Changes in Norway pout (Trisopterus esmarkii)
abundance and distribution under warming
conditions in the Barents Sea
Master of Science in Aquatic Ecology
Department of Biology
University of Bergen
June 2, 2014
Dr. Edda Johannesen, Institute of Marine Research, Bergen
Dr. Geir Odd Johansen, Institute of Marine Research, Bergen
Dr. Espen Johnsen, Institute of Marine Research, Bergen
Professor Arild Folkvord, University of Bergen
Frontpage drawing of Norway pout is adapted from Svetovidov (1948) in
FAO species catalogue: Gadiform fishes of the world (Cohen et al. 1990).
Changes in Norway pout (Trisopterus esmarkii)
abundance and distribution under warming
conditions in the Barents Sea
Master of Science in Aquatic Ecology
Department of Biology
University of Bergen
June 2, 2014
First and foremost I would like to thank all my great supervisors for always being helpful: Dr.
Edda Johannesen, my main supervisor; thanks for reading through drafts of varying quality
and always giving me fruitful and straight-to-the-point feedback. Dr. Geir Odd Johansen, as
head of the research group Ecosystem processes then, you gave me the opportunity to write
this thesis and pushing Edda to have another student; thanks for getting me started, for always
being helpful, and also for fruitful discussions about the secret life of the Norway pout and
other interesting matters of fish. Dr. Espen Johnsen; thanks for helping me out with the part
on diel variation in catchability, and for introducing me to the useful software handling this.
Professor Arild Folkvord, my supervisor at UiB; thanks for giving me really fast feedback on
numerous e-mails, for giving me very useful comments on drafts, and for telling me to hurry
up when deadline was approaching a bit too fast.
I would also like to thank IMR and now head of the research group Ecosystem processes,
Mette Skern-Mauritzen, for giving me the opportunity to write this thesis, providing me with
various practicalities and for letting me join the research group. It has also been nice to get to
know a little bit of the institution, and little bit of all the interesting stuff that goes on here.
Also, thanks to Berit Øglænd, student counselor at UiB, for always helping and answering in
administrative issues etc.
Thanks to my good friend Per Heldal Finne at Directorate of Fisheries for providing me with
some useful and good-looking maps for this thesis, and for great lunches.
I would also really like to thank Thassya Christina dos Santos Schmidt and Ryan James
Dillon at our student room at IMR, for being so helpful and creating a nice atmosphere.
And finally, thanks a lot Mathilde Hjelle Nitter for all your support and caring, and for being
so good-humored, although Norway pout does not seem to be your favorite topic.
29 May 2014
This study uses a 20-year time series of standardized bottom trawl winter survey data (1994 2013) from the Barents Sea, to investigate the changes in abundance and distributional range
of Norway pout (Trisopterus esmarkii) in response to changing sea temperatures. Due to the
boreal Norway pout’s rather limited geographical distribution in the Barents Sea, and that
Norway pout suffers no targeted fishing mortality in the Barents Sea, the species may be a
well suited indicator species of climate and ecosystem change here. Annual Norway pout
abundance indices were adjusted for diel changes in catchability, and an evident increase in
Norway pout abundance was found during the study period, although a marked decrease the
last two years was also evident. The distributional range was also found to increase, especially
northwards from the distinct core area in the southwestern part of the Barents Sea. Although a
rather weak correlation was found when comparing annual Norway pout abundance indices
with annual corresponding sea temperatures (r = 0.32), stronger correlations were found when
abundance indices were compared to sea temperatures which were measured two (r = 0.67)
and three years (r = 0.72) in advance. Reasons for these rather strong lagged (delayed)
correlations are briefly being discussed in this thesis, and may be related to temperature
effects on recruitment, maternal conditions (e.g. fecundity), changes in abundance/distribution
of other species which affect Norway pout abundance (prey, predators or competitors), and/or
a gradual expansion due to increased suitable habitat.
TABLE OF CONTENTS
INTRODUCTION ............................................................................................................ 1
Norway pout ................................................................................................................ 1
The Barents Sea ........................................................................................................... 3
Thesis’ aim and research questions ............................................................................. 4
METHODS ....................................................................................................................... 6
Survey data .................................................................................................................. 6
Calculation of abundance indices ................................................................................ 9
Diel variation in catchability ..................................................................................... 10
Geographical distribution .......................................................................................... 12
Abundance indices and sea temperatures .................................................................. 12
RESULTS ........................................................................................................................ 14
Distribution of catch rates.......................................................................................... 14
Survey coverage ........................................................................................................ 16
Diel variation in catchability ..................................................................................... 16
Abundance indices ..................................................................................................... 18
Geographical distribution .......................................................................................... 19
Sectored N. pout densities .................................................................................. 19
Proportions of catches including N. pout ........................................................... 22
Correlations between sea temperatures and abundance............................................. 23
Overall abundance .............................................................................................. 23
Sectored abundance ............................................................................................ 23
DISCUSSION ................................................................................................................. 25
Data quality................................................................................................................ 25
Diel variation in catchability .............................................................................. 25
Coverage ............................................................................................................. 27
Changes in abundance ............................................................................................... 27
Changes in geographical distribution ........................................................................ 29
Sea temperatures and abundance ............................................................................... 30
Lagged effects of temperature on abundance ..................................................... 32
Temperature and recruitment ............................................................................. 33
Causation or only correlation? ........................................................................... 35
Possible ecosystem effects of a changing N. pout population ................................... 36
Concluding remarks ................................................................................................... 37
REFERENCES ............................................................................................................... 38
APPENDIX ..................................................................................................................... 45
APPENDIX 1 – Data exploration ................................................................................ 46
APPENDIX 2 – General sources of error .................................................................... 54
Norway pout (N. pout, Trisopterus esmarkii) is a boreal species (Andriyashev and Chernova
1995) found south to the English Channel; with the North Sea, Skagerrak and to a lesser
extent, the Norwegian Møre coast, being the major fishing grounds (Cohen et al. 1990).
The Barents Sea represents the northern limit of the distribution of N. pout in the North-East
Atlantic. Due to the species’ historically rather limited abundance and distribution in the
Barents Sea, the species has not been fished nor studied in this area, and therefore its
ecological importance is to a large extent not known in this area.
Due to the N. pout’s more southern origin, it is assumed to be increasing in abundance and
distribution in the Barents Sea when the sea temperature increases. Since the species is not
fished in the Barents Sea, it can be considered a well suited indicator species for the impact of
ocean warming in the Barents Sea.
This thesis should therefore be considered as a preliminary study of today’s status of N. pout
in the Barents Sea, where it tries to cast light on how this boreal species have responded to the
changing temperatures in the Barents Sea of the period 1994 - 2013.
N. pout is a benthopelagic to pelagic species, which is found over muddy bottoms at depths of
50 - 300 m (Cohen et al. 1990). In the North Sea, it has been found to be most abundant at
depths of 100 - 200 m (Cohen et al. 1990); however, along the edge of the Norwegian Trench,
the species have been found deeper than 200 m, although few deeper than 300 m (Albert
1994). In the North Sea, the north-western part is likely to be the principal spawning area
(Nash et al. 2012), but N. pout is also known to spawn in some areas along the edge of the
Norwegian continental shelf; from about 61ºN to about 71ºN (Bakketeig et al. 2014).
Although it is possible that the northernmost spawning area also stretches into the Barents
Sea, signs of spawning in this area have not yet been found. Still, due to the Norwegian
coastal current and the Norwegian North Atlantic Current, it is expected that egg, larvae and
fry of N. pout from as far south as Møre (about 63°N) and Haltenbanken (about 65°N ) also
may drift into the Barents Sea (Sundby et al. 2013, Baranenkova and Khoklina 1968 in Nash
et al. 2012). However, not so much is known about how the northern populations of N. pout
along the Norwegian coast are connected (Nash et al. 2012).
N. pout is a small (less than 20 cm is an ordinary size (Cohen et al. 1990)) and shortlived
species which rarely lives longer than 4 - 5 years (e.g. Sparholt et al. 2002a). The species may
mature at as early as age 1, but maturation at age 2 is considered most common (Raitt 1968a,
Albert 1994). From the ICES stock assessment of N. pout in the North Sea, 10 % of age group
1 and 100 % of age group 2 and age group 3 were considered to mature (ICES 2007).
However, 60 % of the 1964 year class was reported to mature within age group 1 (Raitt 1968b
in Lambert et al. 2009), which made Lambert et al. (2009) suggest a possible densitydependence in growth and a stability in length-at-maturity. Lambert et al. (2009) also found
that the juvenile growth rate is higher when stock density is low, which results in a decrease
in age-at-50%-maturity; and also that the N. pout growth rates seem to be affected by the
abundance of the important predators cod (Gadus morhua), haddock (Melanogrammus
aeglefinus) and whiting (Merlangius melangus). Strong indications of spawning mortality
have been found (e.g. Lambert et al. 2009), and N. pout abundance in the North Sea and
Skagerrak is considered being strongly influenced by variations in recruitment and natural
mortality, such as spawning mortality and predation (Bailey and Kunzlik 1984, Sparholt et al.
2002b, ICES 2007). Summarized; although N. pout generally is short-lived and matures early,
geographical and annual variation in age of maturity and age specific mortality rates seem to
some extent to be common.
N. pout is regarded as an important link in the North Sea ecosystem (e.g. Albert 1994). The
species feeds mostly on planktonic crustaceans (copepods, euphausiids, shrimps,
amphipods), but also on small fish and various eggs and larvae (Cohen et al. 1990). It is
eaten by species such as cod, whiting, saithe (Pollachius virens), haddock and Atlantic
mackerel (Scomber scombrus) (ICES 2007), of which the tree first mentioned species by far
are the main predators on N. pout in the North Sea (Sparholt et al. 2002b). Since N. pout,
when caught, usually represent almost the whole catch (Johannessen et al. 1964 in Raitt
1968a), Raitt (1968a) points out that the likelihood of intraspecific competition being more
important than interspecific competition. Still, whiting caught in the same trawl as N. pout
have been found to feed on the same prey species (Raitt and Adams 1965 in Raitt 1968a).
