Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance

MARSYS-02481; No of Pages 9 Journal of Marine Systems xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Marine Systems journal homepage: www.elsevier.com/locate/jmarsys

Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance K. Campbell a,⁎, C.J. Mundy a, D.G. Barber a, M. Gosselin b a b

Centre for Earth Observation Science, Faculty of Environment, Earth and Resources, 467 Wallace Building, University of Manitoba, Winnipeg, Manitoba R3T 2N2, Canada Institut des sciences de la mer, Université du Québec à Rimouski, 310 Allée des Ursulines, Rimouski, Québec G5L 3A1, Canada

a r t i c l e

i n f o

Article history: Received 30 September 2013 Received in revised form 21 January 2014 Accepted 22 January 2014 Available online xxxx Keywords: Polar oceanography Ice algae Spring bloom Snow Time series Chlorophyll a

a b s t r a c t The bottom ice algae chlorophyll a (chl a)–snow depth (HS) relationship was investigated for first-year sea ice in Allen Bay, Nunavut, from 27 April to 13 June 2011. A transmitted irradiance technique was used to estimate ice algae chl a throughout the period at time series locations covered and cleared of snow. Furthermore, chl a was estimated along transects perpendicular to dominant snowdrift orientation, and at short-term snow clear experimental sites. The association between chl a and most snow depths was characterized by four phases over the spring; light limitation (negative relationship), a transitional period (no relationship), chl a decline associated with higher transmitted irradiance (positive relationship), and a final phase of chl a decline independent from HS (no relationship). Algal chl a under areas cleared of snow was lower, reached zero chl a earlier and declined faster than snow-covered control sites. Results indicated that snow removal caused these chl a responses through photoinhibition, as well as ice melt later in the spring. Based on this research we propose that weather events that can rapidly melt the snowpack could significantly deplete bottom ice chl a and cause early termination of the bloom if they occur late in the spring. © 2014 Published by Elsevier B.V.

1. Introduction Accessibility of light to sea ice primary producers during the spring bloom is largely dependent on radiative transfer through the snow and sea ice cover. Absorption and scattering of photons within the sea ice matrix, composed of ice, brine, air, and sometimes salts, efficiently attenuate light, resulting in an exponential decline of photosynthetically active radiation (PAR; 400–700 nm) transmittance with ice thickness (Ehn et al., 2008; Perovich, 1996). Despite the attenuation properties and thickness of sea ice, it is actually the much thinner layer of overlying snow that is often present, which primarily controls the magnitude of bottom ice PAR because of its high albedo and greater capacity to scatter light. In addition, snow also affects bottom ice temperature, and consequently bottom ice ablation, insulating it from the warming atmosphere during the spring (Sturm and Massom, 2010). As a result of these characteristics, the uneven distribution of snow on sea ice from wind forced displacement of drifts creates a non-uniform light and thermal environment at the ice bottom that translates to a spatially variable distribution of ice algae on the order of 100 m (Gosselin et al., 1986). Algae further vary at the microscale because of ice substructure and brine hole spacing (Mundy et al., 2007a), as well as at much larger scales (kilometers) due to the influences of ocean water salinity, ice thickness, and nutrient

⁎ Corresponding author. E-mail address: [email protected] (K. Campbell).

availability (Gosselin et al., 1986; Granskog et al., 2005; Robineau et al., 1997). The manner in which snow depth (HS) influences algal chl a can change as the growth season evolves. At the beginning of the growing season, when the ice surface is dominated by a relatively thick snow cover (i.e. in late winter or early spring), ice algae are often negatively associated with HS as the energy required for photosynthesis is greatly restricted by decreased transmitted radiation below thicker snowdrifts. During this period, algae acclimate to the low irradiance by modifying their photosynthetic apparatus (Barlow et al., 1988; Robinson et al., 1995) and by producing accessory pigments to enhance light harvesting (Arrigo et al., 2010), but positive net photosynthesis cannot begin until a minimum level of irradiance, on the order of 7.6 μmol photons m−2 s−1, is reached (Gosselin et al., 1985). Later in the spring, snowmelt from rising air temperatures permits an increase in the amount of radiation entering the sea ice. Ice algae can acclimate to the resultant change in bottom ice irradiance by altering their cellular content of light-harvesting pigments and/or reaction centers (Barlow et al., 1988; Falkowski and Raven, 2007; Michel et al., 1988). Despite the ability of algae to acclimate to increasing light levels, high levels of irradiance can still have negative physiological effects resulting in photoinhibition (Barlow et al., 1988; Michel et al., 1988). For this reason, along with the thermal influence of snow as it buffers the sea ice from warming atmospheric temperatures, thereby reducing bottom ice ablation, the negative association between chlorophyll a (chl a) and HS may switch to a positive relationship (Gosselin et al.,

0924-7963/$ – see front matter © 2014 Published by Elsevier B.V. http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

2

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

1986; Sturm and Massom, 2010; Welch and Bergmann, 1989). During this latter period of the spring, algae are also limited by depleted nutrient resources because of algal consumption or may experience a form of light limitation as cells higher in the ice column absorb preferred wavelengths. Over the course of the bloom, ice algae populations can thus shift from a phase characterized by light limited growth and accumulation to that of one or a combination of diurnal light limitation, nutrient limitation, self-shading induced light limitation, photoinhibition, and ice ablation (i.e., habitat erosion; Cota and Smith, 1991; Gosselin et al., 1990; Lavoie et al., 2005). Grazing pressure may also influence ice algae biomass throughout the spring although, grazing by ice meiofauna is thought to be minimal (Nozais et al., 2001) and contributions by zooplankton remain largely unknown, with best estimates relying on a set percentage of total biomass (Lavoie et al., 2005). The bloom ends in the late spring with the release of ice algae into the water column following snow and ice melt. Spring melt in the Arctic is anticipated to occur quicker and earlier due to the warming climate (ACIA, 2005). This may affect the timing and extent of ice algal bloom. Changes in the overall productivity or the timing of bloom could have critical implications for the Arctic icecovered ecosystem (Leu et al., 2011). It is thus important to understand the potential responses of ice algae to rapid decreases in snow thickness (and associated light conditions) at different periods of the spring. A study by Juhl and Krembs (2010) concluded that the response of ice algae to rapid light increases depends not only on the magnitude of change in irradiance but also on the photophysiological state of the algae before the change. Furthermore, their observations suggest that ice algae may therefore have the capacity to acclimate to increased light levels given sufficient time. However, algal exposure to sudden increases in irradiance levels without adequate time to acclimate will likely result in a decline in ice algae chl a (Fortier et al., 2002; Juhl and Krembs, 2010). The objectives of our study were to investigate the relationship between ice algae chl a and HS, and to examine the effects rapidly changing snow depths have on bottom algal chl a across the spring bloom period. The study used a transmitted irradiance technique (Campbell et al., 2014) to estimate bottom ice chl a concentrations at time series sites, along transects, and at locations cleared of snow in the Canadian High Arctic during the spring season. By meeting the objectives, we will provide new information on potential ice algal responses to abrupt warming events expected to occur in the Arctic associated with a warming climate.

