Samples were collected from the following six locations along the European NE Atlantic coast during summer, from north, in the Arctic, to south, on the coast of Brittany: Spitsbergen (SPT), Tromsø (TRM), Bodø (BOD), Helgoland (HLG), Roscoff (ROS) and Quiberon (QUI; Figures 1 , 2 ; for sampling details see Liesner et al., 2020a). Spitsbergen and Quiberon, respectively, correspond to the northern and southern distribution boundaries of L. digitata. Maps were generated using a European Environment Agency coastline shapefile (European Environment Agency, 2021) and QGIS 3.8.2-Zanzibar software (QGIS Development Team, 2019).

Fertile L. digitata sporophytes (n=30-35, for Helgoland n=15), spaced ≥1 m apart, were collected at low tide in the infralittoral fringe at each location, and stored in ambient seawater flow-through tanks for immediate processing not lasting more than three days. Two soral discs per individual (Ø=20 mm) were cut out, briefly rinsed with fresh water, and placed in a 50 ml centrifuge tube containing two microscope slides and 45 ml sterile seawater. They were left for 1-3 days to allow released spores to settle. In samples from Bodø, sporulation was induced by leaving discs in a wet chamber overnight before adding sterile seawater. Saturated germanium dioxide solution (GeO₂) was added (0.5 ml/l) to prevent diatom growth (Shea and Chopin, 2007). Tubes were transported in dark, cooled (<15°C) boxes from the respective locations to the Alfred Wegener Institute in Bremerhaven, Germany, within two days. Samples were stored at 5°C with a maximum irradiance of 5 µmol photons m^-2 s^-1 white light (LED Mitras Lightbar Daylight 150 controlled by a ProfiLux 3 computer, GHL Advanced Technology, Kaiserslautern, Germany), measured using a LI 190SB quantum sensor coupled to a LI-185B photometer (LI-COR-Biosciences, Lincoln, NE, USA) in a 16:8 h light:dark (LD) photoperiod. Microscope slides were then transferred to Petri dishes (Ø=9 cm) filled with iron-free, half strength Provasoli enriched seawater (PES) to inhibit gametogenesis (Provasoli, 1968; Lewis et al., 2013; iodide enrichment according to Tatewaki (1966), double concentration of Na₂glycerophosphate) and GeO₂, and stored at 15°C in 3-4 µmol photons m^-2 s^-1 red LED light to allow germination into vegetative gametophytes until start of experiment 1 (for details see Table S1 ). For experiment 2, pairs of male and female gametophytes, each originating from a single sporophyte, were isolated in 2016 (Spitsbergen) and in 2019 (Tromsø, Quiberon) into unialgal stock cultures and cultivated in 3 µmol photons m^-2 s^-1 red light at 15°C in ½ PES until start of the experiment in 2020 (for details see Table S2 ). This approach was needed, as long-term cultivation of dense, mixed gametophytes is often prone to hidden contamination that causes experimental bias, and because one sex may overgrow the other, resulting in an increasingly skewed sex-ratio (see Martins et al., 2017).

Statistics were performed using R version 3.6.3 (R Core Team, 2020). One-way ANOVA was performed to test initial gametophyte densities in experiment 2 and gametophyte survival at 24°C and 25°C on day 8 (HLG, day 7) in experiment 1. Initial densities in experiment 1 did not meet assumptions of homogeneity, so a non-parametric Kruskal-Wallis test was performed, with subsequent pairwise Wilcoxon rank sum testing and Bonferroni correction of p-values to reduce the probability of type I errors. Sporophyte GR _length, GR _width and L:W ratio were analyzed using ANCOVA, with sporophyte density (day 7) as a covariate, as density may influence growth rates. All other parameters were evaluated by two-way ANOVA. Statistically significant differences were compared with the post hoc Tukey honest significance difference (HSD) test using the “emmeans” package. Prior to analysis, data were tested for homogeneity of variance using the Levene’s test. Data with heterogeneous variances were either square-root, fourth-root or logarithmically transformed (Underwood, 1996). If assumptions of homogeneity were still not met, the significance level was lowered to p<0.01, as there is no non-parametric alternative for two-way ANOVA with post hoc tests. Reports of pairwise comparisons are described by the highest p-value.

