Environmental history shapes ecophysiology and microbiota of Macrocystis pyrifera at contrasting glacier- and ocean-influenced sites in the Magellanic sub-Antarctic ecoregion

Kelp forests are highly productive marine ecosystems that face increasing threats from global climate change, particularly in sub-Antarctic regions, where warming and glacial retreat are altering temperature, salinity, and light availability. In this study, we evaluated the responses of different frond types in Macrocystis pyrifera from two contrasting sites in the Magellanic sub-Antarctic ecoregion: one under glacial influence (Ultima Esperanza, UE) and the other under oceanic influence (Punta Carrera, PC). Ecophysiological responses were assessed under both natural conditions and controlled experiments manipulating temperature (6 and 10 °C) and irradiance (5 and 50 μmol photons m - ² s - ¹). In addition, we conducted an exploratory analysis of the associated epiphytic microbiota. The two sites differed markedly in underwater spectral composition and light attenuation, with UE showing lower light penetration. These site-level differences were accompanied by contrasting ecophysiological responses in natural populations: fronds from UE exhibited a higher optimum quantum yield of photosystem II (F v /F m) and distinct pigment profiles compared with fronds from PC. Experimental responses also varied among frond types: middle fronds were the most resistant in terms of survival and growth, apical fronds were the most sensitive but also showed the greatest pigment plasticity, and sporophylls were the most vulnerable overall. The composition of the epiphytic microbiota was structured primarily by site, with no significant effect of frond type once other factors were accounted for, indicating that local environmental conditions exerted a stronger influence than tissue identity. Overall, our results suggest that variation in M. pyrifera responses to local environmental conditions is expressed simultaneously through the ecophysiological condition of the host, functional differentiation within the thallus, and the structure of the associated microbiota, with site emerging as a common organizing axis. These findings provide a basis for anticipating the likely direction of response of M. pyrifera under environmental change and highlight the need to incorporate population-level environmental context and functional heterogeneity within the thallus into conservation strategies for kelp forests in sub-Antarctic ecosystems.

Kelp forests are highly productive marine habitats of major ecological importance in temperate and cold regions (Wernberg et al., 2019). These ecosystems contribute to atmospheric carbon capture and sequestration, provide shelter, spawning, and feeding grounds, and support high levels of associated biodiversity (Duarte et al., 2017; Filbee-Dexter et al., 2022). However, they are increasingly threatened by global climate change, which is altering temperature, salinity, light, and nutrient availability, ultimately driving their loss, contraction, and transformation into less complex and less productive ecosystems (Hanley et al., 2024).

In the Southern Hemisphere, the kelp Macrocystis pyrifera is widely distributed along the Pacific coast, from northern Peru (~5° S) to the continental shelf of the Diego Ramírez–Drake Passage Marine Park, Chile (~56° S) (Hoffmann and Santelices, 1997; Rozzi et al., 2017; Marambio et al., 2023). In the Magellanic sub-Antarctic ecoregion (MSAE), this species forms perennial, highly productive populations with high reproductive activity, making it a key structural component of southern coastal ecosystems (Mansilla and Ávila, 2005). Its thalli exhibit marked morphological and functional differentiation between vegetative and reproductive fronds, which fulfill distinct roles within the individual and may generate contrasting microenvironments in both ecophysiological and surface-related terms (Marambio et al., 2023).

Within these ecosystems, M. pyrifera inhabits environments characterized by pronounced oceanographic and atmospheric heterogeneity, particularly in salinity, temperature, turbidity, and light availability (Silva and Calvete, 2002; Valdenegro and Silva, 2003; Soto et al., 2012). In the MSAE, glacier-influenced environments typically exhibit low salinity and temperature, high turbidity, and reduced light availability due to freshwater input and suspended sediments associated with glacial melt (Bahamonde et al., 2022). In contrast, environments under stronger oceanic influence, such as those connected to the Strait of Magellan, tend to exhibit more stable conditions, including higher salinity and temperature, lower turbidity, and greater light availability (Valdenegro and Silva, 2003; Iriarte, 2018).

M. pyrifera exhibits high ecophysiological plasticity, enabling it to persist across contrasting environmental conditions (dos Santos et al., 2019; Marambio et al., 2023; Gómez et al., 2025). Nevertheless, glacial influence may constrain this acclimation capacity by reducing light availability and increasing turbidity, thereby affecting the photosynthetic efficiency and pigment composition of its populations (Coral-Santacruz et al., 2024). This environmental variability not only modulates the ecophysiological performance of M. pyrifera but may also influence the distribution and structure of its populations and associated biota (Mansilla and Ávila, 2005; Ríos et al., 2007).

Temperature and irradiance are two major drivers of environmental variation between glacier-influenced and oceanic systems in the MSAE. In M. pyrifera, experimental studies have shown that temperature modulates photosynthetic rates from the earliest stages of the life cycle, affects growth and reproduction, and may compromise survival and tissue integrity when it exceeds optimal thresholds (Rodríguez et al., 2019; Schmid et al., 2020; Le et al., 2022). Irradiance, in turn, determines the energy available for carbon fixation and directly influences photosynthesis, pigment composition, and light harvesting in macroalgae (Martins et al., 2023). In M. pyrifera, supraoptimal irradiance may induce photoacclimation and photoprotective responses and may ultimately trigger photoinhibition and oxidative stress (Koch et al., 2016; Umanzor et al., 2020; Palacios et al., 2021).

