighted to the proportion of fish in each vitality category in the total catch, the estimated
survival in the observation period was 20.6% (14.8-27.9%) (Table 8). Two models were used to
forecast forward from the KM survival plots; when combined across all hauls, because the rate of
mortality of sole had not reached asymptote (Figure 15); the forecast survival estimate was lower at
18.6-20.3% (Table 9).
35
Simulated discarding of 32 fish was conducted to observe evidence of avian predation; 28 fish were
observed to actively swim away (Table 8). The remaining four fish were scored to be in Poor or
Moribund (Dead) condition and sank, unmoving, out of view. There were some seabirds present in
the area, but only one sea bird was observed to show interest in one discarded plaice, but it made
no attempt to pick the fish up. Therefore, there was no evidence of avian predation observed.
Ad hoc measurements of sea-surface temperature, air temperature, salinity and dO2are given in
Annex 3. Salinity at sea was 35ppt, whilst in the onshore tanks it was maintained at 30ppt; the air
temperature varied between 15.7 deg. C and 19.9 deg. C; and dO2 fell to 85% in the onshore tanks
but was often at 100%, and down to 44% in the tubs when fish were moved from the vessels to the
on-shore tanks.
Figure 12 Enmeshed sole in trammel net were moved to the side during the hauling process during
normal sorting
36
Table 5 Details of hauls, including soak time, sea conditions, and sorting times
Haul date
Tide Gear Haul
Haul
Haul
Soak
ICES
Wind
Wind
Swell Height (feet)
Light
Air
Sea
Total
(m)
Type
time
depth
time
rectangle
force
direction
level
temp.
surface
sorting
(m)
(h:m)
(lux)
(°C)
temp.
time
(°C)
(h:m)
18/08/2014 10.9
B
1
10:58
21
24:58
31E6 4 to 5
NW
3 to 4
17
01:04
B
2
12:26
23
25:56
31E6
5
NW
3 to 4
17
00:32
B
3
13:36
25
26:36
31E6 4 to 5
NW
3 to 4
17
00:48
A
4
15:16
23
27:46
32E6 3 to 4
NW
3
17
01:46
A
5
15:33
21
27:33
32E6 3 to 4
NW
3
17
02:34
19/08/2014 10.5
B
6
10:11
22
19:23
31E6
3
NW
2
16
00:28
B
7
11:17
21
22:58
31E6
3
NW
2 to 3
16
00:40
B
8
12:30
21
23:05
31E6
3
NW
2 to 3
16
00:26
20/08/2014
10
B
9
13:10
21
24:48
31E6
2
NW
1 to 1.5
20
00:36
B
10
14:27
22
27:22
31E6
2
NW
1 to 1.5
20
00:46
21/08/2014
9.9
A
11
09:44
18
24:19
31E6 3 to 4
NW
2 to 2.5
16
00:45
A
12
10:55
18
25:20
31E6 3 to 4
NW
2
16
00:38
B
13
12:22
22
22:27
31E6 4 to 5
NW
2 to 3
16
00:35
B
14
13:53
18
22:04
31E6 4 to 5
NW
3
16
01:59
22/08/2014 10.3
A
15
09:45
18
21:35
31E6
3
NW
1
17.2
17.8
00:42
B
16
11:14
19
21:49
31E6 3 to 4
NW
1 to 2
17.2
17.8
00:34
B
17
12:26
18
22:46
31E6
3
NW
1 to 2
17.2
17.8
00:31
23/08/2014 10.9
B
18
11:25
17
26:00
31E6
3
NW
1
16.6
17.7
00:41
A
19
12:29
17
26:29
31E6 3 to 4
NW
2
16.6
17.7
00:33
B
20
13:33
19
24:38
31E6 3 to 4
NW
1 to 2
16.6
17.7
01:46
01/09/2014 11.3
A
21
09:30
29
23:30
31E6
5
NW
1.5
49000
18.5
18
01:10
B
22
11:22
24
25:07
31E6
3
W
0.5
49000
18.5
18
00:43
A
23
12:43
21
26:12
31E6 3 to 4
W
0.5
49000
18.5
18
00:55
37
Haul date
Tide Gear Haul
Haul
Haul
Soak
ICES
Wind
Wind
Swell Height (feet)
Light
Air
Sea
Total
(m)
Type
time
depth
time
rectangle
force
direction
level
temp.
surface
sorting
(m)
(h:m)
(lux)
(°C)
temp.
time
(°C)
(h:m)
B
24
14:20
16
27:35
31E6 3 to 4
W
0.2
49000
18.5
18
01:00
B
25
15:40
14
28:40
31E6 2 to 3
S
0.3
49000
18.5
18
00:20
02/09/2014 10.9
A
26
09:30
29
22:15
31E6
2
SE
0.3
71500
17.7
18.1
01:05
B
27
11:30
24
23:00
31E6
3
SE
0.3
71500
17.7
18.1
00:40
A
28
12:45
24
22:30
31E6
2
E
0.3
71500
17.7
18.1
00:40
B
29
13:50
21
22:05
31E6
2
W
0.3
71500
17.7
18.1
01:00
03/09/2014 10.4
A
30
09:43
27
22:28
31E6
2
W
0.3
46600
20
18.8
01:02
B
31
11:24
23
22:54
31E6
2
W
0.3
46600
20
18.8
00:50
A
32
12:48
23
22:48
31E6
2
W
0.3
46600
20
18.8
00:52
B
33
14:18
22
23:03
31E6
2
W
0.3
46600
20
18.8
00:42
04/09/2014 10.3
A
34
13:52
30
26:45
31E6
3
NW
1 to 2
00:56
B
35
15:57
20
27:25
32E6
3
NW
1 to 2
00:53
A
36
17:50
20
27:40
31E6
3
NW
1 to 2
00:42
05/09/2014 10.2
A
37
09:58
25
21:28
31E6
0
V
0
00:54
B
38
11:31
20
20:01
31E6
0
V
0
00:36
B
39
12:40
20
19:20
31E6
0
V
0
00:43
A
40
14:06
23
20:26
31E6
0
V
0
00:47
06/09/2014 10.7
A
41
09:23
20
22:03
31E6
1
SE
0.5
63300
20.3
18.8
00:45
B
42
10:40
18
22:10
31E6
1
SE
0.5
63300
20.3
18.8
00:34
B
43
11:40
18
21:40
31E6
1
SE
0.5
63300
20.3
18.8
00:34
A
44
12:54
20
21:34
31E6
1
SE
0.5
63300
20.3
18.8
00:53
38
Table 6 Data summary
Study 1
Study 1
Study 2
Bristol Channel
Bristol Channel
Bristol, Channel
Area
ICES VIIf
ICES VIIf
ICES VIIf
Gear
Trammel net
Trammel net
Twin otter trawl
Mesh size: inner; outer
114mm; 610mm
114mm; 610mm
85mm
Target
Dover sole
Dover sole
Mixed demersal
Study period
18 Aug - 6 Sept
18 Aug – 6 Sept
10 Mar – 16 Mar
Fishing days
12
12
3
Hauls
44
44
10
Species
Plaice
Sole
Plaice
Mean length plaice catch
29.8
35.8
23.5
cm
Vitality assessed from
409
455
572
catch n
% plaice catch assessed
53
39
57
as excellent
% plaice catch assessed
10
10
13
as good
% plaice catch assessed
16
15
20
as poor