The Barents Sea
The Barents Sea is an arcto-boreal sea, which is situated between Svalbard, Franz Josef Land,
Novaya Zemlya and Northern Norway. The ocean is one of the shallow shelf seas surrounding
the Arctic Ocean. Average depth is 220 m (Gorshkov 1980 in Ozhigin and Ingvaldsen 2011),
and ranges from 20 m to about 500 m (Ozhigin and Ingvaldsen 2011). Warmer Atlantic water,
mainly from the Norwegian Sea, flows into the southern Barents Sea, whereas colder water
from the Arctic Ocean flows into the northern part of the Barents Sea (Ozhigin and
The Barents Sea, like the Arctic in general, is experiencing elevated sea temperatures. Levitus
et al. (2009) investigated Barents Sea average monthly mean water temperature for the 100 150 m layer for the period 1900 - 2006, and found that interannual to interdecadal variability
was evident, with the warm period of 2002 - 2006 being the warmest registered period in the
Barents Sea so far. Although the recent warming has been positive for fish stocks in the
Barents Sea, the long-term effects of this warming are uncertain (Johannesen et al. 2012).
It is known that climate plays an important role in changing the distribution and production of
fish species in the Barents Sea; where changes in the distribution of species such as cod,
herring (Clupea harengus) and capelin (Mallotus villosus), have been linked to changing sea
temperatures (Drinkwater et al. in 2011). Several studies have also found that there has been
an increase in the abundance of boreal species in the Barents Sea due to higher sea
temperatures (e.g. Johannesen et al. 2012), which is similar to what happened during the
warm period in the middle of the 20th century (Drinkwater et al. 2011).
Future climate change is expected to result in higher phytoplankton production due to loss of
seasonal ice in the Barents Sea (Slagstad and Wassmann 1997). Similar to cod, other boreal
fish species are also expected to extend farther east and north as the Barents Sea gets warmer
(e.g. Stenevik and Sundby 2007); for example eastward extensions of Atlantic mackerel are
possible, which might change the structure and function of the Barents Sea ecosystem, due to
changes in species interactions (Drinkwater et al. 2011).
1.3. Thesis’ aim and research questions
The N. pout catch data have been collected by the Institute of Marine Research (IMR) and
The Polar Research Institute of Marine Fisheries and Oceanography (PINRO) during the
yearly winter survey in the Barents Sea throughout the period 1994 - 2013. The winter survey
started in 1981, it covers the southern ice-free part of the Barents Sea, and it is now the
longest continuous bottom trawl series from the Barents Sea (Johannesen et al. 2009).
Traditionally, the main aim of the winter survey is to investigate spatial distribution and
abundance of the commercially exploited demersal fish species cod and haddock, and
abundance indices from the winter survey are being used in stock assessment of cod, haddock,
golden redfish (Sebastes marinus), deep-sea redfish (Sebastes mentella) and Greenland
halibut (Reinhardtius hippoglossoides) (Jakobsen et al. 1997, Johannesen et al. 2009). Also,
catch weight and catch number of all other fish species, shrimp and king crab have also been
recorded during the winter survey (Johannesen et al. 2009).
In this thesis a time series of N. pout abundance in the Barents Sea is estimated for the first
time. Further three questions about N. pout in the Barents Sea are being addressed: (i) Has the
population size increased during the study period? (ii) Has there been a change in the
geographical distribution (i.e. distributional range) during the study period? (iii) Is there a
correlation between increasing sea temperatures and the abundance of N. pout?
How temperature may affect the abundance and distribution of N. pout in the Barents Sea, and
also possible effects of a changing N. pout population in the Barents Sea are also briefly being
To investigate these questions, two important factors which may affect the results, were taken
into account and discussed; (a) the possible diel variation in N. pout catches, and (b) the
annual variation in survey coverage (i.e. spatial distribution of trawl stations).
(a) Diel vertical migration (DVM) may create diel variation in catch rates, which in turn may
cause a bias when calculating abundance indices (Hjellvik et al. 2002). DVM is found to be a
common phenomenon in many species across a broad range of taxa, and may be driven by
predator avoidance, biochemical signals or environmental conditions (Stafford et al. 2005)
The phenomenon has previously been found in N. pout (Onsrud et al. 2004). In this study, diel
variation in catch rates were tested for, and abundance indices were adjusted accordingly
(hereafter referred to as adjusted abundance indices).
(b) Due to both climatic and political reasons, the area covered during the winter survey has
not been the same from year to year, which may affect the calculated abundance indices. The
potential impact of varying survey coverage was therefore considered when interpreting the
2.1. Survey data
The catch data used in this thesis are stratified catch data from bottom trawling, sampled
annually throughout the period 1994 - 2013, with dates ranging from January 20 to March 15
(annually not the same time interval); sampled throughout the 24 hr cycle. The catch data was
prepared by IMR, and contained in addition to number of N. pout caught in each tow, further
information about each tow; vessel, time of tow (UTC), tow ID, tow number, tow distance
(nm), tow duration (minutes), spatial strata, and latitude, longitude and depth of tow (m). The
winter survey sampling area has since 1996 consisted of seven subareas with 23 strata (figure
1), where the trawl stations have been spaced at regular grids, although with different
densities (Jakobsen et al. 1997). The sun angle (degrees above the horizon), which had been
calculated using a macro based on time, latitude and longitude, was also provided with the
Data from the period 1981 - 1993 were excluded due several changes in methodology of the
survey; such as change in trawling gear (1989), introduced regular trawling station grid
system (1990), expanded survey area (1993), and reduced mesh size (1994) (Johannesen et al.
2009). However, there has also been changes in methodology and implementation of the
survey after 1994; such as strapping of all hauls (1998), varying distance between trawling
stations, changes in Russian contribution (PINRO) and reduced coverage in the Russian part
of the Barents Sea due to political decisions and ice coverage (Johannesen et al. 2009).
Probably the biggest single factor contributing to differences in methodology throughout the
study period 1994 - 2013, has been the varying survey area in the eastern sectors (SE and
NE), with some years lacking coverage of as much as 50 % of the strata of these two sectors
(figure 2). Table 1 show the number of trawl stations with average depths for all years and
strata, while App. figure 1.6 shows the spatial distribution of the trawl stations for all years
Figure 1. The 23 winter survey strata and the area covered by the winter survey. A grouping of the strata into
four sectors (NW, SW, NE and SE; in red) was applied in this thesis (see below). The Fugløya - Bear Island
temperature transect (see below) is marked as a purple line. Figure made by Per Finne, Directorate of Fisheries.
Figure 2. Number of strata in the eastern part of the Barents Sea (NE and SE, in total fourteen strata) covered by
the winter survey throughout the study period.
Table 1. Number of trawl stations for all strata throughout the study period, including average depth (m) of all trawl stations within each stratum.
Average number of trawl stations within each stratum, and sum of all trawl stations within each year is also included.
A Campelen 1800 research shrimp trawl with rockhopper gear, mesh size 80 mm (stretched)
in the front and 22 mm in the cod end, was used throughout the study period, with the length
of the swipe wires being 40 m (Jakobsen et al. 1997). The trawl has a horizontal opening of
17 m and a vertical opening of 4 - 5 m (Wienerroither et al. 2013). Standard towing time was
30 minutes in 1994 (Jakobsen et al. 1997), but was reduced to 15 minutes in 2011 due to very
large catches of cod in some stations (Wienerroither et al. 2013). The standard towing speed
has been 3 knots (Jakobsen et al. 1997). For further information on trawl details, see Jakobsen
et al. (1997).
Calculation of abundance indices
Based on the catch data, annual estimated abundance indices were calculated to quantify the
development of the N. pout population throughout the study period. Annual N. pout density
(N. pout/nm2) at each trawling station (s), referred to as station density (Ds), was calculated:
where Npout is the number of N. pout caught in each trawling station, l (nm) is the length of
the corresponding tow, and w (nm) is the fishing width, which has been set to 25 m (=
0.013499 nm) (based on Asgeir Aglen, pers. comm.).
The average density of N. pout for each stratum (N. pout/nm2), referred to as the stratum
density (Dstratum), was then calculated:
where Ntrawl is the number of trawl stations in each stratum.
Finally, the annual total abundance index for the whole Barents Sea was calculated:
where Astratum,i is the area (nm2) of each stratum (i), and L is the number of strata for each year.
The same operation was also done to calculate the adjusted abundance indices, which are
taking the diel variation in catchability into account (see section “Diel variation in
The annual uncertainty of the estimated abundance indices was quantified by calculating the
standard errors of the mean (SE). This was done by first calculating the sum of variance for
each stratum within each year (si2), according to the following formula:
Where Davg. is the average N. pout density in stratum i, and ni is the number of strata.
The estimated variance of the stratified mean (var(Davg.)) was then calculated:
Finally, SE for each year was found by the following formula:
2.2. Diel variation in catchability
Diel vertical migration (DVM) is found to be a common phenomenon in many species and
taxa (Hutchinson 1967 in Lampert 1989). Anti-predator behavior is regarded as an important
driver of DVM, where prey typically aggregate to avoid predators (e.g. Fuiman and Magurran
1994 in Kerfoot 1985). DVM leads to diel variation in catchability, and may bias the
abundance estimates (Hjellvik et al. 2002). To examine the diel variation in catchability of N.
pout, the R software package DIVA (Hjellvik 2005) was used. DIVA contains several setting
options and build-in functions to estimate and adjust for diel variation in catchability.
The clustered priority of the input data in DIVA was as follows: Year, vessel, month and day.
Regarding the general settings in DIVA; the input data were log-transformed, UTC-time was
used, and observations in clusters with only one observation were moved to the adjacent
cluster. Finally, night catches were adjusted to day-catch level, since a reduced catchability
during night (due to DVM) was expected.
There are several ways to try to correct for the diel variation, whereas the sinusoid and the
logistic functions are the two parametric functions which have been found especially useful
(Hjellvik et al. 2002). Although the logistic and the sinusoid function showed a large
similarity, only the logistic function was chosen to be further investigated in this study (see
Appendix 1 for details). It has been suggested that DVM is being triggered by the light
intensity (e.g. Bohl 1980), and since time of sunrise and sunset varied substantially during the
survey periods (due to large sampling range in time and space), altitude of the sun (s) was
used as a proxy for time; hence the following logistic equation was used to calculate the diel
variation in catchability (gl):
where D determines the diel amplitude of the variation in catchability between day and night,
α determines the length of the transition phases between day and night, and β determines
when the same transition phases occur (i.e. temporal location of the transition phases). In the
analyses, α was fixed, while β and D was estimated.
