Increasing picocyanobacteria success in shelf waters contributes to long‐term food web degradation

Continental margins are disproportionally important for global primary production, fisheries and CO2 uptake. However, across the Northeast Atlantic shelves, there has been an ongoing summertime decline of key biota—large diatoms, dinoflagellates and copepods—that traditionally fuel higher tropic levels such as fish, sea birds and marine mammals. Here, we combine multiple time series with in situ process studies to link these declines to summer nutrient stress and increasing proportions of picophytoplankton that can comprise up to 90% of the combined pico‐ and nanophytoplankton biomass in coastal areas. Among the pico‐fraction, it is the cyanobacterium Synechococcus that flourishes when iron and nitrogen resupply to surface waters are diminished. Our field data show how traits beyond small size give Synechococcus a competitive edge over pico‐ and nanoeukaryotes. Key is their ability to grow at low irradiances near the nutricline, which is aided by their superior light‐harvesting system and high affinity to iron. However, minute size and lack of essential biomolecules (e.g. omega‐3 polyunsaturated fatty acids and sterols) render Synechococcus poor primary producers to sustain shelf sea food webs efficiently. The combination of earlier spring blooms and lower summer food quantity and quality creates an increasing period of suboptimal feeding conditions for zooplankton at a time of year when their metabolic demand is highest. We suggest that this nutrition‐related mismatch has contributed to the widespread, ~50% decline in summer copepod abundance we observe over the last 60 years. With Synechococcus clades being prominent from the tropics to the Arctic and their abundances increasing worldwide, our study informs projections of future food web dynamics in coastal and shelf areas where droughts and stratification lead to increasing nutrient starvation of surface waters.


