Seasonal patterns in the nocturnal distribution and behavior of the mesopelagic fish Maurolicus muelleri at high latitudes

Abstract

Acoustic scattering layers (SL) ascribed to pearlside Maurolicus muelleri were studied in Masfjorden, Norway, using upward-looking echo sounders cabled to shore for continuous long-term measurements. The acoustic studies were accompanied by continuous measurements of surface light and supplemented with intermittent field campaigns. From autumn to spring, young M. muelleri formed an SL in the upper similar to 75 to 150 m in the daytime, characterized by migration to near-surface water near dusk, subsequent 'midnight sinking', followed by a dawn ascent before a return to the daytime habitat. Light levels were similar to 1 order of magnitude lower during the dawn ascent than for ascent in the afternoon, with the latter terminating before fish reached upper layers on similar to 1/3 of the nights from late November to mid-April. Adults showed less tendency of migration during autumn and winter, until the SLs of young and adults merged in late spring, and thereafter displayed coherent migration behavior. The midnight sinking became progressively deeper from autumn to winter but was strongly reduced from mid-May when the darkest nocturnal light intensity (PAR) at the surface was above 10(-3) mu mol m(-2) s(-1). The pearlside took on schooling in upper waters during the even lighter nights in early June, with minimum light of similar to 5 x 10(-3) to 10(-1) mu mol m(-2) s(-1) at the surface. Nocturnal schooling ceased in early July, and midnight sinking reappeared in mid-August. We suggest that the strong variation in nocturnal light intensity at high latitudes provides changing trade-offs between visual foraging and avoiding predators and hence varying time budgets for feeding in the upper, productive layers.

Full text

MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 521: 189–200, 2015
doi: 10.3354/meps11139 Published February 17
INTRODUCTION
Mesopelagic fish are widely distributed in the
world’s oceans (Gjøsæter & Kawaguchi 1980, Dal-
padado & Gjøsæter 1988, Irigoien et al. 2014). These
fish act as a trophic link between zooplankton
(Shreeve et al. 2009) and commercially valuable and
other predators (Giske et al. 1990, Potier et al. 2007,
Doksæter et al. 2008). They may also play a signifi-
cant role in the carbon transport from productive
upper layers to food-deprived deeper layers (Her -
nández-León et al. 2010, Bianchi et al. 2013, Irigoien
et al. 2014). Diel vertical migration (DVM) is common
among both zooplankton and fish, usually explained
as trade-offs between avoiding predators and forag-
ing (Rosland & Giske 1994, Pearre 2003). Various
studies have shown that light acts as proximate ini-
tiator for DVM (Kampa & Boden 1954, Kampa 1970),
and whether DVM behavior relates to the preference
for a specific absolute intensity or isolume (Sweatt &
Forward 1985), the rate of change in the light inten-
sity (Ringelberg 1995, Cohen & Forward 2009), or the
preference for a range of light intensities (Staby &
Aksnes 2011) has been debated.
Most focus on the mesopelagic fish DVM in rela-
tion to light has been on light conditions during the
© Inter-Research 2015 · www.int-res.com*Corresponding author: stein.kaartvedt@ibv.uio.no
Seasonal patterns in the nocturnal distribution
and behavior of the mesopelagic fish Maurolicus
muelleri at high latitudes
Perdana K. Prihartato1, Dag L. Aksnes2, Stein Kaartvedt1, 3,*
1King Abdullah University of Science and Technology, Red Sea Research Center, Thuwal 23955-6900, Saudi Arabia
2Hjort Centre for Marine Ecosystem Dynamics, Department of Biology, University of Bergen, Norway
3University of Oslo, Department of Biosciences, PO Box 1066 Blindern, 0316 Oslo, Norway
ABSTRACT: Acoustic scattering layers (SL) ascribed to pearlside Maurolicus muelleri were stud-
ied in Masfjorden, Norway, using upward-looking echo sounders cabled to shore for continuous
long-term measurements. The acoustic studies were accompanied by continuous measurements
of surface light and supplemented with intermittent field campaigns. From autumn to spring,
young M. muelleri formed an SL in the upper ~75 to 150 m in the daytime, characterized by migra-
tion to near-surface water near dusk, subsequent ‘midnight sinking’, followed by a dawn ascent
before a return to the daytime habitat. Light levels were ~1 order of magnitude lower during the
dawn ascent than for ascent in the afternoon, with the latter terminating before fish reached upper
layers on ~1/3 of the nights from late November to mid-April. Adults showed less tendency of
migration during autumn and winter, until the SLs of young and adults merged in late spring, and
thereafter displayed coherent migration behavior. The midnight sinking became progressively
deeper from autumn to winter but was strongly reduced from mid-May when the darkest noctur-
nal light intensity (PAR) at the surface was above 10−3 µmol m−2 s−1. The pearlside took on school-
ing in upper waters during the even lighter nights in early June, with minimum light of ~5 ×10−3
to 10−1 µmol m−2 s−1 at the surface. Nocturnal schooling ceased in early July, and midnight sinking
reappeared in mid-August. We suggest that the strong variation in nocturnal light intensity at
high latitudes provides changing trade-offs between visual foraging and avoiding predators and
hence varying time budgets for feeding in the upper, productive layers.
KEY WORDS: Behavior · Diel vertical migration · Light levels · Mesopelagic · Acoustics
Resale or republication not permitted without written consent of the publisher