The combined effects of temperature and irradiance on kelp ecophysiology have been increasingly documented, particularly in high-latitude and glacier-influenced systems. A recent global meta-analysis of true kelps found that ocean warming has a strong negative effect across life stages and physiological levels, including growth, reproduction, and survival, with impacts intensifying as the magnitude and duration of exposure increase (Roethler et al., 2025). In polar and sub-polar fjords, where glacial melt drives simultaneous changes in temperature, irradiance, and salinity, multiple-driver experiments have shown that these factors can interact in complex ways. In Arctic kelps, combinations of warming, low irradiance, and low salinity that reproduce near-glacier conditions elicited species-specific physiological and molecular responses rather than a uniform effect (Lebrun et al., 2024), and the combination of reduced salinity with elevated temperature significantly lowered the maximum quantum yield of photosystem II (F v / F m) in the endemic kelp Laminaria solidungula, with drivers interacting additively or antagonistically depending on the response variable (Diehl et al., 2020). A recent structured review of high-latitude kelps identified temperature, salinity, sediment load, and light as the stressors most likely to change and affect these species under climate change, and noted that populations with cold-temperate, subpolar, and polar distributions may face range contractions exacerbated by increased freshwater and sediment input from rapidly retreating glaciers (Farrugia-Drakard et al., 2023). These findings indicate that responses to temperature and irradiance are strongly context-dependent and shaped by the environmental history of each population. In this ecoregion, where glacial influence generates comparable multi-driver gradients, this dependence remains poorly characterized in M. pyrifera.

Recent work in the MSAE has begun to characterize how M. pyrifera responds to the strong environmental heterogeneity of its fjord and channel systems. Along the Strait of Magellan, morpho-functional, bio-optical, and photobiological traits vary among sites following differences in underwater light and tidal regime, providing evidence of ecotypic differentiation among populations (Marambio et al., 2017; Gómez et al., 2025). Photobiological performance also varies within the thallus: in a glacier-influenced fjord, apical and basal fronds differed significantly in maximum relative electron transport rate, electron transport efficiency, and light saturation point, consistent with functional stratification along the thallus (Coral-Santacruz et al., 2024). Studies in land-terminating glacier fjords have further framed these glacier-influenced populations as a window into future conditions under continued glacial retreat (Palacios et al., 2021). Together, these studies indicate that both among-site environmental context and within-thallus differentiation shape the ecophysiology of M. pyrifera, and these responses are generally attributed to phenotypic acclimation rather than to fixed genetic adaptation (Fernández et al., 2021; Solas et al., 2024). Despite these advances, however, it remains unclear whether such acclimation translates into differences in ecophysiological tolerance when individuals from contrasting environments are exposed to controlled changes in temperature and irradiance, or whether these responses vary among frond types within the same individual. This question is particularly relevant in the region, where strong environmental heterogeneity may drive differential responses depending on the populations’ environmental histories and the functional specialization of their tissues.

Macroalgae can also be understood as holobionts, that is, biological units composed of the host and its associated microbiota (Saha et al., 2024). Complex biofilms, primarily dominated by bacteria, develop on the surface of their tissues and participate in key processes such as nutrient recycling, organic matter transformation, production of bioactive compounds, defense against pathogens, and maintenance of host health (Wahl et al., 2012; de Mesquita et al., 2019; van der Loos et al., 2019). In addition to their direct effects on the alga, changes in temperature and irradiance may indirectly influence these microbial communities by altering the host’s ecophysiological state, exudate production, and microenvironmental conditions at the thallus surface (Ramanan et al., 2016; Paix et al., 2021; Qui-Minet et al., 2025). These interactions may also vary among frond types, as each tissue may constitute a distinct microhabitat. However, most studies on M. pyrifera have addressed either host ecophysiology (e.g. Marambio et al., 2017; Coral-Santacruz et al., 2024) or the associated microbiota (e.g. Florez et al., 2021; Soto et al., 2024) in isolation. By contrast, integrated approaches that evaluate both dimensions simultaneously remain scarce, particularly in sub-Antarctic environments under glacial influence. This gap limits a more comprehensive understanding of the mechanisms underlying acclimation, resilience, and tolerance in this species in response to drivers associated with global climate change, as well as variation in these responses among populations and frond types. Such variation may have important consequences for the health, stability, and persistence of kelp forests, as well as for their conservation and the ecosystem services they provide to coastal communities, including artisanal fisheries.

In this context, we compared M. pyrifera from two contrasting sites: one under glacial influence (UE) and one under oceanic influence (PC). We investigated whether these populations differ in their baseline ecophysiological condition, whether frond types differ in their tolerance to experimental changes in temperature and irradiance, and whether epiphytic bacterial communities are structured primarily by site, frond type, or experimental treatment.

Based on this evidence, we tested three hypotheses. First, that individuals from the glacier-influenced site (UE), developing under lower and more strongly attenuated irradiance, would exhibit a low-light photoacclimation signature, reflected in higher F v / F m and pigment profiles distinct from those at PC. Second, that frond types would differ in their tolerance and plasticity in response to experimental changes in temperature and irradiance, consistent with the functional stratification described along the M. pyrifera thallus, with the more exposed apical tissue expected to show the strongest photochemical response and pigment plasticity, and the reproductive sporophylls expected to be the most vulnerable overall. Third, that the epiphytic bacterial communities would be structured primarily by site rather than by frond type or experimental treatment, reflecting the influence of local environmental context on kelp-associated microbiota. By addressing host ecophysiology and associated microbiota together, this study provides an integrated basis for understanding the processes that may shape the performance and persistence of M. pyrifera in sub-Antarctic environments undergoing environmental change.

Study area and measurement of underwater spectral composition

The glacially influenced site in Ultima Esperanza Fjord (UE; Figure 1A, B, red dot) is strongly affected by the Serrano and Balmaceda glaciers. In contrast, the oceanically influenced site at Punta Carrera (PC) is located within the central micro-basin of the Strait of Magellan (Figure 1A, C, blue dot), where Pacific Ocean waters predominate (Valdenegro and Silva, 2003).