Gametophytes from the five tested populations (Quiberon in the south, to Tromsø in the north), spanning a wide distribution range in the NE Atlantic, exhibited a similar response in the upper temperature gradient with a uniform UST of 24°C after a two-week exposure, with no short-term recovery and limited long-term recovery. Gametophytes from Spitsbergen have been investigated before and exhibited the same upper lethal temperature limit (Franke et al., 2021). This is despite local mean summer SSTs differing by ~12-13°C between the southern (Oppliger et al., 2014) and northern distribution margins (Fischer et al., 2019; Figure 2 ). Although none of the populations are likely to experience such high SSTs over 14 days in their natural environment, it should be noted that our experimental priming temperature of 24°C only exceeds the mean summer SST at the southern-most location in Quiberon by ~6°C, but by 18-19°C at the northern-most limit of the species in Spitsbergen (Oppliger et al., 2014; Fischer et al., 2019). Few studies have examined the intra-specific thermal response of microscopic gametophyte and sporophyte stages throughout their distribution gradient (Martinez, 1999; Oppliger et al., 2012; Mohring et al., 2014; Becheler et al., 2022). However, several previous studies have shown that the upper lethal limit with respect to survival and growth of gametophytes of Atlantic Laminaria spp. and Saccharina spp. was also similar across sites with different thermal environments (Bolton and Lüning, 1982; tom Dieck (Bartsch), 1992; Martins et al., 2020); but these authors only worked with single clonal isolates. With recent insights into the genetic heterogeneity of L. digitata, its reduced gene flow in marginal populations (e.g., Billot et al., 2003; King et al., 2020; Liesner et al., 2020a) and latitudinal adaptation of sporophytes towards heat stress (e.g., King et al., 2019; Liesner et al., 2020a), it became questionable whether this single-clonal approach was sufficient to understand the full tolerance width within a single kelp species. By applying a population approach and assuming that we incorporated several genetic lineages per location, our study now confirms that the upper thermal limit of L. digitata gametophytes is conserved along its whole distribution gradient in the NE, irrespective of the local long-term thermal environment. This supports early assumptions of Wiencke et al. (1994) that the upper survival limit of seaweed species is relatively fixed and illustrates that neutral genetic diversity is not always a good indicator for intra-specific performance plasticity (King et al., 2020).

Polar and temperate kelp species exhibit a complex alternation of generations between microscopic gametophytes and macroscopic sporophytes, where each life cycle stage exploits different ecological niches according to its physiological needs. This strategy is thought to provide an adaptive advantage, especially in variable environments (Ladah and Zertuche-González, 2007; Barradas et al., 2011). Microscopic gametophytes of kelps have been shown to be the most heat tolerant stage, regardless of population or species (tom Dieck (Bartsch), 1993; Ladah and Zertuche-González, 2007), highlighting their widespread ability to persist as hardy, long-lived seed banks, that can exist for up to decades without losing their fertility (tom Dieck (Bartsch), 1993; Edwards, 2000; Carney and Edwards, 2006; Edwards, 2022; Martins et al., 2020). The fact that such a high UST has been preserved in L. digitata gametophytes underlines the importance of this parameter as a survival mechanism in which persisting gametophytes can replenish sporophyte populations following die-off events (e.g., Bartsch et al., 2013). This may be especially important for southern populations that have in fact faced large-scale extirpation in the past (Lima et al., 2007; Fernández, 2011; Moy and Christie, 2012; Voerman et al., 2013; Smale et al., 2015).

After several weeks of post-cultivation at 15°C, gametophytes previously exposed to 24°C were observed to grow again. These may have arisen from single living cells in damaged multicellular gametophyte filaments or from unaccounted surviving gametophytes that we missed in our monitoring protocol, where we only sampled random areas of the petri dishes. Therefore, we highlight the importance of allowing long-term cultivation under growth-supporting conditions after exposure to environmental stressors, as we might be underestimating the survival potential of microscopic stages of kelps at near lethal conditions.