| INTRODUC TI ON
The global importance of continental margins arises from sustained nutrient supply via river discharge, dust deposition and exchange with shallow sediments. The high nutrient availability favours the 'classical pelagic food chain' with large primary producers, intensive phytoplankton blooms, abundant mesozooplankton (e.g. copepods) and efficient carbon transfer to higher trophic levels such as fish, seabirds and marine mammals (Kiørboe, 2008). About 80% of the world's wild-captured seafood derives from coastal and shelf seas (Watson, Green, Tracey, Farmery, & Pitcher, 2016) and marine lipids and proteins are in high demand for uses in human consumption, aquaculture, agriculture and health (Chassot et al., 2010;Greenberg, 2018).
However, there are indications that with climate change, the productivity of shelf areas is declining. The well-studied North Sea (Northeast Atlantic) represents a prime example. Here, reductions in euphausiid and copepod abundance and copepod size have been recorded since the 1980s, with negative implications for fish stock recruitment (Beaugrand, Brander, Lindley, Souissi, & Reid, 2003;Capuzzo et al., 2017).
Two mechanisms have been proposed for these changes in zooplankton abundance. On the one hand are direct effects of rising temperature (Beaugrand, Reid, Ibanez, Lindley, & Edwards, 2002) and on the other hand are indirect effects of warming and deeutrophication that act via the overall food availability (Capuzzo et al., 2017). As a direct response to rising temperatures, many terrestrial and marine species shift their biogeographical range towards the poles (Parmesan & Yohe, 2003). North Atlantic warm-water copepod species have moved polewards by up to 260 km per decade between 1958 and 1999, and replaced cool-water copepods in the North Sea (Beaugrand et al., 2002). As the warm-water copepod assemblages have typically lower biomass and contain smaller species, changes in the food web of the North Sea have been linked to these shifts in species distribution (Beaugrand et al., 2003).
In addition to direct physiological effects, rising temperatures can also have indirect effects on food web relationships. Larger heat absorption of the ocean strengthens surface stratification, which hinders nutrient exchange with deeper water and can lead to nutrient limitation of phytoplankton and decreasing net primary production (Bopp et al., 2013). For the North Sea, such a decline in primary production over the last 25 years has been linked to reductions in zooplankton abundance and fish stock recruitment (Capuzzo et al., 2017). A co-occurring shift in the spring bloom phenology, with an earlier formation and termination of the bloom (Desmit et al., 2020;Friedland et al., 2018), is in line with an earlier annual start of water column stratification due to climate change (Holt, Wakelin, Lowe, & Tinker, 2010).
However, nutrient limitation affects not only net primary production but also the phytoplankton community structure.
A key prognosis of climate warming effects, the shift towards smaller primary producers, is backed up by geological records (Finkel et al., 2007), modelling approaches (Dutkiewicz, Scott, & Follows, 2013), in situ monitoring (Agirbas et al., 2015) and satellite observations (Brewin et al., 2012). Advantages of decreased cell size are a thinning of the cell boundary layer and increase in nutrient diffusion per unit of cell volume (Raven, 1998), which rapidly decreases the nutrient concentration required for saturated growth rates (Chrisholm, 1992). Thus, growth of picophytoplankton is saturated at ambient nitrate concentrations of >0.2 µM, while maximum growth rates of diatoms require 0.7-1.0 µM nitrate (Agawin, Duarte, & Agustí, 2000). In stratified shelf seas of the North East Atlantic, nitrate concentrations <0.2 µM and iron concentrations <0.2 nM are not exceptional (Birchill et al., 2017(Birchill et al., , 2019, indicating that species' efficient nutrient uptake and/or storage will be key. However, picophytoplankton is too small to be grazed by copepods (Kiørboe, 2008) and non-diatom phyla often lack essential biomolecules such as omega-3 polyunsaturated fatty acids and sterols (Jónasdóttir, 2019;Ruess & Müller-Navarra, 2019). Therefore, the efficiency and quality of shelf sea food webs in a warming climate will depend on direct temperature effects (Beaugrand et al., 2003), on total primary production (Capuzzo et al., 2017) and also on the size and taxonomic composition of the dominant primary producers (Schmidt, Kähler, & von Bodungen, 1998).
To understand how climate change may impact the community structure at the base of the food web in shelf seas, we approached the problem at two scales. Large temporal and spatial resolution, as provided by satellite earth observations and the Continuous Plankton Recorder (CPR) survey, allowed us to identify (a) differences between longer term (60 years) and shorter term (20 year) changes, (b) the spatial extent of changes and (c) the months that are most affected. At the smaller scale, we use intensive field observations of the phytoplankton community along a coast-shelf-shelf break gradient of nutrient supply as a 'natural experiment' to reveal the mechanisms that drive the success of the various primary producers. By linking the two scales, we provide a conceptual model of why the classical food web is increasingly under threat in temperate coastal and shelf areas.

| Continuous Plankton Recorder survey
The CPR survey has operated in the North Atlantic since 1931 and is currently managed by the Marine Biological Association (Plymouth).
The collection and analysis of CPR samples have followed a consistent methodological approach since 1958, with details given in Richardson et al. (2006). In brief, the CPR collects samples using a high-speed plankton recorder that is towed behind 'ships of opportunity' through the surface layer of the ocean (~10 m depth). Water passes through the recorder, and plankton are filtered by a slow moving silk layer (mesh size 270 µm). A second layer of silk covers the first and both are reeled into a tank containing 4% formaldehyde, and thereby preserved for later analysis.
We collated a 60 year time series  from 16 CPR subareas in the NE Atlantic, covering an area of ~2,000 × 1,500 km ( Figure 1a) and based on 136,903 samples. Our data set comprises only species/genera that have continuously been identified since 1958, including 65 diatoms, 53 dinoflagellates and the seven most abundant copepods. In the case of the Calanus spp., we distinguished between early Calanus copepodites (CI-IV; no species level) and late copepodites (CV/adults of Calanus finmarchicus and Calanus helgolandicus). The smaller copepods were only identified to genus level (Oithona spp., Acartia spp., Paracalanus spp., Pseudocalanus spp.) or the species were merged for analysis (Centropages hamatus and Centropages typicus), with the exception of Temora longicornis.
DOIs of the individual data sets are provided in the References section (Johns, 2019).

