Prevalence, structure and properties of subsurface chlorophyll maxima in Canadian Arctic waters

16 Pages·2010·3.25 MB·English

Vol. 412: 69–84, 2010 MARINE ECOLOGY PROGRESS SERIES
Published August 18
doi: 10.3354/meps08666 Mar Ecol Prog Ser
OPPEENN
ACCCCEESSSS
Prevalence, structure and properties of subsurface
chlorophyll maxima in Canadian Arctic waters
Johannie Martin1,*, Jean-Éric Tremblay1, Jonathan Gagnon1, Geneviève Tremblay2,
Amandine Lapoussière2, Caroline Jose2, Michel Poulin3, Michel Gosselin2,
Yves Gratton4, Christine Michel5
1Québec-Océan & Département de biologie, Université Laval, Québec, Québec G1V 0A6, Canada
2Institut des sciences de la mer (ISMER), Université du Québec à Rimouski, Rimouski, Québec G5L 3A1, Canada
3Canadian Museum of Nature, Ottawa, Ontario K1P 6P4, Canada
4Institut National de Recherche Scientifique, Centre Eau, Terre et Environnement, Québec, Québec G1K 9A9, Canada
5Freshwater Institute, Fisheries and Oceans Canada, Winnipeg, Manitoba R3T 2N6, Canada
ABSTRACT: Comprehensive investigations of the Canadian Arctic during late summer and early fall
revealed the widespread occurrence of long-lived subsurface chlorophyll maxima (SCM) in season-
ally ice-free waters. The vertical position of the SCM corresponded with the depth of the subsurface
biomass maximum (SBM), at least in Baffin Bay, suggesting that SCM could be an important source
of carbon for the food web. Most of these SCM were located well below the pycnocline in close
association with the nitracline, implying that their vertical position was driven mainly by a shortage
of inorganic nitrogen in the upper euphotic zone. The diversity of SCM configurations with respect to
physical properties of the water column complicates the estimation of euphotic-zone chlorophyll and
primary production from surface properties. High photosynthetic yields (F /F ) showed the phyto-
v m
plankton to be photosynthetically competent and well acclimated to conditions of irradiance and
nutrient supply near the surface and at the SCM. A well-defined primary nitrite maximum was
associated with the SCM in the southwest Canadian Arctic, but not in the northeast where nitrite
concentrations were highest much below the euphotic zone. This contrast is consistent with
differences in vertical stratification, the light–dark cycle and, possibly, the physiological state and
taxonomic composition of the phytoplankton community at the SCM. This study demonstrates that
the SCM, once regarded as anecdotal due to under-sampling, are a dominant feature of the Arctic
Ocean that should be considered in remote sensing studies and biogeochemical models.
KEY WORDS: Subsurface chlorophyll maximum· Deep chlorophyll maximum· Subsurface biomass
maximum· Phytoplankton· Arctic· Nutrients· Nitrogen· Photosynthetic yield
Resale or republication not permitted without written consent of the publisher
INTRODUCTION with the lowest requirements for light (i.e. compensa-
tion irradiance; Huisman et al. 2006). The SCM have
The spatial distribution of marine phytoplankton is been found to harbor up to an order of magnitude the
highly heterogeneous. On the vertical, subsurface concentration of chlorophyll a(chl a) found at the sur-
maxima of chlorophyll (SCM) or biomass (SBM) are face (Steele 1964, Anderson 1969, Klausmeier & Litch-
common under stratified conditions (Cullen 1982, man 2001, Sharples et al. 2001) and to be dominated
Coon et al. 1987). Their vertical position is regarded as numerically by few algal species (Coon et al. 1987,
a compromise between nutrient limitation near the Huszar et al. 2003). Their existence poses a challenge
surface and light limitation at depth: a situation that to the remote-sensing estimation of primary produc-
should favor shade-adapted phytoplankton species tion (e.g. Uitz et al. 2006).
*Email: [email protected] © Inter-Research 2010 · www.int-res.com
70 Mar Ecol Prog Ser 412: 69–84, 2010
Phytoplankton biomass at a given depth is the net spring, which rapidly induces nitrogen limitation
result of local production, death by lysis or grazing and above the nutricline (e.g. Franklin Bay, Tremblay et al.
the gains and losses imparted by passive or active 2008). The SCM form within days of the ice break-up
motion of the cells (e.g. Dolan & Marrasé 1995, Klaus- and hypothetically persist throughout summer, medi-
meier & Litchman 2001, Fennel & Boss 2003, Hodges & ating a large portion of the annual nitrate drawdown.