%plaice catch assessed as
20
46
10
dead/moribund
Captive observation
107
96
-
sample number
Captive observation
Onshore
Onshore
-
method
Mean length observed
30.2
36.8
-
cm
Observation period
76-81h
76-81
-
% survival of plaice catch
72.7
50.0
-
assessed as excellent
% survival of plaice catch
36.4
0.0
-
assessed as good
% survival of plaice catch
42.1
6.3
-
assessed as poor
% survival in observation
49.3 (37.1-59.8)
20.6 (14.8-27.9)
-
period for plaice catch
Modelled % survival with
no time constraint for
3.6-39.1
18.6-20.3
-
total plaice catch
39
Figure 13 Study 1 Length frequencies of plaice and sole in trammel net catches and held
for observation
40
Figure 14 Study 1 Semi-quantitative vigour vitality score for plaice and sole trammel net
catches
Vitality (vigour) of total catch
250
200
150
100
50
0
Excellent
Good
Poor
Moribund/dead
Plaice
Sole
Table 7 Study 1 Avian predation observations
Excellent Good Poor Moribund/dead Total
Mean Fish Length (cm)
25.3
25.6
25.3
27.0
Swam Clear
23
5
0
0
28
Bird(s) Interested
0
1
0
0
1
Birds fighting or competing
0
0
0
0
0
Picked up but rejected
0
0
0
0
0
Eaten
0
0
0
0
0
Lost sight of fish
0
0
3
1
4
41
Figure 15 Study 1 Kaplan-Meier estimates of survival probability
The Kaplan-Meier plots show clear separation between the vitality (vigour) categories, with the level
of survival in the expect order – i.e. best survival with Excellent vitality and survival decreasing with
vitality. This finding is supported by the results of the log-rank tests comparing the survivor curves.
Overall, there are statistically significant differences in survivor curves between vitality categories for
plaice and sole between pairs of categories except for Good and Poor. These results demonstrate
that the vitality assessment distinguished the chances of survival of fish assessed as Excellent
compared with other categories.
Plaice
Kaplan-Meier survival estimates by SQA
1.0
E
G
P
0.8
ility
0.6
b
a
b
ro
l p
a
iv
rv
u
0.4
S
0.2
0.0
0
10
20
30
40
50
60
70
Hours
Bristol Channel gill net fishery -
Plaice
Comparison
Chisq
p
E , G
12.8
<0.001
E , P
10.2
0.001
G , P
0.3
0.564
42
Sole
Kaplan-Meier survival estimates by SQA
1.0
E
G
P
0.8
ility
0.6
b
a
b
ro
l p
a
iv
rv
u
0.4
S
0.2
0.0
0
10
20
30
40
50
60
70
Hours
Bristol Channel gill net fishery - Sole
Comparison
Chisq
p
E , G
24.6
<0.001
E , P
19.7
<0.001
G , P
0.1
0.722
43
Table 8 Study 1 Survival of captive fish during observation time period and modelled for extended period
The table gives the overall percentage survival of the captive fish; the survival probability within the observation period with upper and lower 95% CIs from
the K-M analysis and also the predicted percentage survival based on a modelled asymptote in the survival curve from the two extension models. Extension
model 1 (ph) gives the output from a semi-parametric proportional hazards mixture cure model (PHMC) (Cai, Zou et al. 2012); Extension model 2 (Wei) gives
the outputs from a parametric mixture distribution model (Benoit, Hurlbut et al. 2012).
Captive observation results
Percentage
Survival
survival of captive
probability (KM)
Extension model 1 Extension model 2
Species SQA
fish
as percentage
lower 95%
upper 95%
(ph)
(Wei)
Excellent
72.7
72.7
61.3
81.3
56.6
0.0
Plaice Good
36.4
36.4
11.2
62.7
36.0
34.6
Poor
42.1
42.1
20.4
62.5
32.5
0.0
Excellent
50.0
50.0
37.5
61.3
49.2
45.1
Sole
Good
0.0
0.0
0.0
0.0
0.0
0.0
Poor
6.3
6.3
0.4
24.7
6.3
5.5
44
Table 9 Study 1 Estimated discard survival for all plaice and sole on observed trips using vitality as a proxy
The table presents the weighted mean survival proportions of the total catch from the captive observation estimates (Table 7) and the catch vitality profiles
Survival probability Survival probability
Proportion
Survival probability as percentage in
as percentage in
Survival with no
Survival with no
at vitality in as percentage in
obs. period
obs. period
time constraint
time constraint
SQA
total catch
obs. period
Lower 95%
Upper 95%
model 1
model 2
Excellent
0.53
38.8
32.7
43.3
30.2
0.0
Good
0.10
3.7
1.1
6.4
3.7
3.6
Plaice
Poor
0.16
6.8
3.3
10.1
5.2
0.0
Moribund/dead*
0.20
0.0
0.0
0.0
0.0
0.0
Survival rate %
49.3
37.1
59.8
39.1
3.6
Excellent
0.39
19.7
14.7
24.1
19.3
17.8
Good
0.10
0.0
0.0
0.0
0.0
0.0
Sole
Poor
0.15
1.0
0.1
3.8
1.0
0.9
Moribund/dead*
0.46
0.0
0.0
0.0
0.0
0.0
Survival rate %
20.6
14.8
27.9
20.3
18.6
*Moribund/dead individuals not assessed for survival in captive observation experiment; assumed 0% survival of fish assessed as moribund/dead in catch
45
Study 2 – Otter trawl, plaice
In total, 10 hauls of a commercial otter trawl were made during three days in March 2015. The tows
were conducted in depths ranging from 19m to 34m (mean 26m), for durations of between 2hr
45min and 4hr 45min (mean 3hr 52min). After hauling, the time taken for the catch to be sorted,
and the maximum amount of time fish were exposed to the air, ranged from 15min to 30min (mean
23min).