Rodríguez (2026) characterized the environmental conditions at both sites from depth-integrated vertical profiles recorded within the M. pyrifera forest with a HANNA HI 9829 multiparameter probe (five independent profiles per site), spanning the depth range inhabited by the species (0–5 m at UE; 0–3 m at PC). UE showed lower salinity (12.23) and temperature (6.88 °C), and higher turbidity (17.10 Formazin Nephelometric Units, FNU), whereas PC showed higher salinity (30.93) and temperature (9.43 °C), and lower turbidity (0.96 FNU). Depth-integrated photosynthetically active radiation (PAR), measured with the hyperspectral radiometer described below, beneath the kelp canopy, was likewise lower at UE (8.40 μmol photons m - ² s - ¹) than at PC (17.85 μmol photons m - ² s - ¹).

Underwater spectral composition was measured in triplicate (n = 3) at each sampling site on the same day as algal collection using a UV-VIS hyperspectral radiometer equipped with a 256-channel photodiode array and a 2π cosine-corrected collector (RAMSES ACC; TriOS Mess- und Datentechnik GmbH, Rastede, Germany) (Rautenberger et al., 2009). All measurements were taken within the M. pyrifera forest, beneath the canopy, at 1-m depth intervals from 1 to 5 m at UE and from 1 to 3 m at PC, corresponding to the depth range inhabited by the species at each site. Underwater irradiance was measured from 316 to 700 nm and analyzed using the manufacturer’s software (MSDA_XE, v7.0, TriOS Mess- und Datentechnik GmbH). Spectral irradiance products were calculated as described by Rautenberger and Bischof (2016).

Collection sites and algal material

Vegetative (apical and middle) and reproductive (sporophyll) fronds of M. pyrifera were collected by scuba diving at two sites in the MSAE: one under glacial influence in Ultima Esperanza Fjord (UE; 51°33′34.82″S, 72°49′11.05″W) (Figure 1A, B), and another under oceanic influence at Punta Carrera, Strait of Magellan (PC; 53°34′44.13″S, 70°55′59.81″W) (Figure 1A, C). Sampling was conducted at depths of 0–5 m at UE and 0–3 m at PC (Figure 1D) during the summers of 2022 and 2023, respectively.

Experimental design and setup

A controlled culture experiment evaluated the effects of temperature and irradiance on the ecophysiology and associated microbiota of M. pyrifera from the UE and PC sites. Fronds were transported from each collection site to the laboratory in coolers filled with native seawater, kept cool and shaded to minimize stress during transit. Upon arrival, twelve individuals were selected from each locality, and one apical, one middle, and one sporophyll frond was collected from each individual. Tissue discs (2.2 cm in diameter) were excised from the middle section of each frond using a hollow punch, which was disinfected with 70% (v/v) ethanol between samples.

Prior to the experiment, tissue discs were acclimated for 7 days in a cold chamber, in 1 L containers with native seawater from the respective collection site, thereby preserving the in situ salinity of each population (UE and PC). During acclimation, discs were maintained at 8 °C under a 16:8 h light:dark photoperiod and an irradiance of 15 μmol photons m - ² s - ¹.

For each frond type and locality, a factorial experiment was established with two temperature levels (6 and 10 °C) and two irradiance levels (5 and 50 μmol photons m - ² s - ¹; hereafter I5 and I50, respectively), yielding four treatment combinations (Figure 2). The experimental temperature and irradiance levels were selected to represent ecologically relevant conditions of the study region rather than arbitrary values. The two temperatures correspond, respectively, to the mean winter and summer seawater temperatures recorded in glacier-influenced environments of the MSAE (Bahamonde et al., 2022), thus bracketing the seasonal thermal range experienced by M. pyrifera in the area. The irradiance levels were defined in relation to the underwater light conditions measured at the study sites, where in situ PAR was low (UE, 8.40; PC, 17.85 μmol photons m - ² s - ¹), consistent with strong light attenuation beneath the canopy and with glacial turbidity. I5 represents a light-limiting condition comparable to that occurring beneath the canopy and in turbid waters. I50, although low in absolute terms for brown macroalgae, represents a substantial increase relative to the in situ irradiance, close to six times the PAR recorded at UE, and was not intended to reproduce the conditions of any particular site. This 2 × 2 factorial design therefore crossed the two main environmental drivers that distinguish the study sites, allowing their individual and interactive effects to be assessed. The use of two levels per factor was intended to contrast extreme seasonal and light conditions rather than to model response curves or optimal values.

Within each locality and frond type, the 12 individuals were distributed across the four treatment combinations, with three independent individuals per treatment. Each replicate unit consisted of 11 tissue discs excised from the same frond type of a single individual and maintained together in 1 L of native seawater. Tissue discs from a given individual were not split across treatments; therefore, discs within each replicate unit were considered subsamples rather than independent replicates. Thus, for each locality and frond type, the experimental unit consisted of tissue from a single individual, resulting in 12 independent replicate units per frond type. Cultures were maintained for 15 days under constant aeration within closed culture chambers, under a 16:8 h light:dark photoperiod. Air was not filtered prior to delivery; although the closed chambers limited exposure to general laboratory air circulation, the potential introduction of airborne bacteria cannot be entirely excluded and is therefore considered when interpreting the exploratory microbiota results. Each replicate unit was maintained in native seawater from its respective collection site throughout the experiment. No external nutrients were added; a 20% renewal with native seawater was performed on each measurement day to partially replenish the medium and limit the accumulation of metabolic waste.