Interestingly, the few gametophytes that survived the 24°C priming treatment were primarily male. These results suggest differential survival capacities between sexes, with female gametophytes being more sensitive to heat than male gametophytes. In several species of the genus Laminaria, male gametophytes were reported to survive better in high temperatures than female gametophytes, (tom Dieck (Bartsch), 1993). Likewise, Arctic isolates of the morphologically similar species L. digitata and Hedophyllum nigripes, revealed higher heat tolerance in male than female gametophytes (Franke et al., 2021). Bolton and Lüning (1982) also found a 1°C higher UST in male L. digitata gametophytes compared to females, but not for the other N-Atlantic kelp species L. hyperborea, S. latissima and Arctic Laminaria solidungula. One explanation for the differential survival capacity of sexes is the ability of male gametophytes to reproduce vegetatively through fragmentation (Destombe and Oppliger, 2011), giving an illusion of higher survival. This is thought to act as a dispersal mechanism by reducing male to female distances (to <1 mm) to facilitate pheromone signaling and thereby enhancing reproductive success (Reed et al., 1997; Destombe and Oppliger, 2011). In the wild, more male survivors may increase the chance of females being fertilized, as an oogonium and antheridium only produce one egg and sperm, respectively (Destombe and Oppliger, 2011). Why sex-ratio is especially male-skewed at high temperatures (see also Lee and Brinkhuis, 1988; Izquierdo et al., 2002), however, is not clear, but may be linked to better stress resistance of males. In fact, gene expression profiles in vegetative S. latissima gametophytes have been shown to be different between temperatures and sexes (Monteiro et al., 2019). Only a small proportion (7-12%) of differentially expressed genes were consistent across temperatures, indicating that temperature strongly influences development. In general, functional categories linked to responses to stress and external and abiotic stimuli were over-represented in females. Females also responded more strongly at higher temperatures than males, suggesting a lower heat tolerance of females (Monteiro et al., 2019). These hypotheses may explain the significantly slower gametophyte mortality of the Tromsø population at 24°C and 25°C within the priming period compared to all other populations, as only this population was male-dominated (1.74 × more males than females on day 0 - an artificial effect of inconsistent seeding), which could have provided a survival advantage.

In contrast to the uniform lethal limit at high temperatures across populations, we discovered that gametophyte survival from three populations in the lower temperature gradient between 0-15°C was clearly reflective of the thermal environment from which they originated. The southern-most population of Quiberon showed a positive relationship of survival with increasing temperature, whereas the northern-most population Spitsbergen showed a negative relationship with increasing temperature. Gametophytes from both populations survived best near their in situ annual average SSTs at 15°C (Quiberon) and 0-3°C (Spitsbergen). Gametophytes from Tromsø, located in between, where the mean annual SSTs are ~7°C (Chepurin and Carton, 2012; Figure 2 ), revealed rather constant survival over temperatures, with significantly lower survival only at 3°C and 12°C compared to 15°C. This suggests that L. digitata populations have adapted in response to local long-term environmental conditions and that selection pressure towards local mean temperatures was higher than for USTs. A study by Mohring et al. (2014), that examined the thermal performance of microscopic stages of the warm-temperate kelp Ecklonia radiata from three regions across its latitudinal distribution in southern Australia, came to similar conclusions. Gametophyte survival (density) and growth (area of gametophytes) was optimal in conditions reflective of the environment in which they persist. These findings support the hypothesis that variability in the thermal performance of kelps with broad latitudinal distributions is influenced by biogeography and ecotypic adaptation to local temperature regimes, but that this adaptation is only reflected in naturally-occurring temperature ranges, shaping optimal responses towards local conditions rather than altering physiological thresholds.

North Atlantic kelps are generally quite young. The Arctic endemic kelp L. solidungula likely originated from the first major migration event of Pacific Laminaria through the Bering Strait around 5.3 Mya (Rothman et al., 2017). The cold-temperate species, such as L. digitata, followed later from a second migration event, which also gave rise to the warm-temperate species (Rothman et al., 2017). So, despite their ancestors crossing the Arctic from the Pacific to the Atlantic, North Atlantic kelps are mostly not Arctic in their temperature profile, except for the more distantly related L. solidungula (tom Dieck (Bartsch), 1992; tom Dieck (Bartsch), 1993; Roleda, 2016). The uniform UST of L. digitata gametophytes shown here, in combination with the ecotypic adaptation at lower temperatures, indicate that populations of this species are likely in an early stage of adaptation to cold environments, due to their relatively short cold-water history since around 3 My (sensu Wiencke et al., 1994).

Phylogeographic studies of NE Atlantic L. digitata have identified genetically distinct northern and southern clusters (King et al., 2020; Liesner et al., 2020a; Neiva et al., 2020). This genetic structure corresponds to the differentiation of thermal traits in L. digitata sporophytes (King et al., 2019; Liesner et al., 2020a) and might also explain the location-specific gametophyte survival at lower temperatures (this study). This divergence can be traced back to different post-glacial redistribution histories after the Last Glacial Maximum (20,000 YBP; Assis et al., 2017a; Neiva et al., 2020). Potential glacial refugia have been suggested near the unglaciated coastlines of the Armorican and Celtic Seas (Brittany and SW Britain; Maggs et al., 2008), and close to Ireland (Neiva et al., 2020; Schoenrock et al., 2020), which was at the border of former, heavily glaciated regions. Repeated range contractions into and expansions from these refugia may have maintained unique gene pools and facilitated genetic diversification across glacial cycles (Assis et al., 2017a; Neiva et al., 2020).