5°N

54' Cornwallis Island

48'

Ice Camp

42'

Resolute Bay

Resolute Passage

36' Griffith Island 30' 96°W

40'

20'

95°W

40'

20'

94°W

Fig. 1. Location of 2011 ice camp in Allen Bay, Nunavut, Canada (74° 43.165′ N; 95° 10.099′ W).

2.2. Measurement of spectral transmitted irradiance Transmitted irradiance (Ed(z, λ)) was measured approximately 0.15 m beneath the sea ice at core locations for calibration of the normalized difference index (NDI) (see Eq. (1) below), along transects, and at snow clear sites. This was achieved using a dual head visible-near infrared spectrometer (Analytical Spectral Devices Inc.©) with a cosine receptor (180° field of view) that measured spectral irradiance (W m−2 nm−1) over wavelengths from 350 to 1050 nm at a 1.4 nm bandwidth. The underwater sensor, which automatically applies an immersion correction, was positioned beneath the ice by attachment to a mechanical arm, deployed through an auger hole, which positioned the sensor 1.5 m south of the hole. Underwater spectra were integrated over 4.35 s. While the instrument recorded irradiance, a solid foam device covered the hole to minimize light contamination on the measurement. Coincident with all Ed(z, λ) measurements, incident spectral irradiance was also recorded using the reference ASD sensor and was integrated over 17 ms. PAR was estimated by integrating spectral data over 400 to 700 nm and a PAR transmittance (i.e. Ed(z, PAR) / Ed(0, PAR)) was calculated from these measurements (Campbell et al., 2014).

2. Materials and methods 2.1. Study site Data were collected between 27 April and 13 June 2011 at a field station located in Allen Bay, Nunavut, Canada (Fig. 1). The region was characterized by smooth landfast first-year sea ice (FYI) approximately 1.3–1.7 m thick overlying a 60 m water depth. For sampling purposes snow depths were categorized as low (b 10 cm), medium (10 to 18 cm), or high (N18 cm) in the early spring. During advanced melt, areas of low HS are predisposed to form ponds (Iacozza and Barber, 1999); therefore, melt ponds in this study were considered extensions of the low snow depth category, and areas of snow cover persisting into late stages of spring were categorized as high. Meteorological variables including air temperature (Vaisala HMP 45C212), wind speeddirection (RM Young 05103), incident downwelling PAR (Kipp & Zonen PAR-Lite), and incident upwelling and downwelling shortwave radiation from wavelengths 305 to 2800 nm (Kipp & Zonen CNR1) were recorded every minute at a weather station installed on the ice surface throughout the entire research period. Albedo was calculated using the upwelling and downwelling measurements from this location to provide a descriptor of surface change over time. All variables from this meteorological site were averaged on a daily basis.

2.3. Extracted ice core chl a measurements Cores from high, medium, and low snow sites in the early spring, or high and low during advanced melt, were taken using a 9 cm Mark II Kovacs core barrel at new locations every four days following Ed(z, λ) measurement around 10:00 h central daylight time. The bottom 10 cm of each core was collected and melted over a period of 18 to 24 h in 0.2 μm filtered seawater that was added at a ratio of 3:1. Pseudoduplicate subsamples for fluorometric analysis were obtained from filtration of the melted cores (Whatman GF/F filters) followed by extraction of pigments by placement of the filter in 10 ml of 90% acetone at 5 °C for approximately 24 h. Fluorometric measurements (Turner Designs Fluorometer 10-005R) were then taken on the subsamples before and after acidifying the solution with 5% HCl (Parsons et al., 1984). Areal chl a concentration was calculated based on these measurements and the equations of Holm-Hansen et al. (1965). 2.4. Determining chl a from transmitted irradiance Ice algae chl a was estimated in this study using the transmitted irradiance technique described by Mundy et al. (2007b) and presented

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

in Campbell et al. (2014). The method requires calibration of transmitted irradiance (Ed(z, λ)) to core based chl a estimates using a NDI ratio of optimal wavelengths to estimate chl a. NDI has the form: NDI ¼ ½Ed ðz; λ1Þ–Ed ðz; λ2Þ=½Ed ðz; λ1Þ þ Ed ðz; λ2Þ:

ð1Þ

The best NDI wavelength combination to estimate chl a for the 2011 calibration dataset (NDIx) was determined to be 478 and 490 nm (Campbell et al., 2014). See Campbell et al. (2014) for a full description of NDI methodology. 2.5. Time series sampling sites

3

2.7. Termination of time series and snow clear experiments The use of transmitted irradiance to estimate algal chl a during the late spring can result in falsely high values because of absorption by algal derived particulate and dissolved material in the water column between the sensor and ice subsurface (Campbell et al., 2014). To prevent falsely high ice algae chl a estimates from absorption by these materials, measurements at time series and snow cleared sites (Sections 2.5 and 2.6) were ceased once zero chl a was reached. Zero chl a was not reached during the first 10 days of snow clear experiments A, B and C and consequently, measurements for the desired 10-day length of sampling were used. However, sample days used in analysis of sites D and E were less than the desired 10 days as chl a declined to zero on sample days 6 and 3, respectively. In addition, based on the 13 June date at which all snow-covered time series reached concentrations of zero chl a, no measurement at any sampling site was used past this date, serving to shorten snow clear site F to 8 days total.