The catch composition was dominated by Lesser Spotted Dogfish, rays and plaice. All unwanted
plaice caught were recorded by length (to the nearest cm below; Figure 16). Only a few plaice, 3% of
the total number, were retained by the vessel; the mean length of the unwanted plaice was 23.6 cm.
Of the 572 plaice assessed, 18 (3%) were categorised as dead at the point that they would be
discarded. The majority of plaice (57%) were assessed as being in Excellent condition with the
remaining scored as either Good or Poor (Figure 17).
Simulated discarding of 70 plaice was conducted to observe evidence of avian predation. 39 fish
were observed to actively swim away (Table 11), most of which were assessed as Excellent or Good.
The observers lost sight of the remaining 31 fish which were assessed as mostly Moribund (Dead).
Therefore, no evidence of avian predation was observed.
Figure 16 Study 2 Length frequency of plaice caught by a Bristol Channel otter trawler
46
Figure 17 Study 2 semi-quantitative vigour vitality score for plaice catches in a Bristol
Channel otter trawler
Plaice Study 2 - Bristol Channel otter trawl
500
450
400
350
300
250
200
150
100
50
0
Excellent
Good
Poor
Moribund
Dead
Table 10 Study 2 Avian predation observations
Excellent Good Poor Moribund/dead Total
Mean Fish Length (cm)
22.4
23.6
23.9
23.0
Swam Clear
16
13
0
10
39
Bird(s) Interested
0
0
0
0
0
Birds fighting or competing
0
0
0
0
0
Picked up but rejected
0
0
0
0
0
Eaten
0
0
0
0
0
Lost sight of fish
1
4
8
18
31
47
Table 11 Study 2 inferring discard survival for plaice on observed trips on-board a Bristol Channel otter trawler using vitality as a proxy
Captive observation data from comparable otter trawl
Study 2 results
fishery in ICES VIIe*
Inferred discard survival %
Survival
Survival
Survival
probability
probability
probability
as
as
Survival
Survival
Survival
as
percentage
percentage
with no
with no
probability
Extension
Extension
percentage
in obs.
in obs.
time
time
Number
(KM) as
lower upper model 1
model 2
in obs.
period
period
constraint
constraint
Vitality
(raised)
Percentage percentage
95%
95%
(ph)
(Wei)
period
Lower 95%
Upper 95%
model 1
model 2
Excellent
324 (463)
81%
90.2
82
94.8
84.6
90.2
73
66
77
68
73
Good
73 (81)
14%
70.4
59.7
78.8
40.6
71.3
10
8
11
6
10
Poor
117 (138)
24%
28.7
18.8
39.5
2.3
18.3
7
5
10
1
4
Moribund
40 (42)
7%
5
0.9
14.8
4.7
4.6
0
0
1
0
0
Dead
18 (22)
4%
0
0
0
0
0
0
0
0
0
0
Total
572 (746)
100
90
79
99
75
88
* Other survival studies were conducted in separate contract at the same time as this project. Data from these studies were generated using the same
methodology by Cefas in Defra funded project MF1234 (Catchpole, unpubl. 2015, Tom Catchpole, Peter Randall, Robert Forster, Sam Smith, Stuart
Hetherington, Victoria Bendall, Frank Armstrong. Estimating the discard survival rates of selected commercial fish species (plaice - Pleuronectes platessa) in
four English fisheries (MF1234/C6160), May 2015, Cefas report)
48
Factors influencing discard survival
The effect of impaired reflexes
The binomial GLM model in case study1, the trammel net fishery, showed that plaice with impaired
orientation had significant higher mortality that the unimpaired plaice. The orientation impairment
was the only reflex that showed significant association with the proportion of dead: alive fish. For
sole, no impaired reflexes showed significance in affecting the proportion of dead and alive fish
(Table 12). In the case study 2, the otter trawl fishery, only plaice was assessed for vitality, no captive
experiment was conducted. Therefore, it was not possible to make any analysis on the effect of
impairments on the mortality of plaice, around one third of fish displayed impairment in all reflex
tests. Summary of vitality scores are presented in Table 13.
Table 12 – Summary data for case study 1 – Trammel net fishery, with the number of fish dead and
alive in the experiment, when impaired and unimpaired for each vitality reflex, percentage (%) of
dead fish impaired, percentage (%) of alive fish impaired,
p value from binomial GLM. Number of
impaired/ unimpaired and proportion of impaired plaice and sole in the total catch.
Experiement
Population
Species
Reflex name
Reflex response
% of dead fish % of alive fish
Alive
Dead
p-value Number
Proportion impaired
impaired
impaired
unimpaired
64
34
253
Body flex
13%
6%
0.973
12%
impaired
4
5
34
unimpaired
66
38
271
Operculum
3%
3%
0.277
6%
impaired
2
1
16
Plaice
unimpaired
67
35
262
Startle touch
10%
1%
0.293
9%
impaired
1
4
25
unimpaired
63
27
235
Orientation
31%
7%
0.012*
18%
impaired
5
12
52
unimpaired
33
58
214
Body flex
6%
3%
0.994
12%
impaired
1
4
29
unimpaired
33
59
217
Operculum
5%
3%
0.992
11%
impaired
1
3
26
Sole
unimpaired
33
54
194
Startle touch
13%
3%
0.993
20%
impaired
1
8
49
unimpaired
33
52
193
Orientation
16%
3%
0.993
21%
impaired
1
10
50
Table 13 – Summary data for case study 2 – otter trawl fishery, with the number of
impaired/unimpaired fish and proportion impaired in the population.
Proportion
Species
Reflex name
Reflex response
Number
impaired
unimpaired
402
Body flex
30%
impaired
170
unimpaired
434
Operculum
24%
impaired
138
Plaice
unimpaired
389
Startle touch
32%
impaired
183
unimpaired
386
Orientation
33%
impaired
186
49
The effect of injuries
In the trammel net case study, for plaice and sole net marks were observed in 70% and 79% of fish,
respectively, and was the most prevalent injury. Abrasion and scale loss were also frequently seen,
in both species but were more common in sole (Table 14). The same analyses with the binomial GLM
was applied to the injuries observed for each species in the trammel net case study. The GLM results
for plaice showed the injuries that had the most significant association on the proportion of dead
fish were internal organs exposure (p < 0.01) fin fraying and abrasion (p = 0.05)(Table 14). On the
other hand, the injuries that caused significantly higher proportion of dead sole were net marks and
abrasion.