Mar Ecol Prog Ser 521: 189–200, 2015
day (Baliño & Aksnes 1993) or differences between
day and night (O’Driscoll et al. 2009, Klevjer et al.
2012). However, light levels also vary at night. Fish
vertical distribution, predation, and foraging activi-
ties can significantly be affected by nocturnal lights
in relation to moonlight (Gliwicz 1986, Benoit-Bird et
al. 2009) but also by seasonal cycles in the nocturnal
light climate (Sameoto 1989, Rasmussen & Giske
1994, Kaartvedt 2008). Masfjorden, Norway, is
located at a high latitude (~60° N) and represents a
relatively deep, protected, semi-enclosed water
body, providing a natural laboratory to study meso-
pelagic fish. Maurolicus muelleri, or pearlside, which
is the focus species of this study, is the prevailing
acoustic target in the upper ~200 m of Masfjorden
during daytime, being distributed shallower during
the night (Giske et al. 1990, Baliño & Aksnes 1993,
Rasmussen & Giske 1994, Kaartvedt et al. 1998,
Aksnes et al. 2004, Staby & Aksnes 2011, Staby et al.
2011). However, not all parts of the population
appear to take part in DVM at all times (Staby et al.
2011).
Several studies have indicated that the vertical dis-
tribution of pearlside is related to a certain range of
light levels (Baliño & Aksnes 1993, Kaartvedt et al.
1996, Staby & Aksnes 2011). Pearlside appear to take
advantage of the so-called anti-predation window
(Clark & Levy 1988) and forage in upper waters at
dusk and dawn. Particularly juveniles migrate to the
surface in the afternoon and spend a short period
near the surface, followed by midnight sinking and a
subsequent dawn rise before returning to their day-
time depth (Giske et al. 1990, Baliño & Aksnes 1993,
Rasmussen & Giske 1994, Staby et al. 2011, 2013).
According to the anti-predation window hypothe-
sis (Clark & Levy 1988), small planktivorous fish seek
out intermediate light levels that are sufficient for
obtaining food while at the same time being suffi-
ciently low to be relatively safe from visually search-
ing piscivores. ‘Midnight sinking’ follows as it
becomes too dark for foraging at night. However, at
high latitudes, the timing of the anti-predation win-
dow in upper waters may change, as nights get much
lighter during spring and summer. This pattern may
prevent vertically migrating fish from entering sur-
face layers for feeding if there is too much light
(Sameoto 1989, Kaartvedt 2008). On the other hand,
the dusky light levels of summer nights might cause
an extended, shallow anti-predation window, per-
mitting foraging in likely food-rich upper waters
throughout the night. Some previous studies in Nor-
way have suggested that midnight sinking may be
skipped entirely in summer (Rasmussen & Giske
1994, Kaartvedt et al. 1998, Staby et al. 2011). M.
muelleri may even initiate schooling in upper layers
during light summer nights (Kaartvedt et al. 1998).
However, such reports are mostly from short-term
observations, and light levels associated with
switches between the different behaviors have not
been established.
In this study, we take advantage of the unique
opportunity for a long-term study of M. muelleri
offered by a deep locality in close proximity to land.
We applied upward-looking echosounders cabled to
shore that provided continuous data on vertical
migration behavior from autumn throughout the sub-
sequent summer. This research was accompanied
with continuous measurements of surface light and
intermittent field campaigns, including measure-
ments of water-column light extinction. We here
address the seasonal patterns in DVM behavior of M.
muelleri, emphasizing the nocturnal distribution,
particularly focusing on the relatively light summer
nights.
MATERIALS AND METHODS
Acoustic measurements
The study was carried out at a ~370 m deep loca-
tion in Masfjorden, Norway (~60° 50’ N, ~5° 30’ E),
from 7 October 2010 to 15 August 2011. Continuous
acoustic measurements were made near the location
of former studies (Staby et al. 2011, Dypvik et al.
2012a,b; see Fig. 1a). We used 3 upward looking
SIMRAD EK60 split beam echo sounders (7.1° beam
widths), mounted on the bottom (38 kHz; ~370 m)
and on rigs anchored to the bottom, floating at
~250 m (120 kHz) and ~90 m (200 kHz) (Fig. 1). Echo
sounder depths were selected to give high- resolution
data of different segments of the water column. The
echo sounders were calibrated at the surface using
standard methods (Foote et al. 1987). Pulse lengths
and ping rates were 128 µs and 1 to 2 pings s−1
(200 kHz), 256 µs and 1 to 2 pings s−1 (120 kHz), and
512 µs and 1 ping s−1 (38 kHz), respectively.
The submerged transceivers were kept in pres-
sure-proof casings and cabled to shore for power and
transmittance of digitized signals to laptop comput-
ers, where data were stored in raw format for later
analysis. The laptops were connected to the Internet
for delivery of real-time echograms via web-based
interfaces as well as to remotely control the echo
sounders. This linkage allowed remote access for
restarting the echo sounders after periods of power
190

Prihartato et al.: Nocturnal distribution and behavior of Maurolicus muelleri
failures, which occurred due to a periodically unsta-
ble electrical line.
Observed DVM patterns were largely similar at 38
and 120 kHz. This similarity between the 2 frequen-
cies suggests that the observed DVM patterns can be
ascribed to fish because most plankton will not be
detected at 38 kHz with the settings applied here
(Kaartvedt et al. 2008). Moreover, Maurolicus muel-
leri mainly occurs in the upper ~250 m of the water
column. Hence, we only present data from the float-
ing echo sounders at 120 and 200 kHz, although we
also refer to findings at 38 kHz. Data were obtained
for 276 d and 269 d for 120 kHz and 200 kHz, respec-
tively.
Processing and visualization of acoustic data
Echograms were visualized using MATLAB (ver-
sion 2012b). Seasonal patterns of DVM were pre-
sented as monthly averaged daily 24 h echograms,
with bin size of 2 min ×5 cm, for the section of ~0 to
250 m. We organize the monthly echograms into 4
seasonal categories: autumn (October to November),
winter (December to February), spring (March to
May), and summer (June to August).
To display specific behavior patterns of organisms
forming nighttime acoustic scattering layers (SLs), 4
high-resolution examples of SLs were chosen from
both echo sounders to represent events that occurred
in the course of the registration period: interrupted
ascent (6 to 8 January), termination of midnight sink-
ing (14 to 16 May), schooling behavior at light sum-
mer nights (18 to 20 June) and resumption of mid-
night sinking (14 to 16 August). This selection was
done for the whole days at 120 kHz, and with finer
resolution for better visualization at 200 kHz.
The seasonal patterns of nocturnal scattering were
visualized by importing all nocturnal data from
120 kHz and 200 kHz into respective echograms
spanning the whole study period. Night was defined
as the period after sunset and before sunrise at local
time (UTC + 1 h; UTC + 2 h during daylight saving
time from 28 March to 30 October 2011). Echo data
were intermittently affected by different sources of
noise, so prior to importing the data for final analysis,
supervised noise removal was conducted to exclude
‘bad echoes’ using the LSSS software (Korneliussen
et al. 2009). This process involved filtering of a 9-point
moving average window to remove irregular spikes.
The output value of this process was the scattering
area (SA), which was then translated into logarithmic
value of backscattering volume (SV; Knudsen 1990).
SAwas integrated at a resolution of 15 min ×2 m. The
upper 0–4 m was deleted to avoid surface-generated
noise.
Composition of scattering layers
SLs in Masfjorden corresponding to those referred
to in this paper have repeatedly been ascribed to M.
muelleri. Previous studies conclude that the young-of
the year form SLs in the upper 100 m from autumn to
spring, and the adults stay at greater depths (Baliño
& Aksnes 1993, Rasmussen & Giske 1994, Staby et al.
2011). Results from extensive trawling with a 100 m2
pelagic trawl at the start (7 October 2010) and end
(15 August 2011) of the present study were in accor-
dance with these previous findings, with daytime
191
Fig. 1. (A) Map of Masfjorden depicting locations of the 3 echo sounders. The upward-looking echo sounders were mounted
on the bottom ~370 m (38 kHz) and floating in the water column at ~250 m (120 kHz) and 90 m (200 kHz). (B) Sketch of the
experimental set up