To monitor changes in chl a at one point location over time, three snow-covered areas of approximately 9 m2 were selected to represent each of the snow depth categories. Hereafter they are referred to by their HS classification as the low, medium or high time series sites. An auger hole was excavated in the middle of the north end of each snow covered area (permitting sensor placement in the middle of the site), after which Ed(z, λ) was recorded daily following the method outlined in Section 2.2. Precautions were taken in an effort to maintain the natural integrity of the snow surfaces, such as careful re-excavation of the auger hole during the cold period when ice would reform between measurements, and no traffic was permitted on the sample area. Daily measurement of Ed(z, λ) at each time series site was followed by the sampling and averaging of five snow depths on a nearby comparable surface. On 19 May all 3 time series were contaminated by soot fallout from the ice camp furnaces upwind hence, three new sites of the same characteristics were selected approximately 50 m away (at locations out of the contaminated area). For the purposes of analysis, the new high, medium, and low sites were loosely considered extensions of the previous locations where the two sampled areas of each snow depth category were treated as one continuous time series.

To characterize spatial relationships, a total of 11 transects running in an East–West direction, perpendicular to lengthwise snowdrift orientation, were randomly constructed approximately every 4 days over the sample period (Table 1). This directionality was chosen to capture the potential influence of snow depth on ice algae chl a (Gosselin et al., 1986). Holes for operating the mechanical arm were made at two meter intervals along each transect, followed by recording of Ed(z, λ) measurements between 14:00 and 16:30 h. Snow or melt pond depth along each transect was then measured every meter at the approximate location Ed(z, λ) was sampled (1.5 m south of hole). The total length of transects (m) was a function of time, such that measurements had to be completed before the sun was in a far west position, and incident downwelling irradiance became too low for optimal Ed(z, λ) sampling. As a result of weather and equipment failures affecting sampling time, half of all transects sampled were approximately 30 m in length, while the remainders were 60 m.

2.6. Experimental sampling sites

3. Results

Snow cover on patches of sea ice was removed to assess the response of ice algal chl a to rapid increases in transmitted irradiance over the spring. These sites are referred to as the snow clear or experimental sites throughout the manuscript, and were sequentially assigned letters for identification. A total of six snow-cleared regions (A through F) were created, all of which were kept free of snow for a period of approximately 10 days, apart from the first site (A) that was maintained from 27 April to 2 June as the snow cleared time series site. Experimental site A is therefore analyzed as both a short-term snow clear site (when only the first 10 days after snow removal were considered), as well as a time series site (when the total length of sampling was assessed) in this manuscript. Sampling of these experiments began with the measurement of Ed(z, λ) for one day under sites of undisturbed medium snow cover, with the exception of one low snow clear site (D). An area of 9 m2 immediately south of the augered hole for deployment of the radiometer was then marked, sampled for average HS, and cleared that evening. This size of cleared area was deemed sufficiently large to exclude any influence of horizontal scattering from neighboring snow covered ice, comparable to areas used in similar experiments such as that by Gradinger et al. (1991). Following HS removal, Ed(z, λ) was measured daily between 09:00 and 11:00 h at the snow cleared sites, immediately after clearing the ice surface, while ensuring the underwater sensor was centered under the cleared area. These experimental treatments were created until the onset of melt ponds, which corresponded to the disappearance of most snow from the ice surface.

3.1. Site characteristics

2.8. Spatial sampling

Daily averaged air temperature increased approximately linearly over the sampling period from a minimum of − 21.8 ± 3.3 °C on 27 April to a maximum of 1.4 ± 0.6 °C on 10 June. Ed(0, PAR) also increased during most of this period, until peaking on 2 June and declining slightly thereafter (Fig. 2a, c). During the same period, the snow-covered sea ice transitioned from an average snow depth of about 14 cm to a surface of mixed coverage comprised of melt ponds, snow mounds, and surface drained white ice. Albedo remained relatively constant around 0.84 Table 1 Summary of transect construction including sample date, transect length and number of transmitted irradiance (chl a) and snow depth (HS) measurements. Transect ID

Sample date

Length (m)

1 2 3 4 5 6 7 8 9 10 11

29 April 3 May 9 May 13 May 17 May 21 May 25 May 29 May 3 June 6 June 11 June

24 30 24 60 28 30 60 60 60 60 30

Observations (n) Ed(z, PAR)

HS

13 16 13 31 15 16 31 31 31 31 16

25 31 26 61 61 31 61 61 61 61 31

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

4

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

5

a) Snow clear site A

a)

35

Air Temperature (°C)

0 -5

20 15

-10

10 -15

5 0

-20

-5

-30 1

b)

0.9

Chl a (mg m-2) or HS (cm)

0.8 0.7

Albedo

Chl a HS

25

-25

0.6 0.5 0.4 0.3 0.2 0.1 0 800

Incident Downwelling PAR (µmol photons m-2 s-1)

Modelled chl a

30

c)

750 700

35 30 25 20 15 10 5 0 -5 35 30 25 20 15 10 5 0 -5

650 600 550 500 450 400 27 April

7 May

17 May

27 May

6 June

Date

35 30 25 20 15 10 5 0 -5

b) Low HS

c) Medium HS

d) High HS

7 May

17 May

27 May

6 June

Date Fig. 2. Daily average values of (a) air temperature, (b) albedo and (c) incident downwelling photosynthetically active radiation (Ed(0, PAR)) measured at the meteorological station over the sample period. In (a) and (b), bars represent the standard deviation.

until 6 June, after which values declined sharply, particularly after 10 June, to a low of 0.30 ± 0.02 on 13 June (Fig. 2b). The period of rapid change in albedo and thus surface conditions was a likely result of a storm event on 10 June that brought strong winds and rain. Bottom ice temperatures increased as the spring progressed, coincident with increasing under-ice current velocities that followed a bi-weekly tidal cycle (Campbell et al., 2014).