Table 14 - Summary data for case study 1 with the number of fish dead and alive in the experiment,
when injured and not injured for each injury, percentage (%) of dead fish injured, percentage (%) of
alive fish injured,
p value from binomial GLM. Number of injured/not injured and proportion of
impaired plaice in the total catch.
Experiement
Population
Species
Injury
Response
% of dead
% alive fish
Alive
Dead
p- value Number
Proportion injured
fish injured
injured
not injured
28
9
87
Net marks
77%
59%
0.744
70%
injured
40
30
200
not injured
63
29
257
Internal organs exp
26%
7%
0.022*
10%
injured
5
10
30
not injured
62
26
236
Fin fraying
33%
9%
0.056·
18%
injured
6
13
51
not injured
68
36
277
Wounding
8%
0%
0.993
3%
injured
0
3
10
not injured
52
30
214
Plaice
Scale loss
23%
24%
0.461
25%
injured
16
9
73
not injured
53
21
183
Abrasion
46%
22%
0.051·
36%
injured
15
18
104
not injured
68
36
283
0%
0.993
1%
Exophthalmia
injured
0
3
8%
4
not injured
64
37
275
6%
0.932
4%
Bleeding
injured
4
2
5%
12
not injured
68
37
284
0%
0.994
1%
Mucus loss
injured
0
2
5%
2
not injured
16
8
50
Net marks
87%
53%
0.005*
79%
injured
18
54
193
not injured
32
55
218
Internal organs exp
11%
6%
0.29
10%
injured
2
7
25
not injured
30
42
197
Fin fraying
32%
12%
0.289
19%
injured
4
20
46
not injured
32
55
214
Wounding
11%
6%
0.537
12%
injured
2
7
29
not injured
16
17
100
Sole
Scale loss
73%
53%
0.563
59%
injured
18
45
143
not injured
17
13
70
Abrasion
79%
50%
0.048*
71%
injured
17
49
173
not injured
34
61
242
0%
0.994
0%
Exophthalmia
injured
0
1
2%
1
not injured
32
52
213
6%
0.375
13%
Bleeding
injured
2
10
16%
33
not injured
34
61
241
0%
0.994
1%
Mucus loss
injured
0
1
2%
2
In the otter trawl case study, scale loss was observed in 70% of fish, and was the most prevalent
injury, followed by bruising (41%) and bleeding (34%) (Table 15).
50
Table 15 – Summary data for case study 2 – otter trawl fishery, with the number of injured/not injured
fish and proportion injured fish in the population.
Population
Species
Injury
Response
Number
Proportion injured
not injured
561
Net marks
2%
injured
11
not injured
567
Internal organs exp
1%
injured
5
not injured
550
Fin fraying
4%
injured
22
not injured
538
Wounding
6%
injured
34
not injured
173
Scale loss
70%
injured
399
Plaice
not injured
496
Abrasion
13%
injured
76
not injured
375
34%
Bleeding
injured
197
not injured
446
22%
Mucus loss
injured
126
not injured
335
41%
Bruising
injured
237
not injured
462
19%
Scratches
injured
110
Factors influencing survival
Annex 5 summarizes the fit of multinomial models to the counts by vigour assessment category
(vigour), using each variable singly. This analysis was conducted for plaice only. In terms of the
criteria p<0.05, 13 of the 19 variables considered to improve the description of the vigour categories
by haul compared to using the same proportions for all hauls. However, care in interpretation is
required as many of the variables are linked and analysis of large numbers of variables can generate
a proportion of spurious results.
The model results show some effects interest that can be observed with a visual analysis. The wind
strength, seas-state and swell height were all associated with vigour category as the proportion of
Excellent plaice was higher for hauls undertaken in calm weather conditions (Table 16 and Figure
18). There was also an indication that the two different gear types were associated with different
vitality category with more Excellent fish caught in Mono-filament nets than in the Multi Mono-
filament nets (Table 16).
51
Sea State
1.00
0.75
A
Q
S
SQA
in
n
E
0.50
G
rtio
o
P
p
D
ro
P
0.25
0.00
SMOOTH CALM
CALM-SLIGH
S T
LIGHT
SLIGHT-MOD
MOD
MOD-ROUGH
Sea State
Swell Height
1.00
0.75
A
Q
S
SQA
in
E
n
0.50
G
rtio
o
P
p
D
ro
P
0.25
0.00
0
0.2 0.3 0.5
1 1-1.5 1-2 1.5
2
2-3
3
3-4
Swell Height (feet)
Wind Force
1.00
0.75
A
Q
S
SQA
in
n
E
0.50
G
rtio
o
P
p
D
ro
P
0.25
0.00
0
1
2
2-3
3
3-4
4-5
5
Wind Force
Figure 18 The proportion of fish in each vigour assessment category for each haul vs sea state, swell
height and wind force; to provide visual analysis.
52
Table 16 Vitality (vigour) associated with wind force, seas state, swell height and gear type; defined
as end of hauling to end of sort time.