F I G U R E 1 Summer reductions in Chl
a, diatoms, dinoflagellates and copepods, but an increasing picophytoplankton fraction across temperate shelf areas of the NE Atlantic. Results from Continuous Plankton Recorder (CPR) data  and satellite-derived ocean colour data . (a) Picophytoplankton fraction of Chl a during summer (July), climatology 1998-2018. The grid shows the 16 subareas that were considered for our data analysis. (b-i) Slope values for the regression between monthly median abundance over the last 20 years (i.e. =satellite era) and over the last 60 years (i.e. =CPR era). The vertical boxes and error bars indicate the median, 10th, 25th, 75th and 90th percentiles of the data (n = 16). (b) Chl a concentrations, (c) picophytoplankton fraction of Chl a, (d) diatoms , (e) diatoms ), (f) dinoflagellates , (g) dinoflagellates , (h) copepods , (i) copepods . The late spring to summer period, where most of the changes occur, is highlighted in yellow. For this period (May-August), nonparametric Wilcoxon signed rank tests showed that slope values were overall significantly different from zero.

| Satellite data
Chlorophyll data were extracted for the same 16 subareas of the NE Atlantic as the CPR survey. The regional monthly chlorophyll climatol-

| Sampling at the coastal monitoring site in the English Channel
The Samples for flow cytometry analysis were collected from 2, 10, 25 and 50 m depth using 10 L Niskin bottles, returned to the laboratory within 3 hr and processed immediately. Samples for the enumeration of phytoplankton (≥10 µm) were collected at 10 m depth. The Tamar River is the main source of freshwater inflow to the Western English Channel, with a range of 5-140 m 3 /s at its mouth. Daily mean flow data for the Tamar River at Gunnislake were kindly provided by the National River Flow Archive (2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017) and the UK Environment Agency (2018).

| Sampling in the Celtic Sea
Sampling of 20 stations across the inner shelf, outer shelf and shelf break took place during two cruises in 2015 (April, July) on board the R.R.S. Discovery. An overview of the stations and sampled parameters is given in Table S1. Vertical profiles of temperature, salinity, depth and Chl a were measured using a CTD system (SeaBird 9/11+).
Seawater for macronutrient analyses was collected from Niskin bottles attached to the CTD/Rosette system, and taken into acid clean, 'aged', 60 ml HDPE (Nalgene) sample bottles. Samples were kept cool and in the dark until analysis as soon as possible after arrival.
Seawater for trace metal analyses was collected from Niskin bottles attached to the titanium CTD/Rosette system and taken into acid clean LDPE (Nalgene) sample bottles (full details in Birchill et al., 2017). Clean sampling and handling techniques were used. For vertical profiles of phytoplankton <20 µm analysed by flow cytometry, seawater was sampled at 5-13 depth intervals between 2 and 100 m using a rosette of Niskin bottles mounted on a CTD system.
The samples were stored at 4°C in the dark until analysis within 2 hr.
At a reduced number of stations and depth horizons, additional identification of phytoplankton was carried out via light microscopy.

| Macronutrient analysis
The micromolar nutrient analysis was carried out using a SEAL Analytical 5-channel (nitrate, nitrite, phosphate, silicate, ammonium) AAIII segmented flow, colorimetric, autoanalyser, using classical proven analytical techniques (Woodward & Rees, 2001). Detection limits for nitrate, nitrite and phosphate were 0.02 µmol/L, and for ammonium was 0.05 µmol/L. The concentration of silicate was always within the detection limit of the analyser. The accuracy of the measurements was 1%-2%. Nitrate and nitrite concentrations were combined and presented as 'nitrate'.