3.2. Time series analysis Data for ice algae chl a accumulation and loss over time, at each of the four time series (i.e. snow clear site A, and high, medium, and low snow covered sites), were best fit by log normal or Weibull distribution functions (SigmaPlot Version 12.5; Fig. 3). Peaks of these functions, which marked the division between algal accumulation and decline phases, differed with snow depth category, where the snow clear experimental site peaked at 14 May (log normal; r2 = 0.589), 30 May under high snow (Weibull; r2 = 0.568), 17 May under medium (log normal; r2 = 0.677), and 15 May under low snow depths (log normal; r2 = 0.732). The increasingly later peak chl a date with deeper snow covers indicated that under deeper snow, algal populations were able to maintain a chl a accumulation phase longer into spring. However, this trend

Fig. 3. Time series of bottom ice chlorophyll a (chl a), fitted with a log normal (a, b, d) or Weibull (b) function, and snow depth (HS) at (a) the snow clear site A and (b) under low, (c) medium and (d) high snow cover sites. Chl a was calculated using the normalized difference index (see Section 2.5 for details), with outliers removed on the 30 May under high snow cover (d) and 10 June under low snow cover (b). Negative snow depths represent depth of melt ponds.

did not translate to greater chl a accumulation as the modeled maximum daily chl a values at high, medium, and low time series were nearly equivalent at 22.4, 21.4, and 23.2 mg m−2 (Fig. 3b). Net accumulation and loss rates (mg chl a m− 2 d− 1) were calculated on data prior to and after the peak of each time series distribution function, respectively (see Riedel et al., 2008 for accumulation rate calculation). Accumulation of chl a during the beginning of the sampling period was quickest under low snow sites (1.33 mg chl a m − 2 d − 1 ), followed by medium (0.75 mg chl a m− 2 d− 1 ), and high snow (0.54 mg chl a m− 2 d− 1). The opposite was observed for chl a loss associated with bloom termination at these sites where the decrease appeared to be most gradual under low (− 1.19 mg chl a m− 2 d− 1), followed by medium (− 1.23 mg chl a m− 2 d− 1), and high snow sites (− 3.18 mg chl a m− 2 d− 1). Despite these differences in chl a response over the bloom, all snow covered locations ended (reaching zero chl a) on 13 June. The snow clear time series (site A) was classified as having a medium snow cover prior to the removal of snow and therefore, the medium snow time series site was treated as a control for this experiment

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

3.3. Snow clear experimentation Chl a (C) was calculated using the normalized difference index at the short-term snow clear experimental sites for sample day 1 (snow-covered) through n (day 2 onwards; snow-cleared). The changes in chl a between consecutive sample days (Cn − Cn − 1) for each snow clear experiment site were averaged over the site's duration in days (N). These average rates of chl a change following snow removal (RX): Xn RX ¼

2

ðC n −C n−1 Þ

ð2Þ

N

for sites A through F ranged between 0.06 ± 3.90 mg chl a m−2 d−1 on 27 April at site A, and −6.35 ± 6.27 mg chl a m−2 d−1 on 31 May at site E (Fig. 4). Overall, RX became increasingly negative for experiments done progressively later in the spring, with site E having an uncharacteristically large negative response in relation to the overall trend.

a) 27 April - 6 May 20 18 16 14 12 10 8 6 4 2 0

800 750

Chl a

700

E (0, PAR)

650 600 550 500 450 400

1

Chl a (mg m-2)

R = 0.06 ± 3.90 R = 0.13 ± 0.58

2

3

4

5

6

c) 21 - 29 May

20 18 16 14 12 10 8 6 4 2 0

7

8

9

R = -1.28 ± 4.08 R = -0.28 ± 0.54

800 750

600 550 500 450 400 3

4

5

e) 31 May - 2 June

20 18 16 14 12 10 8 6 4 2 0

6

7

8

750 700 650 600 550 500 450 400

1

2

3

ð3Þ

N

R = -1.07 ± 3.82 R = -0.18 ± 0.65

800 750 700 650 600 550 500 450 400

2

3

4

5

6

d) 27 May - 1 June

20 18 16 14 12 10 8 6 4 2 0

7

8

9

10

R = -1.90 ± 3.61 R = 1.13 ± 2.43

800 750 700 650 600 550 500 450 400

1 800

ðC Tn −C Tn−1 Þ

b) 10 - 20 May

20 18 16 14 12 10 8 6 4 2 0

9

R = -6.35 ± 6.27 R = 0.46 ± 0.51

2

were used as controls to which the change at snow clear sites (RX) could be compared. With the exception of site D which used the low snow depth time series (Fig. 3b) as a control, the daily change in chl a at the medium time series site (CTn − CTn − 1) was calculated and averaged for the sampling period of each snow clear experiment (N) using Eq. (3) (Fig. 3c). Comparison of chl a rates of change under snowcovered (RX) and cleared sites (RTS) indicated that algal response under snow free areas was less positive at site A (Fig. 4a), and more negative at sites B through F (Fig. 4). Similar to the assessment of snow clear site A time series, the removal of snow was again seen to hinder the accumulation of bottom ice algal chl a. Daily averaged albedo on sample day 1 of each snow clear experiment did not change over the sample period (r2 = 0.004, P = 0.909; Fig. 2b) as all snow clear experiments were carried out prior to ponding. However, the average magnitude of Ed(0, PAR) on sample day 1 of each site increased significantly over the sampling period (r2 = 0.678, P b 0.05; Fig. 2c). This indicated that the magnitude of radiation entering the sea ice increased over the melt period as insolation increased seasonally. Based on the 17 May peak for the time series of snow clear site A (Fig. 3a), short-term snow clear experiments prior to this date were

1

650

2

Xn RTS ¼

10

700

1

The average rates of change at snow covered time series sites (RTS):

20 18 16 14 12 10 8 6 4 2 0

2

f) 6 - 13 June

3

4

5

6

R = -2.17 ± 4.08 R = -1.09 ± 3.25

Ed(0, PAR) (µmol photons m-2 s-1)

(Fig. 3a, c). Comparison of these sites indicated significant differences in the magnitude of chl a (ANOVA, P b 0.05), chl a suppression at the snow clear location (maximum 9.2 mg chl a m−2) and earlier timing of peak chl a on 15 May. Furthermore, the timing of bloom termination, illustrated by zero chl a, was nearly two weeks earlier than any other time series site (Fig. 3).