Three different models for the diel variation in catchability were investigated: The simple
model (model 1, with two parameters: β and D), which assume no annual or depth dependent
diel variation in catchability; the annual model (model 2, with 21 parameters: β and year
specific D), which was carried out to investigate the annual variation of the diel variations in
catchability; and finally, the depth model (model 3, with three parameters: β, D_depth and
D_intersect), which was carried out to investigate how diel variation in catchability varied by
When the catch rates had been adjusted with DIVA, the adjusted abundance indices were
calculated the same way as the unadjusted abundance indices were calculated. Finally, to
check for the difference in the temporal trend between adjusted and unadjusted annual
estimates, the correlation coefficient (r) between them was calculated.
R version 1.9.1 was used with the DIVA runs.
The geographical distribution was investigated in three different ways: (i) By mapping
summarized catches within four 5-year periods; (ii) by calculating the sectored N. pout
abundance densities, where N. pout density was a preferred measure since the four sectors has
different areas; and (iii) by calculating the proportion of catches which include N. pout, which
was done for both the whole study area and for the four sectors.
(i) A spatial presentation of the distribution and abundance of N. pout catches was made by
using Spatial Analyst Tools in ArcMap (ArcGIS Desktop), where the Kernel point density
method was used, weighted for the number of N. pout. Output cell of catches was set to 10
km, and search radius of catches was set as high as 50 km to include adjacent trawl stations in
the statistical smoothing. This part was done by Per Finne, Directorate of Fisheries.
(ii) and (iii) To study potential changes in distribution of N. pout, the 23 strata were divided
into four sectors (NW, NE, SW and SE), according to figure 1, and changes over time were
Abundance indices and sea temperatures
To test for influence of sea temperatures on N. pout abundance, correlation analyses between
annual mean sea temperatures from the Fugløya - Bear Island transect (71° 30'N - 73° 30'N,
referred to as the FB transect) and the adjusted N. pout abundance indices were carried out.
The FB transect annually measures the sea temperatures at depths of 50 - 200 m (mean
temperatures have been used), normally on 20 different stations (see figure 1 for location of
The N. pout catch data and the abundance indices used in the correlation analyses were from
the period 1994 - 2013, and only those abundance indices which take the diel variation in
catchability into account (i.e. the adjusted abundance indices) were used. Temperature and
abundance indices from the same year were compared, but also the potential lagged (delayed)
effect of temperature on abundance was investigated by comparing temperatures from year n1 to n-3 with abundance indices and catch data from year n.
R version 3.0.2 (“Frisbee sailing”) was used for the Pearson correlation analyses.
Distribution of catch rates
The catch rates of N. pout during the study period varied considerably, with zero catches
accounting for 57.8 % of all catches (n = 6159), and with the average number of N. pout for
all non-zero catches being 277 (n = 2602, range 1 - 23663). Still many catches included few
N. pout, which lead to the median of all non-zero catches being only 28 N. pout pr. catch.
Proportion of N. pout zero catches have generally been declining throughout the study period
(annual median = 0.57, annual range = 0.39 - 0.80) (figure 3).
Figure 3. N. pout zero catches (i.e. catches without N. pout) in proportion to total number of trawl stations
throughout the study period (n = 6159).
The catch size distribution was also briefly investigated, and when defining larger catches as
≥ 1000 N. pout, these catches were found to account for only 6.6 % of all non-zero catches (n
= 2602). Still, these relatively few catches accounted for 64.6 % of the total number of N.
pout caught (n = 721 418) (figure 4).
Figure 4. Development of smaller catches (n = 2431) and larger catches (n = 171) throughout the study period.
Depths of all N. pout catches (n = 2602) were also briefly investigated (figure 5); where the
bulk of the catches (69 %) were taken within the depth interval 200 - 299 m; average depth of
all catches was 277 m (range = 58 - 571 m).
Figure 5. Proportional distribution of the amount of N. pout caught at different depth intervals, grouped in all
catches (n = 2602) and larger catches (n = 171).
During the study period, the survey coverage (i.e. trawl station coverage) has varied within
the winter survey area. The year with the highest number of trawl stations was defined as
maximum coverage; however, this might be different years when studying the survey area as
a whole, the different sectors, or the different strata.
The whole survey area had an average coverage of 308 trawl stations, which was 76 % of the
number of stations trawled in 2002, which was the year with the highest number of trawl
stations (405). The year with the lowest number of trawl stations (1998) had 200 trawl
stations, which is 44 % of the maximum coverage. The survey coverage varied also within the
four sectors (in the following denoted by average annual coverage and minimum annual
coverage within each sector): SW had average = 80 % and minimum = 45 %, NW had
average = 87 % and minimum = 64 %, SE had average = 57 % and minimum = 50 %, and NE
had average = 37 % and minimum = 3 %. The survey coverage varied also within each of the
23 strata; with stratum 11 (in NE) being the highest covered strata (average coverage = 85 %,
minimum coverage = 46 %), and stratum 20 (in NE), being the least covered strata (average
coverage = 12 %, minimum coverage = 0 %).
From the numbers above (and Table 1), it becomes clear that the eastern sectors (SE and NE)
have lower coverage than the western sectors (SW and NW).
Diel variation in catchability
Model 1 (the simple model) had R2 = 0.67; parameter α = 2 was fixed (see App. 1 for details),
β was significant and estimated to -4.53 (p < 0.0001, SE = 0.92), and D was significant and
estimated to 0.48 (p < 0.0001, SE = 0.07). The shape of the function relating catch rates to sun
altidtude is showed graphically in figure 6, where α, β and D from model 1 was used. From
figure 6 night level catches were defined when sun angle was ≤ -10º above the horizon, while
day level catches were defined when sun angle was ≥ 0º above the horizon. Model 1 was
found to increase the night catches by this factor:
e0 / e-0.48 = 1.6
For comparison, when calculating the difference between day and night catches (when
excluding zero catches) for the study period as a whole, average day catches (n = 801) were
found to be 2.2 times higher than the average night catches (n = 1323).
Model 2 (the annual model) had R2 = 0.68. The parameter α was fixed = 2, estimated β was
-3.53 (p < 0.0001). Median D throughout the whole period 1994 - 2013 was 0.56 (range = 0.74 - 1.15, median p-value = 0.08, range p-value = 0.002 – 0.90, median SE = 0.30, range SE
= 0.23 – 0.52).
R2 for model 2 was similar to model 1, but p-values for D were highly variable, with only five
of these annual p-values being significant (p < 0.05). The number of estimated parameters of
model 2 was 21 (D for 20 years and β), compared to 2 (D and β) for model 1.
In addition, SE values for D in model 2 were high (median SE = 0.30, range = 0.23 – 0.52).
Based on that model 2 used 19 more parameters than model 1 and only achieved a slightly
higher R2 (cf. Principle of Parsimony), and, in addition, had high SE-values for D, model 2
was not used for further adjustments of abundance indices.
Figure 6. The function gl (s) from model 1, with α = 2 (fixed), β = -4.53 (estimated) and D = 0.48 (estimated).
gl(s) represents the difference between expected day-time catches (gl(s) = 0) and expected night-time catches
(gl(s) = -0.48), where s is the altitude of the sun. According to the assumption that night-time catchability is
lowered due to DVM, the model elevates night-time catches to the same level as day-time catches.
Model 3 (the depth model) had R2 = 0.67, β was estimated to -4.49 (p < 0.0001), D_intersect
(D at 0 m depth) was estimated to 0.34 (p = 0.28), and D_depth (slope) was estimated to
0.0005 (p = 0.65). Since p-values for D_intersect and D_slope were found not to be
significant (p > 0.05), model 3 was not used for further adjustments of abundance indices.
When the development of both the unadjusted and adjusted abundance indices throughout the
study period were investigated (figure 7); an increase in the abundance indices especially for
the period 2003 - 2009 is evident, followed by a flattening until 2011, and furthermore a quite
strong decrease in 2012 and 2013. The correlation coefficient (r) between unadjusted and
adjusted abundance indices was found to be 0.998, showing that the trends of the two time
series were almost identical. Annual adjusted abundance indices were on average 27 % higher
than unadjusted abundance indices (annual range = 12 - 44 %).
Figure 7. Unadjusted and adjusted abundance indices of N. pout (± SE) throughout the study period.
The estimated median annual abundance indices was 0.99 billion individuals (range = 0.04 4.21 billons) when accounting for the diel variation in catchability. The median adjusted
abundance indices of the last five years of the study period (2009 - 2013) was found to be
13.3 times higher than the median adjusted abundance indices of the first five years (1994 1998); while the average adjusted abundance indices were 7.3 times higher for the last five
years compared to the first five years of the study period.
N. pout catches have increased in size in the species core area (mainly within the sector SW)
throughout the study period, and also, the geographical range of where N. pout have been
caught has expanded towards north and east, where the latter applies for both the species’ core
area and the more peripheral area (figure 8). Most of the larger catches (≥ 1000 N. pout pr.
catch) were made in SW (91.8 %), while NW (5.3 %), SE (1.8 %) and NE (1.2 %) had
relatively few larger catches.
Sectored N. pout densities
As for the proportion of trawl stations including N. pout, the adjusted N. pout density indices
have also generally increased throughout the study period, with the trend being rather similar
for all four sectors; with a first peak in 1997 and a second, and higher, peak within the period
2007 - 2012 (figure 9).
SW was found to clearly have the highest N. pout density (annual median = 32441 ind/nm 2,
annual range = 1638 - 134478 ind/nm2), while NE was found to clearly have the lowest N.
pout density (annual median = 200 ind/nm2, annual range = 0 - 3905 ind/nm2). When
comparing average sectored N. pout densities from the first five years and the last five years
of the study period, the increase was highest in NW, with 33.2 times higher N. pout density
the last five years compared to the first five years. The equivalent increase in SW, SE and NE
was 5.7, 4.2 and 2.0 times, respectively.