Mar Ecol Prog Ser 521: 189–200, 2015
catches in the upper 150 m being completely domi-
nated by M. muelleri. As the identity of these SLs is
already well documented through previous work,
details of trawl catches are not presented for the pur-
pose of this paper. Additional plankton targets were
apparent, mostly at 200 kHz.
Light measurements
Photosynthetically active radiation (PAR, 400 to
700 nm) was continuously recorded about 2 m above
the surface (referred to as surface irradiance) from 10
December 2010 to 15 August 2011 with a calibrated
LI-190 quantum sensor and data stored on a LI-1400
data logger. Measurements were averaged and
stored every 15 min. The lower threshold of the sen-
sor was 0.0001 µmol m−2 s−1. The surface irradiance at
the darkest period at night (minimum nocturnal light
levels) was higher than this sensitivity threshold from
28 April until records ended on 15 August 2011. Prior
to April 28, the sensitivity of the sensor was not ade-
quate to measure light during night. A few inci-
dences of relatively high nocturnal light measured
during winter were removed from the dataset be -
cause these incidents were ascribed to artificial light-
ing at the shore station where the logging was con-
ducted.
We used the results of surface irradiance as a proxy
for the irradiance at the top of the SL (as visible in the
high-resolution 200 kHz echogram when using an SV
threshold of −75 dB) when the SL approached the
surface in the morning (right before dawn descent)
and evening (dusk ascent). However, sometimes
interrupted ascents of the SL were observed in the
afternoon, i.e. the top of the SL was then located well
below the surface. In these cases, results from under-
water light measurements were used to approximate
the ambient irradiance at the top of the SL (explained
below).
Underwater irradiance was measured using a
RAMSES ACC hyper-spectral radiometer (Trios-opti-
cal sensors, Oldenburg, Germany) around noon on
26 January, 22 February, 11 April, 16 June, and 16
August 2011. These measurements were taken at 1,
5, and 10 m and then every 10 m down to ~90 m
depth. For each of these depths, we calculated the
attenuation coefficient (Kz)for downwelling irradi-
ance (PAR) between the surface (i.e. 1 m depth) and
the depth (z) in question according to the following
expression:
Kz= −ln(E1/Ez)/(z − 1) (1)
where E1and Ezare the measured downwelling irra-
diance at 1 and zm respectively. Simultaneous regis-
tration of surface irradiance was obtained for each of
the underwater measurements so that the Kzesti-
mates were based on simultaneous measurements
of E1and Ez. As noted above, for the interrupted
ascents, we approximated the ambient irradiance
(ETSL), on the order of magnitude, for the depth (ZTSL)
where the upward migration of the SL halted accord-
ing to Eq. (2):
ETSL =E0exp[−KZTSL] (2)
where E0is the measured surface irradiance at the
time of the interrupted ascent, and Kis the estimated
attenuation coefficient at the date closest to the inter-
rupted ascent and for the depth closest to ZTSL.
In accordance with the anti-predation window
hypothesis (and previous studies on pearlside), we
hypothesize that M. muelleri will exploit dim light to
forage while concurrently avoiding visually search-
ing piscivores (Clark & Levy 1988, Rosland & Giske
1994, Staby et al. 2013). The extension of the anti-
predation window (by migration) is according to this
hypothesis important for their foraging and survival
success. M. muelleri (particularly juveniles) appear
to mainly forage in upper waters during dusk and
dawn (Staby et al. 2011), and we used the time dur-
ing which surface light values were between 10−3
and 1 µmol m−2 s−1 as a rough proxy to assess how the
duration of potential foraging in near-surface waters
would vary throughout the year. These values were
selected based on the ranges measured in the course
of this investigation and also encompassing the val-
ues given for the top of the Maurolicus SL at any time
of day and year by Staby & Aksnes (2011).
RESULTS
Nocturnal light
Surface light could be detected even during the
darkest part of the night from 28 April to the end of
the registration period (15 August 2011). The mini-
mum measured nocturnal irradiance spanned 2
orders of magnitude, from 0.00016 µmol m−2 s −1 on
28 April to 0.061 µmol m−2 s−1 on 22 June. Thereafter,
it decreased to the end of the records, yet with some
variation between nights (Fig. 2).
Surface light was always above the registration
threshold when the upper SL of Maurolicus muelleri
appeared close to the surface at dusk and dawn
(Fig. 3). Due to the shallow distribution, these light
192