5

800 750 700 650 600 550 500 450 400

1

2

3

4

5

6

7

8

Sampling Day Fig. 4. Temporal changes in bottom ice chlorophyll a (chl a) and daily average incident downwelling photosynthetically active radiation (Ed(0, PAR)) at snow clear sites (a) A, (b) B (c) C, (d) D, (e) E and (f) F. Sample day one is snow covered, sample day two and onwards represent measurements following snow removal. Chl a was calculated using the normalized difference index (see Section 2.5 for details). The rate of change of chl a following snow removal at the snow clear sites (RX) and for the same period at the time series site under medium snow cover (RTS) was calculated according to Eqs. (2) and (3), with the exception of site D where RTS was calculated from the time series site under low snow cover. Standard deviations of mean daily RX and RTS values are also provided.

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

6

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

hypothesized to be in a positive accumulation phase, while those following should have experienced a negative accumulation phase, marked by chl a decline. This hypothesis is supported by the shortterm snow clear experiments, which show a change in RX from a positive value during the accumulation phase (Fig. 4a) to negative values during the loss phase (Fig. 4c–f), with the exception of site B where this experiment encompassed 17 May (Fig. 4b).

3.4. Transect analysis The distributions of chl a, HS, and Ed(z, PAR) along transects sampled over the study are summarized in Fig. 5. Spatial variability of chl a, indicated by the spread of the box plots, was low during the initial and final transects sampled, implying that small levels of chl a were distributed relatively evenly across the bottom ice during early and late spring

35

a)

30

Chl a (mg m-2)

25 20 15 10 5 0

b) 60 50

HS (cm)

40 30 20 10 0

(Fig. 5a). The trend of median transect chl a agreed well with the time series datasets from the medium and high snow sites (Fig. 3c, d). Furthermore, the majority of snow depths measured along transects fell within medium and high snow classes (38% medium and 40% high), which likely explained this agreement. The transects sampled on 17 May, 25 May, and 3 June appeared to deviate from the general peak distribution trend of chl a (Fig. 5a). HS was positively skewed on 17 May and on 25 May as a number of very high snow depths were sampled (Fig. 5b). The uniqueness of these transects could be explained by inadvertent sampling across an area of thick snow cover such as a ridge or patch of rough ice which are known to accumulate deeper snow (Iacozza and Barber, 1999). However, the low median chl a on 3 June was also observed at low and medium snow time series sites (Fig. 3b, c) and thus, was not a result of skewed sampling as the range of HS did not appear to deviate from the average (Fig. 5b). A widespread early sloughing event may instead explain this result. In comparison to chl a, HS and Ed(z, PAR) did not exhibit the same seasonal trend. Instead, median values and spatial variability remained largely consistent over time with the exception of HS on 17 and 25 May (as described above) and Ed(z, PAR) on 11 June (Fig. 5b, c). The large increase in Ed(z, PAR) median and range values between 6 and 11 June illustrated the effect the severe weather event, as discussed earlier, and increasing air temperatures (Fig. 2a) had on the sea ice light environment, as HS decreased and ponds formed. Relationships between transmitted irradiance derived chl a and HS were evaluated for each transect, using linear regression analysis (Fig. 6). The relationship between chl a versus HS changed from negative to positive between 29 May and 3 June (Fig. 6h, i). Algal chl a measured under depths of snow greater than 30 cm was consistently low over the spring despite the observed changes in chl a under other snow depths. The low chl a associated with these very high snow depths was exemplified in transects sampled on 21 May and 3 June, where chl a under low snow had begun to decline resulting in a peak chl a at medium snow depths and overall polynomial distribution (Fig. 6f, i). For all transects sampled, chl a under snow covers of ≥ 30 cm averaged 6.12 mg m − 2 and linearly declined with increasing H S (r = − 0.514, P b 0.05; data not shown). This concentration is much lower than chl a observed under snow less than 30 cm, which averaged 14.55 mg m−2 (Fig. 6). Therefore, chl a accumulation under snow covers deeper than 30 cm was likely light limited over the entire spring, the extent of which was dependent on small variations within this category of HS. However, these deep snow covers only accounted for 13% of the total depths sampled. 4. Discussion

c)

4.1. Photoacclimation

PAR Transmittance

20

15

10

5

0 29 April 3 May 9

13

17

21

25

29

3 June 6

11

Ice algae have the capacity to acclimate to small increases in irradiance levels given enough time (Juhl and Krembs, 2010). Such light acclimation is characterized by a decrease in cellular chl a content, as well as a production of more photoprotective pigments (Barlow et al., 1988; Falkowski and Raven, 2007; Michel et al., 1988). The gradual rise of Ed(0, PAR) over this study (Fig. 2c) would have resulted in algal populations later in the spring exhibiting a higher degree of high light acclimation than those early on. We recognize that the lower chl a values documented later in the spring, at time series, snow clear sites and along transects, could be a function of decreasing cellular chl a content. Nevertheless, the extent of documented change in chl a as well as the seasonal relationships between chl a and HS warrant further discussion.

Date 4.2. Ice algae chl a response to snow removal Fig. 5. Temporal changes and variability in (a) bottom ice chlorophyll a (chl a), (b) snow depth (HS) and (c) integrated transmittance of photosynthetically active radiation (PAR) from 29 April to 11 June along transects. Chl a was calculated using the normalized difference index (see Section 2.5 for details). Boxes represent the median and interquartile ranges, vertical dashed lines the sample range, and outliers are indicated by crosses.