Figure 8. Summarized abundance and distribution of N. pout catches over four 5-year intervals; 1994 - 1998, 1999 - 2003, 2004 - 2008 and 2009 - 2013. Figure made by Per Finne,
Directorate of Fisheries.
Figure 9. Sectored development of log transformed adjusted N. pout densities.
From figure 10 (and figure 8) it is clear that the N. pout population’s center is within the
sector SW, although there are strong indications of that this center has moved towards the
sector NW, and possibly also slightly towards the sector NE, during the study period.
Figure 10. Trend of moving averages (over 4-year intervals) of adjusted sectored N. pout abundance indices in
proportion to adjusted total abundance indices.
Proportions of catches including N. pout
Overall, the proportion of trawl stations including N. pout increased over the study period
(figure 3), indicating that the distribution area of N. pout increased.
The temporal trend of the proportions of catches including N. pout within the four sectors
(SW, NW, SE, NE) was also investigated (figure 11). As expected, SW clearly had the largest
proportion of catches including N. pout throughout the study period (annual median = 85 %,
annual range = 71 - 99 %).
When further studying figure 11, it is clear that all four sectors generally have experienced an
increase in the presence of N. pout in catches during the study period; with the largest
increases seemingly in the peripheral sectors NW (annual median = 0.22, annual range = 9 86 %), SE (annual median = 34 %, range = 11 - 88 %), and NE (annual median = 18 %, range
= 0 - 43 %); where NE still in 2013 being close to its maximum level, while the other three
sectors peaked in 2007 (SE) and 2008 (SW and NW). It is also worth pointing out that SE and
NW showed very evident peaks in the proportions of N. pout in catches in 2007 and 2008,
respectively (figure 11).
Figure 11. Annual proportion of catches including N. pout within each of the four sectors SW (n = 1410), NW
(n = 1314), SE (n = 2378) and NE (n = 1057) throughout the study period., where n refers to number of catches
within each sector.
Correlations between sea temperatures and abundance
In general, sea temperatures (FB transect) have increased during the period 1991 - 2013
(figure 12). Average temperature of the last five years (5.75 ºC) was 0.17 ºC higher than
average temperature of the first five years (5.58 ºC) of the study period. The five-year period
with the highest average was 2004-2008 (6.09 ºC), which was 0.81 ºC higher than the fiveyear minimum 1994-1998 (5.27 ºC).
When comparing adjusted abundance indices from the study period (1994 - 2013), with the
sea temperatures from the Fugløya-Bear Island transect from the period 1991-2013, there
seems to be a strong, but delayed correlation (figure 12).
The correlations between N. pout abundance indices and sea temperatures were found to be
weak when both comparing abundance indices and temperatures from the same year (r =
0.32) and when comparing abundance indices with sea temperatures one year in advance (r =
0.39), where the latter is referred to as 1-year-lag. However, when comparing abundance
indices with sea temperatures two and three years in advance (referred to as 2-year-lag and 3year-lag, respectively), the correlations between abundance indices and sea temperatures were
found to be stronger (r = 0.67 and r = 0.72, respectively). Correlation coefficients (r) are
presented in table 2.
The correlations between sea temperatures and N. pout abundance indices within each of the
four sectors were also investigated (see Table 2 for an overview of the correlation
coefficients). NW and SW showed the same pattern, with increasing correlations with
increasing lags, especially from 1-year-lag to 2-year-lag; however, the overall correlation was
stronger in SW (r = 0.30 - 0.73) than in NW (r = 0.13 - 0.56). When comparing NE and SE,
these two sectors also showed a quite similar level and pattern of correlation; however, NE
showed an increased correlation level from 1-year-lag (r = 0.44) to 3-year-lag (r = 0.64). The
strongest correlations were found in SW (r range = 0.30 - 0.73) and NE (r range = 0.44 0.64), with both maximum correlations found for 3-year-lags.
Figure 12. Development of adjusted total abundance indices of N. pout, and of temperature measured through
the FB transect, throughout the period 1991 - 2013.
Table 2. An overview of the correlation coefficients (r) between sea temperatures (FB transect) and adjusted
total N. pout abundance indices of the whole study area (referred to as Total) and the four sectors (SW, NW, SE,
NE). Abundane indices were compared to temperatures from same year (n), one year in advance (n-1), two year
in advance (n-2), and three years in advance (n-3).
In short, both N. pout abundance and distributional area in the Barents Sea was found to
increase during the study period. Correlation coefficients between abundance and sea
temperatures lagged with 2 - 3 years were found to be higher than correlations between
abundance and temperatures from the same or previous year (true for the study area as a
whole and for all four sectors).
Diel variation in catchability, probably caused by DVM, was found to cause an
underestimation of N. pout abundance, but the temporal trend was still almost identical for
adjusted and unadjusted abundance indices.
I have chosen to focus on the following aspects in the discussion chapter of this thesis: Data
quality, the changes in N. pout abundance and geographical distribution, possible effects of
changing sea temperatures on N. pout abundance, and finally, an attempt to briefly discuss
how a changing N. pout population in the Barents Sea may affect other parts of the
Diel variation in catchability
DVM has previously been found in many species in many taxa (e.g. Stafford et al. 2005).
DVM in N. pout was studied in the Oslofjord, where N. pout was found to approach krill from
below during night time (Onsrud et al. 2004). The rationale behind adjusting catch rates for
such behavior in this current study is that N. pout catchability of the bottom trawl is likely to
decrease during night, as N. pout move upwards in the water column to forage.
Adjusting for diel variation in catchability will usually be a trade-off between, on one hand,
getting more correct abundance or catchability estimates, and on the other hand, adding more
uncertainty to these estimates (Hjellvik et al. 2002). The added uncertainty is considered more
serious if the diel variation varies significantly from one year to another (Hjellvik et al. 2002).
Although model 1 (the simple model) nearly had the same R2 as model 2 (the yearly model)
(0.67 and 0.68, respectively), the annual median standard error of D in model 2 was much
larger than the standard error of D in model 1, and also larger than standard errors of D found
in Hjellvik et al. (2002). This is expected, since there is more uncertainty related to
calculating D for each year, than calculating D for the whole period. Such a tradeoff between
the amount of model bias (underfitting) on one hand, which is a likely result when using a
model which is over-simplified, and the level of sampling variation (overfitting), on the other
hand, which is a likely result in a model with more parameters, is a well known dilemma. The
Principle of Parsimony (Goodman 1984 in Burnham and Anderson 1992) is about accuracy
(bias) versus precision, and suggests that when a simple model (i.e. a model with few
parameters, which is likely to be inaccurate) can explain as much as a more complicated
model (i.e. a model with more parameters, which is likely to have low precision), the simple
model should be chosen. Thus, choosing model 1 over model 2 has support in the principle of
Abundance indices were adjusted up due to the decreased catchability of N. pout during night
time. The difference which was found between adjusted and unadjusted abundance indices is
interesting, since it has previously been stated that vertical migrations of N. pout should be
too limited to influence the catch rates significantly (Albert 1995). The estimated diel
amplitude D of N. pout found in this study is at the same level as found for cod in Hjellvik et
al. (2002), and is therefore likely to be realistic. Also, the standard error of the mean of D
found in this study (model 1) is at the same level as found for cod and haddock in Hjellvik et
The level of adjusted abundance indices was higher than the level of the unadjusted indices,
and as expected, the absolute difference between unadjusted and adjusted abundance indices
was found to increase with increasing abundance indices. Although there was a clear annual
difference between the adjusted and the unadjusted abundance indices, the correlation (r)
between the two time series of abundance indices was close to 1, which means that the two
abundance indices followed the same temporal trend, and thus, that the adjustment has a
negligible effect on the correlation analyses carried out between N. pout abundance and sea
The winter survey coverage has to some extent varied in geographical range, and in general,
the coverage has been higher in the western sectors (SW and NW) than in the eastern sectors
(SE and NE).
There are several ways that varying survey coverage may affect the results in a study like this;
for instance low survey coverage may lead an increased inaccuracy of N. pout abundance
estimates (at survey area level, sector level or stratum level). For example, an overall
overestimation of N. pout abundance estimate is likely in years where strata with low N. pout
density have less coverage. Also, varying survey coverage may affect the proportion of
catches including N. pout; for instance, if there is less coverage in strata where N. pout is less
than average distributed, it may lead to an overestimation of the proportion of catches
including N. pout. As an example of the latter, the low coverage in the eastern sectors (SE and
NE) in 1997 and 1998 could have contributed to the relatively low proportions of zero catches
in the same two years; and also the low coverage in SE in 1997, 1998 and 2007 could explain,
at least partly, the high proportions of catches of N. pout in SE during the same years.
However, the similarity in trends of abundance density, which is found within the four
sectors, may suggest that the varying coverage within the two eastern sectors have not
affected the sectored abundance estimates to a large extent. Also, the abundance of N. pout in
the eastern sectors is low compared to the western sectors, which means that the impact of the
eastern N. pout abundance on the overall N. pout abundance, in any case, is relatively low.
Changes in abundance
Both N. pout abundance and the proportion of trawl stations including N. pout were found to
increase during the study period. The relative increase in N. pout abundance (i.e. density) was
highest in the periphery of the core distribution area, also indicating an increase in spatial
distribution. When accounting for diel variation in catchability, the abundance of N. pout in
the peak year (2011) was found to be 4.2 billion individuals. For comparison, the similar sized
capelin also had a high biomass year in 2011 with an estimated stock of 454.1 billion
individuals (Anon. 2011).
Abundance is in general determined by three factors; (i) mortality, (ii) migration and (iii)
(i) Regarding mortality it is common for harvested marine resources to distinguish between
fishing mortality and natural mortality, but since there is no targeted fishing on N. pout in the
Barents Sea, potential changes in fishing mortality could be ruled out. Although N. pout is
regarded as an important prey species for other gadoids in the North Sea (ICES 2007), N. pout
natural mortality has never been studied in the area, which means that the effect of natural
mortality on N. pout in the Barents Sea remains unknown.