Prihartato et al.: Nocturnal distribution and behavior of Maurolicus muelleri
levels approximately represent light at the top of the
near-surface scattering layer (SL), although the
migrations sometimes stopped at a few meters depth,
apparently hampered by a pycnocline (acoustically
visible and likely related to low-salinity surface
layer). Light levels proved to differ significantly
between dusk and dawn (Mann-Whitney Utest, p <<
0.001), regardless of season. On average, the surface
light at initiation of descent in the morning was rela-
tively stable at 0.032 µmol m−2 s−1. This was an order
of magnitude lower than the average light level
when fish reached the surface layers in the afternoon
(0.21 µmol m−2 s−1). However, as will be outlined
below, on a number of nights, fish interrupted their
ascents in the afternoon and did then not experience
these high light intensities (Fig. 3B).
Behavioral pattern of M. muelleri scattering layers
In autumn and winter, at least 2 SLs ascribed to M.
muelleri were recorded, referred to here as the deep
scattering layer (DSL) and shallow scattering layer
(SSL), often structured into several sub-layers, which
to some extent are masked by the monthly average
presented in Fig. 4. The DSL was situated at 150 to
200 m day and night, with some but only a limited
range of vertical migration subsequent to October
and until March/April. Thereafter, all prevailing
acoustic scatter in the upper ~200 m performed
coherent DVM until records ended in August.
The younger M. muelleri, which formed the SSL,
stayed at depths of ~75 to 125 m during daytime but
were located progressively deeper through the regis-
tration period, so that daytime depth reached 150 m
in spring (Fig. 4). These M. muelleri always carried
out DVM, ascending to the surface at dusk followed
by ‘midnight sinking’ with a subsequent ascent at
dawn before descending to their daytime depths.
The nocturnal distribution of the SSL subsequent to
the descent from near-surface waters at dusk deep-
ened throughout autumn to winter, from the upper
~30 to 50 m early in the registration period toward 70
to 100 m at the end of winter and in early spring,
thereafter reversing to a shallower distribution
(Figs. 4 & 5).
193
Fig. 2. Minimum nocturnal surface
irradiance (PAR) as measured with a
LiCor quantum meter and fish behav-
ior derived from echograms. There
are no light measurements prior to 28
April due to insufficient sensitivity of
the PAR sensor. Dots are measured
minimum light values, with the line
representing the moving average (n =
23). Color and symbol depict fish
behavior on the respective nights
Fig. 3. Downwelling irradiance at the
surface when the upper SL of Mauroli-
cus muelleri is located closest to the sur-
face. (A) Prior to descent in the morn-
ing; (B) after ascent in the afternoon.
Blue points represent the calculated
underwater light intensity at the top of
the SL during interrupted ascent

Mar Ecol Prog Ser 521: 189–200, 2015
194
Fig. 4. Monthly averaged echograms at 120 kHz, displaying 24 h images for the upper ~280 m. Averages are based on the fol-
lowing numbers of days: Oct (26 d), Nov (30 d), Dec (21 d), Jan (29 d), Feb (19 d), Mar (30 d), Apr (30 d), May (29 d), Jun (30 d),
Jul (17 d), and Aug (15 d). Color scale refers to backscattering strength (SV) values (dB). Time is given in UTC (local stan-
dard time − 1 h)
Fig. 5. Echograms depicting the nocturnal backscatter for the whole study period at (A) 120 kHz and (B) 200 kHz. Gaps in the
datasets are from periods without records. Color scale refers to backscattering strength (SV) values (dB)

Prihartato et al.: Nocturnal distribution and behavior of Maurolicus muelleri
From late autumn to spring, there was an asyn-
chrony between the dusk and dawn ascent. In the
afternoon, individuals forming the SSL repeatedly
arrested their ascent at various depths (mean = 24.6 ±
17.4 m) before reaching upper layers (e.g. Fig. 6A).
Such interrupted ascents were recorded on 46 of the
158 dates from 27 November 2010 to 18 April 2011,
particularly frequently in March. The estimated
underwater irradiance at the top of the SL during
interrupted ascent ranged between 10−3 and 10−1
µmol m−2 s−1 and was about 1 to 2 orders of magni-
tude lower than during the afternoons they contin-
ued to the surface (Fig. 3B). Interrupted ascents were
never recorded in the morning, so that dawn rise to
surface waters was documented on every date dur-
ing the same period.
A change in the nocturnal distribution took place
from mid-May. The midnight sinking behavior then
tended to cease, with a subsequently increased por-
tion of the nocturnal SL inhabiting near-surface
waters (upper ~25 m) throughout the night (Fig. 6B).
The minimum surface irradiance at night at the time
of initiation of this behavior was ~10−3 µmol m−2 s−1
(Figs. 2 & 7A,B). The structure of the nocturnal SLs
subsequently changed in early June. On the first
dates of the month, fish were schooling upon reach-
ing the surface early at night, while schooling per-
sisted throughout the short nights in mid-summer
(Figs. 2, 6C & 7C,D). During this time, minimum noc-
turnal surface irradiance was mostly between ~5 ×
10−3 and 10−1 µmol m−2 s−1, while the fishes reached
upper waters at levels of ~1 µmol m−2 s−1 (Fig. 7C,D).
Schooling subsided in early July. Fish thereafter con-
tinued to stay in upper layers throughout night until
mid-August (Fig. 6D), when midnight sinking re -
appeared as nocturnal surface irradiance apparently
(irregular light-data at this time) again dropped
toward 10−3 µmol m−2 s−1 (Figs. 2 & 7E).
The duration of periods with nocturnal surface
light between 10−3 and 1 µmol m−2 s−1 (i.e. a coarse
proxy for the nocturnal extension of the anti-predator
window in upper layers) varied through the year
195
Fig. 6. Selected echograms for 3 consecutive days representing different diel migration patterns. (A) Interrupted ascent in win-
ter (6−8 January 2011), (B) termination of midnight sinking in late spring (14−16 May), (C) schooling in mid-summer (18−20
June), and (D) resumption of midnight sinking in late summer (14−16 August). Color scale refers to backscattering strength
(SV) values (dB)