Suppressed chl a and early termination of the bloom at the snow free time series site A (Fig. 3a), along with the negative chl a responses documented at other snow clear experimental sites (Fig. 4), indicated an

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

a)

10 9 8 7 6 5 4 3 2 1 0 10

15

20

25

30

d)

35

b)

29 April r = -0.465

35

40

13 May r = -0.717

30

15

10

10

5

5

9 May r = -0.769

0

0 5

10

15

20

10

25

17 May r = -0.280

e)

15

25

20

21 May r = -0.694

f)

35 30

20

25

25 20

15

20

15

10

15 10

10

Chl a (mg m-2)

c)

20

3 May r = -0.530

15

25

7

5

5 0 0

10

20

30

g)

25

40

25 May

r = -0.750

0 10 35

5 0 15

20

25

30

35

40

0

29 May r = -0.504

h)

15

10

10

0

5

5

-5

0 5

5

25

45

j)

35

65 6 June r = 0.521

30 25 20 15 10 5 0 5

10

15

20

25

30

10 9 8 7 6 5 4 3 2 1 0 -5

3 June

r = 0.376

20

15

5

40

25

20

10

30

30

25

15

20

i)

40 35

30

20

10

15

25

35

k)

0

0

45

5

15

25

35

45

11 June r = 0.327

5

10

15

20

25

HS (cm) Fig. 6. Linear regressions between bottom ice chlorophyll a (chl a) and snow depth (HS) for transects sampled from the 29 April to 11 June. Open circles represent sites covered with ≥30 cm of snow. Chl a was calculated using the normalized difference index (see Section 2.5 for details). Solid lines indicate slope significantly different from zero (P b 0.05), while dashed lines represent non-significant relationships (P N 0.05).

overall negative effect of snow removal on ice algae chl a accumulation during our study. Clearing of snow cover increases both light transmission into the ice and conductive heat flux between the atmosphere and sea ice. In turn these changes affect ice algal chl a via photo-physiological processes (e.g. photoacclimation, photoprotection and photoinhibition) or through cell loss due to ice melt processes. Ice melt is associated with i) the increased conductive heat flux during spring and, or ii) light absorption by sea ice, as well as iii) the absorption and conversion of radiation to heat by algal cells, referred to as biological melt (Juhl and Krembs, 2010; Lavoie et al., 2005; Zeebe et al., 1996). Ice melt may also indirectly cause cell removal following enhanced brine drainage (Campbell et al., 2014; Mundy et al., 2005). Bottom ice chl a of shade-acclimated populations can decline in part because of photoinhibition when subjected to large jumps in irradiance, while light-acclimated cells may not be affected (Juhl and Krembs, 2010). Based on this reasoning, greater chl a suppression and loss rates due to algal photoinhibition were expected for snow clear experiments conducted on shade-acclimated algae of the early spring in

comparison to the light-acclimated populations later on. However, this hypothesis was not supported by the less negative RX values as well as chl a levels well above zero earlier in the spring (snow clear sites A, B, and C), while rapid chl a losses often reaching zero were documented later in the season (snow clear sites D, E, and F; Fig. 4). These observations suggest that chl a suppression at snow clear sites (relative to their snow covered control sites) may be attributed to photoinhibition however, mechanisms other than photoinhibition were also highly influential on ice algae chl a late in the spring. The strength of the negative linear association between RX (without the site E outlier) and the bottom ice temperature gradient on day 1 of sampling obtained from Campbell et al. (2014; r2 = 0.930, P b 0.05), supports the hypothesis that factors associated with ice melt were important controls on chl a. Indeed the large differences in the rate of chl a change between snow-cleared (RX) and snow-covered (RTS) sites from 27 May to 2 June (Fig. 4d, e) indicate that algal exposure to large increases in transmitted irradiance, at a time when the bottom ice is less stable from warming temperatures, elicits a more negative response.

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

8

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

The decline in chl a to zero (or nearly zero) during the last three experiments also shows the potential for early algal bloom termination following snow clearing of ‘warm’ ice. In contrast, when the ice is sufficiently cold to support a stable ice matrix even after snow removal (the first three experiments), cells may recover and avoid early bloom termination. Therefore, we suggest that the greater severity of chl a response to snow removal in the late spring may be explained by ice melt. It follows that overall algal chl a during this period was negatively influenced by the combination of photoinhibition and ice melt. However, the removal of snow will not result in bottom ice ablation unless atmospheric temperatures are high enough to significantly warm the ice (Juhl and Krembs, 2010). Cold atmospheric (Fig. 2a) and bottom ice temperatures (ranging from about − 6.2 to − 1.85 °C; Campbell et al., 2014) present during early sampling were therefore unlikely conditions for the loss of algae via ice melt even following snow removal at that time. Similar to the snow removal experiments by Juhl and Krembs (2010), we suggest that the negative response of the shade-acclimated algae to rapid snow clearing during the early spring in this study was may be primarily attributed to photoinhibition alone. 4.3. Seasonality of the chl a–HS relationship The change in association between chl a and HS along transects is summarized in Fig. 7, where the Pearson correlation statistic (r) determined for locations with HS less than 30 cm is plotted. The result is an approximately linear and significant increase of r values from −0.855 to 0.761 over time (r2 = 0.720, P b 0.05). This plot was used to divide the spring into four periods of chl a–HS association: significant negative (29 April to 13 May), not significant (14 to 29 May), significant positive (30 May to 6 June), and not significant (11 June and later). Early in the spring (29 April to 13 May), the significant negative association between chl a and HS was likely a result of light limitation caused by low insolation (and low Ed(0, PAR); Fig. 2c), and the strong attenuating properties of snow which are represented by the negative r values describing HS and PAR transmittance in Fig. 7. Following this period, a transitional phase (14 to 29 May) between low light and other environmental factors occurred, represented by the low negative and insignificant correlations of chl a with both HS and Ed(z, PAR). In their modeling efforts, Lavoie et al. (2005) documented a similar period of