Rudnick 2004, Holm-Hansen & Hewes 2004, Huisman Despite their implications for food webs, biogeochemi-
et al. 2006, Beckmann & Hense 2007, Sharples et al. cal fluxes and the accuracy of remote sensing esti-
2007). Vertical decoupling of the SBM and SCM is not mates of primary production, SCM have only been
rare and has been ascribed to photoacclimation, briefly mentioned in observational studies of subarctic
whereby the ratio of chl a to carbon increases with and Arctic primary production (e.g. Martini 1986,
depth to optimize light harvesting (Steele 1964, Kiefer Hirche et al. 1991, Cota et al. 1996, Heiskanen & Keck
et al. 1976, Cullen 1982, Falkowski & Kiefer 1985, Fen- 1996, Lee & Whitledge 2005), and their overall func-
nel & Boss 2003). Most observational studies investi- tion, structure and significance have not been
gate only the SCM with high-resolution profiling of assessed.
chlorophyll fluorescence and ignore the SBM since the Here we report on the large-scale incidence and
visual estimation of biomass from microscopic enumer- properties of SCM in the Canadian High Arctic, the
ation and sizing of phytoplankton is tedious, and dis- subarctic Hudson Bay and a few Labrador fjords. Sur-
crete sampling may miss the SBM altogether. Trans- veys were conducted during late summer and early fall
missometers can be used to pinpoint the SBM in clear to ensure that all regions had lost their seasonal ice
waters where particulate beam attenuation (c ) in the cover and the pelagic growth season was underway.
p
red part of the light spectrum is strongly influenced by Our working hypothesis is that SCM are photosynthet-
microbial organisms (Chung et al. 1998), but the proce- ically competent and associated with favorable condi-
dure is unreliable under the influence of river dis- tions of allochthonous nitrate supply, not merely a
charge or sediment resuspension in coastal waters. change in chl acontent per unit of carbon. This hypoth-
SCM are persistent in perennially stratified tropical esis is validated here by a near-synoptic comparison of
and subtropical waters (Huisman et al. 2006, Mann & intrinsic SCM characteristics (i.e. vertical position, chl a
Lazier 2006) and seasonal at high latitudes in the concentration, taxonomic composition and photosyn-
Southern Ocean (Holm-Hansen & Hewes 2004) and thetic competency) with the physical structure of the
the boreal North Atlantic, where the extensive mixing water column and the vertical distribution of oxygen
of the water column during fall and winter replenishes and macronutrients in contrasted environments. A sec-
the surface with nutrients (Mann & Lazier 2006). This ond objective was to assess if the inventory of chl ain
phenomenon permits the development of a spring the euphotic zone can be predicted from surface values
bloom in the upper euphotic zone and a seasonal in the context of remote sensing.
succession whereby a transient SCM community
replaces the fast-growing bloomers once nutrients are
exhausted from the surface (Mann & Lazier 2006, MATERIALS AND METHODS
Pommier et al. 2009). In the High Arctic, the seasonal
ice cover and extreme solar cycle restrict the produc- Sampling. The 2005 (16 August to 16 October) and
tive period to a few months (Sakshaug 2004). Unlike 2006 (4 September to 4 November) expeditions of the
the North Atlantic, the periodic renewal of nutrients in RV ‘CCGS Amundsen’, covered the entire latitudinal
the upper euphotic zone is often tempered by the ice and longitudinal swath of the Canadian Archipelago,
cover and the strong vertical stability imparted by sea- including Baffin Bay, the Northwest Passage, the
sonal melt and the horizontal inputs of freshwater from Beaufort Sea, Foxe Basin, Hudson Bay and 3 Labrador
large rivers and low-salinity water from the Pacific fjords (Fig. 1) at a total of 219 stations. Vertical profiles
Ocean (Jones et al. 2003, Stein & Macdonald 2004, were obtained with a CTD rosette equipped with
Tremblay et al. 2008). sensors to measure in vivo fluorescence (SeaPoint
In the Beaufort Sea, convection and winds have a Chlorophyll Fluorometer), transmissivity (WET Labs
minor disrupting effect on stratification during fall and CST671DR), dissolved oxygen (Sea-Bird SBE43), ni-
winter (Tremblay et al. 2008). Exceptions to this pat- trate (SATLANTIC ISUS V1), photosynthetically active
tern are found in productive polynyas (e.g. the North radiation (PAR; Biospherical QCP-2300), and tempera-
Water; Tremblay et al. 2002) and along the margin of ture and salinity (Sea-Bird SBE-911plus).