Mar Ecol Prog Ser 521: 189–200, 2015
(Fig. 8) and roughly comprised the time fish resided
in upper waters. It was relatively short (~2 h) during
winter and early spring, as confined to dusk and
dawn periods. The time-window for this light range
increased in mid-May, being 5 to 6 h in late May and
early June. The period thereafter became shorter as
night became shorter toward the end of June (~4 h),
before subsequently increasing with increasing
length of the night until the end of the registration
period in mid-August, when the duration would drop
due to resumption of midnight sinking.
DISCUSSION
We have unveiled seasonal variations in DVM-
behavior of scattering layers (SLs) ascribed to Mau-
rolicus muelleri (sampling from this study and pre -
vious sampling; see references in ‘Material and
methods’) by exploring long-term, high-resolution
acoustic data. Emphasis has been on the nocturnal
vertical distribution in a high-latitude ecosystem
where nocturnal light conditions vary strongly
throughout the year. Individuals in the deeper SL
196
Fig. 7. Echograms of the upper 40 m showing
nocturnal SL (200 kHz) with corresponding sur-
face light intensities depicting (A,B) termination
of midnight sinking (15−16 May), (C,D) school-
ing behavior (18−19 June) and (E) resumption of
midnight sinking (15 August; E). Color scale
refers to backscattering strength (SV) values (dB)
Fig. 8. Nocturnal extension of the
duration of the anti-predation win-
dow, with grey circles depicting
observations and the line indicating
the moving average (n = 23). The
antipredation window relates to
intermediate light levels sufficient for
obtaining food, while at the same
time being sufficiently low to offer
relative protection from visually
searching piscivores. We selected the
periods with surface light values
being between 10−3 and 1 µmol m−2
s−1 based on the light ranges meas-
ured between Maurolicus muelleri
reaching the surface in the afternoon
and initiation of midnight sinking

Prihartato et al.: Nocturnal distribution and behavior of Maurolicus muelleri
(adult part of the population) did not appear to
migrate extensively during fall and winter, yet the
mid-water vertical distribution was somewhat shal-
lower at night. Corresponding patterns emerge from
previous studies (Giske et al. 1990, Staby & Aksnes
2011). Different behavior in juveniles and adults has
been interpreted as different trade-off between for-
aging and predator avoidance in the 2 groups
(Rosland & Giske 1994), as also found for other
pelagic taxa (De Robertis 2002, Pearre 2003). Adults
seem to have no or negative growth during winter
(Rosland & Giske 1997) yet may forage in their day-
time depth on plankton, such as overwintering cope-
pods (Bagøien et al. 2001). By largely remaining in
deep water, adult pearlside strongly increase their
probability of survival to the spawning season in
spring (Rosland & Giske 1997).
Individuals of the shallowest SL always carried out
DVM, as was also the case for the adults during
spring and summer, again in accordance with previ-
ous studies (Staby et al. 2011). However, the migra-
tion pattern varied and comprised migrations with
and without midnight sinking, interrupted ascents in
the evening and shifts from occurrence in diffuse
nocturnal SL through most of the year to nocturnal
near-surface schooling behaviors during the lightest
part of the year.
Behavioral responses to nocturnal lights
It is well established that the vertical distribution of
pearlside is related to light intensity in daytime and
during diel vertical migrations (Baliño & Aksnes
1993, Rasmussen & Giske 1994). We here unveil how
changes in nocturnal light levels also affect the verti-
cal distribution and behavior of pearlside. We largely
refer to surface light, and it must also be noted that
the underwater light intensity we have estimated
involves interpolation of the light attenuation coeffi-
cient over relatively large time periods. This interpo-
lation has obviously reduced the accuracy of our esti-
mates of underwater light but nevertheless provides
useful information concerning the order of magni-
tude of the ambient light intensities experienced by
the fishes.
For the whole period of measurements, estimated
light intensity at the top of the SL typically spanned 4
orders of magnitude (10−4 to 1.6 µmol m−2 s−1; the lat-
ter being surface value when the fish reached their
shallowest distribution). The additional light extinc-
tion from top to bottom of the SL should also be con-
sidered. Therefore, although the vertical positioning
of M. muelleri is tightly related to light, this is not a
fixed light value. In line with Staby & Aksnes (2011),
our results suggest that pearlside follow preferred
ranges of light intensity rather than a constant fixed
isolume and that these ranges span several orders of
magnitude. This finding concurs with conclusions
from studies of mesopelagic fish in other systems
(Roe 1983, Benoit-Bird et al. 2009).
At shorter time scales, the vertical distribution
appears to be coupled to a narrower range of light
levels (Baliño & Aksnes 1993, Staby & Aksnes 2011),
although there were notable variations from day to
day. This variation particularly related to afternoons
with interrupted ascents, but other aspects of the
behavior also did not fully match the variation in light
levels (e.g. Fig. 2). Some of this variation might
reflect methodological constraints, as interpretations
from echograms were subjective, but evidently other
factors than light will also affect behavior. Finally,
our calculation of underwater light intensity has not
accounted for variations in Kcaused by variations in
the radiant field, such as the change in the angle of
incoming sunlight during the day, cloudiness, and
wave action.
Within days, the surface light intensity was on
average ~1 order of magnitude weaker when the SL
reached the surface during the morning ascent than
during the corresponding ascent in the afternoon.
This pattern corroborates findings by Staby & Aksnes
(2011). The weaker light when fish reached the sur-
face in the morning would be in line with fish being
motivated for early feeding after a long night without
foraging. In contrast, this behavior is contrary to the
assumption that hungry fish would be more prone to
undertake risky behavior (Dill 1983) and therefore
extend the duration of their dawn ascent in suppos-
edly food-rich waters into higher light intensities.
Alternatively, this behavior might have a pure physi-
ological cause if dark-adapted fish are more efficient
in detecting prey at low light levels in the morning
than in the evening (this hypothesis might relate both
to the pearlside and their predators). The dark-to-
light adaptation in the retina is indeed much more
rapid than the light-to-dark adaptation (Ferwerda et
al. 1996). This hypothesis has been suggested as one
possible explanation for the emergence of small
planktivorous coral fishes from their nocturnal shel-
ter in the morning at lower light levels than the levels
at which they return to their shelter in the evening
(Rickel & Genin 2005). Note however, that the trade-
off between feeding and predator avoidance is very
different in that setting. Small planktivorous fish on
coral reefs are safest in bright light since they spot
197