Measurements were taken on days 0, 1, 3, 7, 12, and 15. Within each replicate, one tissue disc was marked at the beginning of the experiment and repeatedly measured throughout the experimental period to assess the optimum quantum yield of photosystem II (F v /F m) and growth rate; these measurements were performed under sterile conditions to avoid contamination. Survival was monitored throughout the experiment, and tissue was recorded as dead when it showed pigment degradation (bleaching) together with the absence of a detectable F v /F m signal (both criteria required). On day 15, one additional tissue disc per replicate was used to determine final pigment concentration, and another was collected for characterization of the associated microbiota. Because microbiota was analyzed only at the end of the experiment (day 15), sequencing was restricted to surviving samples, and mortality therefore conditioned the final sampling structure for microbial analyses.

Growth rate

The maximum diameter (longest axis in the plane of the blade) of the marked tissue disc was measured at each sampling time using a millimetre ruler. Specific growth rate (SGR) was calculated from this dimension using the following equation:

where Li and Lf are the initial and final maximum diameter, respectively, and tf − ti ​ is the time interval between measurements (days) (Kopczak et al., 1991).

Photosynthetic efficiency and pigment analysis

Fv/Fm and pigment concentrations were evaluated in apical, middle, and sporophyll fronds of M. pyrifera from seven individuals in natural populations, as well as in tissue discs obtained from the culture experiment. Fv/Fm was measured in vivo after 15 min of dark adaptation using a pulse-amplitude-modulated chlorophyll fluorometer (MINI-PAM-II, Heinz Walz GmbH, Effeltrich, Germany) (Marambio et al., 2017). Measurements were performed using the following settings: measuring light intensity 1, saturation pulse intensity 10, saturation pulse width 0.6 s, gain 2, and damping 2. Pigment concentrations were determined following the methodology described for brown algae by Mansilla et al. (2016) and Marambio et al. (2017). For pigment analysis, samples were dried over silica gel, which was replaced regularly, and stored dry, cool, and in darkness until biochemical analyses, following Niedzwiedz et al. (2025); biochemical measurements were therefore based on dry weight (DW). Briefly, 0.125 g DW of tissue per sample was incubated in 1 mL of dimethyl sulfoxide (DMSO, ≥99%) for 15 min. Chlorophyll a (Chl a), chlorophyll c (Chl c), and fucoxanthin (Fucox) were then quantified spectrophotometrically (Genesys UV) at room temperature. Absorbance at 750 nm was used as a correction for light scattering. Pigment concentrations were calculated according to the equations of Seely et al. (1972) and expressed as mg g - ¹ DW (Marambio et al., 2023).

Sampling of epiphytic bacteria on M. pyrifera: DNA extraction and sequencing of the 16S rRNA gene V1–V3 region

Epiphytic bacterial communities were sampled from tissue discs excised from apical, middle, and sporophyll fronds of M. pyrifera at the end of the experiment (day 15). For each site, treatment, and frond type, two tissue discs were sampled for bacterial community analysis whenever available. Because the microbiota was assessed only at the end of the experiment, only samples that survived until day 15 were included. Each tissue disc was placed in a pre-cleaned, UV-sterilized plastic bag, wrapped in aluminum foil, cleaned with 70% (v/v) ethanol, flash-frozen in liquid nitrogen, and stored at −20 °C for approximately 24 h until DNA extraction. Before DNA extraction, an additional detachment step was performed to recover epiphytic bacterial cells from the tissue discs using a modified version of the protocol described by Morán et al. (2008) and Florez et al. (2019). Briefly, circular tissue discs were transferred to pre-cleaned plastic bags sterilized by UV exposure and rinsed with 70% (v/v) ethanol, and 5 mL of 1× TBE buffer (89 mM Tris, 89 mM boric acid, 2 mM EDTA; pH 8.0) was added to each sample. Samples were then sonicated in an ultrasonic bath (Baku Ultrasonic Cleaner BK-2300, China) at 45 kHz for 15 min to detach surface-associated microbial cells. The suspension was centrifuged (Hettich MIKRO 200R, Tuttlingen, Germany) at 15,000 rpm for 15 min at room temperature (~20 °C), and the pellet was immediately processed for DNA extraction.

DNA was extracted using the DNeasy PowerSoil Pro DNA Kit (Qiagen, Hilden, Germany). Genomic DNA concentration was quantified with a Qubit fluorometer (Invitrogen, Carlsbad, CA, USA) using the Qubit dsDNA HS Assay Kit (Invitrogen, Life Technologies, Carlsbad, CA, USA). The V1–V3 region of the 16S rRNA gene was amplified with the primer pair 27F (5′-AGRGTTTGATCMTGGCTCAG-3′) and 519R (5′-GTNTTACNGCGGCKGCTG-3′) (Florez et al., 2019). Metabarcoding libraries targeting the 16S rRNA gene were prepared at AUSTRAL-omics (Valdivia, Chile) following the Illumina 16S Metagenomic Sequencing Library Preparation protocol. Details of library purification, quality control, pooling, and paired-end sequencing (2 × 300 bp) are provided in Rodríguez (2026).

Bioinformatic analysis of sequences

Microbial community analyses were conducted in QIIME 2 (version 2024.2) (Bolyen et al., 2019). Paired-end reads were merged, denoised, screened for chimeras, and used to infer amplicon sequence variants (ASVs) with the DADA2 pipeline (Callahan et al., 2016, 2017). Low-frequency ASVs (<5 total occurrences) were removed. Taxonomic assignment was performed using a pre-trained SILVA 138 database (Quast et al., 2013) and a naïve Bayes classifier (Bokulich et al., 2018; Pedregosa et al., 2011) (Supplementary Table 1). ASVs assigned to mitochondria and chloroplasts were excluded.