shallow continental shelves when upwelling-favorable At a subset of 140 stations (55 in 2005 and 85 in
winds blow under conditions of reduced ice cover 2006), water samples for nutrient determinations (ni-
(Williams & Carmack 2008). The low initial inventories trate [NO –] + nitrite [NO –]), phosphate (PO 3–) and
3 2 4
of nitrate in the surface mixed layer are readily used in silicic acid (Si(OH) ) were taken with 12 l Niskin type
4
Martin et al.: Subsurface chlorophyll maxima in the Canadian Arctic 71
Fig. 1. Location of sampling stations with the presence (blue
circles) or absence (red squares) of a subsurface fluorescence
maximum in the Canadian Arctic during 2005 (left-hand pan-
els) and 2006 (right-hand panels). Four oceanographic sec-
tions are identified within boxes: southeast Beaufort Sea (BS;
5–6 Oct 2006), Amundsen Gulf (AG; 29 Sep–18 Oct 2006),
Barrow Strait/Lancaster Sound (LS; 20–25 Sep 2006) and
northern Baffin Bay (BB; 16–22 Aug 2005). Stars represent the
starting point of the sections presented in Figs. 5 and 6
bottles attached to the CTD rosette at standard depths in the upper mixed layer (5 m), in the lower and upper
(5, 10, 20, 30, 40, 50, 60, 70, 80, 100, 125, 150, 175, 200, trails of the SCM, at the SCM, and below the euphotic
250, 300 m and then every 100 m) unless the Arctic zone. Samples for phytoplankton identification and
halocline was identified on the CTD downcast. In this enumeration were collected at the depth of the SCM
case, sampling in the 100–200 m range occurred at at 35 stations in 2005 (all regions) and at 15 stations
every 20 m and at a salinity of 33.1 to capture the nutri- in 2006 (Baffin Bay, Northwest Passage, and Beaufort
ent maximum. Out of those 140 stations, 64 (35 in 2005 Sea).
and 29 in 2006) were also sampled for ammonium Nutrients. Samples for nutrient determination were
(NH +), chl a, and photosynthetic yield. Samples for collected into acid-cleaned polyethylene tubes after
4
chlaand photosynthetic yield were taken at 5 depths: thorough rinsing and filtration through a 5 µm polycar-
72 Mar Ecol Prog Ser 412: 69–84, 2010
bonate filter inserted in a filter holder. This step Sensor calibrations and data transformations. De-
insured the removal of the large particles (e.g. clay, tailed vertical profiles of water temperature, salinity,
mud) and organisms that may interfere with the analy- transmissivity, PAR, oxygen and in vivo fluorescence
ses. Samples were stored at 4°C in the dark and ana- were analyzed for 219 stations. The CTD and
lyzed within a few hours for NO –, NO –+ NO –, PO 3– attached sensors were factory calibrated prior to each
2 3 2 4
and Si(OH) using standard colorimetric methods expedition. Analytically determined NO – concentra-
4 3
(Grasshoff et al. 1999) adapted for the AutoAnalyzer 3 tions were used to post-calibrate the optical nitrate
(Bran+Luebbe). NH + was determined manually with probe and generate high-resolution vertical profiles.
4
the sensitive fluorometric method of Holmes et al. Due to problems with the batteries of the probe and
(1999). The working reagent was added within min- with problematic calibrations at a few stations,
utes of sampling. The detection limit for NH +analysis detailed NO –profiles were only available for 147 sta-
4 3
was 0.02 µM or better. tions. The output of the oxygen sensor was frequently
Extracted chlorophyll and photosynthetic compe- calibrated against Winkler titrations (modified as in
tency. Concentrations of chl a were determined using Carpenter 1965, and automated as described in
the fluorometric method described by Parsons et al. Knapp et al. 1990) and proved to be reliable and sta-
(1984). Samples were filtered onto Whatman GF/F fil- ble over time. The degree of oxygen saturation was
ters and extracted with 90% acetone for 18 h at 4°C in calculated using in situ concentration and theoretical
the dark. The fluorescence was measured before and solubility based on temperature and salinity data
after acidification with a Turner Designs Model 10-AU (Weiss 1970). The Brunt-Väisälä (or buoyancy) fre-
fluorometer. quency (N2 = {[g/ρ dρ/dz]1/2}2; in s–2 and where g is
The photosynthetic competency (i.e. maximum photo- the gravitational acceleration) was estimated from the
chemical quantum yield of photosystem II = F /F ) of the difference in potential density (ρ) between consecu-
v m
algae was estimated by pulse-amplitude-modulated tive depth (z) intervals. The pycnocline and nitracline
fluorometry (WALZ Phyto-PAM). This method is based were defined as the depth where N2 and the vertical
on the induction and detection of chlorophyll fluores- gradient in NO – concentration (dNO –/dz) were
3 3
cence, which provides the minimum (F ; near-darkness highest, respectively. The depth of the SCM was
o
condition) and maximum (F ; saturation pulse of 200 µs defined as the depth where the in vivo fluorescence
m
at 4000 µmol quanta m–2s–1) fluorescence required for was at a maximum, while its thickness was estimated
the computation of variable fluorescence (F = F – F ). as the zone of elevated fluorescence between areas
v m o
Samples were dark-adapted for 30 min at ~4°C (Ban et where the mean vertical gradient in in vivo fluores-
al. 2006; no significant correlation was observed be- cence (d(in vivo fluorescence)/dz) was zero over 5
tween F /F and in situ temperature at the depth of consecutive depth bins.