| Flow cytometry analysis
Phytoplankton smaller than ~20 µm were analysed by flow cytom-  Tarran and Bruun (2015). The median cell volume of each of these categories was calculated from median diameter measurements by size-fractionating seawater samples through successive polycarbonate membrane filters (Poreticsis ® ) from 10 µm down to 0.2 µm. A carbon conversion factor of 0.22 pg C/µm 3 was used (Booth, 1988).

| Light microscopic analysis of phytoplankton
Samples for the enumeration of phytoplankton ≥10 µm were immediately fixed with 2% Lugol's iodine and stored in cool, dark conditions. Taxonomic analysis using light microscopy followed the British  (2011,2015,2018) and years with low proportions of picophytoplankton . For the environmental parameters, the mean value during the stratified period (May-August) is represented, while ratios between picocyanobacteria (Synechococcus, SYN) and picoeukaryotes (PEUK) or nanoeukaryotes (NEUK) are based on mean values in August and September, when picophytoplankton is most prominent at this site. Sampling depths: pico-and nanophytoplankton (10 m and European standard protocol (BS EN 15204:2006). Mean cell measurements of individual taxa were used to calculate cell biovolume (Olenina et al., 2006) and converted to carbon (pg C/cell) using the equations of Menden-Deuer and Lessard (2000).

| Combining satellite and CPR time series
To provide the large-scale spatiotemporal context of change, we considered an area of ~2,000 × 1,500 km in the NE Atlantic, centred on shelf regions adjacent to the UK and divided into 16 subareas ( Figure 1a). In combination, both satellite and CPR data show similar changes over the longer term  and recently (1997-2017/2018). Between May and August/September, our indices of Chl a, diatoms, dinoflagellates and total copepods have all declined, while the proportion of picophytoplankton to total Chl a has increased (Figure 1b-i). Unifying features of the trends in both phytoplankton and copepods are first, that they are primarily a summer phenomenon, and second, that they are largely consistent across the 16 subareas that we tested (Figure 1), even though the subareas include shelf and oceanic areas and summer surface temperature varies from ~9°C in the north to ~19°C in the south (Locarnini et al., 2010). The total copepod abundance shows a ~50% decline when comparing median summer values across all 16 subareas in recent years  with those from earlier years since the beginning of the time series (1958-1996; Figure S1). Despite this overall decline in copepod abundance, some species have increased ( Figure S1). These are the more carnivorous genera that select for motile food (e.g. Centropages spp., Djeghri et al., 2018), whereas more herbivorous genera with a preference for diatoms (e.g. Oithona spp., Para-and Pseudocalanus spp., Djeghri et al., 2018) experienced the strongest decline. The two Calanus congeners show opposite trends with the cold-water C. finmarchicus being largely replaced by the warm-water C. helgolandicus ( Figure S1).
However, the abundance of the combined offspring of these two species, Calanus copepodites I-IV, declined by >60%, in line with the overall loss of copepods across the study area ( Figure S1).   Figure S2). In line with these regional differences in summer nutrient resupply, Synechococcus reached abundances of 2 × 10 5 cells/ml on the nutrient-starved shelf, while at the shelf break, abundances were an order of magnitude lower even though temperatures were higher (Figure 3b,d). Instead, the shelf break was characterized by higher abundances of diatoms, indicated by cell counts (Table S2) and enhanced NO 3 :Si(OH) 4 ratios in subsurface waters (Figure 3c).

| Synechococcus abundances along a gradient of summer nutrient stress
Depth-integrated Chl a concentrations were about twice as high at the shelf break compared to the shelf (Figure 3b), confirming the overall higher productivity of this region (Sharples et al., 2009).