Relative abundance values were calculated from the non-rarefied, quality-filtered ASV table after removal of chloroplast and mitochondrial sequences. For diversity analyses, a rarefaction depth of 6,674 sequences per sample was selected based on inspection of alpha rarefaction curves to ensure adequate sampling coverage (Supplementary Figures 1A, B). Alpha diversity metrics (Shannon index, observed richness, and Chao1) were calculated from the rarefied ASV table. Due to the low sample size within each treatment combination, these metrics were used for exploratory purposes only, and no formal inferential tests were performed. Beta diversity was assessed on the rarefied ASV table using marginal PERMANOVA (Type III sums of squares) on Bray-Curtis distances (relative abundance) and Jaccard distances (presence/absence), each with 9,999 permutations. Site, frond type, temperature, and irradiance were included as factors, and each term was evaluated after accounting for all others. This approach was chosen because the design was unbalanced owing to differential mortality and because there was no a priori justification for a specific ordering of the factors. Homogeneity of multivariate dispersions was verified for each factor using betadisper with permutation tests (9,999 permutations). Pairwise comparisons among frond types were performed with pairwise PERMANOVA and Benjamini-Hochberg correction; pairwise tests were not required for the remaining factors, which had only two levels. Principal coordinate analysis (PCoA) was used to visualize differences in community structure. These analyses were performed in R using the vegan (Oksanen et al., 2012) and pairwiseAdonis packages (Martínez-Arbizu, 2020).

Putative functional profiles of the microbial communities were inferred with the FAPROTAX v1.1 database (Louca et al., 2016) based on genus-level taxonomic assignments. Functional annotation frequencies were expressed as percentages and visualized as heatmaps using ggplot2 (Wickham, 2016).

Data analysis

To evaluate the effects of study site (UE and PC) and frond type (apical, middle, and sporophyll) on F v /F m and photosynthetic pigment concentrations (Chl a, Chl c, and Fucox) in individuals from natural populations, two-way analyses of variance (ANOVA) were performed with site and frond type as fixed factors, including their interaction. For photosynthetic pigments, a separate two-way ANOVA was fitted for each variable. Prior to analysis, the assumptions of normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test) were verified using the car package (Fox and Weisberg, 2018) (Supplementary Table 2); additional visual diagnostics are described below.

Under experimental conditions, F v /F m was analyzed separately for each frond type using linear mixed-effects models fitted with the lme function in the nlme package (Pinheiro and Bates, 2025). In all models, site, temperature, irradiance, and day were included as fixed effects, and replicate identity was included as a random intercept. Because the aim was to assess temporal change, day was treated as a numeric predictor, centered prior to analysis, and both linear and quadratic temporal trends were evaluated. Linear and quadratic models were initially fitted using maximum likelihood (ML) and compared using likelihood ratio tests, the Akaike information criterion (AIC) (Burnham and Anderson, 2002), and the Bayesian information criterion (BIC). The best-supported model was then refitted using restricted maximum likelihood (REML) for final inference (Zuur et al., 2009).

Photosynthetic pigment data under experimental conditions were analyzed separately for each frond type using one-way multivariate analysis of variance (MANOVA), with treatment group (defined as the combination of site, temperature, and irradiance) as the fixed factor and the log-transformed concentrations of Chl a, Chl c, and Fucox as response variables. Multivariate significance was assessed using Pillai’s trace and, when significant, followed by univariate ANOVAs (Johnson and Field, 1993). The effects of treatment on the pigment ratios Chl c /Chl a and Fucox/Chl a were evaluated separately for each frond type using one-way ANOVAs, with the combined treatment factor (site, temperature, and irradiance) included as a fixed factor.

For ANOVAs, these assumptions were further checked by inspecting histograms, Q-Q plots, and residuals versus fitted values. For mixed-effects models, assumptions were assessed by visual inspection of normalized residuals versus fitted values, normal Q-Q plots, histograms, and boxplots of residuals (Zuur et al., 2009; Schielzeth et al., 2020). The significance of main effects and interactions was evaluated in the final models. For ANOVAs, Tukey’s honestly significant difference (HSD) test was used for post hoc comparisons when appropriate. All statistical analyses were conducted in R version 4.3.2 (R Core Team, 2023).

Environmental conditions and natural ecophysiology of M. pyrifera

Underwater spectral composition

The spectral composition of underwater irradiance differed markedly between the two sampling sites, UE and PC, and across the depths evaluated. At both sites, irradiance decreased with depth; however, pronounced differences were observed in both absolute irradiance and spectral distribution. At UE, irradiance at 1 m depth peaked at 20.9 mW m - ² nm - ¹ within the 500–600 nm range and declined sharply with depth, approaching zero at 5 m (Figure 3A). Shorter wavelengths in the 400–500 nm range were strongly attenuated with depth. At PC, irradiance was higher at all depths, reaching a maximum of 28.8 mW m - ² nm - ¹ at 1 m within the 550–600 nm range (Figure 3B). The decline with depth was more gradual than at UE, resulting in greater light penetration, particularly across the 500–650 nm range.

Optimum quantum yield of photosystem II (F v /F m)

F v /F m was generally higher at UE than at PC across the three frond types (Figure 4). F v /F m differed significantly between sites (higher at UE; p = 0.007) and among frond types (p = 0.011), being highest in sporophylls and lowest in apical fronds; in the site × frond-type comparisons, sporophylls from UE showed the highest values, differing significantly from apical and middle fronds from PC (Figure 4; Supplementary Tables 3, 4).

Photosynthetic pigment concentrations

Photosynthetic pigment concentrations varied among frond types and, in the case of Chl a and Fucox, also between sampling sites (Table 1; Supplementary Table 5). Overall, Chl a concentrations were higher than those of Chl c and Fucox at both sites (Table 1). The lowest pigment concentrations were observed for Fucox at UE and for Chl c at PC (Table 1).