v m
collection) to allow relaxation of fluorescence quenching. Due to highly variable weather conditions (cloud
A blank was assessed at each station with SCM water cover) and because the ship did not stay at any station
filtered through a 0.2 µm syringe filter. To minimize the for more than a few hours, a comparison of sampling
effect of taxonomic variability, fluorescence was mea- sites on the basis of absolute irradiance at the SCM
sured at 3 specific wavelengths (470, 520, and 645 nm). was not practical. Instead, we made these comparisons
Emissions at 645 nm (wavelength for allophycocyanin using the coefficient of diffuse light attenuation (k),
and phycocyanin: specific pigments of cyanobacteria) which is a more stable property of the water column
were close to background noise so that only the 470 and over time scales of a few days at least. The percentage
520 nm emissions (wavelengths for chl a, band c, fuco- of incident PAR available at the SCM was calculated
xanthin and carotenoids; specific pigments of diatoms, from the value of k determined between the surface
dinoflagellates and green algae) were averaged for the and the SCM using vertical PAR profiles. In this paper,
calculation of F /F . There was no relationship between we arbitrarily define the base of the euphotic zone as
v m
chl aconcentration and F /F . the 1% of surface irradiance to facilitate comparisons
v m
Phytoplankton abundance and taxonomic composi- with the literature.
tion. For the identification and enumeration of phyto- Ordinary least-squares regressions (model I linear
plankton, samples were preserved with acidic Lugol’s regression) were used to determine predictive rela-
solution (Parsons et al. 1984). Cells ≥4 µm were identi- tionships (e.g. equations used in the reconstruction of
fied to the lowest possible taxonomic rank using an chl a profiles; post-calibrations) and geometric mean
inverted microscope (WILD Heerbrugg) equipped with regressions (considering error on both variables;
phase contrast optics (Lund et al. 1958). For each sam- model II linear regression) were used to assess func-
ple, a minimum of 300 cells was counted. The main ref- tional relationships between 2 variables (Laws &
erences used for phytoplankton identification were Archie 1981, Wallace et al. 1995, Calbet & Prairie
Tomas (1997)and Bérard-Therriault et al. (1999). 2003).
Martin et al.: Subsurface chlorophyll maxima in the Canadian Arctic 73
RESULTS 0.14
Subsurface fluorescence maxima were widespread 0.12
in the Canadian High Arctic, Hudson Bay and
n 0.10
Labrador fjords during late summer 2005 and early fall o
2006 (Fig. 1). Out of 140 stations, 85% clearly showed ati
nu 0.08
a subsurface maximum, which ranged broadly in verti- e
cal position (7 to 67 m; median: 29 m) and thickness att
(2 to 74 m; median: 18 m). The other stations (15%) m 0.06
a
showed vertically homogenous fluorescence (shallow Be 0.04
Foxe Basin; 7 to147 m) or a surface maximum (western
Baffin Bay, Mackenzie Shelf, inner Canadian Archi- 0.02
pelago, Hudson Bay and Labrador fjords).