Similarly, as for survival at the lower temperature end, gametophytes from Tromsø, Roscoff and Quiberon showed ecotypic responses in relative microscopic sporophyte formation per female gametophytes at the upper temperature end. This was apparent after thermal priming in high (sub)-lethal temperatures with subsequent transfer to fertility-inducing 15°C. While all three populations showed 87% to 99.8% reproductive success at continuous 15°C, indicating an optimum response, the limiting priming temperature for reproduction was 23°C in all populations. Gametophytes from the southern populations Quiberon and Roscoff exposed to priming temperatures of 20-22°C exhibited significantly higher reproductive success during recovery in optimal conditions compared to the northern population of Tromsø. Furthermore, female gametophytes from the southern-most population Quiberon showed no decline in sporophyte formation in these priming temperatures, whereas Roscoff revealed a clear negative relationship with increasing priming temperature. This variation towards thermal priming at high temperatures points towards thermal adaptation to local conditions. The impact of thermal priming on subsequent trait performance seems to be vital in explaining some of the observed trait plasticity in kelps. Recent discoveries revealed that L. digitata gametophytes from the North Sea and the Arctic (same locations as in this study) also behaved differently after recovery from short heatwave priming treatments (20-25°C), indicating ecotypic responses (Martins et al., 2020).

Within the same set of results, we additionally noted that reproductive success in gametophytes exposed to the 21°C and 22°C priming treatments was significantly different among all three populations in latitudinal sequence. Given the close geographic proximity (~1° latitude apart) and similarity in thermal environments (2°C difference in maximum monthly mean SSTs) of the two French sites, it is surprising that such a considerable difference was observed between the two. For comparison, Tromsø is situated roughly 22° latitude further north and the local annual average SST is approximately 7-8°C lower than in northern France (Chepurin and Carton, 2012; Oppliger et al., 2014). Therefore, it is likely that intraspecific variation not only plays a role between populations from the two opposite distribution edges, but also among neighboring populations themselves (Billot et al., 2003). Gene flow is low among L. digitata populations from Roscoff and Quiberon (Billot et al., 2003; Robuchon et al., 2014; Liesner et al., 2020a), which potentially facilitated differentiation between populations. The southern-most European L. digitata population of Quiberon might be under especially high selection pressure.

In the lower temperature gradient, relative sporophyte formation per female gametophyte also varied between populations, but not in a pattern that indicated adaptation to local temperatures. In fact, female gametophytes from the high-latitude population of Spitsbergen had a high relative sporophyte formation between 6-15°C, with optima at 9-12°C, compared to the low-latitude population of Quiberon which produced most sporophytes between 3-12°C, with optima at 6-9°C. The fact that reproduction in L. digitata gametophytes appears to be adapted to such a wide temperature window (3-15°C; tom Dieck (Bartsch), 1992; Müller et al., 2008; Martins et al., 2017), but not to low temperatures of 0°C, further supports the idea that the migration of L. digitata to the Arctic may have been a rather recent event (sensu Wiencke et al., 1994). On the other hand, absolute sporophyte development showed a different pattern. The Quiberon population produced more sporophytes per area at 3-9°C than the Spitsbergen population. This was likely influenced by the initial gametophyte density which was >100 gametophytes cm^-2 higher in the former compared to the latter population ( Table S3 ), providing Quiberon with more female cells that could potentially develop sporophytes. Aside from this difference between the two populations, and contrary to their relative ability to form sporophytes, absolute sporophyte development clearly peaked in a narrow range of intermediate temperatures (Quiberon at 9°C and Spitsbergen at 9-12°C). A similar mismatch between speed of gametogenesis and recruitment optima has been described before (Martins et al., 2017), pointing towards differential selection pressure on proximate ontogenetic processes, distinctively controlled by temperature.

Linear GR _length and GR _width of microscopic sporophytes from different populations of L. digitata at 0-15°C were not obviously related to the thermal environment in which the populations occur. Although we found differences in GR _length and GR _width among populations (Spitsbergen, Tromsø and Quiberon), the general growth patterns were similar, with >4-fold higher GRs _length and GRs _width at 9-15°C than at 0-3°C. Thus, sporophyte growth at cool conditions is clearly not favored, irrespective of location. The higher GR _length of the Tromsø population compared to Spitsbergen and Quiberon was likely attributable to the low reproductive success and consequently low sporophyte density of the Tromsø population, providing more space and nutrients for growth. Our results are in line with former studies that consistently showed maximum growth of juvenile L. digitata sporophytes between 10°C and 15°C (Bolton and Lüning, 1982; tom Dieck (Bartsch), 1992) and therefore do not suggest prominent ecotypic responses for this trait. The only other study investigating juvenile kelp sporophyte growth along latitudes in a common-garden approach dealt with Lessonia nigrescens. The reported thermal ecotypes in growth of microscopic sporophytes from three Chilean locations spanning 21° latitude (Martinez, 1999), were probably due to the presence of two cryptic species (Oppliger et al., 2012) and thus also do not suggest intraspecific, ecotypic adaptation of this trait.