Mar Ecol Prog Ser 521: 189–200, 2015
their predators by sight and can retract to nearby
shelters when threatened as well as when light
decreases (Holbrook & Schmitt 2002). Mesopelagic
fish, in contrast, rely on hiding in dim light.
Interrupted ascent
Interrupted ascent behavior was frequently
recorded from late autumn to mid-March yet at dif-
ferent depths. In spring, the estimated underwater
irradiance at the top of the SL during interrupted
ascent was about 1 to 2 orders of magnitude lower
than during the afternoons that the fish continued to
the surface (Fig. 3). This pattern might suggest some
relation to predator avoidance behavior. Previous
studies have shown instantaneous diving responses
among mesopelagic fish to the presence of predators
(Kaartvedt et al. 2012), and such responses were also
recorded for M. muelleri during this study (e.g. Godø
et al. 2014). However, there were no systematic
records of excessive numbers of predators associated
with such events, and the interrupted ascents were
recorded coherently by all echo sounders, located
several hundred meters away from each other. Alter-
natively, as the copepod Calanus ascends from over-
wintering in winter and early spring, it is possible
that satiation following feeding during the ascent
reduces motivation for further migration during this
time of the year (Staby et al. 2011). However, such
reaction to satiation would expectedly be an individ-
ual response (Pearre 2003) and not the population
response indicated in the echograms. Migrations
might also be stopped by gradients in temperature
and salinity, yet such gradients are not expected at
depth and would not explain the consistent differ-
ence between dawn and dusk ascents. Overall, the
most likely reason for the interrupted ascents in the
afternoon seems to have some relation to perceived
risk of predation, but this behavior is still poorly
understood.
Midnight sinking
M. muelleri does not forage in darkness, and mid-
night sinking apparently takes place when the condi-
tions are too dark for visual detection of prey (Giske
et al. 1990). Accordingly, midnight sinking was initi-
ated as nights became darker in late summer (Fig.
7E). Resumption of midnight sinking in August is also
documented by Staby et al. (2011). In parallel to the
darker nights in August, there may have been
changes in the distribution of potential prey, as the
copepod Calanus finmarchicus tends to start des cen -
ding for overwintering during this time of the year,
with potential effects on the diel migration pattern of
mesopelagic fish (Kaartvedt et al. 2009, Dypvik et al.
2012b).
The nocturnal distribution became deeper
through fall and winter (Fig. 5). Giske & Aksnes
(1992) suggested that pearlside was seeking warm
temperature at night for more rapid digestion of the
afternoon meal, in this way maximizing growth (cf.
Wurtsbaugh & Neverman 1988). We do not have
temperature profiles through winter to correlate
with the observed distributions, but the subsurface
temperature maximum will become progressively
deeper due to cooling from above (Bagøien et al.
2001). This pattern would be in accordance with
the observations. Results from Staby et al. (2011)
did not fully support the importance of temperature
profiles for pearlside that perform midnight sinking
behavior, and these authors referred to predator
avoidance from visually searching piscivores as an
alternative explanation. We could not measure sur-
face light during the dark winter nights due to lack
of instrument sensitivity, but light intensities at the
upper SL during midnight sinking would have
been 3 to 5 orders of magnitude less than surface
light (based on measurements of extinction), i.e.
<10−8 µmol m−2 s−1. In early spring, when nocturnal
surface light could be detected, midnight sinking
occurred when this light decreased to ~10−3 µmol
m−2 s−1, then translating into <10−6 µmol m−2 s−1 at
the depth of midnight sinking. Gadoids appear to
be the most important predators on M. muelleri in
Masfjorden (Giske et al. 1990, Staby & Aksnes
2011). Ryer & Olla (1999) showed that juveniles of
the gadoid walleye pollock could forage success-
fully on Artemia at very low light intensities (5 ×
10−7 µmol m−2 s−1). This result suggests that mid-
night sinking of M. muelleri actually may be bene-
ficial to avoid nocturnal predators.
Termination of midnight sinking
From mid-May to mid-August, M. muelleri re -
mained in upper layers throughout the nights, with
limited evidence of midnight sinking. This time rep-
resents a period of the year when the concentration
of zooplankton peaks in upper waters of Masfjorden
(Aksnes et al. 1989, Rasmussen & Giske 1994). Sea-
sonally fluctuating vertical distribution and abun-
dance of zooplankton appear to affect the migration
198