(ii) The known spawning grounds of N. pout closest to the Barents Sea, are in the Norwegian
Sea, on the shelf outside the counties of Nordland and Troms in Northern Norway (Sundby et
al. 2013). Still, there could also be spawning grounds of N. pout in the Barents Sea, which has
not been discovered yet. Although the dynamics and connectivities of the northern
populations of N. pout to a large extent is unknown (Nash et al. 2012), it is likely that eggs
and larvae that are being spawned off the north-western coast of Norway generally are
transported to the Barents Sea by the Norwegian coastal current (Baranenkova and Khokhlina
1968 in Nash et al. 2012). Although it is likely that the Barents Sea N. pout population
undertakes regular spawning migrations towards the coast of Nordland and Troms, it is also
possible that the Barents Sea N. pout population is a sink population, which benefits only
from an increase in the influx of N. pout larvae due to improved spawning conditions outside
the Barents Sea and/or increased influx of Atlantic water into the Barents Sea. Indeed, an
increase in Atlantic water inflow has been found to explain a shift towards higher volume of
Atlantic water in the Barents Sea (Zhang et al. 1998).
(iii) N. pout is a relatively short-lived species; Raitt (1968a) reported that the maximum age
recorded from the North Sea is four years, and that the majority of this population is within
age group 1. The short life span has been linked both to high rates of predation mortality (e.g.
Raitt 1960 in Raitt 1968a), and also significant spawning mortality rates (e.g. Lambert et al.
2009). The species is found to mature at early age, where the majority first-time-spawners
commonly are within age group 2 (e.g. Raitt 1968a, Albert 1994). The presumed high
turnover rate within N. pout populations suggests that recruitment conditions play an
important role in determining the abundance of the species. Indeed, N. pout is known to have
strong varying recruitment (e.g. Kempf et al. 2009, Bakketeig et al. 2014), and the species
may therefore experience a strong increase in abundance when the conditions are good; e.g. a
study in the North Sea indicated considerable variation in the abundance of larvae (Munk et
al. 1999). Since there is no fishing mortality in the Barents Sea, and since N. pout recruitment
is considered an important factor for the species’ abundance in the North Sea (e.g. Munk et al.
1999, Bakketeig et al. 2014), it is reasonable to suggest that recruitment plays an important
role in determining the abundance of N. pout also in the Barents Sea. Finally, recruitment is
generally regarded as an important factor affecting species’ abundance, and the link between
temperature and recruitment will be discussed more thoroughly further in the discussion.
However, when studying the increase in abundance indices found in this study throughout the
period 1999 - 2009, a gradual increase is evident, which might suggest that the increase in
abundance is due to a gradual increase in geographical distribution caused by post-larval
movements, rather than evident changes in recruitment. Still, a combination of improved
recruitment and an increased geographical distribution might perhaps be the most likely
explanation for the observed increase in N. pout abundance. Changes in geographical
distribution are discussed later in the thesis.
The evident increase in N. pout abundance presented in this thesis could of course have
several reasons (both proximate and ultimate reasons). However, since N. pout is a boreal
species which in this study has been found to increase its abundance and distributional within
the northern part of its distributional range, it is natural to think that there somehow is a link
between a warming Barents Sea and the changes in the Barents Sea N. pout population.
Changes in geographical distribution
N. pout occurs only on the continental shelf on the eastern side of the North Atlantic, and is
regarded as most common in the North Sea, Skagerrak, west and north of the Scottish coasts,
and along the Norwegian continental shelf (Raitt 1968a, Albert 1994). The Barents Sea
includes the periphery of the species’ northeastern range (e.g. Cohen et al. 1990).
N. pout population has increased its overall distributional range in the Barents Sea during the
study period, which also is expected from the increase in abundance, since a positive
correlation between average population densities and geographic range previously has been
found for many other species (Brown 1984).
It is also clear that N. pout is most abundant in the sector SW; which is probably linked to the
adjacent N. pout population further southwest. The sector SW should therefore be regarded as
the centre of the species range in the Barents Sea, although this centre has been found to move
slightly northwards during approximately the last half of the study period. However, this trend
seems to have stopped the last years of the study period, which is also supported by the fairly
constant perimeter of the distribution for the periods 2004 - 2008 and 2009 - 2013.
The most evident increase in distribution, regarding areas with both high and low N. pout
densities, has been during the period 2004 - 2008, which also corresponds quite well to the
trend of the summarized catches over the same 5-year-periods. Still, although the summarized
catches show the largest increase during the period 2004 - 2008, the development of the
abundance indices (which takes the trawled area into account), shows that the increase in
abundance is likely to be rather equal for the period 2009 - 2013, which then again may
suggest that the last years (2009 - 2013) increase in abundance have been within the same
area as the 2004 - 2008 distributional range, thus, suggesting an increasingly denser
population within this period.
It is evident that the proportion of catches including N. pout increased more within the three
peripheral sectors than within SW. Still, this seems rather likely since the species already is
widely distributed within SW, and may therefore not have a lot of unused preferred habitat
left in this sector.
The proportion of catches including N. pout was also found to vary more within the peripheral
sectors, especially within SE and NW, than within SW, which is statistically expected
according to the hypothesis that density is greatest near the center of a species distributional
range (e.g. Brown 1984).
Sea temperatures and abundance
Higher N. pout numbers and increased northern distribution have also previously been
associated with higher sea temperatures; where Svetovidov (1948, in Raitt 1968a) related N.
pout records from e.g. Bear Island and the western part of the Barents Sea to a general
warming along the shores of the Scandinavian peninsula, and Baranenkova (1960, in Raitt
1968a) related the large numbers of N. pout in the southern Barents Sea in 1959 to the
intensity of the Murmansk current that year, and high sea temperatures from March and
As previously mentioned, a positive correlation has been found between abundance and
distribution for other species (Brown 1984), which implies that when studying how
temperature affects a species’ abundance, it is also important to know something about both
the species’ vertical distribution and the species’ patchiness (i.e. the size distribution of the
schools). The period 2006 - 2012 showed a higher amount of larger catches (≥ 1000 N. pout
pr. catch) compared to both previous (except 1997) and later years. This increase in large
catches has mainly been in the sector SW, and strongly indicates that more abundant N. pout
schools have become more common in the same period within SW.
Albert (1994) refers to previous studies (Mason 1960, Raitt 1968b) which have found that
depth was the most significant environmental variable which explained the variation in
abundance of N. pout in the North Sea, and also that there was no effect of temperature nor
salinity on the observed variation in abundance. Most N. pout in this current study were
caught in the depth interval 200 - 300 m, which is considerably deeper than indicated in Raitt
(1968a), where the highest N. pout catches in the North Sea were found at depths of 100 - 200
m with bottom temperatures of 6 - 9 °C; and also slightly deeper than the preferred depth of
N. pout found in the Norwegian Trench (approximately 200 m) (Albert 1995). The Barents
Sea has an average depth of 220 m (Gorshkov 1980 in Jakobsen and Ozhigin 2011), while the
North Sea has an average depth of 95 m, which of course could explain why the species seem
to prefer deeper habitats in the Barents Sea than in the North Sea. However, the Norwegian
trench has maximum depths varying between 280 and 700 m (Albert 1995), which makes the
seemingly preference of shallower habitats here compared to the Barents Sea harder to
explain. In conclusion, the reason why N. pout seem to prefer deeper habitats in the Barents
Sea than further south may simply be related to the depths of suitable habitats; still the
preferred temperature interval could potentially also influence the differences found in
preferred depth (i.e. N. pout has to go deeper in the relatively cold Barents Sea to find its
optimal temperature interval).
Lagged effects of temperature on abundance
There are many pathways of which climate may impact marine populations; there may for
instance be direct, indirect, lagged and unlagged responses of climate, which makes it difficult
to distinguish and recognize the connection between climate and the ecological responses
(Ottersen et al. 2010). Lagged (delayed) effects may originate in climatic events affecting the
critical survival of the early life stages of entire cohorts, where favorable climatic conditions
may produce larger cohorts with larger individuals which may survive at a higher rate
(Ottersen et al. 2010). Ottersen et al. (2010) also states that “it appears that ecosystem
responses to bottom-up forcing include both quick and short term responses at low trophic
levels and slower and more persistent responses at high trophic levels”.
Based on both presented length measurements of N. pout during the Barents Sea winter
survey (Wienerroither 2013), and on presented maximum lengths of different age groups in
the North Sea (Lambert et al. 2009), it is likely that the majority of N. pout caught during the
winter survey are of age group 1 and age group 2. This means that possible effects of climate
on N. pout recruitment, are likely to be most visible in N. pout catches after one and two
years. However, the correlations between sea temperatures and N. pout abundance which
were found in this study were highest after two and three years, indicating even more lag
(delay) in this system.
This further lag could for instance be related to possible maternal effects of climate, i.e. that
higher sea temperatures not only creates suitable conditions for the early life stages of N.
pout, but also for the parental generation prior to spawning. This is not unlikely, since N. pout
is a boreal species, which is likely to profit on higher temperatures, both during the adult stage
and the early life stages. Such positive maternal effects of changing climate have previously
been found in other boreal fish species off the Norwegian coast, such as cod (Kjesbu et al.
1998) and herring (Óskarsson et al. 2002).
Effects of environmental change, such as temperature, may also travel through the food chain,
and thus have lagged effects on populations. Such lagged food web effects, where
environmental factors affect prey availability, have previously been found in for instance
marine seabird populations (e.g. Sandvik et al. 2005, Wanless et al. 2007). However, such
lagged food web effects may also affect species through competition or predation; an example
of the latter is the lagged response of temperature which has been found in capelin in the
Barents Sea (Hjermann et al. 2004). Still, this lagged response is the opposite of the one found
in N. pout, where higher temperatures subsequently lead to a decreasing capelin population,
which was suggested to be due to increased predation of stronger year classes of cod and
herring (i.e. higher temperatures improves recruitment in these species), making it an indirect
and delayed bottom-up effect of temperature on the capelin population (Hjermann et al.
2004). However, although N. pout respond opposite of capelin (i.e. subsequently increasing
population with increasing temperatures), N. pout may somehow profit on changes in other
populations (i.e. predators, prey or competitors) caused by changes in sea temperature.