Predicted proportions in Vigour category
Wind
Force
E
G
P
D
0
0.89
0.00
0.11
0.00
1
0.85
0.00
0.07
0.08
2
0.44
0.16
0.21
0.19
2-3
0.00
0.00
0.00
1.00
3
0.57
0.13
0.11
0.18
3-4
0.48
0.11
0.15
0.25
4-5
0.34
0.11
0.21
0.34
5
0.28
0.06
0.33
0.33
Swell
Predicted proportions in Vigour category
Height (ft)
E
G
P
D
0
0.89
0.00
0.11
0.00
0.2
0.50
0.08
0.17
0.25
0.3
0.48
0.13
0.20
0.19
0.5
0.80
0.05
0.05
0.09
1
0.38
0.12
0.50
0.00
1-1.5
0.38
0.17
0.21
0.24
1-2
0.54
0.13
0.13
0.20
1.5
0.37
0.13
0.25
0.25
2
0.74
0.11
0.11
0.04
2-3
0.53
0.16
0.11
0.21
3
0.34
0.10
0.16
0.40
3-4
0.23
0.10
0.29
0.39
Predicted proportions in Vigour category
Sea State
E
G
P
D
SMOOTH
0.89
0.00
0.11
0.00
CALM
0.60
0.10
0.15
0.16
CALM-
SLIGHT
0.50
0.50
0.00
0.00
SLIGHT
0.55
0.12
0.16
0.17
SLIGHT-
MOD
0.31
0.12
0.19
0.37
MOD
0.41
0.14
0.14
0.32
MOD-
ROUGH
0.28
0.06
0.33
0.33
Predicted proportions in Vigour category
Gear
E
G
P
D
A
0.62
0.07
0.12
0.18
B
0.47
0.13
0.19
0.22
53
Discussion
The project delivered its aim to generate discard survival estimates for selected species for (Welsh)
commercial fisheries operating in the Bristol Channel. The structure of the project dictated the
method that could be used, and this was developed within the project and in parallel with the ICES
Workshop on Methods to Estimate Discard Survival (WKMEDS). Therefore, this project has provided
a testing ground for the methods and concepts developed from that ICES group and observations
from the project have fed back to improve the guidance on how best to conduct these experiments.
The approach selected was to use vitality assessments during a representative range of conditions
and combining this with the captive observation of individuals with different vitality levels to
generate an overall weighted mean discard survival estimate.
It is recognised in the literature that not all discards die. Although there is considerable variation in
estimates of discard survival within and between studies, research has shown that in some
circumstances the proportion of discarded fish that survive can be substantial. Studies that have
looked at flatfish species, including plaice and sole, show variable results, with survival rates
presented in the range of ~40-80%, although zero survival was observed in some experiments (STECF
14-19). These results studies differed in the fishery, operational characteristics and method, making
it difficult to compare between studies. Moreover, the results often report only a component of
discard mortality, either by not including predation or by presenting estimates that do not account
for the full time period over which discarded fish may die.
For many European fisheries-species combinations (particularly regulated species) there are no
discard survival estimates available. A recent literature review of studies on discard survival rates for
STECF (Revill, 2012) showed there are few experiments for passive gears, and no reported discard
survival data for plaice and sole in European gill and trammel net fisheries. Here we present the first
estimated discard survival of plaice and sole caught in gill nets. This study demonstrated that after
an observation period of 76-81h, the percentage of discarded plaice surviving after normal
commercial fishing practice was 49.3% (37.1-59.8%). For Dover sole, after this period, discard
survival was 20.6% (14.8-27.9%). This lower survival reflected that 46% of caught sole were assessed
as dead/moribund at the point of discarding and only fish assessed as being vigorous at the point of
discarding survived. Although, the observation period was necessarily limited, an attempt was made
to forecast a final survival estimate that would take account of all discard mortality, based on
changes in the discard mortality rate over the observation period. Two models were applied,
providing discard survival estimates of 3.6-39.1% for plaice and 18.6-20.3% for sole. The difference
in the plaice modelled estimates was driven by the death of a few individuals at the end of the
observation period, demonstrating the sensitivity of the model to these data.
In the otter trawl fishery, only plaice was investigated. For this fishery only vitality, reflex and injury
data were generated. To generate an estimate of survival for this fishery required the application of
survival rates from another source. Using the same method, during the same period, survival
estimates by vitality category were generated from an otter trawler working in a neighbouring ICES
sub division. The estimate generated inferred that survival was the same for each vitality as was
observed in a closely related otter trawl fishery. This assumed that the stresses endured by the fish
in one fishery were the same as those of the other and that health vitality assessments were
consistent across both studies. The vessel was smaller, the fishing gear lighter, depths shallower and
54
towing times shorter, suggesting that the survival rates are not likely to be lower than inferred. The
estimated survival for otter trawl caught plaice, accounting for these assumptions, is 75-88%
survival.
There are a number of factors that are known to affect the survival of discarded fish and these can
be classified into three broad categories: technical (e.g. fishing method, catch size and composition,
handling practices on deck), environmental (e.g. changes in temperature, depth, light conditions)
and biological (e.g. species, size age, physical condition, occurrence of injuries) (Davis, 2002;
Broadhurst, Suuronen et al. 2006).
All fishing methods induce stress and cause a degree of injury to captured fish (e.g. internal and
external wounding, crushing and scale loss) (Davis and Ryer, 2003). Fish that are captured in
trammel nets may be ‘gilled’ by the meshes of the inner wall, or they may be ‘bagged’ by pushing
the inner wall through the larger meshes of the outer wall to create a pocket. In this experiment it
was observed that the two species of flatfish studied meshed differently to one another. The
majority of sole were ‘gilled’ by the inner wall of the net. As a product of their body shape, the
viscera of the sole were visibly compressed by the twine (often multiple meshes) and the net
marking injuries observed reflected this. The majority of plaice were observed to be ‘bagged’ by the
outer wall. The forward movement of plaice through a trammel net causes inner meshes to be
picked up (possibly on the spine behind the anus) and carried through the outer wall to create a
pocket or ‘bag’ from within which they are unable to escape. Once captured, the fish remained
‘gilled’ or ‘bagged’ in the net until the tier of nets was hauled and cleared. Generally speaking, the
longer fish are exposed to the fishing gear, the more severe the stress, leading to exhaustion and
increased physical damage. For the fishing operations in this study, captured fish remained in the net
for a maximum of between 19hr 20min and 28hr 52min.
Other species were caught in the nets and may have contributed to the stress or injuries if they were
closely meshed, particularly the substantial catches of large elasmobranchs recorded in this study.
The hauling process involved the nets being mechanically raised from the sea bed to the surface,
involving a change in depth (14-30m to surface), hence pressure, and a change in sea temperature.
As the nets are hauled, the headrope and the footrope are brought together and may twist, so the
fish may suffer from compression injuries, especially as the nets are hauled over the rollers of the
net hauler.
From the moment the fish came out of the water, they were subjected to stressors associated with
air exposure. The nets were placed in a pile on the deck until the hauling process was complete and
catch sorting began, although, as part of normal commercial practice, care was taken to prevent the
sole and plaice from being crushed by placing these fish to the edge of the pile (Figure 10). Exposure
to air is an integral part of fish capture and is directly related to the sorting and handling times on
deck. On this vessel, sorting times varied from a few seconds, for fish that were un-meshed
immediately, to more than 2hrs, when poor weather led to catch sorting being delayed until the
vessel was in port. The process of un-meshing the individual fish also varied from a few seconds to
several minutes, depending on how the fish were meshed. Due to their body shape and the way that
they ‘gilled’ in the inner meshes, the removal of sole from the net involved gripping the head and
pulling the fish through the meshes, further compressing the gut and potentially injuring the head.
