Chlorophyll fluorescence-based estimates of photosynthetic electron transport in Arctic phytoplankton assemblages

We employed Fast Repetition Rate fluorometry for high-resolution mapping of marine phytoplankton photophysiology and primary productivity in the Lancaster Sound and Barrow Strait regions of the Canadian Arctic Archipelago in the summer of 2019. Continuous ship-board analysis of chlorophyll a variable fluorescence demonstrated relatively low photochemical efficiency over most of the cruise-track, with the exception of localized regions within Barrow Strait where there was increased vertical mixing and proximity to land-based nutrient sources. Along the full transect, we observed strong non-photochemical quenching of chlorophyll fluorescence, with relaxation times longer than the 5-minute period used for dark acclimation. Such long-term quenching effects complicate continuous underway acquisition of fluorescence amplitude-based estimates of photosynthetic electron transport rates, which rely on dark acclimation of samples. As an alternative, we employed a new algorithm to derive electron transport rates based on analysis of fluorescence relaxation kinetics, which does not require dark acclimation. Direct comparison of kinetics- and amplitude-based electron transport rate measurements demonstrated kinetic-based estimates were, on average, 2-fold higher than amplitude-based values. The magnitude of decoupling between the two electron transport rate estimates increased in association with photophysiological diagnostics of nutrient stress. Discrepancies between electron transport rate estimates likely resulted from the use of different photophysiological parameters to derive the kinetics- and amplitude-based algorithms, and choice of numerical model used to fit variable fluorescence curves and analyze fluorescence kinetics under actinic light. Our results highlight environmental and methodological influences on fluorescence-based productivity estimates, and prompt discussion of best-practices for future underway fluorescence-based efforts to monitor phytoplankton photosynthesis.