Using the average output of the fluorometer during 0.00
0 100 200 300 400
bottle closure, the relationships between in vivofluores-
Phytoplankton biomass (µg C l–1)
cence and extracted chl aconcentration were assessed
separately for each region and expedition (e.g. for Baffin Fig. 2. Relationships between particulate beam attenuation
Bay in 2006: y= 1.05x– 0.08; n = 39, r2= 0.92, p < 0.0001). (cp) and phytoplankton carbon estimates in northern Baffin
The small residuals of the regressions showed an even Bay (data from late summer 1998; Booth et al. 2002). Model II
linear regressions: y= 0.0003x+ 0.03, n = 53, r2= 0.88 (diatom
dispersal for concentrations ranging at least 2 orders of
carbon; black circles, dotted line) and y= 0.0002x+ 0.03, n =
magnitude, and intercepts were statistically undistin- 53, r2= 0.88 (total autotrophic carbon; gray circles, solid line)
guishable from zero, indicating negligible interferences
or quenching near the surface. Region-specific relation-
ships were used to construct high-resolution vertical the SCM ranged from 0.001 to 48% of surface values
profiles of chl a. To determine if c (λ ≈ 660 nm) is an (daily mean absolute irradiance corresponding to
p
acceptable surrogate for phytoplankton biomass in 0.0006–75.12 µmol quanta m–2 s–1 assuming that the
Baffin Bay, the biovolume-based estimates of total mean irradiance along the ship track during a day is
autotrophic carbon biomass reported by Booth et al. representative of a given station), 83% (181 out of 219
(2002)(determined by epifluorescence microscopy enu- CTD stations) of the SCM were located below the 10%
meration and sizing) were compared with concomitant light level and the mode of the distribution (52% of the
transmissometer data (Fig. 2). Strong linear relationships stations) was in the 3 to 10% range (Fig. 4A). Water
were observed between c and the carbon biomass of to- temperature ranged from –1.5 to 2.0°C at 83% of the
p
tal phytoplankton and diatoms. This exercise was not at-
tempted in the shallow Canadian Archipelago and the 50 50
coastal Beaufort Sea due to the proximity of the Macken-
zie River and resuspension on shallow shelves.
The depths of the SCM, the SBM and the oxygen sat- 40 40
uration maximum (SOM) were compared in northern
Baffin Bay (Fig. 3), where river runoff has a negligible m) 30 30 m)
influence on particle load. In 2006 (when the oxygen M ( M (
probe worked consistently and the spatial coverage of B O
northern Baffin Bay was extensive), the vertical depth S 20 20 S
of the SCM closely matched those of the SBM (when
detectable; i.e. when the particle load in surface waters 10 10
was negligible) and of the SOM. A positive relation-
ship between the depths of the SCM and SOM was
0 0
also observed in the Beaufort Sea but was much 0 10 20 30 40 50
weaker (r2= 0.24, n = 61) than in Baffin Bay. SCM (m)
Fig. 3. Relationships between vertical position of the subsur-
face chlorophyll maximum (SCM), subsurface biomass maxi-
General conditions at the SCM mum (SBM; gray circles) and oxygen saturation maximum
(SOM; black circles) for northern Baffin Bay stations in 2006.
Model II linear regressions between SBM and SCM depths
The frequency distributions of physical properties
(solid line: y = 0.96x – 0.93; n = 10, r2 = 0.98) and between
and nitrogenous nutrient concentrations at the SCM
SOM and SCM depths (dotted line: y = 1.06x– 3.57; n = 10,
depth are presented in Fig. 4. Although irradiance at r2= 0.88)
74 Mar Ecol Prog Ser 412: 69–84, 2010
100 50
A B
80 40
y
nc 60 30
e
u
q
re 40 20
F
20 10
0 0
0.001 0.003 0.01 0.03 0.1 0.3 1 3 10 30 100 –2 –1 0 1 2 3 4 5 6 7 8 9
Irradiance (% of incident at surface) Temperature (°C)
35 35
C D
30 30
25 25
y
c
n 20 20
e
u
eq 15 15
r
F
10 10
5 5
0 0
0 1 2 3 4 5 6 7 8 9 10 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
NO – (µM) NH + (µM)
3 4
Fig. 4. Frequency distributions of (A) percentage of surface irradiance, (B) water temperature, (C) NO–and (D) NH+concentra-
3 4
tions at the depth of the subsurface chlorophyll maximum (SCM). In (C), gray bar = samples below the analytical limit of
detection (i.e. <0.05 µM). All sampling stations and years were pooled
stations (181 out of 219 CTD stations), with a mode Baffin Bay [BB]: 1.39 to 16.65 µg l–1) (Fig. 5). There was
between –1.5 and –1.0°C (at 21% of stations; Fig. 4B). no consistent regional difference in the depth of the
Extreme values of –1.8 and 9°C were observed in SCM, which most often occurred between 20 and 35 m.
isolated instances. Nitrate concentrations generally Conspicuous exceptions were noted off the shelf break
ranged from 0.5 to 3.0 µM with values <0.5 µM or on the BS section (40 to 62 m), at the entrance of LS
>3 µM at 22 and 20% of the stations, respectively (70m) and at 3 stations on the LS and BB sections (12to
(Fig. 4C). NH + concentrations varied between the 15 m). The SCM was at or well above the 1% surface
4
detection limit (0.02 µM) and 1.40 µM with a mode of irradiance in all but 4 stations on the sections. Chl a
0.20 µM at 52% of the stations (Fig. 4D). concentration generally declined to very low values
(<0.1 µg l–1) below the SCM, except in northern BB
where they remained moderate (>0.5 µg l–1) in the
Oceanographic sections upper 80 m of the water column.