For Chl a, a significant site × frond-type interaction was detected (p = 0.040), in addition to significant main effects of site (p = 0.004) and frond type (p < 0.001) (Supplementary Table 5). Because of this interaction, the difference between sites depended on the frond type considered: UE and PC differed mainly in middle fronds (Table 1). Chl a therefore did not follow a single spatial gradient along the thallus. For Chl c, no significant difference was detected between sites (p = 0.110), and significant differences occurred only among frond types (p = 0.009), with apical fronds differing from sporophylls at both sites (Table 1; Supplementary Table 5). For Fucox, significant differences were detected both between sites (p = 0.021) and among frond types (p < 0.001), with apical fronds differing from middle fronds and sporophylls (Table 1; Supplementary Table 5). Thus, Chl c and Fucox followed a consistent gradient along the thallus, being higher in apical tissue, whereas Chl a did not.

Experimental responses to temperature and irradiance

Survival and growth rate

Survival of M. pyrifera under experimental conditions varied by frond type, treatment, and site. In general, apical and middle fronds maintained high survival across most conditions, whereas sporophylls were the most sensitive (Figure 5). At UE, survival remained high in most treatments, although apical fronds at 10 °C-I50 and sporophylls at 10 °C-I5 and 6 °C-I50 reached complete mortality by the end of the experiment. At PC, mortality was concentrated in specific treatment combinations, with marked declines in apical fronds at 6 °C-I50 and in sporophylls at 10 °C-I50 (Figure 5). Middle fronds were comparatively more resistant, although survival decreased under some treatments, particularly at PC toward day 15 (Figure 5).

Specific growth rate (SGR) also varied among frond types and across experimental conditions. Overall, the three frond types exhibited distinct growth responses, indicating that elongation responses were not uniform along the thallus. In apical fronds, SGR values were generally low and ranged from slightly positive to negative depending on treatment and site (Figure 6A). The highest positive values were observed at UE under 10 °C-I5, whereas negative growth was recorded at PC under 10 °C-I50. In middle fronds, SGR remained positive under most treatments, although values were lower under 6 °C-I5 and 10 °C-I50, and the highest growth rates were observed under 6 °C-I50 and 10 °C-I5 (Figure 6B). Overall, middle fronds showed the most stable and consistently positive growth across treatments. In sporophylls, SGR showed the strongest treatment response and the greatest variability, including negative values under 6 °C-I5 and markedly positive values under 10 °C-I5 (Figure 6C). Compared with apical and middle fronds, sporophylls showed the clearest contrast between unfavorable and favorable experimental conditions.

Optimum quantum yield of photosystem II (F v /F m)

Differences in F v /F m were observed among frond sections, and temporal responses varied by frond type. Model comparison indicated that temporal changes were better described by a quadratic trend in apical and middle fronds, whereas a linear trend was sufficient for sporophylls (Supplementary Table 6). In apical fronds, F v /F m declined over time, with lower values under high irradiance (I50) and higher temperature (10 °C) (Supplementary Figure 2). This response was primarily associated with temperature (p = 0.016) and irradiance (p = 0.036) (Table 2), and the temporal trend was well described by a quadratic term (p = 0.001). In middle fronds, F v /F m also changed over time, but the temporal trajectory depended on the specific combination of factors: it differed between UE and PC (p = 0.004) and between irradiance treatments (p < 0.0001) (Table 2; Supplementary Figure 3). In sporophylls, the temporal trend was less pronounced (Supplementary Figure 4), and no significant effects or interactions were detected, although irradiance showed a marginal trend (p = 0.053).

Photosynthetic pigment responses

Photosynthetic pigment composition varied among frond types. In apical fronds, the multivariate pigment profile differed significantly (p < 0.001), and this difference was driven primarily by variation in Chl a (p = 0.002) and Fucox (p = 0.002) (Table 3; Supplementary Figure 5A). In middle fronds, the multivariate pigment profile also differed significantly (p = 0.002) (Table 3; Supplementary Figure 5B), although no individual pigment showed a significant effect when analyzed separately. In sporophylls, pigment composition did not differ significantly (p = 0.134) (Table 3). However, compared with apical and middle fronds, pigment composition appeared more stable across treatments (Supplementary Figure 5C).

Pigment ratios also varied among frond types, indicating that responses differed among treatment combinations along the thallus (Figure 7A, B; Supplementary Table 7). In apical fronds, both Chl c /Chl a (p = 0.012) and Fucox/Chl a (p < 0.001) differed significantly among treatment combinations (Table 4). For Chl c /Chl a, the clearest difference was observed between PC at 10 °C-I5 and UE at 6 °C-I5, with higher values in the former (p = 0.005) (Figure 7A; Supplementary Table 7). For Fucox/Chl a, differences among treatment combinations were more pronounced: UE at 10 °C-I5 showed lower values than UE at 6 °C-I5, PC at 6 °C-I5, PC at 10 °C-I5, PC at 6 °C-I50, and PC at 10 °C-I50 (p < 0.001 in all cases) (Figure 7B; Supplementary Table 7). In addition, PC at 10 °C-I5 (p < 0.001) and PC at 10 °C-I50 (p = 0.030) showed higher Fucox/Chl a values than UE at 6 °C-I5 (Supplementary Table 7). In middle fronds, both pigment ratios showed significant differences among treatment combinations, although only a few pairwise contrasts reached significance (Figure 7A, B; Supplementary Table 7). Chl c /Chl a differed significantly (p = 0.020) (Table 4), with PC at 10 °C-I50 showing higher values than UE at 10 °C-I50 (p = 0.018) (Figure 7A; Supplementary Table 7). Fucox/Chl a also differed significantly among treatment combinations (p = 0.026) (Table 4), with higher values in PC at 10 °C-I50 than in UE at 6 °C-I5 (p = 0.023) (Supplementary Table 7). In sporophyll fronds, both pigment ratios differed significantly among treatment combinations (Figure 7A, B; Supplementary Table 7). Chl c /Chl a differed significantly (p = 0.011) (Table 4), with higher values in PC at 6 °C-I5 (p = 0.008) and PC at 6 °C-I50 (p = 0.048) than in UE at 6 °C-I5 (Figure 7A; Supplementary Table 7). Fucox/Chl a also differed significantly among treatment combinations (p = 0.026) (Table 4), with both PC at 6 °C-I5 (p = 0.043) and PC at 6 °C-I50 (p = 0.032) showing higher values than UE at 6 °C-I5 (Figure 7B; Supplementary Table 7). Overall, the significant contrasts in sporophylls were fewer and involved mainly higher ratio values in specific PC treatment combinations under 6 °C. However, several treatment combinations could not be compared because of total mortality (Figure 7A, B).