change between phases where a low light phase was followed by a period characterized by alternating diurnal light and nutrient limitation, before reaching the final period dominated by low nutrient levels. During more advanced stages of melt (30 May to 6 June) in this study, algae populations became inhibited by factors related to a thinning snowpack, resulting in significantly positive correlation coefficients. The greater loss of chl a associated with greater bottom ice temperatures in snow clear experiments suggested that populations of algae were able to persist late in the spring under higher snow depths as a result of a delay in bottom ice warming and likely associated ice melt. Timing of the transitional period in Fig. 7 is further reinforced by the modeled peaks of all time series sites in our study (Fig. 3), which fell within this phase. The progressively later peak dates for deeper snow depths can be explained by the rates of chl a accumulation, where faster growth of the ice algal community under lower snow depths could have resulted in a more rapid use of available nutrients, leading to earlier nutrient limitation. Różańska et al. (2009) demonstrated that maximum chl a in Arctic first-year landfast ice was a function of sea surface nitrate concentrations. Although nutrients were not specifically sampled in this research, our observations could reflect their regulation on the maximum chl a reached by the bottom ice algae. It also implies that algal populations under different snow depths could simultaneously experience slightly different limitation states. The shift from a negative to positive relationship between chl a and HS occurred approximately two weeks later in our study than in the Hudson Bay study by Gosselin et al. (1986). In general, Hudson Bay undergoes melt onset in May, much earlier than the higher latitude location of Resolute Bay in the Canadian Archipelago where ponding does not occur until June (Fequet et al., 2011). This observation, along with the early termination documented at the snow clear time series illustrates that the earlier snowmelt occurs, the sooner bloom termination will likely occur. The important role snow serves in delaying the end of the ice algal spring bloom is thus supported (Lavoie et al., 2005). The fourth and final phase contained the 11 June transect, which had no statistically significant r values for both chl a–HS and chl a–Ed(z, PAR) associations. We suggest that ice algae chl a at this date and later were declining as a result of bloom termination independent of any association with HS or Ed(z, PAR). That is, any spatial variability present in ice

1 0.8

Correlation Coefficient

0.6

chl a-HS

chl a-PAR transmittance

0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 29-April

PAR transmittance-H

4-May

9-May

14-May

19-May

24-May

29-May

3-June

8-June

Date Fig. 7. Temporal changes in Pearson correlation coefficient (r) between bottom ice chlorophyll a (chl a), snow depth (HS) and integrated transmittance of photosynthetically active radiation (PAR) for transects sampled from 29 April to 11 June. Chl a was calculated using the normalized difference index (see Section 2.5 for details). Note that chl a–Ed(z, PAR) and chl a–HS relationships were based on sites with HS b 30 cm. Solid circles indicate significant correlations (P b 0.05).

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008

K. Campbell et al. / Journal of Marine Systems xxx (2014) xxx–xxx

algae was a relic of previous conditions as all populations were subjected to critical levels of widespread ice melt and, or nutrient limitation. It is important to note that the phases discussed above were a function of only those sites where snow depth was 30 cm or less. Ice algae at sites with snow ≥30 cm remained within a light limited phase throughout the spring, the extent of which was linearly dependent on small variations in snow depth. 5. Conclusions The association between ice algae chl a and snow depth was investigated in this study. Analysis of time series revealed that the nature of response and total accumulation of chl a during the spring bloom was dependent on the class of overlying snow depth (none, low, medium, or high). For algae under less than a 30 cm of snow cover, the nature of chl a–HS association was summarized by four periods; light limitation in the early spring, a transitional period, losses driven by high levels of irradiance and potentially nutrient limitation, and snow independent decline. Algae under snow greater than 30 cm depth remained primarily light limited over the entire spring bloom. These snow depth dependent trends in chl a accumulation and loss undoubtedly contribute to the bottom-ice spatial distribution of ice algae that is often documented. Ice temperature, and a rapid change in Ed(z, PAR), were important factors in causing low bottom ice chl a in snow clear experiments, likely through their influence on photoacclimation, photoinhibition and ice melt. The combined influence of photoinhibition and ice melt was not as significant earlier in the spring when ice temperatures were still cold enough to essentially retain algae in the ice matrix. These observations along with previous documentation of an ice temperature threshold by Campbell et al. (2014) suggest that the severity of algal response to rapid exposure of high irradiance is dependent on the time of spring. It follows that rain events and warm spells, which are anticipated to increase in frequency in the Arctic as a result of climate change (ACIA, 2005), would have greater negative effects if they were to occur later in the spring. Sooner bloom termination from these events in the early spring is unlikely, since ice algae during our study showed the capacity to recuperate during periods of low to zero ice melt. Acknowledgments The authors would like to recognize the support provided by a Northern Scientific Training Program grant and the Natural Sciences and Engineering Research Council of Canada (NSERC) graduate scholarship to KC, and by the NSERC Discovery grants to CJM, DGB, and MG and the Polar Continental Shelf Program (PCSP) of Natural Resources Canada. This work represents a contribution to the research programs of ArcticNet, the Canada Research Chair to DGB, and the Canada Excellence Research Chair unit at the Centre for Earth Observation Science (CEOS) at the University of Manitoba. We also wish to thank Søren Rysgaard, Kristy Hugill and Matt Gale for their contributions, as well as the two anonymous reviewers who improved the manuscript. References Arrigo, K.R., Mock, T., Lizotte, M., 2010. Primary producers and sea ice, In: Thomas, D.N., Dieckmann, G.S. (Eds.), Sea Ice, 2nd ed. Wiley Blackwell Publishing, Malaysia, pp. 283–325. ACIA, 2005. Arctic Climate Impact Assessment. Cambridge University Press, New York. Barlow, R.G., Gosselin, M., Legendre, L., Therriault, J.-C., Demers, S., Mantoura, R.F.C., Llewellyn, C.A., 1988. Photoadaptative strategies in sea-ice microalgae. Mar. Ecol. Prog. Ser. 45, 145–152.