The pycnocline (i.e. depth with the maximum value
The positioning of the SCM relative to vertical pro- of N2) was shallow nearly everywhere (14 to 32 m) but
files of nutrients and the physical structure of the water the stratification was stronger in the western sections
column was explored in detail using 4 oceanographic (i.e. BS and AG) than in the eastern ones (i.e. LS and
sections (see Fig. 1 for locations). Concentrations of BB) (Fig. 6A). The pycnocline was systematically
chlaat the SCM were generally lower in the western located in the upper euphotic zone and most often
sections (Beaufort Sea [BS]: 0.32 to 0.74 µg l–1; Amund- above the SCM. Oxygen saturation was generally
sen Gulf [AG]: 0.34 to 2.50 µg l–1) than in the eastern >100% at the SCM, with maximum values generally
sections (Lancaster Sound [LS]: 0.53 to 4.52 µg l–1; observed at or close to the SCM (Fig. 6B). NO –
3
Martin et al.: Subsurface chlorophyll maxima in the Canadian Arctic 75
concentrations were generally near the analytical Photosynthetic competency
detection limit above the SCM, which was typically
associated with the top of the nitracline (Fig. 6C). Note Considering the entire sampling area, data from
that Si(OH) and PO 3–concentrations (not shown) fol- 2005 and 2006 generally displayed high F /F at the
4 4 v m
lowed the same vertical patterns as NO –, but were not SCM (median: 0.56) although values varied between
3
depleted in the surface layer. Maximum NO –concen- 0.32 and 0.70 (Fig. 8). At the surface, F /F ranged
2 v m
trations were much higher in the western sections than from 0.08 to 0.71 with a median value of 0.55 (Fig. 8).
in the eastern ones (Fig. 6D). A well-defined primary The scatter of values was much lower at the SCM than
NO –maximum (PNM) was visible only in the western at the surface (where F /F was frequently <0.45) and
2 v m
sections, where it generally tracked the SCM. In the there was no statistically significant difference be-
eastern sections, the layer of elevated NO –concentra- tween the 2 sampling depths (Wilcoxon signed rank
2
tion was relatively diffuse and thick with no apparent test, p = 0.542).
relationship with the SCM. On all sections, NH +
4
concentrations were uniformly low in surface waters
and generally showed a subsurface maximum below Chl a (µg l–1)
BS
the SCM (Fig. 6E). Subsurface concentrations were 3
particularly high in the eastern sections (up to 1.24 µM)
20
and elevated NH4+concentrations extended far below m)
the euphotic zone in northern BB. h( 40 0.3
pt
e 60
D
Light attenuation, nutrients and vertical position of 80 x x x x xxx x x x x x x 0.03
0 50 100 150
the SCM
AG 3
The relationship between the vertical position of the
SCM and kfor all stations is shown in Fig. 7A. The range 20
of kwent from a minimum of 0.046 m–1at deep stations m)
( 40 0.3
to a maximum of 0.475 m–1at neritic stations influenced h
pt
by large rivers (Fig. 7A). To remove this influence, a e 60
D
model II linear regression was adjusted for stations with
80 xxx x x x x x x x 0.03
k< 0.15 m–1and showed a very weak (r2= 0.14), negative 0 100 200 300 400 500
relationship between the 2 variables (Fig. 7A).
LS
The SCM was deeper than the pycnocline at 86% of 5
the stations and no significant correlation was ob-
20
served between their respective vertical positions m)
(Fig. 7B). However, 68% of the stations showed a ( 40
h 0.5
vertical match between the depth of the SCM and the pt
e 60
nitracline within a margin of ±10 m (Fig. 7C). This D
match extended to 90% of the stations for a margin of 80 x x x x x x x x 0.05
±20 m. A linear regression fit trough all data evi- 0 100 200 300 400 500 600
denced 16 large residuals (standardized residuals ≥2; BB
20
belonging to stations with poorly defined SCM or nitra-
cline) that belonged to either very weakly stratified 20
stations (maximum N2 ≤ 0.0006) or sites where the m)
( 40 2
nitracline was more than 12 m below the 1% of surface h
irradiance (Fig. 7C). The regression was largely ept 60
D
improved by removing the outliers from the analysis.