Prihartato et al.: Nocturnal distribution and behavior of Maurolicus muelleri
patterns of the mesopelagic fish in this system
(Staby et al. 2011, Dypvik et al. 2012b), yet as visual
predators, M. muelleri need sufficient light to see
their prey. The termination of midnight sinking
occurred when the minimum nocturnal surface light
exceeded 10−3 µmol m−2 s−1. We interpret the shift in
behavior as a sign that the pearlside now had suffi-
cient light for foraging throughout the night (Ras-
mussen & Giske 1994, Kaartvedt et al. 1998). The
light summer nights therefore represented a marked
increase in the time available for foraging in upper
waters (Fig. 8). This result suggests that summer
may be a particularly important period for growth,
not only due to higher plankton concentrations and
warmer surface waters at this time but also due to
the seasonal light cycle. The importance of the sea-
sonal light cycle at high latitudes has been reported
for fish in other settings. Suthers & Sundby (1996)
found enhanced growth rates of cod larvae at high
latitudes in summer, which they ascribed to the long
period for visual foraging at times with midnight
sun.
Schooling in light summer nights
As nights became even lighter toward mid-sum-
mer, M. muelleri took on schooling in upper layers.
This pattern occurred from early June until ceasing
the first half of July concordant with minimum noc-
turnal surface light above 5 ×10−3 µmol m−2 s−1 (Figs.
2, 6 & 7). The pearlside often schooled at 5 to 20 m
depth, with light intensities at these depths being 1 to
2 orders of magnitude lower than at the surface.
Schooling is well acknowledged as an anti-predator
strategy (Magurran 1990), and mesopelagic fish
taking up schooling behavior have previously been
reported for fish being chased by tuna during day-
light hours (Alverson 1961, Marchal & Lebourges
1996).
The change in nocturnal anti-predator behavior
shows behavioral flexibility, which may permit
extension of the oceanic habitat available for M.
muelleri. However, there are likely limitations for the
seasonal variation in light conditions that the meso-
pelagic fish can handle because they appear to be
scarce in the polar regions. Kaartvedt (2008) sug-
gested that this could relate to the extreme light
climate at high latitudes; very light summer nights
with midnight sun will prevent the fish from seeking
upper layers in cover of darkness, while total dark-
ness in winter hampers their feeding during that
season.
Acknowledgements. We thank Thor A. Klevjer and Anders
Røstad for invaluable help during the acoustic studies. This
study was funded by King Abdullah University of Science
and Technology.
LITERATURE CITED
Aksnes DL, Aure J, Kaartvedt S, Magnesen T, Richard J
(1989) Significance of advection for the carrying capacity
of fjord populations. Mar Ecol Prog Ser 50: 263−274
Aksnes DL, Nejstgaard J, Sødberg E, Sørnes T (2004) Opti-
cal control of fish and zooplankton populations. Limnol
Oceanogr 49: 233−238
Alverson FG (1961) Daylight surface occurrence of myc-
tophid fishes off the coast of Central America. Pac Sci 15:
483
Bagøien E, Kaartvedt S, Aksnes DL, Eiane K (2001) Vertical
distribution and mortality of overwintering Calanus.
Limnol Oceanogr 46: 1494−1510
Baliño BM, Aksnes DL (1993) Winter distribution and migra-
tion of the sound scattering layers, zooplankton and
micronekton in Masfjorden, western Norway. Mar Ecol
Prog Ser 102: 35−50
Benoit-Bird KJ, Au WW, Wisdom DW (2009) Nocturnal light
and lunar cycle effects on diel migration of micronekton.
Limnol Oceanogr 54: 1789−1800
Bianchi D, Galbraith ED, Carozza DA, Mislan K, Stock CA
(2013) Intensification of open-ocean oxygen depletion by
vertically migrating animals. Nat Geosci 6: 545−548
Clark CW, Levy DA (1988) Diel vertical migrations by juve-
nile sockeye salmon and the antipredation window. Am
Nat 131: 271−290
Cohen JH, Forward RB Jr (2009) Zooplankton diel vertical
migration: a review of proximate control. Oceanogr Mar
Biol Annu Rev 47: 77−110
Dalpadado P, Gjøsæter J (1988) Feeding ecology of the
lanternfish Benthosema pterotum from the Indian
Ocean. Mar Biol 99: 555−567
De Robertis A (2002) Size-dependent visual predation risk
and the timing of vertical migration: an optimization
model. Limnol Oceanogr 47: 925−933
Dill LM (1983) Adaptive flexibility in the foraging behavior
of fishes. Can J Fish Aquat Sci 40: 398−408
Doksæter L, Olsen E, Nøttestad L, Fernø A (2008) Distribu-
tion and feeding ecology of dolphins along the Mid-
Atlantic Ridge between Iceland and the Azores. Deep-
Sea Res II 55: 243−253
Dypvik E, Klevjer TA, Kaartvedt S (2012a) Inverse vertical
migration and feeding in glacier lanternfish (Bentho -
sema glaciale). Mar Biol 159: 443−453
Dypvik E, Røstad A, Kaartvedt S (2012b) Seasonal variations
in vertical migration of glacier lanternfish, Benthosema
glaciale. Mar Biol 159: 1673−1683
Ferwerda JA, Pattanaik SN, Shirley P, Greenberg DP (1996)
A model of visual adaptation for realistic image synthe-
sis. In: Wesley A, Rushmeier H (eds) SIGGRAPH 97 Conf
Proc. Ann Conf Ser, ACM SIGGRAPH, p 369−378
Foote KG, Knudsen HP, Vestnes G, MacLennan DN, Sim-
monds EJ (1987) Calibration of acoustic instruments for
fish density estimation: a practical guide. ICES Coop Res
Rep 144: 1−69
Giske J, Aksnes DL (1992) Ontogeny, season and trade-offs:
vertical distribution of the mesopelagic fish Maurolicus
muelleri. Sarsia 77: 253−261
199
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤

Mar Ecol Prog Ser 521: 189–200, 2015
Giske J, Aksnes DL, Baliño BM, Kaartvedt S and others
(1990) Vertical distribution and trophic interactions of
zooplankton and fish in Masfjorden, Norway. Sarsia 75:
65−81
Gjøsæter J, Kawaguchi K (1980) A review of the world
resources of mesopelagic fish. FAO Fish Tech Pap 193:
1−151
Gliwicz ZM (1986) A lunar cycle in zooplankton. Ecology 67:
883−897
Godø OR, Handegard NO, Browman HI, Macaulay GJ and
others (2014) Marine ecosystem acoustics (MEA): quanti-
fying processes in the sea at the spatio-temporal scales
on which they occur. ICES J Mar Sci 71: 2357−2369
Hernández-León S, Franchy G, Moyano M, Menendez I,
Schmoker C, Putzeys S (2010) Carbon sequestration and
zooplankton lunar cycles: Could we be missing a major
component of the biological pump? Limnol Oceanogr 55:
2503−2512
Holbrook SJ, Schmitt RJ (2002) Competition for shelter
space causes density-dependent predation mortality in
damselfishes. Ecology 83: 2855−2868
Irigoien X, Klevjer T, Røstad A, Martinez U and others (2014)
Large mesopelagic fishes biomass and trophic efficiency
in the open ocean. Nat Commun 5: 3271
Kaartvedt S (2008) Photoperiod may constrain the effect of
global warming in arctic marine systems. J Plankton Res
30: 1203−1206
Kaartvedt S, Melle W, Knutsen T, Skjoldal HR (1996) Verti-
cal distribution of fish and krill beneath water of varying
optical properties. Mar Ecol Prog Ser 136: 51−58
Kaartvedt S, Knutsen T, Holst JC (1998) Schooling of the
vertically migrating mesopelagic fish Maurolicus muel-
leri in light summer nights. Mar Ecol Prog Ser 170:
287−290
Kaartvedt S, Torgersen T, Klevjer TA, Røstad A, Devine JA
(2008) Behavior of individual mesopelagic fish in
acoustic scattering layers of Norwegian fjords. Mar Ecol
Prog Ser 360: 201−209
Kaartvedt S, Røstad A, Klevjer TA, Staby A (2009) Use of
bottom-mounted echosounders in exploring behavior of
mesopelagic fishes. Mar Ecol Prog Ser 395: 109−118
Kaartvedt S, Staby A, Aksnes DL (2012) Efficient trawl
avoidance by mesopelagic fishes causes large underesti-
mation of their biomass. Mar Ecol Prog Ser 456: 1−6
Kampa EM (1970) Underwater daylight and moonlight
measurements in the eastern North Atlantic. J Mar Biol
Assoc UK 50: 397−420
Kampa EM, Boden BP (1954) Submarine illumination and
the twilight movements of a sonic scattering layer.
Nature 174: 869−871
Klevjer TA, Torres DJ, Kaartvedt S (2012) Distribution and
diel vertical movements of mesopelagic scattering layers
in the Red Sea. Mar Biol 159: 1833−1841
Knudsen HP (1990) The Bergen echo integrator: an intro-
duction. ICES J Mar Sci 47: 167−174
Korneliussen RJ, Heggelund Y, Eliassen IK, Johansen GO
(2009) Acoustic species identification of schooling fish.
ICES J Mar Sci 66: 1111−1118
Magurran AE (1990) The inheritance and development of
minnow anti-predator behaviour. Anim Behav 39:
834−842
Marchal E, Lebourges A (1996) Acoustic evidence for
unusual diel behaviour of a mesopelagic fish (Vinciguer-
ria nimbaria) exploited by tuna. ICES J Mar Sci 53:
443−447
O’Driscoll RL, Gauthier S, Devine JA (2009) Acoustic esti-
mates of mesopelagic fish: as clear as day and night?
ICES J Mar Sci 66: 1310−1317
Pearre S (2003) Eat and run? The hunger/satiation hypothe-
sis in vertical migration: history, evidence and conse-
quences. Biol Rev Camb Philos Soc 78: 1−79
Potier M, Marsac F, Cherel Y, Lucas V, Sabatié R, Maury O,
Ménard F (2007) Forage fauna in the diet of three large
pelagic fishes (lancetfish, swordfish and yellowfin tuna)
in the western equatorial Indian Ocean. Fish Res 83:
60−72
Rasmussen OI, Giske J (1994) Life-history parameters and
vertical distribution of Maurolicus muelleri in Masfjor-
den in summer. Mar Biol 120: 649−664
Rickel S, Genin A (2005) Twilight transitions in coral reef
fish: the input of light-induced changes in foraging
behaviour. Anim Behav 70: 133−144
Ringelberg J (1995) Changes in light-intensity and diel ver-
tical migration: a comparison of marine and fresh-water
environments. J Mar Biol Assoc UK 75: 15−25
Roe HSJ (1983) Vertical distribution of euphausiids and fish
in relation to light intensity in the Northeastern Atlantic.
Mar Biol 77: 287−298
Rosland R, Giske J (1994) A dynamic optimization model of
the diel vertical distribution of a pelagic planktivorous
fish. Prog Oceanogr 34: 1−43
Rosland R, Giske J (1997) A dynamic model for the life his-
tory of Maurolicus muelleri, a pelagic planktivorous fish.
Fish Oceanogr 6: 19−34
Ryer CH, Olla BL (1999) Light-induced changes in the prey
consumption and behavior of two juvenile planktivorous
fish. Mar Ecol Prog Ser 181: 41−51
Sameoto D (1989) Feeding ecology of the lantern fish Ben-
thosema glaciale in a subarctic region. Polar Biol 9:
169−178
Shreeve RS, Collins MA, Tarling GA, Main CE, Ward P,
Johnston NM (2009) Feeding ecology of myctophid
fishes in the northern Scotia Sea. Mar Ecol Prog Ser 386:
221−236
Staby A, Aksnes DL (2011) Follow the light — diurnal and
seasonal variations in vertical distribution of the meso-
pelagic fish Maurolicus muelleri. Mar Ecol Prog Ser 422:
265−273
Staby A, Røstad A, Kaartvedt S (2011) Long-term acoustical
observations of the mesopelagic fish Maurolicus muelleri
reveal novel and varied vertical migration patterns. Mar
Ecol Prog Ser 441: 241−255
Staby A, Srisomwong J, Rosland R (2013) Variation in DVM
behaviour of juvenile and adult pearlside (Maurolicus
muelleri) linked to feeding strategies and related preda-
tion risk. Fish Oceanogr 22: 90−101
Suthers IM, Sundby S (1996) Role of the midnight sun: com-
parative growth of pelagic juvenile cod (Gadus morhua)
from the Arcto-Norwegian and a Nova Scotian stock.
ICES J Mar Sci 53: 827−836
Sweatt AJ, Forward RB (1985) Diel vertical migration and
photoresponses of the chaetognath Sagitta hispida
Conant. Biol Bull (Woods Hole) 168: 18−31
Wurtsbaugh WA, Neverman D (1988) Post-feeding thermo-
taxis and daily vertical migration in a larval fish. Nature
333: 846−848
200
Editorial responsibility: Alejandro Gallego,
Aberdeen, UK
Submitted: May 12, 2014; Accepted: November 28, 2014
Proofs received from author(s): January 23, 2015
➤
➤
➤
➤
➤
➤➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