49
Phytoplankton productivity in polar marine waters is constrained by nutrient and light 50 availability, which fluctuate dramatically across seasonal cycles and shorter time and space 51 scales [1]. In late summer, when sea ice cover is at a minimum and the mixed layer is shallow 52 and highly stratified, phytoplankton are exposed to high solar irradiances and low nutrient 53 concentrations [2]. Under these conditions, the growth and photosynthetic efficiency of Arctic 54 phytoplankton becomes nitrogen-limited [3]- [5]. However, localized regions of elevated 55 productivity can persist where various processes transport nutrients into the mixed layer, 56 including upwelling, tidal mixing, and freshwater input from rivers and glaciers [6]- [8]. 57 Modelling studies suggest that Barrow Strait in the Canadian Arctic Archipelago is one such 58 productivity hotspot, with strong tidal currents and shallow sills driving vertical mixing in a 59 region where Pacific and Atlantic-derived water masses converge [9], [10]. Additionally, Barrow 60 Strait receives glacial and land-derived nutrients from the Cornwallis and Devon Island rivers 61 [11], [12]. Rapid climate change in the Arctic is expected to have complex effects on these 62 nutrient delivery mechanisms through the intensification of coastal erosion [13], increasing river 63 inputs [14] and reduced vertical mixing due to intensifying stratification [15], [16]. At present, it 64 is unclear how phytoplankton productivity will respond to these anticipated perturbations. 65 Assessing phytoplankton productivity in physically-dynamic marine waters requires high spatial 67 resolution measurements that cannot be obtained from traditional discrete bottle incubation 68 methods, such as 14 C uptake experiments. For this reason, oceanographic field studies have 69 increasingly employed continuous sampling of surface water properties using variable 70 chlorophyll a (Chla) fluorescence from Fast Repetition Rate Fluorometry (FRRf) and other 71 related methods to rapidly and autonomously assess phytoplankton photophysiology and 72 productivity (e.g. [17]- [21]). Such variable fluorescence techniques rely on the inverse 73 relationship between Chla fluorescence and photochemistry. These processes, along with heat 74 dissipation, represent the three energy dissipation pathways of absorbed light energy within 75 Photosystem II (PSII) [22]. Fast Repetition Rate Fluorometry operates by supplying rapid 76 excitation light pulses to progressively saturate the photosynthetic pathway and simultaneously 77 induce a measurable Chla fluorescence response -often referred to as a fluorescence transient 78 same principles of light harvesting and photosynthetic electron transport, but arrive at ETRPSII 88 estimates using slightly different, but theoretically equivalent, combinations of 89 photophysiological metrics. As a result, different algorithms confer different field-sampling 90 advantages and challenges (see [31]). The so-called 'amplitude' based approach (abbreviated 91 ETRa; sometimes also referred to as the sigma-algorithm), calculates ETRPSII as the product of 92 photosynthetically available radiation (PAR), the functional absorption cross section of PSII in 93 the dark-acclimated state (σPSII), and the photochemical efficiency of PSII normalized by the 94 dark-acclimated maximum photochemical efficiency of PSII [27]. This approach reduces 95 uncertainty in ETRPSII estimates by using σPSII measurements made in the dark-acclimated state, 96 which are subject to less noise than σPSII' measurements made in the light [ phytoplankton photophysiological stress. We relate the spatial patterns in our observations to 142 regional and fine-scale patterns in hydrography and nutrient supply in the eastern Canadian 143 Arctic Archipelago, and discuss the potential effects of different data analysis approaches on 144 ETRPSII-based productivity estimates. Results from our work will inform future ship-based 145 deployments of FRRf and related techniques to understand spatial patterns in phytoplankton 146 productivity. saturation, ΔO2/Ar, was measured as a metric of net community production using a membrane 170 inlet mass spectrometer (MIMS) following the approaches outlined by Tortell et al. [39], [40]. 171 These gas measurements were made continuously on seawater obtained from the same underway 172 lines that supplied the FRRf system. Briefly, seawater was circulated at a constant flow rate past 173 the mass spectrometer's inlet cuvette consisting of a 0.18 mm thick silicone membrane. 174 Measurements of the mass-to-charge ratios at 32 (O2) and 40 (Ar) atomic mass units were 175 obtained at approximately 20 s. intervals. Air standards, consisting of filtered seawater (<0.2 µm) 176 incubated at ambient sea surface temperature and gently bubbled using an aquarium air pump, 177 were run periodically by automatically switching the inflow water source every 45 -90 minutes.
Here, PAR (μmol quanta m -2 s -1 ) is the total actinic light provided by the FRRf LEDs, σPSII (Å 2 254 quanta -1 ) is the PSII functional absorption cross section measured in dark acclimated samples, 255 and Fqʹ/Fmʹ divided by Fv/Fm (dimensionless) is the PSII photochemical efficiency measured 256 under actinic light normalized by the dark-measured maximum photochemical efficiency. The 257 constant 6.022*10 -3 converts σPSII units from Å 2 RCII -1 to m 2 RCII -1 and PAR from mol quanta 258 to quanta. optimal signal-to-noise ratios were re-analyzed using a 3-component multi-exponential model to 262 describe Qa reoxidation kinetics. This numerical procedure was applied as a fitting option in the 263 (3) 268 ETRk was then derived at each light step as 289 Here, PARmax is a super-saturating light level chosen as a value three-fold higher than the light 293 saturation parameter, Ek, derived from photosynthesis irradiance curves.
is the PSII 294 photochemical efficiency measured under PARmax. The photosynthetic turnover rate, We quantified NPQ as a measure of the relative increase in heat dissipation of absorbed energy 300 by PSII between samples exposed to low light and 150 m 2 s -1 , following Bilger 301 and Bjorkman [43] as: Pair-wise statistical relationships between measured variables were determined using Spearman 313

Rank correlation tests. Samples from two regions of our transect (Lancaster Sound and Barrow 314
Strait) were compared using Kruskal-Wallis test (Fig 8). Lilliefors test rejected the null 315 hypothesis that underway data were normally distributed, so we report median rather than mean 316 values of all photophysiological variables. Deviation from the median was determined as the 317 median absolute deviation. Continuously measured, in-situ parameters such as temperature and salinity were not corrected for autocorrelation. All analyses were completed using Matlab 319 (R2020a). conditions. Small-scale features in ΔO2/Ar distributions were observed across hydrographic 342 frontal regions with only weak correlation to salinity or temperature measured along the cruise 343 track (Table 2). 344