0.2
To verify if light penetration through the upper water 80 x x x x x x x x x
0 50 100 150
column improved the relationship, a multivariate lin-
Distance (km)
ear regression model considering the depth of the
SCM, k (x ) and the depth of the nitracline (x ) was Fig. 5. Vertical variations of calibrated chl a concentration
1 2
adjusted to stations where k < 0.15 m–1 (minus the 16 (right-hand color key) along sections in BS, AG, LS and BB
(see Fig. 1 for exact locations). Dashed and solid lines: depths
outliers). A relationship was obtained (y = –36.67x +
1 of the subsurface chlorophyll maximum and of the euphotic
0.775x + 10.92; r2= 0.67, p = 0.2260, p < 0.0001), but
2 1 2 zone (defined here as 1% of surface irradiance), respectively.
the regression coefficient for kwas not significant. Position of each sampling station: ‘x’ on the bottom axis
76 Mar Ecol Prog Ser 412: 69–84, 2010
4 3 2 1 0 110 100 90 80 15 10 5 0 0.3 0.2 0.1 1 0.8 0.6 0.4 0.2 0 au-
e
B
x ns in and
x150 ctiones
x g seof li
Baffin Bay xxxxx10050 concentrations alonFig. 5 for definition
+H4ee
x0 E) Nm). S
Lancaster Sound xxxxxx200300400500600 ––sat), (C) NO, (D) NOand (322cale for Nand Osat (0–80 2bottom of water column
xxxx0100500 Distance (km) ygen saturation (O2he different depth so data; Grey areas:
Amundsen Gulf xxxx200300400 2cy (N), (B) percent oxern Baffin Bay. Note tmbols. White areas: n
0 nhy
x10 quenorts
xxxx0150 Väisälä freound and
-S
ntr
ue
x Brast
a x00 A) nc
ort Se xx1 ns of (ulf, La
Beauf xxxx50 variatiodson G
2) x al un
A 20 40 N260–3–(x10 s80 B 20 40 satO 602(%)80 C 50 100–NO3(µM)150 D 50 100–NO2(µM)150 E 50 100+NH4(µM)xxx1500 Fig. 6. Verticfort Sea, Am
Depth (m)
Martin et al.: Subsurface chlorophyll maxima in the Canadian Arctic 77
Taxonomic composition at the SCM
80
A
70
The taxonomic composition of the phytoplankton at
60 the SCM was investigated for both sampling years
(Fig. 9). The relative abundance of flagellates and
m) 50 dinoflagellates was 21 and 38% lower, respectively, in
M ( 40 2006 than in 2005 in all regions. During both years, the
C BS presented lower percentage of diatoms and higher
S 30
percentages of flagellates and dinoflagellates than
20 northern BB. Flagellates and dinoflagellates were
numerically dominant in the Hudson Bay system in
10
2005 and made up, on average, 90% of the total phyto-
0 plankton abundance. In all regions, centric diatoms
0.0 0.1 0.2 0.3 0.4 0.5
k (m–1) were, on average, 3 to 27 times more abundant than
pennate diatoms. Except in the Hudson Bay system,
80 Chaetoceros spp. were the most abundant centric
B diatoms at the SCM in 2005 and 2006 (76 to 99% of
70
total centric diatoms on average). Chaetoceros socialis
60 Lauder was present at 56% of the stations and repre-
sented up to 30% of the total Chaetocerosabundance.
m) 50 However, C. socialiswas less abundant in 2006 than in
M ( 40 2005. Diatoms of the genus Thalassiosira were scarce
C
S 30 throughout the Canadian Arctic (0 to 0.5%).
Pearson’s product moment correlations (PPMC) were
20 used to evaluate relationships between environmental
10 variables and the relative and absolute abundance
data. Only the significant correlations are described
0
here. At the SCM, the absolute abundance of flagel-
0 10 20 30 40 50 60 70 80
lates increased with NH + concentration (r = 0.48, p <
Pycnocline (m) 4
0.01). The relative abundance of dinoflagellates in-
80 creased with N2(r = 0.32, p < 0.05), whereas the rela-
C tive abundance of pennate diatoms increased with
70
NO – concentration (r = 0.39, p < 0.01). The relative
2
60 dominance of diatoms over flagellates decreased with
water temperature (r = –0.34, p < 0.05) and increased
m) 50 with ambient NO – concentration (r = 0.31, p < 0.05).