The ‘bagged’ plaice were generally removed from the net by prising the meshes back over the head,
55
as the diamond/wedge body shape and the size of the plaice prevented them from being pulled
through the meshes. It was therefore observed that removing the fish from the net would have
induced substantial stress on the fish.
Previous studies have shown that air exposure is one of the greatest contributors to discard
mortality rates (Davis 2002, Broadhurst, Suuronen et al. 2006) and that reducing handling time and
exposure to air could be a useful measure to increase discard survival (Benoit
et al., 2010; Davis and
Ryer, 2003). Effects of air exposure on deck may be exacerbated by simultaneous exposure to direct
sunlight and increased temperatures, which can lead to rapid dehydration. Fish may already have
suffered skin injuries and scale loss as a result of the capture process; the exposure to air (wind) and
sunlight will have synergistic effects. This study was conducted during the commercial sole fishery in
summer, at a time when air temperatures on deck ranged from 16°C to 20°C and conditions were
generally bright and sunny, with up to 21kts of wind and sea surface temperature was 18°C.
Temperature has been identified as an important factor effecting survival assessments, with high
temperatures, of the water and air associated with lower rates of survival.
As conditions in the fishery change (e.g. seasons, areas fished), so can the resulting discard mortality
rates (Benoit
et al., 2011). In our opinion, the presence of observers on board the vessel did not
influence the catch handling process and the conditions experience by the fish in our study were
consistent with the types of condition present in the fishery at this time. The vessel used in case
study 1 was operated by one experienced crew member and the process of boarding the nets and
un-meshing the fish was consistent with normal routine. The only difference was that at the point of
normal discarding, the fish were handed to the observers for assessment. Due to the number of trips
and timing of the study it was not possible to investigate different sorting practices that might
reduce air exposure or investigate the potential effect of different water temperatures. However,
there were some indications of factors that did influence survival, namely the weather conditions
and the fishing gear construction material. Survival was observed to be lower during poor weather
(quantified as wind force, swell height and sea state and when using multi-monofilament nets
compared with monofilament nets. Multi-monofilament netting is a cross between multifilament
netting and monofilament netting and is composed of several strands of monofilament twine loosely
twisted together. There is an indication that this design of netting, may induce more stress on the
fish compared with monofilament leading to lower survival rates. Therefore, different netting
construction designs offer one potential mean to increase the survival rates.
This study generated discard survival estimates based on captive observation and vitality
assessments as a proxy. We found no evidence of avian predation, but without the time or the
resources within this project to conduct tagging experiments, the levels of marine predation remain
unknown. Discarded fish may be susceptible to increased predation risk due to impaired swimming
abilities (e.g. loss of orientation, reduced swimming speed) as a result of injuries or post-traumatic
behaviour. Davis and Ryer (2003) found that behavioural impairment, in the fish species studied,
lasted at least 2hrs with fish recovering with 24hrs, and that behavioural impairment was correlated
to the magnitude of the stress. Increased risk of infection, as a result of scale loss or skin injuries,
may also eventually induce mortality in the medium to long term.
Captive observation studies that exclude predation and do not account for delayed mortality
resulting from injuries or infection, are likely to represent over estimates of actual discard survival
56
under commercial fishing conditions. Conversely, unless suitable experimental controls are
employed, stresses associated with handling during transfer and the holding of fish in captivity can
induce mortality and could lead to under estimates of discard survival (Benoit et al. 2010, Depestele
et al. 2014) Portz
et al., 2006). Ideally, the mortality associated with captive conditions would be
estimated using control fish that had not been subjected to the capture and handling processes but
held in identical conditions to the treatment fish.
The control experiments at the laboratory with the on-shore holding tanks demonstrated zero
mortality and unimpaired behavioural reflexes in the control fish, indicating that the design of the
system did not induce mortality or negatively affect vitality. The treatment fish were held in a
different location, however, the tanks were supplied with a constant supply of temperature and
salinity regulated seawater. It was not logistically feasible to conduct control experiments on-board
the vessel and during the transit from the vessel to the shore unit. The captive fish will have
undergone a number of stresses additional to the capture and discard process. It can, therefore, not
be fully determined whether it was the treatment or the method that was responsible for the
observed mortality of fish in the holding tanks. Physical damage caused by being held in tanks on
board a moving vessel, changes in light, salinity, temperature, water quality and being held in close
proximity with other fish, all exert stress. Where these stressors are occurring, they will likely have
additive effects to the treatment stressors already encountered and reduce the observed survival
rates. The on-board tanks were filled with fish from the bottom up, therefore, any increasing
mortality rates through the stack of tanks would indicate an experimental effect of the time spent in
the tanks, the position in the stack of the fish or to different qualities of the seawater. The potential
for an on-board tank effect was explored by ranking the proportion of deaths in each tank and
conducting a Spearmen’s rank correlation test. The absence of any significant difference between
tanks (Spearman’s Rank Correlation, 0.2, -0.4 for Excellent plaice and sole; number of survivors were
insufficient for other categories) indicates that the on-board tanks had limited effect on survival.
The port of operation in this study presented logistical challenges for locating the onshore tanks. The
cessation of water flow to the on-board tanks as the vessel approached the port, and the time taken
to pass through the lock and in transit to the onshore tanks, led to reduced dO2 and increased water
temperatures. While healthy individuals may be able to tolerate worsening conditions, these
additional stressors may have increased mortality of the already stressed captive fish. Portz
et al. (2006) stated that water quality is one of the most important contributors to fish health and stress
level, and that short term exposure to poor water quality can result in permanent damage or
mortality if physical or chemical variables combine to reach lethal levels.
Experiments conducted by Davis and Ryer (2003) showed increased stress and mortality in fish that
were sequentially subjected to increased seawater temperature and air exposure, following a
simulated trawl process. The treatment fish in our experiment were subjected to water temperature
up to 19.9°C, increased from a sea-surface temperature of around 18°and a reduction in dO2 down
to 44% during the transit phase. The dO2 concentration of water decreases with increasing
temperature, and in these experiments would have decreased further as a result of increased
metabolic activity and oxygen consumption of stressed fish. We did make attempts to aerate the
water in the transportation containers with battery operated air stone pumps, but it was believed
that the noise and vibrations generated by the pumps in the close confinement of the vehicle may
have added to the stress levels and been counterproductive.