Another possible reason for the lagged relationship between N. pout abundance and sea
temperatures is also worth considering, namely the gradual expansion of the distributional
area which seem to have happened during the study period. When considering that it takes
time for fish to expand into new areas, to make spawning migrations etc., this gradual
expansion may to some extent explain both the gradual increase in abundance which is found
during most of the study period, and to some degree the lagged relationship which has been
found between abundance and sea temperatures.
The last two years of the study period reveals a marked decrease in N. pout abundance
indices. This may be related to a delayed response of the decrease in sea temperatures in the
period 2008 – 2011, which were in contrast to the high temperatures in 2006 and 2007 (time
series maxima). Also, interestingly enough, the proportion of catches including N. pout has
shown a relatively strong variation the last four years of the study period, with a relatively
high proportion of zero N. pout catches in 2011 and 2013, which may indicate that the
decrease in abundance being related to a decrease in distribution. Still, there could of course
be other explanations for the recent decline in abundance, for instance possible higher
predation pressure from other boreal species such as cod. However, to conclude in this matter,
further investigations of both predator and prey species of N. pout in the Barents Sea, is
4.4.2. Temperature and recruitment
How environmental factors (e.g. temperature) may affect recruitment is an interesting and old
question which has engaged many scientists over a long time span, and within many species
and areas (e.g. Cushing 1982 in Myers 1998, Myers 1991, Ottersen et al. 1994 and Myers
1998). Myers (1998) retested correlations between abiotic factors and recruitment in many
species, and found that such correlations generally only were statistically significant in
populations close to the limits of their range, in which moderated environmental conditions
were related to increased recruitment. These findings supported the hypothesis of Huffaker
and Messenger (1964 in Myers 1998), a hypothesis which also was supported by Myers
(1991), in where recruitment of cod, haddock and herring was found to be more variable at
the limit of these species range.
N. pout recruitment has not been studied in the Barents Sea, but Kempf et al. (2009) studied
the species’ recruitment in the North Sea and Skagerrak in the period 1992 - 2006, and
suggest that sea surface temperature during spring determined the overall level of N. pout
recruitment, with lower temperatures yielding higher recruitment. However, when sea surface
temperatures exceeded 8.5 °C, the same relationship between temperature and recruitment
was not recognized (Kempf et al. 2009).
The relationship between temperature variability and recruitment in N. pout has generally not
been well investigated, which makes it useful to discuss the results from this thesis in the light
of similar studies on other species. Ottersen et al. (1994) studied the influence of temperature
variability on recruitment of cod and haddock in the Barents Sea, where also these species
have their distributional range limit, and found that the difference in recruitment strength
between colder and warmer years was statistically significant for cod and haddock for the
period 1965 - 1992. The results also showed that the influence of temperature on cod
recruitment had increased during the latter 25 years compared to previous decades, which was
suggested to be due to increased sensibility to environmental variations as the spawning stock
was declining, and also the change in the age composition of the stock.
Sætersdal and Loeng (1987) concluded that conditions which increase survival of cod larvae
in the Barents Sea is related to the occurrence of a larger and warmer Atlantic component of
the Norwegian coastal current, a hypothesis which they found support for in the high temporal
similarity in survival of cod, haddock and herring in the Barents Sea. Mukhina et al. (1987, in
Nakken 1994) also found increased larval transport into the Barents Sea in years of abundant
year classes, linking larger inflow of Atlantic water and cod larvae into the Barents Sea with
increased survival of cod. Nakken (1994) summarizes possible ways that temperature may
affect cod recruitment, where the timing of sufficient prey for cod larvae (Eilertsen et al.
1989) such as older copepodite statges and adult Calanus (Folkvord et al. 1993) is likely to
increase growth and survival through reduced cannibalism (Folkvord et al. 1993) and reduced
starvation (Sundby et al. 1989) is regarded as essential. Also, the growth of older and adult
cod have been found to increase with increasing temperatures (Nakken 1994).
In conclusion, abiotic factors such as temperature, are likely to affect N. pout recruitment
conditions also in the Barents Sea and on the spawning grounds off the coast of Nordland and
Troms; in where increasing temperatures are likely to have positive effects on growth and
survival of the early life stages of N. pout, as well as possibly affecting the growth and
survival of N. pout spawners, which if so will affect the amount and quality of eggs and
larvae. However, it is important to also consider the probable correlation between increased
inflow of warmer Atlantic water and increased inflow of N. pout larvae from the Norwegian
Sea, which underlies the importance of not confusing correlation with causation, which is
4.4.3. Causation or only correlation?
The first main premise of this thesis is that the Barents Sea population of N. pout has been
found to increase throughout the study period (1994 - 2013). The other main premise is that
the sea temperature of the Barents Sea also has increased throughout the study period. Still, as
mentioned, the increasing Barents Sea N. pout population may at least partly be due to an
increased inflow of warmer Atlantic water into the Barents Sea. As previously mentioned,
increased Atlantic water inflow has been found to explain the shift towards higher volume of
Atlantic water in the Barents Sea (Zhang et al. 1998), and also, Mukhina et al. (1987, in
Nakken 1994) found increased larval transport into the Barents Sea in years of abundant year
classes, linking larger inflow of Atlantic water and cod larvae into the Barents Sea with
increased survival of cod.
Johannesen et al. (2012) underlines that changes, both in relationships between species in the
Barents Sea, and between temperature and various biological parameters, makes it
challenging to predict effects of future climate change based on previous relationships in the
dynamic system, which the Barents Sea is. Although the N. pout population generally has
experienced an increase in abundance and distribution at the same time as sea temperatures
generally also have been increasing, it is hard to determine how much of the correlation that is
related to causation, and how much which is not.
I have previously suggested that the increase in N. pout abundance in the Barents Sea is
linked to the increase in sea temperatures, although with a time lag. Still, both biotic and
abiotic factors have to be taken into account, and it is likely that many other factors than
temperature also affect the N. pout population in various ways, both directly and indirectly;
e.g. competing species, predators, spatiotemporal variation in prey and physical properties
such as currents, salinity and oxygen may work both directly and indirectly at the same time.
Indeed, Myers (1998) concluded that when factors such as mortality across life stages and
density-dependent mortality in juvenile stages are combined, the ability to predict recruitment
from environmental factors is limited.
4.5. Possible ecosystem effects of a changing N. pout population
N. pout Sea is regarded as an important link in the North Sea ecosystem, and an important
prey species for larger predators such as cod in the North Sea (Albert 1994). N. pout have also
been found to be more common in stomach samples from cod in the Barents Sea in recent
years (Edda Johannesen, pers. comm.), and should therefore be regarded as a potential
important link and prey species also in the southwestern part of the Barents Sea, the area of
the Barents Sea where it has been found to be most abundant.
Based on the cod stomach samples, the increasing cod stock in the Barents Sea might have a
controlling effect on N. pout. Although N. pout in the North Sea has been found to mainly
prey on crustaceans (e.g. Albert 1994), the possible role of N. pout as predator on cod and
other gadoid larvae and juveniles should also be considered. Also, age group 0 of N. pout in
the Barents Sea may for instance be an important competitive species to age group 0 of other
gadoids, such as cod, haddock and saithe in the southwestern part of the Barents Sea. Indeed,
common prey niches have previously been found among N. pout and other gadoids in the
North Sea (Bromley et al. 1997). Although the role of N. pout both as prey, predator and
competitive species remain unclear in the Barents Sea, it is natural to think that an increasing
N. pout population with an increasing distribution area will have increasing impact on other
Finally, it is worth mentioning that it is not an easy task to suggest how a changing N. pout
population might affect other species in the Barents Sea, simply due to lack of data
concerning this N. pout population. Also, many factors vary both temporally and spatially in
an ecosystem, which makes it hard to distinguish ultimate from proximate factors, and
equally, as previously discussed, easy to mix correlation with causation.
4.6. Concluding remarks
Johannesen et al. (2012) states that although good time series data are lacking on many
species, there are indications of increasing distribution range of southerly warm-water species
(i.e. boreal species), with simultaneously indications of decreasing distribution ranges of
Arctic species, in periods of a warmer Barents Sea; still, Kjesbu et al. (2014) points out that
fishing also have a large impact on the abundance of such species. N. pout, together with
species such as cod and haddock, belongs to the first species category, and is therefore
expected to increase in abundance and distribution range in periods with high Barents Sea
Indeed, the Barents Sea N. pout population has in this study been found to increase in
abundance and spatial distribution, and higher sea temperatures seem to have had a positive
lagged effect on the abundance of the population. Due to that N. pout is a boreal species
which has never been commercially exploited in the Barents Sea, it can be regarded as a well
suited indicator fish species for climate change in the Barents Sea. In the North Sea, N. pout is
regarded as an important link in the ecosystem, between prey such as invertebrates and small
fish, and predators such as larger fish (e.g. Albert 1994); thus, I suggest that it is important to
know how this particular species may respond to future climate changes in the Barents Sea,
and further, how it may affect other species in this ecosystem. Little is so far known about the
N. pout population of the Barents Sea, but the influence of this population on the Barents Sea
ecosystem seem likely to increase if the recent warming of the Barents Sea continues.
Albert, O. T. 1994. Biology and ecology of Norway pout (Trispterus esmarkii Nilsson, 1855)
in the Norwegian Deep. ICES Journal of Marine Science, 51: 45 - 61.
Albert, O. T. 1995. Diel changes in food and feeding of small gadoids on a coastal bank. ICES
Journal of Marine Science, 52: 873 - 885.
Andriyashev, A. P. and Chernova, N. V. 1995. Annotated list of fishlike vertebrates and fish
of the Arctic seas and adjacent waters. Journal of Ichthyology, 35: 81 - 123.
Anon. 2011. Survey report from the joint Norwegian/Russian ecosystem survey in the Barents
Sea August-October 2011. IMR/PINRO Joint Report Series, No. 3. ISSN 1502-8828.
Bailey, R. S., and Kunzlik, P. A. 1984. Variation in growth and mortality rates of Norway
pout Trisopterus esmarkii (Nilsson). ICES Document CM 1984/G: 70.
Bakketeig I.E., Gjøsæter H., Hauge M., Sunnset B.H. and Toft K.Ø. (Eds.) 2014.
Havforskningsrapporten 2014. Fisken og havet, No. 1 (Special issue). pp 213.