Results of our study reveal that low-latitude populations of L. digitata, might rely on faster reproduction and sporophyte growth to minimize exposure of gametophytes to unpredictable thermal fluctuations in summer, whereas colder, high-latitude populations, that generally experience more stable thermal conditions, may benefit from delayed maturation and increased vegetative growth of gametophytes, resulting in multicellular gametophytes that can bear more offspring (Destombe and Oppliger, 2011; Oppliger et al., 2012). If slow recruitment is better in the cold, and fast recruitment better in warm temperatures, maybe it is this plasticity that is selected for. As temperatures of >9°C have not yet been encountered in the Arctic, the high temperature recruitment optimum of the Spitsbergen population may suggest that this trait has not been under selection pressure yet.

In contrast to GRs, sporophyte morphology, expressed as L:W ratio, might suggest local adaptation. The southern range edge population of Quiberon showed the widest range in L:W ratio, with a maximum at 9°C. Sporophytes from Tromsø were less variable over the thermal treatments but peaked at 6°C. Spitsbergen sporophytes on the other hand, displayed a constantly low L:W ratio across all temperatures. Morphologically, this may imply that adult sporophytes from Quiberon generally grow long and thin at intermediate temperatures, whereas sporophytes from Spitsbergen grow wider rather than longer blades. It is, however, unclear whether these differences reflect local adaptation, or why a broad range in sporophyte morphology may be advantageous in warm environments as opposed to consistent morphology in cold environments. Nonetheless, whether this feature is plastic or genetically determined may have significant implications in selecting seedstocks for farming and domestication.

Intraspecific variation in thermal responses is prevalent among widely-distributed kelp species (Mohring et al., 2014; Saada et al., 2016; King et al., 2018; King et al., 2019; Liesner et al., 2020a; Martins et al., 2020). Yet, the specific mechanisms acting on different life history stages are not well-understood. We have shown that temperature-dependent responses of microscopic stages of L. digitata vary among life history stages and populations. The clear pattern of local adaptation we have identified in some traits, but not in others, may shed light on the crucial developmental processes under selection pressure and the fundamental niche of the species.

Gametophyte survival in lower (sub-) optimal temperatures and sporophyte formation after thermal priming at upper temperatures between 15°C and 23°C showed ecotypic responses across populations ( Table 5 ). Phenotypic divergence in these traits may highlight their importance for survival in the natural environment. On the other hand, the upper survival limit of gametophytes was uniform across populations. Adapting the UST does not seem to have been under selection pressure in the past, and current evidence suggests that ocean warming is outpacing the adaptive capacity of kelp forests (Vranken et al., 2021). Sporophyte formation and growth at low temperatures showed a random response, potentially concurring with the overall low levels of local adaptation found in adult sporophytes (Liesner et al., 2020a). Consequently, variation manifests itself in different traits both in haploid and diploid stages. Examining only single parameters or life cycle stages might therefore lead to incomplete or even false interpretations, and so whole-life-cycle approaches, using a range of different fitness-related traits, are vital to reveal the complex survival strategies of kelps.

In conclusion, our study demonstrates trait-dependent ecotypic adaptation in the thermal performance of different microscopic life history stages of L. digitata across its entire NE Atlantic distribution. In the future, kelp forests will struggle to thermally acclimate, adapt, or shift their distribution ranges fast enough to keep up with the rising temperatures and marine heatwaves (Merzouk and Johnson, 2011). While local adaptation may delay the regional contractions of marginal populations, it is unlikely that they will cope with global warming in the long term (Liesner et al., 2020a; Neiva et al., 2020; Vranken et al., 2021). The potential for fast adaptation of kelps via thermal priming or trans-generational and epigenetic inheritance, is only starting to be understood (Liesner et al., 2020b; Gauci et al., 2022; Scheschonk et al., 2022). Therefore, the development of strategies to future-proof and restore kelp forests is of utmost importance (Wood et al., 2019; Coleman et al., 2020; Eger et al., 2020; Jueterbock et al., 2021).