57
The potential stressors on the captive fish associated with the methodology in this study, are likely
to have resulted in experimental induced mortality and therefore underestimated survival. The
stressors of temperature, salinity and dO2 are factors known to increase mortality in fish. Specifically
these stressors included:
Handling fish to conduct the vitality assessments, length measurements and to put fish
into the on-board tanks
Captivity in the on-board tanks (movement caused by vessel movement; proximity with
other fish; serial flow of water from top to bottom tank)
Stopping water flow to on-board tanks on approach to port until docked (reducing dO2)
Transfer of fish into tubs (handling of fish)
Carrying tubs off the vessel and transporting, by van, to onshore holding tanks
(increased temperature, reduced dO2, movement)
Handling the fish to transfer into onshore tanks
Adjusting to salinity and temperature change in the onshore tanks
Monitoring fatalities using tail grab
To be able to use the assessments of fish vitality as a proxy for survival when combined with captive
observation results, two conditions were required. Firstly, scientific fieldworkers had to be able to
assess the vitality of fish consistently, in time and in different conditions. Secondly, to be able to use
vitality assessments as a proxy for survival, there must be a significant relationship between survival
and vitality score. The first condition was considered to have been met in the trammel net study, by
having only one fieldworker making all the health assessments. The second condition was also
considered to have been met in the trammel net study because it could be demonstrated that there
were statistically significant differences in survivor curves between vitality categories for plaice and
sole between pairs of categories except for Good and Poor, which had low levels of survival. These
results demonstrated that the vitality assessment effectively distinguished the chances of survival,
and therefore could be used as a proxy for survival.
For survival estimates to be representative of the fishery, vitality data should be generated for fish
discarded during all conditions of a fishery. However, because conditions are constantly changing,
without a continuous vitality monitoring programme, the survival estimates may be representative
only for the trips from which vitality data have been collected. To extrapolate the results from this
study to the fishery, it must be assumed that the combination and strength of stressors on the
discarded fish are the same on all trips as those from which vitality data were collected. It can be
stated that the trips from which these data were generated were conducted under normal
representative commercial fishing conditions.
Conducting survival studies on small commercial vessels in remote ports is technically and logistically
challenging. The vessels are restricted in deck space and can hold only small numbers of fish in
suitable tanks, and these must be transferred to shore when fishing for less than one day; this meant
that the use of controls had to be limited and there were unavoidable additional stressors exerted
on the fish. The survival estimates should, therefore, be interpreted as minimum discard survival
estimates that do not account for experimental induced mortality, that exclude marine predation
but do include avian predation. Here we present the first estimates from a static net fishery in
Europe of plaice and sole discard survival rates. The observed survival estimates was 49.3% for plaice
58
and 20.6% for sole after 77-81hrs; and modelled estimates produced survival rates of 3.6-39.1% for
plaice and 18.6-20.3% for sole. Using captive observation results from a similar neighbouring otter
trawl fishery, produced inferred discard survival estimates for plaice caught by an otter trawler in
the Bristol Channel of 75-88%.
59
Conclusions
The project achieved its aim to generate discard survival estimates for selected species in fisheries
operating off the Welsh coast in the Bristol Channel. Better health condition of plaice was
significantly associated with higher survival, validating the integrated method of combining the
assessed vitality of fish from the catch with the survival probability associated with those vitalities.
The project generated both experimental estimates within a defined observation period, and
modelled results to account for predicted mortalities beyond the observation period.
This study demonstrated that after an observation period of 76-81h, the percentage of discarded
plaice surviving after normal commercial fishing practice was 49.3% (37.1-59.8%). For Dover sole,
after this period, discard survival was 20.6% (14.8-27.9%). Modelling the predicted final rates
beyond the observation gave discard survival estimates of 3.6-39.1% for plaice and 18.6-20.3% for
sole. Using captive observation results from a similar otter trawl fishery in a parallel study combined
with health assessment, produced inferred discard survival estimates for plaice caught by an otter
trawler in the Bristol Channel of 75-88%.
All estimates excluded marine predation, but include avian predation, of which none was observed.
Furthermore, the stressors exerted on the fish from the method, including temperature differences,
handling, confinement, close proximity to other fish and dissolved oxygen depletion, were likely to
have induced some experimental mortality. Therefore, the results presented here should be
interpreted as minimum estimates of discard survival, excluding marine predation.
There were many factors with the potential to effect survival and the relatively low number of
replicates of the treatment making it difficult to identify the key influencing variables. However,
some initial analysis of the factors that influence survival showed that lower survival was associated
with poor weather conditions, and the netting construction type was also a possible factor. There
was an indication that higher survival was associated with monofilament nets compared with multi-
monofilament nets, suggesting that changing the net design could provide a mechanism to increase
survival rates.
The survival estimates generated here are representative of the observed trips. Assumptions must
be made to extrapolate the data to vessel and fleet levels. However, this evidence is considered to
provide scientifically robust estimates of discard survival and will inform fisheries managers of the
appropriateness and potential to develop proposals to gain exemption from the European landing
obligation under the high survivability provision.
60
Acknowledgments
The authors would like to thank the skippers of the two fishing vessels that took part in this study –
We would also like to thank the University of Wales, Swansea,
specifically
for allowing us to house the onshore tanks on their premises. Thanks also go
to the participants of ICES WKMEDS, who have collectively been developing the methods during the
course of the project. This project was funded by the Welsh Government.
61
References
(2010). "Assessing the factors influencing discard mortality of
demersal fishes using a semi-quantitative indicator of survival potential." Fisheries Research
106(3):
436-447.
(2012). "Estimating fishery-scale rates of discard
mortality using conditional reasoning." Fisheries Research
125: 318-330.
(2006). "Estimating collateral mortality from towed
fishing gear." Fish and Fisheries
7(3): 180-218.
(2012). "Smcure: An R-package for estimating semiparametric
mixture cure models. ." Computer methods and programs in biomedicine,
108: 1255-1260.
(2002). "Key principles for understanding fish bycatch discard mortality." Canadian
Journal of Fisheries and Aquatic Sciences
59(11): 1834-1843.
, and
(2003). "Understanding Fish Bycatch Discard And Escapee Mortality."