Baranenkova, A. S. 1960. Results of the estimation of young cod and haddock in the Barents
Sea in the winter season 1959-1960. ICES, C. M., Gadoid Committee, Doc. No. 100. 7
Baranenkova, A. S. and Khoklina, N. S. 1968. Distribution of eggs, larvae, and adults of
Norway pout off north-western Norway and in the Barents Sea. Rapports et Procésverbaux des Réunions du Conseil Permanent International pour l’Exploration de la
Mer, 158: 90 - 100.
Bohl, E. 1980. Diel pattern of pelagic distribution and feeding in planktivorous fish.
Oecologica, 44: 368 - 375.
Bromley, P. J., Watson, T. and Hislop, J. R. G. 1997. Diel feeding patterns and the
development of food webs in pelagic 0-group cod (Gadus morhua L.), haddock
(Pollachius virens L.), and Norway pout (Trisopterus esmarkii Nilsson) in the
northern North Sea. ICES Journal of Marine Science, 54: 846 - 853.
Brown, J. H. 1984. On the relationship between abundance and distribution of species.
American naturalist, 124: 255 - 279.
Burnham, K. P. and Anderson, D. R. 1992. Data-based selection of an appropriate biological
model: The key to modern data analysis. In McCullough, D. R. and Barrett, R. H.
(Eds.) Wildlife populations: Populations. Elsevier Applied Science, London and New
York (pp. 16 - 30).
Cohen, D.M., Inada, T., Iwamoto, T. and Scialabba, N. 1990. FAO species catalogue. Vol. 10.
Gadiform fishes of the world (Order Gadiformes). An annotated and illustrated
catalogue of cods, hakes, grenadiers and other gadiform fishes known to date. FAO
Fish. Synop. 125(10). Rome: FAO. 442 pp.
Cushing, D.H. 1982. Climate and Fisheries. Academic Press, London. 373 pp.
Drinkwater, K., Loeng, H., Titov, O. V. and Boitsov, V. D. 2011. Global warming and
climate change. In Jakobsen, T. and Ozhigin, V. K. (Eds.). The Barents Sea:
Ecosystem, resources and management – Half a century of Russian – Norwegian
cooperation (pp. 777 - 807). Tapir Academic Press, Trondheim.
Eilertsen, B., Fossum, P., Solemdal, P and Sundby, S. 1989. Realtion between temperature
and survival of eggs and first-feeding larvae of northeast Arctic cod (Gadus morhua
L.). Rapports et Proces-Verbaux des Reunions du Conseil International pour
I'Exploration de la Mer, 191: 209 - 219.
Ekman, S. 1967. Zoogeography of the Sea. Sidgwick and Jackson, Dublin. 417 pp.
Folkvord, A., Blom, G., Dragesund, O., Johannessen, A., Nakken, O. and Nævdal, G. 1993. A
conseptual framework for enhancing and stabilizing recruitment of marine stocks.
Aquaculture and Fisheries Management, 25, Supplement 1, 245 - 258, conceptual.
Fuiman, L. A. and Magurran, A. E. 1994. Development of predator defences in fishes.
Reviews in Fish Biology and Fisheries, 4: 145 - 183.
Goodman, L. A. 1984. The analysis of cross-classified data having ordered categories.
Harvard University, Cambridge, Massachusetts. 414 pp.
Gorshkov, S. G. (Ed.) 1980. Atlas of the Oceans. Arctic Ocean. USSR Ministry of Defense,
Voenno-Morskoy Press, Moscow. 184 pp.
Hjellvik, V. 2005. DIVA V2.0.3 An application for calculating and adjusting for diurnal
variation of marine populations. Unpublished user manual. 22 pp.
Hjellvik, V., Godø, O. R. and Tjøstheim, D. 2002. Diurnal variation in bottom trawl survey
catches: does it pay to adjust? Canadian Journal of Fisheries and Aquatic Sciences,
59: 33 - 48.
Hjermann, D. O., Stenseth, N. C. and Ottersen, G. 2004. Indirect climatic forcing of the
Barents Sea capelin: a cohort effect. Marine Ecology Progress Series, 273: 229 - 238.
Huffaker, C. B. and Messenger, P. S. 1964. The concept and significance of natural control.
Chapman and Hall, London. 306 pp.
Hutchinson, G. E. 1967. A treatise on limnology, Vol. 2. Introduction to lake biology and the
limnoplankton. Wiley, New York. 1115 pp.
ICES. 2007 Report of the working group on the assessment of demersal stocks in the North
Sea and Skagerrak. ICES Document CM 2007/ACFM: 07. 60 pp.
Jakobsen, T., Korsbrekke, K., Mehl, S. and Nakken, O. 1997. Norwegian combined acoustic
and bottom trawl surveys for demersal fish in the Barents Sea during winter. ICES CM
1997/Y: 17. 26 pp.
Jakobsen, T. and Ozhigin, V. K. 2011. The Barents Sea – Ecosystem, resources and
management. Tapir Academic Press, Trondheim. 825 pp.
Johannesen, E., Ingvaldsen, R. B., Bogstad, B., Dalpadado, P., Eriksen, E., Gjøsæter, H.,
Knutsen, T., Skern-Mauritzen, M., Stiansen, J. E. 2012. Changes in Barents Sea
ecosystem state, 1970-2009: climate fluctutations, human impact, and trophic
interactions. ICES Journal of Marine Science, 69: 880 - 889.
Johannesen, E., Wenneck, T. L., Høines, Å., Aglen, A., Mehl, S., Mjanger, H., Fotland, Å.,
Halland, T. I. and Jakobsen, T. 2009. Can the winter survey be useed to detect changes
in the Barents Sea fish community? – An overview of methods and data from 19812007. Fisken og havet, No. 7. 29 pp.
Johannessen, J. L., Olsen, S., Stålesen, O. 1964. The Norwegian fisheries for Norway pout.
ICES C. M. Gadoid Committee, Doc. No. 120: 5 pp.
Kempf, A., Floeter, J. and Temming, A. 2009. Recruitment of North Sea cod (Gadus morhua)
and Norway pout (Trisopterus esmarkii) between 1992 and 2006: the interplay
between climate influence and predation. Canadian Journal of Fisheries and Aquatic
Sciences, 66: 633 - 648.
Kerfoot, W.C. 1985. Adaptive value of vertical migration: comments on the predation
hypothesis and some alternatives. Contributions in Marine Science, 27 (Suppl.): 91 113.
Kjesbu, O. S., Bogstad, B., Devine, J. A., Gjøsæther, H., Howell, D., Ingvaldsen, R. B., Nash,
R. D. M. and Skjæraasen, J. E. 2014. Synergies between climate and management for
Alantic cod fisheries at high latitudes. Proceedings of the National Academy of
Sciences of the United States of America, 111: 3478 - 3483.
Kjesbu, O. S., Witthames, P. R., Solemdal, P. and Greer Walker, M. 1998. Temporal response
in the fecundity of Arcto-Norwegian cod (Gadus morhua) in response to natural
changes in food and temperature. Journal of Sea Research, 40: 303 - 321.
Lambert, G., Nielsen, R., Larsen, L. I. and Sparhold, H. Maturity and growth population
dynamics of Norway pout (Trisopterus esmarkii) in the North Sea, Skagerrak, and
Kattegat. ICES Journal of Marine Science, 66: 1899 - 1914.
Lampert, W. 1989. The adaptive significance of diel vertical migration of zooplankton.
Functional ecology, 3: 21 - 27.
Levitus, S., Matishov, G., Seidov, D. and Smolyar, I. Barents Sea multidecadal variability.
Geophysical Research Letters, 36. DOI: 10.1029/2009GL039847.
Mason, J. 1960. A report on the distribution of Gadus esmarkii Nilsson in the North Sea and
adjacent waters. ICES CM. Gadoid Committee, Doc. No. 41. 4 pp.
Mukhina, N. V., Mukhin, A. I. and Dvinina, E. A. 1987. Oceanographic conditions and
reproduction of Arcto-Norwegian cod of the Barents Sea in 1980-1985. In Loeng
(Ed.). The effect of oceanographic conditions on distributions and population
dynamics of commercial fish stocks in the Barents Sea. Proceedings of the Third
Soviet-Norwegian Symposium, Murmansk, 26 - 28 May, 1986. Institute of Marine
Research, Bergen, 1987: 145 - 158.
Munk, P., Larsson, P. O., Danielssen, D. S and Moksness, E. 1999. Variability in frontal zone
formation and distribution of gadoid fish larvae at the shelf break in the northeastern
North Sea. Marine Ecology Progress Series, 177: 221 - 233.
Myers, R. A. 1991. Recruitment variability and range of three fish species. NAFO Science
Council Studies, 16: 21 - 24.
Myers, R. A. 1998. When do environment-recruitment correlations work? Review in Fish
Biology and Fisheries, 8: 285 - 305.
Nakken, O. 1994. Causes of trends and fluctuations in the Arcto-Norwegian cod stock. ICES
Marine Science Symposia, 198: 212 - 228.
Nash, R. D. M., Wright, P. J., Matejusova, I., Dimitrov, S. P., O’Sullivan, M., Augley, J. and
Höffle, H. Spawning location of Norway pout (Trisopterus esmarkii Nilsson) in the
North Sea. ICES Journal of Marine Science, 69: 1338 - 1346.
Onsrud, M. S. R., Kaartvedt, S., Røstad, A. and Klevjer, T. A. 2004. Vertical distribution and
feeding patterns in fish foraging on the krill Meganyctiphanes norvegica. ICES
Journal of Marine Science, 61: 1278 - 1290.
Óskarsson, G. J., Kjesbu, O. S. and Slotte, A. 2002. Predictions of realised fecundity and
spawning time in Norwegian spring-spawning herring (Clupea harengus). Journal of
Sea Research, 48: 59 - 79.
Ottersen, G., Kim, S., Huse, G., Polovina, J. J. and Stenseth, N. C. 2010. Major pathways by
which climate may force marine populations. Journal of Marine Systems, 79: 343 360.
Ottersen, G., Loeng, H. and Raknes, A. 1994. Influence of temperature variability of cod in
the Barents Sea. ICES Journal of Marine Science, 198, 175 - 193.