| Traits that allow Synechococcus to outcompete picoeukaryotes
During our studies, Synechococcus and picoeukaryotes were of nearly identical size both at the monitoring site L4 (1.7 vs. 1.8 µm) and in the Celtic Sea (1.3 vs. 1.4 µm; Table S3). This allows us to examine traits, other than small size, that give Synechococcus a the Synechococcus-to-nanoeukaryotes ratio (Figure 4a; Figure S4).
Across the Celtic Sea shelf break and shelf, Synechococcus-topicoeukaryote ratios rose from 0.7 to 15 as the surface water became increasingly stratified (Figure 3b,e), which reflects their superior abilities to outcompete picoeukaryotes under nutrient shortage.
Based on our data, two traits can be identified that likely contribute to Synechococcus' success in nutrient-starved waters. First, Synechococcus cope better with low light levels than co-occurring pico-or nanoeukaryotes. This is seen in the enhanced Synechococcusto-picoeukaryote and Synechococcus-to-nanoeukaryotes ratios in deeper water (25-80 m) at L4 and the Celtic Sea (Figure 4a,b).
Another prerequisite for photosynthesis at low irradiance is a high affinity to iron, indicated by the major increase in Fe:N and Fe:P ratios of phytoplankton when grown under low irradiance ( Figure S5 adapted from Finkel et al., 2006). In the Celtic Sea, enhanced macronutrient-to-iron ratios [NO 3 :dFe, Si(OH) 4 :dFe, PO 4 :dFe] occur in subsurface waters of ~20 to 60 m depth where Synechococcus account for up to 80% of the combined pico-and nanophytoplankton volume (Figure 4b). This implies that Synechococcus is efficient at iron uptake.

| The Synechococcus 'strategy': Structural investment for sustained resource acquisition
Our study shows that Synechococcus can best outcompete pico-and nanoeukaryotes in subsurface waters where light levels are as low as 2% surface irradiance (Figure 4a,b; Figure S6). Occupying these darker waters enhances the chances of nutrient uptake near the pycnocline; however, it requires efficient harvesting of both light and iron to sustain photosynthesis (Finkel et al., 2006).
Compared to the dominant picoeukaryote in the study area, the prasinophyte Micromonas pusilla (Not et al., 2004), Synechococcus' suite of light-harvesting pigments is superior for light absorption in subsurface coastal waters ( Figure 5). Synechococcus populations from the Western English Channel are capable of Type IV chromatic acclimation (Humily et al., 2014), which means that they can change the ratio of blue light-absorbing phycourobilin (A max = 495 nm) versus green light-absorbing phycoerythrobilin (A max = 550 nm) depending on the ambient light colour (Grébert et al., 2018). In contrast, chlorophyll b, the light-harvesting pigment of M. pusilla has its maximum absorption in darker blue light (A max = 480 nm; Kunugi et al., 2016), which is less suitable in subsurface coastal waters where green light prevails ( Figure 5).
Other studies have shown that, in contrast to their oceanic counterparts, coastal Synechococcus are well adapted to fluctuating iron concentrations (Mackey et al., 2015;Palenik et al., 2006). Their ability to use or even produce strong iron chelators (e.g. siderophores; Hutchins, Witter, Butler, & Luther, 1999;Ito & Butler, 2005;Wilhelm & Trick, 1994) and their enhanced iron storage capacities (Mackey et al., 2015) may give Synechococcus a competitive edge over picoeukaryotes ( Figure 5). Laboratory experiments have shown that iron uptake rates of the siderophore-producing coastal Atlantic isolate Synechococcus PCC7002 are ~3 times higher than those of a non-siderophore-producing oceanic strain (Lis, Kranzler, Keren, & Shaked, 2015). However, it is currently unknown how widespread siderophore production is among coastal picocyanobacteria (Hopkinson & Morel, 2009).
On the downside for Synechococcus, both the phycobi lisomebased light-harvesting system and siderophore transport systems are costly cell structures (Lis et al., 2015;Ting Rocap, King, & Chisholm, 2002), which may explain the exceptionally high nitrogen demand and relatively low growth rates of coastal Synechococcus ( Figure 5). For a given carbon content, the Synechococcus PCC7002 strain contains 64% more nitrogen than M. pusilla (Blanco-Ameijeiras et al., 2018;Maat, Crawfurd, Timmermans, & Brusaard, 2014). In line with this, the protein investment for the phycobilisome-based light-harvesting system of Synechococcus is >3 times higher than for the chlorophyll a/b-based antennae of green algae (Ting, Rocap, King, & Chisholm, 2002). A less demanding structural composition may enable M. pusilla to grow about twice as fast as Synechococcus under nutrient replete conditions (Marañón et al., 2013).
Slow grow rates make Synechococcus vulnerable to grazing control. In the laboratory, Synechococcus growth shows a positive relationship to temperature (Agawin, Duarte, & Agustí, 1998), but at L4, their abundances reach the annual minimum not in the coldest month February, but in May when their mixotrophic grazers (Tsai, Chiang, Chan, Lin, & Chang, 2007) find ideal growth conditions. Similar size picoeukaryotes seem to be less affected by grazing, as the low Synechococcus-to-picoeukaryote ratios in May indicate.
Likewise, at the nutrient-replete shelf break of the Celtic Sea, Synechococcus remained under grazing control during summer, even though temperatures were higher than on the shelf.
Overall, we suggest that the strategy of coastal Synechococcus consists of high structural investments for sustained resource acquisition (light, nutrients), with the trade-off of low growth rates and high vulnerability to grazing. This strategy pays off when competing pico-and nanoeukaryotes are severely nutrient limited and their grazing pressure eases.