9

Campbell, K., Mundy, C.J., Barber, D., Gosselin, M., 2014. Response of remotely estimated ice algae biomass to the environmental conditions during spring melt. MS 13-125 in press. Arctic 67. Cota, G., Smith, R.E.H., 1991. Ecology of bottom ice algae: III. Comparative physiology. J. Mar. Syst. 2, 297–315. Ehn, J., Papakyriakou, T., Barber, D., 2008. Inference of optical properties from radiation profiles within melting landfast sea ice. J. Geophys. Res. 113, C09024. http:// dx.doi.org/10.1029/2007JC004656. Falkowski, P.G., Raven, J.A., 2007. Aquatic Photosynthesis, 2nd ed. Princeton Univ. Press, Princeton, New Jersey. Fequet, D., Hache, L., McCourt, S., Langlois, D., Dicaire, C., Premont, B., Jolicoeur, A., Minano, A., 2011. Sea Ice Climatic Atlas for the Northern Canadian waters 1981–2010. Environment Canada. http://www.ec.gc.ca/glaces-ice/default.asp?lang=En&n= 4B35305B-1 (Accessed 21 July 2012). Fortier, M., Fortier, L., Michel, C., Legendre, L., 2002. Climatic and biological forcing of the vertical flux of biogenic particles under seasonal Arctic sea ice. Mar. Ecol. Prog. Ser. 225, 1–16. Gosselin, M., Legendre, L., Demers, S., Ingram, R.G., 1985. Responses of sea-ice microalgae to climatic and fortnightly tidal energy inputs (Manitounuk Sound, Hudson Bay). Can. J. Fish. Aquat. Sci. 42, 999–1006. Gosselin, M., Legendre, L., Therriault, J.C., Demers, S., Rochet, M., 1986. Physical control of the horizontal patchiness of sea-ice microalgae. Mar. Ecol. Prog. Ser. 29, 289–298. Gosselin, M., Legendre, L., Therriault, J.C., Demers, S., 1990. Light and nutrient limitation of sea-ice microalgae (Hudson Bay, Canadian Arctic). J. Phycol. 2, 220–232. Gradinger, R., Spindler, M., Henschel, D., 1991. Development of Arctic sea-ice organisms under graded snow cover. Polar Res. 10, 295–307. Granskog, M.A., Kaartokallio, H., Kuosa, H., Thomas, D.N., Ehn, J., Sonninen, E., 2005. Scales of horizontal patchiness in chlorophyll a, chemical and physical properties of landfast sea ice in the Gulf of Finland (Baltic Sea). Polar Biol. 28, 276–283. Holm-Hansen, O., Lorenzen, J., Holmes, R.W., Strickland, J.D., 1965. Fluorometric determination of chlorophyll. ICES J. Mar. Sci. 30, 3–15. Iacozza, J., Barber, D.G., 1999. An examination of the distribution of snow on sea-ice. Atmos. Ocean 37, 21–51. Juhl, A., Krembs, C., 2010. Effects of snow removal and algal photoacclimation on growth and export of ice algae. Polar Biol. 33, 1057–1065. Lavoie, D., Denman, K., Michel, C., 2005. Modeling ice algal growth and decline in a seasonally ice-covered region of the Arctic (Resolute Passage, Canadian Archipelago). J. Geophys. Res. 110, C11009. http://dx.doi.org/10.1029/2005JC002922. Leu, E., Soreide, J.E., Hessen, D.O., Falk-Peterson, S., Berge, J., 2011. Consequences of changing sea-ice cover for primary and secondary producers in the European Arctic shelf seas: timing, quantity, and quality. Prog. Oceanogr. 90, 18–32. Michel, C., Legendre, L., Demers, S., Therriault, J.-C., 1988. Photoadaptation of sea-ice microalgae in springtime: photosynthesis and carboxylating enzymes. Mar. Ecol. Prog. Ser. 50, 177–185. Mundy, C.J., Barber, D.G., Michel, C., 2005. Variability of snow and ice thermal, physical and optical properties pertinent to sea ice biomass during spring. J. Mar. Syst. 58, 107–120. Mundy, C.J., Barber, D.G., Michel, C., Marsden, R.F., 2007a. Linking ice structure and microscale variability of algal biomass in Arctic first-year sea ice using an in situ photographic technique. Polar Biol. 30, 1099–1114. Mundy, C.J., Ehn, J.K., Barber, D.G., Michel, C., 2007b. Influence of snow cover and algae on the spectral dependence of transmitted irradiance through Arctic landfast first-year sea ice. J. Geophys. Res. 112, C03007. http://dx.doi.org/10.1029/2006JC003683. Nozais, C., Gosselin, M., Michel, C., Tita, G., 2001. Abundance, biomass, composition and grazing impact on the sea-ice meiofauna in the North Water, northern Baffin Bay. Mar. Ecol. Prog. Ser. 217, 235–250. Parsons, T.R., Maita, Y., Lalli, C.M., 1984. A Manual of Chemical and Biological Methods for Seawater Analysis. Pergamon Press, New York. Perovich, D., 1996. The optical properties of sea ice. US Army Corps of Engineers. CRREL Monogr. 96–1, 1–23. Riedel, A., Michel, C., Gosselin, M., LeBlanc, B., 2008. Winter–spring dynamics in sea-ice carbon cycling in the coastal Arctic Ocean. J. Mar. Syst. 74, 918–932. Robineau, B., Legendre, L., Kishino, M., Kudoh, S., 1997. Horizontal heterogeneity of microalgal biomass in the first-year sea ice of Saroma-ko Lagoon (Hokkaido, Japan). J. Mar. Syst. 11, 81–91. Robinson, D.H., Arrigo, K.R., Iturriaga, R., Sullivan, C.W., 1995. Microalgal light-harvesting in extreme low-light environments in McMurdo Sound, Antarctica. J. Phycol. 31, 508–520. Różańska, M., Gosselin, M., Poulin, M., Wiktor, J.M., Michel, C., 2009. Influence of environmental factors on the development of bottom ice protist communities during the winter–spring transition. Mar. Ecol. Prog. Ser. 386, 43–59. Sturm, M., Massom, R.A., 2010. Snow and sea ice, In: Thomas, D.N., Dieckmann, G.S. (Eds.), Sea Ice, 2nd ed. Wiley Blackwell Publishing, Malaysia, pp. 154–204. Welch, H., Bergmann, M.A., 1989. Seasonal development of ice algae and its prediction from environmental factors near Resolute, N.W.T, Canada. Can. J. Fish. Aquat. Sci. 46, 1793–1804. Zeebe, R.E., Eicken, H., Robinson, D.H., Wolf-Gladrow, D., Dieckmann, G.S., 1996. Modeling the heating and melting of sea ice through light absorption by microalgae. J. Geophys. Res. 101, 1163–1181.

Please cite this article as: Campbell, K., et al., Characterizing the sea ice algae chlorophyll a–snow depth relationship over Arctic spring melt using transmitted irradiance, J. Mar. Syst. (2014), http://dx.doi.org/10.1016/j.jmarsys.2014.01.008