Ozhigin, V. K., Ingvaldsen, R. B., Loeng, H., Boitsov, V. D. and Karsakov, A. L. 2011. The
Barents Sea. In Jakobsen, T. and Ozhigin, V. K. (Eds.). The Barents Sea: Ecosystem,
resources and management – Half a century of Russian – Norwegian cooperation (pp.
39 - 76). Tapir Academic Press, Trondheim.
Raitt, D. F. S. 1960. Preliminary studies on the age and growth of Gadus esmarkii Nilsson.
ICES 1960 Gadoid Committe, Doc. No. 40: 7 pp.
Raitt, D. F. S. 1968a. Synopsis of biological data on the Norway pout Trisopterus esmarkii
(Nilsson, 1855). FAO Fisheries Synposis No. 33, Rev. 1. 36 pp.
Raitt, D. F. S. 1968b. The population dynamics of Norway pout in the North Sea. Marine
Research, 5: 1 - 24.
Raitt, D. F. S. and Adams, J. A. 1965. The food and feeding of Trisopterus esmarkii (Nilsson)
in the Northern Sea. Marine Research, 3: 1 - 28.
Sandvik, H., Erikstad, K. E., Barrett, R. T. and Yoccoz, N. G. 2005. The effect of climate on
adult survival in five species of North Atlantic seabirds. Journal of Animal Ecology,
74: 817 - 831.
Slagstad, D and Wassmann, P. 1997. Climatic change and carbon flux in the Barents Sea: 3-D
simulations of ice-distribution, primary production and vertical export of particulate
organic carbon. Memoirs of National Institute of Polar Research, 51 (Special issue):
119 - 141.
Sparholt, H., Larsen, L. I., and Nielsen, J. R. 2002a. Non-predation natural mortality of
Norway pout (Trisopterus esmarkii) in the North Sea. ICES Journal of Marine
Science, 59: 1276 - 1284.
Sparholt, H., Larsen, L. I. and Nielsen, J. R. 2002b. Verification of multispecies interactions
in the North Sea by trawl survey data on Norway pout (Trisopterus esmarkii). ICES
Journal of marine science, 59: 1270 - 1275.
Stafford, K. M., Moore, S. E. and Fox, C. G. 2005. Diel variation in blue whale calls recorded
in the eastern tropical Pacific. Animal Behaviour, 69: 951 - 958.
Stenevik, E. K. and Sundby, S. 2007. Impacts of climate change on commercial fish stocks in
Norwegian waters. Marine Policy, 31: 19 - 31.
Sundby, S., Bjørke, H., Soldal, A. and Olsen, S. 1989. Feeding conditions of ArctoNorwegian cod larvae compared with the Rothschild-Osborn theory on small-scale
turbulence and plankton contact rates. Rapports et Proces-Verbaux des Reunions du
Conseil International pour I'Exploration de la Mer, 191: 351 - 358.
Sundby, S., Fossum, P., Sandvik, A., Vikebø, F. B., Aglen, A., Buhl-Mortensen, L., Folkvord,
A., Bakkeplass, K., Buhl-Mortensen, P., Johannessen, M., Jørgensen, M. S.,
videreutvikling av kunnskapsgrunnlaget for mengde, fordeling og transport av
kommersielle fiskearter fra gytende fisk til egg, larver, yngel og ungfisk i nordnorske
sokkelområder med vekt på Lofoten og Vesterålen. Fisken og havet, No. 3. 186 pp.
Svetovidov, A. N. 1948. Treskoobraznye [Gadiformes]. Fauna SSSR, Zoologicheskii lnstitut
Akademi Nauk SSSR (n.s.) 34, Ryby [Fishes] 9,4: 1 - 222 [in Russian; English
translation, 1962, Jerusalem: Israel Program for scientific translations, 304 pp].
Sætersdal, G. and Loeng, H. 1987. Ecological adaptation of reproduction in Northeast Arctic
cod. Fisheries Research, 5: 253 - 270.
Wanless, S., Frederiksen, M., Daunt, F., Scott, B. E. and Harris, M. P. 2007. Black-legged
kittiwakes as indicators of environmental change in the North Sea: Evidence from
long-term studies. Progress in Oceanography, 72: 30 - 38.
Wienerroither, R., Johannesen, E., Dolgov, A., Byrkjedal, I., Aglen, A., Bjelland, O.,
Drevetnyak, K., Eriksen, K. B., Høines, Å., Langhelle, G., Langøy, H., Murashko, P.,
Prokhorova, T., Prozorkevich, D., Smirnov, O and Wenneck, T. 2013. Atlas of the
Barents Sea Fishes based on the winter survey. IMR-PINRO Joint Report Series, No.
2. 220 pp.
Zhang, J. L., Rothrock, D. A. and Steele, M. 1998. Warming of the Arctic Ocean by a
strengthened Atlantic inflow: Model results. Geophysical Research Letters, 25: 3541 3541.
APPENDIX 1 - Data exploration
APPENDIX 2 - Sources of error
APPENDIX 1 – Data exploration
The distribution of N. pout zero catches (i.e. no N. pout in the catches) for the whole period
1994 - 2013 was investigated; a total of 57.8 % of all catches (n = 6159) were zero catches
(App. figure 1.1). When studying the variation in diel catchability it is important to
investigate zero catches carefully (Hjellvik et al. 2002): An uneven distribution of zero
catches could be an indication of diel variation in catchability, and omitting the zero catches
could in such cases lead to an underestimation of the diel variation (Hjellvik et al. 2002).
However, the zero catches in this study were not found to vary significantly over the 24 h
cycle (p = 0.24) (App. figure 1.2); thus, all zero-catches could be omitted.
App. figure 1.1. Distribution of N. pout numbers in all trawl catches throughout the study period (n = 6159),
with zero catches (n = 3557) and smaller catches (n = 2431) to the left, and larger catches (n = 171) to the right,
of the dotted line.
Indeed, model 1 explained more of the total variation without the zero-catches (R2 = 0.67)
compared to when all catches taken in strata with at least 50% non-zero-catches were
included (R2 = 0.56), which in turn lead to a preference of omitting zero-catches compared to
this 50 % alternative. However, when all zero-catches were included, R2 was even larger (R2 =
0.74) than when omitting all zero-cathes. Still, since the zero catches were not found to vary
significantly over the 24h cycle, and since including including all zero-catches gave a poor
normal Q-Q plot of residuals (which means that the residuals of the model don’t fit the normal
distribution of the actual data) (App. figure 1.3), compared to the Q-Q plot when zero-catches
were omitted (App. figure 1.4), all zero-catches were omitted from models 1, 2 and 3.
Time (2 hr intervals)
App. figure 1.2. Distribution of the number of zero catches (i.e. catches without N. pout) in 2hr intervals over
24hr period; first bin from hr 00.00 - 01.59, etc. (output figure from DIVA).
App. figure 1.3. Q-Q-plot of residuals, visualizing how model 1 fits the data when all zero catches have been
included (output figure from DIVA).
App. figure 1.4. Q-Q plot of residuals, visualizing how model 1 fits the data when all zero catches have been
omitted (output figure from DIVA).
Shape of the function of diel variation in catchability
Both the logistic and the sinusoid function were tested in model 1 (the simple model), with
both of the functions having the same R2 = 0.67. In the logistic function, β and D was
estimated to -4.53 (p < 0.0001, SE = 0.92) and 0.48 (p < 0.0001, SE = 0.07), respectively. In
the sinusoid function, D was estimated to 0.57 (p < 0.0001, SE = 0.09) (there is no β in the
sinusoid function). Although the logistic and the sinusoid function showed a large similarity,
only the logistic function was chosen to be further investigated. This was due to that the
logistic function was considered to best describe the diel vertical migration; with an
approximately constant night level, another constant day level, and a transition phase between
them. Also, the standard error of D was lower in the logistic function than in the sinusoid
In the logistic function, α and β determines the length of the dial transition phase, and the time
(or in this case sun angle) of the middle of the transition phase, respectively. Following
Hjellvik et al. (2002) α = 2 was fixed, which corresponds to a transition phase of
approximately 3 hours two times during a 24h period. β was not fixed, hence it had to be
estimated from the models. D describes the diel amplitude of the variation in catchability
between day and night.
App. figure 1.5. Number of N. pout in catches distributed within 2 hr intervals: 0 = h 00.00 – 01.59, 2 = h 02.00
– 03.59, etc.
An annual spatial overview of all trawl stations throughout the study period is shown in App.
figure 1.6. Average trawling depth for all stations has been 266 m (range 52 - 720 m).
App. figure 1.6. Annual spatial oveview of all trawl stations (n = 6159) throughout the study period 1994 2013. Figure made by Edda Johannesen, Institute of Marine Research.
APPENDIX 2 – General sources of error
This appendix briefly discusses other sources error than diel variation in catchability and
survey coverage. The most obvious source of error when calculating abundance indices is the
uneven geographical distribution and the movement of the fish, which leads to a random
variation in catch data within different years, different sectors and different strata. The strata
and sector divisional system is also large and irregular, which makes it impossible to detect
small-scale differences and changes in the N. pout population.
The winter survey design is regular (although it has varied) and stratified, where catch weight
and catch number of all fish species, shrimp and king crab have been recorded (Johannesen et
al. 2009). Length of all species has also been measured at all stations (Johannesen et al. 2009).
However, the data should be used carefully due to changes in how the winter survey has been
conducted, where some of the main changes have been that the survey area has increased
during the sample period, and also that there has been a reduction in towing time, as well as
there have been changes in gear and mesh size (Johannesen et al. 2009). Other factors have
been poorly staffed and equipped commercial vessels which have participated in the survey,
and also that one has started to use new equipment, such as electronic measure boards during
the time series (in 1997) (Johannesen et al. 2009).
A major source of error to keep in mind is that the vertical capture efficiency of the bottom
trawl, is fish size dependent (Jakobsen et al. 1997). Another source of error is that abundance
and distribution not necessarily are two independent variables, this due to the greater
probability of catching fish also in the outskirts of a population, when the abundance of a
population increases (Johannesen et al. 2009).
Worth mentioning is also the fact that the N. pout catches constitute of very many very small
catches, and very few very large catches, which makes the abundance indices especially
vulnerable to coincidence, and the statistical methods less robust.