| Synechococcus are extending their global distribution
Picocyanobacteria of the genus Synechococcus are among the most important and widespread marine primary producers (Flombaum et al., 2013). This is enabled by the existence of >20 genetically distinct clades, some of which inhabit cold, mesotrophic waters, others warm, oligotrophic open ocean or sites with permanently low iron availability (Paulsen et al., 2016;Sohm et al., 2016). Some strains are specifically adapted to the dynamic coastal environments, for example, via their pigment types and capacity to sense and respond to changes in iron availability (Grébert et al., 2018;Mackey et al., 2015;Palenik et al., 2006).

Niche models project a 14% increase in global cell numbers of
Synechococcus by the end of the 21st century, which involves their spatial expansions (Flombaum et al., 2013). While all four of their model projections agree that these increases will occur in low latitudes, some models project also up to 50% increase in Synechococcus numbers for mid-and higher latitudes (40-60°N; Flombaum et al., 2013). Our study shows that, already in the present day, Synechococcus can dominate the phytoplankton biomass in temperate shelf and coastal areas if certain conditions are met. Crucial are not high temperatures per se, but rather a reduced resupply of nutrients to surface waters. We identify intense stratification and low river discharge as conditions that can lead to this nutrient starvation. Both macronutrients and dissolved iron are affected by this lack of resupply, with iron stress only recently recognized to extend onto the NW European shelf (Birchill et al., 2017(Birchill et al., , 2019. Our process studies suggest that Synechococcus is highly competitive under nitrogen-iron colimitation, a state of nutrient limitation that is pervasive throughout ~50% of the global surface ocean (Browning et al., 2017). is complexed by naturally occurring organic ligands (L). These ligands can facilitate or impede iron uptake by phytoplankton depending on their identity: L 1 -strong ligands (e.g. siderophores), L 2 -moderate ligands (e.g. humics, biological degradation/excretion products), L 3 , L 4 -weak ligands (e.g. saccharides). Prokaryotes and eukaryotes seem to differ in their efficiency to access these various iron chelates 5 . Compared to other phytoplankton species, Synechococcus has a high nitrogen demand 6-8 and low growth rates. 9 Like all cyanobacteria, Synechococcus lacks PUFA 1 , which are an essential diet component for higher trophic levels. The given C:N ratios derived from nutrient replete cultures. The protein investment is normalized to one tetrapyrrole of the light-harvesting system (LSH). 1. Six et al. (2007). 2. Kunugi et al. (2016). 3. Humily et al. (2014). 4. Not et al. (2004). 5. Hutchins et al. (1999). 6. Geider and La Roche (2002). 7. Bertilsson, Berglund, Karl, & Chisholm (2003). 8. Blanco-Ameijeiras et al. (2018). 9. Marañón et al. (2013). 10. Patil et al. (2007). 11. Grébert et al. (2018). 12. Ting et al. (2002). 13. Lis et al. (2015). 14. Ito and Butler (2005). 15. Mackey et al. (2015). 16. Palenik et al. (2006 suggest that these conditions will increasingly benefit the advance of Synechococcus into shelf areas.