362
The photophysiological parameters Fv/Fm and σPSII fluctuated over diel cycles throughout our 363 sampling period (Fig 3a-b), showing daily maxima during the night, and decreasing during 364 daylight hours. Both of these variables exhibited a significant negative correlation with actinic 365 surface PAR intensity averaged over the 5 min window prior to sample measurements (r = -0.50 and r = -0.72, respectively, p < 0.001 and n = 481 for both ; Fig 4a- Fmʹ)/Fm, were negatively correlated to surface PAR (Fig 4c). This surprising result can be 390 explained by the derivation used for here for NPQ, which measures the fractional change in NPQ 391 between light and dark (low light) measurements, rather than total NPQ. As a result, this 392 approach does not account for any residual quenching present in samples after five minutes of 393 low light acclimation. To address this limitation, we used the ratio (F ʹq/F ʹm)/(Fv/Fm) to estimate 394 the extent of residual quenching present in samples after five minutes of NPQ relaxation. As expected, this derived variable was well correlated to surface PAR (Fig 4d, r = 0 photosynthetic potential in this region (Fig 3b). In contrast, night-time σPSII did not significantly 409 vary between Lancaster Sound (250 ± 17.2) and Barrow Strait (241.1 ± 15.3) (Fig 3a). These 410 absolute σPSII values are somewhat lower than those reported in previous studies, likely reflecting 411 our use of simultaneous excitation flashlets centered around 445, 470, 505, 535, and 590 nm. 412 Relative to blue light, not all of these wavelengths are efficiently absorbed by phytoplankton, 413 resulting in an apparent decrease σPSII [48]. As discussed below, σPSII values are also subject to 414 physiological, taxonomic and environmental effects [49].  432 We used light response curves to compare FRRf-based ETRa and ETRk estimates. In this 433 approach, ETRa (Eq. 1) was plotted against actinic irradiance (Fig 6) to derive the maximum rate 434 of charge separation at RCII (ETRmax), the light-dependent increase in the charge separation rates 435 ( ), and the saturating light intensity (Ek). Fit parameters from these curves varied considerably, 436

Eq. 5), and compared with ETRa values. This comparison revealed a strong correlation between 447
the ETR values (r = 0.81, p < 0.001; Fig 7). However, the kinetics-based algorithm produced 448 consistently higher results than ETRa, with values 1.96 ± 1.2 times greater, on average, than 449 ETRa (Fig 6). The primary focus of our work was to quantify phytoplankton photophysiology and productivity 474 along our ship-track using active Chla fluorescence methods. With this in mind, we applied a 475 sampling and analysis strategy to support both amplitude-based and kinetic analysis of FRRf 476 data to derive ETR estimates. In the following, we first discuss the spatial patterns in FRRf- rivers that discharge into Barrow Strait (Fig 9; [11], [54] ). In this region, we observed low 507 surface water salinity and elevated Fv/Fm, suggesting a link between river input and increased 508 photo-efficiency, which we ascribe to nutrient inputs. Additionally, the greatest mixed layer 509 depths were found at the two CTD profiling stations situated between Cornwallis Island and 510 Devon Island (Table 2) We observed strong residual NPQ effects after five minutes of low light acclimation (Figs. 3, 4). 540 Notably, the extent of these quenching effects was a predictable function of the short-term light 541 history experienced by in situ phytoplankton assemblages (Fig 3d). Previous studies examining 542 the drivers of NPQ variability [58]- [60] suggest that the magnitude of NPQ effects at a given light level is tied to a number of environmental factors (e.g. temperature and CO2 544 concentrations), phytoplankton taxonomy and physiological status. Given these sources of 545 variability, NPQ relaxation times needed for robust Fv/Fm measurements are expected to differ 546 significantly across ocean regimes. In cold waters, such as those encountered along our ship 547 track, NPQ relaxation is slower [61], and this may have contributed to the longer-lived 548 quenching observed in our low-light samples. As the spatial and temporal resolution of Fv/Fm 549 and ETRa measurements are constrained by such acclimation periods, it is recommended that 550 future field deployments of FRRf conduct experiments using natural assemblages to determine 551 the regional minimum relaxation period necessary to achieve steady-state dark-acclimation. This 552 acclimation step can then be incorporated into underway FRRf protocols, resulting in more 553 robust ETRa estimates, albeit with reduced measurement frequency. Such routine determinations 554 of the minimum NPQ relaxation time requirements have not been commonly carried out for 555 marine phytoplankton [25]. As a result, there is little systematic knowledge of the global 556 variability of NPQ relaxation times. Adopting such pre-study tests (or within protocol) as 557 standard practice would improve current understanding of environmental and taxonomic controls 558 on NPQ relaxation kinetics. Moreover, it may be necessary to adjust the length of the dark-559 acclimation period to reflect changing conditions over the duration of a cruise. We thus 560 recommend that future work incorporate semi-regular assessments of dark-acclimation times into 561 field-sampling protocols. 562 563