Microbial community structure

A total of 36 algal samples were successfully sequenced, including 16 from UE and 20 from PC (Supplementary Table 8). Because microbiota was characterized only on day 15, the dataset was restricted to samples that survived until the end of the experiment. Consequently, differential mortality among treatment combinations determined the final sampling structure for microbial analyses. High-throughput sequencing of epiphytic bacterial communities associated with M. pyrifera tissue discs generated 5,356,732 raw paired-end reads. After quality filtering, 653,200 high-quality sequences were retained, with an average length of 479.5 bp (± 16.5 SD). These sequences resulted in 2,724 ASVs assigned to 13 phyla, 21 classes, 61 orders, 94 families, and 199 genera. Diversity analyses were based on the rarefied ASV table (6,674 sequences per sample), which retained 2,650 ASVs across the 36 samples. Alpha diversity was evaluated descriptively only because of the limited sample size. Communities from PC tended to show higher Shannon diversity, observed richness, and Chao1 than those from UE in several cases, particularly in apical and middle fronds (Supplementary Figures 6–8), although this tendency was not consistent across all frond types and environmental conditions. Beta diversity was structured mainly by site of origin. In the marginal PERMANOVA (each factor adjusted for the others), site had a strong and highly significant effect under both Bray-Curtis (pseudo-F = 22.40, R² = 0.35, p = 0.0001) and Jaccard distances (pseudo-F = 11.86, R² = 0.24, p = 0.0001). Temperature had a smaller but significant effect (Bray-Curtis, R² = 0.06, p = 0.004; Jaccard, R² = 0.05, p = 0.006), and irradiance a marginal one (Bray-Curtis, R² = 0.03, p = 0.048; Jaccard, p = 0.054). In contrast, frond type had no significant effect once the other factors were accounted for (Bray-Curtis, p = 0.108; Jaccard, p = 0.146) (Supplementary Table 9), a result confirmed by pairwise comparisons among frond types, none of which were significant after Benjamini-Hochberg correction (all adjusted p > 0.34) (Supplementary Table 10). Tests of multivariate dispersion (betadisper) were non-significant for all factors (all p > 0.13) (Supplementary Table 11). Ordination (PCoA) showed a clear separation between UE and PC across all treatment combinations (Figure 8; Supplementary Figure 9).

Order-level relative abundance analyses revealed differences in the relative abundance of bacterial orders between sites, whereas variation among frond types and experimental treatments was less pronounced (Figure 9). In most samples, Alteromonadales was the dominant order, frequently accounting for more than 50% of the relative abundance (Figure 9). This dominance was especially evident at UE, where communities associated with apical, middle, and sporophyll fronds were often enriched in Alteromonadales under both temperature treatments (Figure 9). In contrast, communities from PC were more heterogeneous, generally showing lower relative dominance of Alteromonadales and greater contributions from Flavobacteriales and, to a lesser extent, Rhodobacterales; this pattern was most pronounced at 10 °C, where the combined contribution of these two orders increased markedly relative to the more Alteromonadales-dominated profile observed at 6 °C (Figure 9). This temperature-associated shift was particularly apparent in apical and middle fronds, whereas sporophyll-associated communities remained largely dominated by Alteromonadales regardless of temperature, with only secondary contributions from Flavobacteriales and Rhodobacterales (Figure 9). At UE, Rhodobacterales showed the opposite trend, contributing more at 6 °C than at 10 °C in middle fronds, indicating that the effect of temperature on these taxa was site-dependent (Figure 9).

Despite these taxonomic differences between sites, predicted functional profiles were broadly similar and dominated by heterotrophic functions (Supplementary Figures 10, 11). FAPROTAX analysis identified 31 functional groups in UE-associated communities and 32 in PC-associated communities across the experimental treatments. At both sites, predicted functional profiles were consistently dominated by chemoheterotrophy (>25%), which accounted for the largest proportion of annotated functions across frond types and temperature-irradiance combinations. Secondary predicted functions were mainly associated with nitrate reduction (3.4–20.4%) and fermentation (2.9–20.0%), with moderate contributions across several treatments. In addition, sulfur-related functions, including dark sulfite oxidation, dark sulfur oxidation, and dark oxidation of sulfur compounds (0.7–7.1%), were more consistently represented in PC-associated communities than in UE. Other predicted functions, including phototrophy-related categories, were present at very low relative abundances across treatments (approximately 1%).