| Implications for energy flow through pelagic food webs
Zooplankton such as copepods are considered beacons of climate change (Richardson, 2008) and can be more sensitive indicators than the environmental variables themselves, due to their nonlinear responses that can amplify subtle environmental changes (Taylor, Allen, & Clark, 2002). Moreover, it has been suggested that trophic amplification can lead to significantly larger changes in fishery resources than implied by net primary production changes alone (Chust et al., 2014). In this respect, we have to interpret the overall ~50% decline in copepod abundance over the last 60 years ( Figure S1) and the large spatial extent of this decline (Figure 1 Biogeographical and phenological shifts due to increasing temperature are known to be occurring in the north Atlantic (Beaugrand et al., 2002). However, alternative (non-mutually incompatible) mechanisms have also been proposed, namely declining primary production as a cause of the copepod decline in the North Sea (Capuzzo et al., 2017).
Based on an extensive network of time series, Bedford et al. (2020) show that this long-term copepod decline extends from inshore to offshore areas right across the NW European shelf. Our study spans an even larger area and by detailing the species composition and seasonal timing of the copepod decline, we shed light on the potential causes. First, the copepod decline is observed across the whole NE Atlantic and fringing shelves; too large an area to be solely explainable by the observed magnitude of range shifts (Chivers, Walne, & Hays, 2017). Second, it occurs for both large and small copepod species and across those with different temperature preferences. Third, it is a summer phenomenon, at the time when the food composition has changed the most (Figure 1). We conclude that, in addition to the direct effects of rising temperature, significant changes are occurring in the food environment of these copepods.
Even though Synechococcus can reach near-bloom concentrations in subsurface layers (Chl a ~ 1 mg/m 3 , Figure 3b), this biomass is of limited use for copepods as they cannot capture cells <5 µm efficiently (Kiørboe, 2008). Moreover, like all cyanobacteria, Synechococcus lack biomolecules such as omega-3 polyunsaturated fatty acids and sterols (Jónasdóttir, 2019;Patil, Källqvist, Olsen, Vogt, & Gislrød, 2007;Ruess & Müller-Navarra, 2019). These compounds are essential for both copepods and fish to acquire from their diet in order to sustain egg production and growth. Thus, under summer conditions, the 'classical food chain' from diatoms or dinoflagellates via copepods to fish is severely reduced, giving way to the 'microbial food web' from Synechococcus via flagellates and ciliates to copepods and fish. However, this microbial pathway is less efficient, both because of the increased number of trophic steps and the lower nutritional quality of the main primary producer ( Figure S7).
While we highlight the role of diminishing food quality, other factors contribute to summer becoming an increasingly stressful period of the year for pelagic consumers. These include a spring bloom that occurs earlier in the year (here indicated by a Chl a increase in April and decrease in May, Figure 1b) and the overall decrease in phytoplankton biomass in summer ( Figure 1b). Together, these all lengthen the summer period of low food quality and quantity for zooplankton.
Winter food shortage is easier manageable due to low temperatures and therefore low basal metabolism, but respiration costs are high in summer and prolonged lack of food at this time of year has more adverse effects. Such a nutrition-related mismatch during summer will increasingly challenge temperate shelf and coastal areas to maintain their historically important food chains.

ACK N OWLED G EM ENTS
This study has been conducted using E.U. Copernicus Marine Service The authors have no conflicts of interest to declare.