564
We observed significantly higher ETRk compared to ETRa across our study region (Fig 6), with 566 the magnitude of ETRk and ETRa decoupling varying strongly in response to phytoplankton 567 photophysiological conditions (Fig 7). The ratio between ETRa and ETRk depends on a number derive ETRa and ETRk can lead to discrepancies between these two productivity metrics. 587 Going forward, further investigation of ETRk and ETRa divergence is critical to inform our 589 understanding of electron requirements for carbon assimilation and biomass production, 590 particularly under nutrient limiting conditions [28]. For instance, Schuback et al. [36] found that 591 iron-limited phytoplankton assemblages exhibited elevated ETRa and greater decoupling 592 between ETRa and C-assimilation rates as compared to iron-enriched assemblages. This result 593 was attributed to the higher σPSII values in iron limited samples. Since ETRk does not directly 594 include σPSII, we speculate that C-assimilation will show less decoupling with this productivity 595 metric under low iron conditions. This hypothesis remains to be tested in future studies. irradiance and resulted in a markedly shorter photosynthetic turnover time. Our own FRRf-derived τQa values displayed a weak relationship with applied actinic irradiances (r = 0.17, p < 633 0.01, n = 203). It is possible that applying the FIRe model fit to our own data may have also 634 yielded slower photosynthetic turnover times, and therefore lower ETRk estimates, but it is 635 unclear to what extent the alternative model may have affected our ETRk results and the observed 636 decoupling between ETRk and ETRa. 637 638 Going forward, it will be important to separate physiological drivers of ETRk and ETRa 639 decoupling from offsets resulting from the use of different mathematical approaches to data 640 analysis. Discrepancies between ETRk and ETRa that cannot be explained by different 641 derivations of 9! must be attributable to differences between the two ETRPSII algorithms 642 themselves. This raises the important question of which approach is most accurate, as neither has 643 been established as a "gold standard". Addressing this issue will require parallel independent 644 measurements of PSII activity, such as gross oxygen evolution measurements from 18 O 645 experiments, as preformed previously for ETRa (e.g. [49]) but not ETRk. Such fluorescence-646 independent validations of ETRa and ETRk are critically lacking, and will help elucidate the 647 taxonomic and environmental influences on FRRf-based productivity measurements (Hughes et 648 al., 2018). This, in turn, will be of significant practical utility to FRRf users seeking to derive 649 ship-based primary productivity estimates. 650

652
Fast Repetition Rate Fluorometry offers a means to rapidly assess both physiological status and 653 photosynthetic electron transport rates of phytoplankton. The aim of this study was to evaluate an autonomous protocol for high resolution FRRf measurements of phytoplankton physiology, 655 and to compare two alternative models for deriving primary productivity estimates from FRRf 656 data. Our results demonstrate significant residual NQP effects after five minutes of low light 657 acclimation, suggesting the need for extended low light acclimation periods, which would 658 significantly decrease measurement frequency. In contrast, in-situ surface PAR had no effect on