M ( 40 3
C
S 30 35
20
30
10
%) 25
00 10 20 30 40 50 60 70 80 y ( 20
c
Nitracline (m) n
e
Fig. 7. Relationships between depth of the subsurface chloro- u 15
q
phyll maximum (SCM) and (A) the coefficient of diffuse light e
r
attenuation (k), (B) the pycnocline, and (C) the nitracline in F 10
2005 (black circles) and 2006 (gray circles). In (A), vertical
dash-dotted line: k= 0.15 m–1(see ‘Results: Light attenuation, 5
nutrients and vertical position of the SCM’ for details); and
dashed line: model II linear regression for stations, where 0
k < 0.15 m–1 (y = –594.5x + 96.18; r2 = 0.14). In (B) and (C), 0.1 0.2 0.3 0.4 0.5 0.6 0.7
solid line indicates 1:1 match; and dashed line: model II linear F /F
v m
regression between SCM and pycnocline depths (y= 1.18x+
13.86; r2= 0.04) and between SCM and nitracline depths (y= Fig. 8. Frequency distribution of F/F at the surface (black)
v m
0.98x – 0.16; r2 = 0.66 when excluding 16 outliers). Dash- and at subsurface chlorophyll maximum (gray). All sampling
dotted lines in (C) represent 1:1 ±20 m stations and years were pooled
78 Mar Ecol Prog Ser 412: 69–84, 2010
Beaufort Sea Canadian Archipelago Baffin Bay Hudson Bay
2005 1.7 0.1
4.8 0.1 0.2 0.5 0 5.0
1.1 5.9 04.8 0.2 2.1 0.1 3.6 2.0 3.9
11118888....8888 11119999....0000
8888....1111 7777....2222
1.0
3.3
88883333....3333 66667777....3333 66668888....7777 888777...333
66668888....7777%%%%
88883333....3333%%%%
2006
8.8 3.1 4.1
Dinoflagellates
0.4 0
7.7 6.9 0.5 13.3 Flagellates
Pennate diatoms
20.3 29.9 Chaetoceros socialis
35.4 45.2
37.1 Chaetoceros spp.
0 8.9 55.9 Thalassiosira spp.
8.0
7.1 Other centric diatoms
7.1
2.8 4.8
Fig. 9.Mean percent abundance of the major phytoplankton groups observed at the depth of the subsurface chlorophyll maxi-
mum in the Beaufort Sea, inner Canadian Archipelago, northern Baffin Bay and Hudson Bay in 2005 and 2006. Hudson Bay was
not sampled in 2006
NO – concentration was positively correlated to NO – weakly stratified (western BB, Hudson Strait) and shal-
3 2
concentration (r = 0.52, p < 0.001) and salinity (r = 0.30, low (Canadian Archipelago, eastern Hudson Bay,
p < 0.05). Mackenzie Shelf and Labrador fjords) stations. Other-
wise, chl a concentrations at the SCM were in agree-
ment with sparse reports for the same regions or other
DISCUSSION sectors of the North-American Arctic (Cota et al. 1996,
Vidussi et al. 2004, Lee & Whitledge 2005).
This study provides the first near-synoptic assessment Concentrations of chl a at the SCM were relatively
of the incidence and properties of SCM through the low (0.1–2.5 µg l–1) in LS, AG and the southeast BS,
Canadian Arctic, including the subarctic Hudson Bay where values generally increased from deep offshore
and Labrador fjords, during late summer and early fall. waters to the shelf. The depth of the SCM was also
Results show that SCM are almost ubiquitous and imply quite variable. Off the shelf break, where NO – was
3
that they persist throughout the ice-free period since depleted in the upper euphotic zone, SCM were as
they are known to appear early in the growth season deep as 62 m. This result can be explained by the
(Booth et al. 2002, Tremblay et al. 2008). Now that the strong stratification, which is not broken down during
prevalence of SCM is established, their characteristics winter and limits the upward supply of nutrients all
will be discussed with respect to the physico-chemical year long (Tremblay et al. 2008). Near the coast, some
structure of the water column, potential repercussions SCM were as shallow as 8 to 20 m where nutrient con-
for the estimation of primary production by remote- centrations in the upper part of the water column were
sensing and biogeochemical implications. elevated due to enhanced local mixing (e.g. internal
The absence of SCM was noted only near rivers and waves; Sharples et al. 2007) and mild shelf-break
at shallow locations (<100 m) where vertical mixing or upwelling (Carmack & Chapman 2003)(Fig. 6). By con-
upwelling is important (Fig. 1). For example, the shal- trast, chl aconcentrations at the SCM in BB were gen-
low water column (<100 m) on the eastern side of Foxe erally high (1.4 µg l–1 in the east to 16.7 µg l–1 in the
Basin is completely mixed by 8 m tides (Prinsenberg west), even over relatively deep waters. The most
1986). Chl awas usually highest at the surface at very diffuse SCM on the vertical occurred in eastern BB