AFSC Quarterly Report.
and
(2014). "Short-term survival of
discarded target fish and non-target invertebrate species in the "eurocutter" beam trawl fishery of
the southern North Sea." Fisheries Research
154: 82-92.
EU (2013). Regulation 1380/2013 on the Common Fisheries Policy (basic regulation), published in the
Official Journal 28 December 2013.
ICES (2014). Report of the Workshop on Methods for Estimating Discard Survival (WKMEDS), 17–21
February 2014, ICES HQ, Copenhagen, Denmark. ICES CM 2014/ACOM:51. 114 pp.
and
(2006). "Stress-associated impacts of short-term holding
on fishes." Reviews in Fish Biology and Fisheries
16(2): 125-170.
(2012). "Survival of discarded fish. A rapid review of studies on discard survival rates." DG
MARE A2. Request For Services Commitment No. S12.615631
and
(2013). "Mortality of adult plaice, Pleuronectes
platessa and sole, Solea solea discarded from English Channel beam trawlers." Fisheries Research
147: 320-326.
STECF (2014). Scientific, Technical and Economic Committee for Fisheries (STECF) – Landing
Obligation in EU Fisheries - part II (STECF-14-01). 2014. Publications Office of the European Union,
Luxembourg, EUR 26551 EN, JRC 88869, 67 pp.
Venables, W. N. and B. D. Ripley (2002). Modern Applied Statistics with S, 4th edn. Springer, New
York, 495pp.
62
Annexes
Annex 1 STECF EWG 14-11 Table of survival estimates
Relevant species for which discard survival estimates are available, the gear and location of the
study, the literature reference, the time period of observation from the point of discarding and the
minimum and maximum levels of survival observed in the study.
Max of
Min of
discard
discard
survival
Common
Observation
survival
rate higher
Species
name
Gear
Location
Reference
period
lower limit
limit
Pleuronectes
English
Revill
et
al.
platessa
Plaice
Beam trawl
Channel
(2013)
3 days
37.3
79.6
Pleuronectes
Beam trawl
Depestele et al.
platessa
Plaice
("eurocutter")
Belgium
(2014)
77h
48
69
Pleuronectes
platessa
Plaice
Otter trawl
Germany
Kelle (1976)
7 days
12
70
Pleuronectes
Berghalm et al.
platessa
Plaice
Otter trawl
North Sea
(1992)
5 days
0
100
Pleuronectes
The
van Beek et al.
platessa
Plaice
Otter trawl
Netherlands
(1990)
3.5 days
0
48
Pleuronectes
van Marlen et
71h; 133-158h;
platessa
Plaice
Pulse beam trawl
North Sea
al. (2013)
157h
0
80
Pleuronectes
van Marlen et
platessa
Plaice
Pulse beam trawl
North Sea
al. (2005)192h
12
59
Berghalm et al.
Solea solea
Sole
Otter trawl
North Sea
(1992)
5 days
71
100
English
Revill et al.
Solea solea
Sole
Beam trawl
Channel
(2013)
3 days
53.1
76.4
Beam trawl
Depestele et al.
Solea solea
Sole
("eurocutter")
Belgium
(2014)
91h
14
29
Solea solea
Sole
Demersal trawl
Germany
Kelle (1976)
7 days
33
59
Berghalm et al.
Solea solea
Sole
Demersal trawl
North Sea
(1992)
5 days
71
100
The
van Beek et al.
Solea solea
Sole
Demersal trawl
Netherlands
(1990)
3.5 days
4
37
van Marlen et
36h; 72h; 133-
Solea solea
Sole
Pulse beam trawl
North Sea
al. (2013)
158h; 204h
27
70
van Marlen et
Solea solea
Sole
Pulse beam trawl
North Sea
al. (2005)
192h
17
54
Rays and
Enever et al.
Elasmobranch
skates
Otter trawl
U.K
(2009)
3 days
55
55
Rays and
Revill et al.
Elasmobranch
skates
Beam trawl
U.K
(2005)
2.5 days
92
100
Rodriguez-
Rays and
Cabello et al.
Elasmobranch
skates
Fish trawl
Spain
(2005)
1 hour
78
78
Rays and
Hueter et al.
Elasmobranch
skates
Gillnet
U.S.A
(2006)
Tagging
60
69
Gurshin and
Rays and
Szedlmayer
Elasmobranch
skates
Hook and line
U.S.A
(2004)
6 hours
90
90
Rays and
Enever et al.
Elasmobranch
skates
Otter trawl
U.K
(2010)
2 days
55
67
Mandelman
Rays and
and Farrington
Elasmobranch
skates
Otter trawl
U.S.A
(2006)
3 days
80
100
Rays and
Falkland
Laptikhovsky
Elasmobranch
skates
Squid trawl
Islands
(2004)
3 hours
0
71
63
Annex 2 Ad hoc physical environmental measurements
Observation
Water
Flow
D. Oxygen
Salinity
Time
Tank
date
temp.
rate
(%)
(ppt)
22/08/2014
16:50
4
16.8
2
90
30
22/08/2014
16:50
8
15.9
2
90
30
22/08/2014
16:50
2
15.9
2
90
30
23/08/2014
12:15
ONBOARD
17.7
2
93
35
23/08/2014
15:35
bucket
0
44
23/08/2014
17:55
10
16.5
2
88
30
24/08/2014
08:45
6
15.7
2
89
30
24/08/2014
18:30
3
16.8
2
96
30
24/08/2014
18:30
9
16.4
2
98
30
24/08/2014
18:30
1
16.1
2
96
30
25/08/2014
07:30
5
16.2
2
98
30
25/08/2014
07:30
7
15.8
2
102
30
25/08/2014
07:30
10
15.7
2
102
30
25/08/2014
20:00
5
16.6
2
96
30
25/08/2014
20:00
9
16.3
2
93
30
01/09/2014
19:00
7
16.7
2
104
30
02/09/2014
13:30 ONBOARD 1
18.1
2
98
35
02/09/2014
13:30 ONBOARD 5
18.1
2
89
35
02/09/2014
16:40
ONBOARD
19.5
0
86
35
02/09/2014
19:00
2
16.5
2
107
30
03/09/2014
19:00
11
16.8
2
95
31
05/09/2014
19:00
9
17.2
2
94
30
06/09/2014
19:00
bucket
19.9
0
51
30
07/09/2014
19:00
1
16.7
2
85
30
08/09/2014
19:00
5
16.8
2
90
30
64