The responses of M. pyrifera were mainly shaped by two factors: local environmental conditions and, in the case of the host, functional differentiation among frond types. The contrasting underwater light environments of the two sites provide the framework for interpreting these responses. At UE, the strong attenuation of irradiance with depth and the loss of blue-violet wavelengths reflect the suspended particles derived from glacial input, a condition typical of sub-Antarctic environments where turbidity constrains both the quantity and the spectral quality of the light available for photosynthesis (Huovinen et al., 2020; Coral-Santacruz et al., 2024). At PC, in contrast, light penetrated to greater depth and the spectrum was less altered, with a greater representation of the green-yellow range usable by accessory pigments such as Fucox (Marambio et al., 2017). Because light availability along the vertical profile is a main driver of photoacclimation in this species (Colombo-Pallotta et al., 2006; Marambio et al., 2017; Gómez et al., 2025), the greater light availability at PC would not necessarily translate into better photosynthetic performance; on the contrary, it may have increased photoinhibitory pressure, contributing to the lower F v /F m values recorded there. Notably, UE and PC also differ in salinity and nutrient availability, variables not manipulated here, which could additionally contribute to the site-level responses.

F v /F m was higher at UE than at PC, suggesting a more favorable photophysiological state at UE, whereas individuals from PC may have experienced greater photoinhibitory pressure or a higher sensitivity of the photosynthetic system to local conditions (Gevaert et al., 2002). F v /F m also varied among frond types, with higher values in sporophylls and lower values in apical fronds, consistent with the ecophysiological stratification described along the thallus in M. pyrifera, where surface tissues are exposed to greater irradiance and environmental variability, whereas deeper tissues develop under more shaded conditions (Colombo-Pallotta et al., 2006; Edwards and Kim, 2010; Celis-Plá et al., 2021). In this context, the lower potential efficiency of photosystem II in apical fronds is consistent with a greater demand for photoacclimatory regulation under highly exposed conditions (Skene, 2004; Hanelt et al., 2003). The pigment responses varied along the thallus with different magnitudes and sensitivities, and their interpretation requires distinguishing between variables that showed a consistent gradient and those that did not. Chl c and Fucox varied mainly with frond type, being higher in apical tissues; this constitutes a consistent spatial gradient along the thallus, in agreement with the classic photoacclimation response in M. pyrifera, in which these pigments increase toward the more exposed tissues (Koch et al., 2016; Dambeck and Sandmann, 2014; dos Santos et al., 2019). Fucox additionally differed between sites, consistent with its role in photoprotection and light harvesting under high irradiance (Marambio et al., 2023). In contrast, Chl a was affected by site, frond type, and their interaction, so that the difference between sites depended on the frond type considered (the sites differed mainly in middle fronds). For this reason, Chl a should not be described as a single spatial gradient; rather, its variation was contingent on the combination of site and frond type, indicating site-specific photoacclimation in particular tissues (Marambio et al., 2017; Celis-Plá et al., 2021).

The comparison between natural populations and experimental responses reveals an apparent discordance that is, in fact, informative. In natural populations, individuals from UE exhibited the highest F v /F m values and sporophylls showed the highest photochemical efficiency of all frond types, suggesting a favorable photophysiological condition. However, these same categories proved to be the most vulnerable under experimental conditions: UE tissues had the highest mortality under the most demanding treatments, and sporophylls were the most sensitive overall. Far from contradicting each other, these observations point to a mismatch between the conditions to which each population is acclimated and those imposed experimentally. A high baseline photochemical efficiency reflects successful acclimation to the prevailing local conditions, but it does not imply a broad tolerance range. Tissues finely tuned to a restricted environmental envelope, such as the low-light and low-temperature conditions of a glacier-influenced site, may be the least able to respond when that envelope shifts. Consequently, the ecophysiological condition measured in situ should not be interpreted as a direct indicator of resilience to environmental change, and assessments based solely on field indicators such as F v /F m may overestimate the robustness of populations that appear to be in good condition within their own environmental context. A similar argument is made by Britton et al. (2024) in Ecklonia radiata: populations at the cool edge of the distribution are commonly assumed to be more heat-tolerant, or even to benefit from warming as they approach their thermal optimum, yet in one such population the modelled thermal optimum was significantly lower than in warm-edge populations, reflecting acclimatization or local adaptation to the thermal conditions of the site, which implies a narrower thermal safety margin under future warming rather than greater tolerance.

This mismatch requires considering the environmental history of the populations of origin. The experimental irradiance of 50 μmol photons m - ² s - ¹ does not represent a high irradiance in absolute terms for brown macroalgae; however, it represents a nearly six-fold increase relative to the in situ PAR recorded at UE (8.40 μmol photons m - ² s - ¹), a site characterized by low temperature and strong light attenuation. Similarly, 10 °C, although corresponding to the mean summer temperature recorded in glacier-influenced environments of the MSAE (Bahamonde et al., 2022), represents an increase relative to the in situ temperature measured at UE at the time of collection (6.88 °C). In this context, the higher mortality observed in UE tissues under the most demanding combinations is consistent with a mismatch between the condition of the population of origin and the conditions imposed experimentally, rather than with a non-specific experimental stress.

Several lines of evidence support the interpretation of an organized physiological response, and not a handling artifact. Mortality was structured according to site of origin and the combination of temperature and irradiance, a pattern inconsistent with a generic deterioration due to transport or handling, which would be expected to be more homogeneously distributed among treatments. The pigment adjustments reinforce this reading: the Fucox/Chl a ratio, governed more by site identity and temperature than by irradiance in isolation, reached its lowest value in apical fronds from UE under 10 °C-I5, immediately before the total mortality recorded under 10 °C-I50. This sequence, in which the deterioration of pigment adjustment precedes mortality, agrees with the findings of Chaloub et al. (2025) in Sargassum natans, who found that the capacity for non-photochemical photoprotective dissipation (NPQ) collapses at high temperatures before F v /F m declines. The authors identify this loss of photoprotection, and not the decline in F v /F m, as the event that triggers photosynthetic collapse. Although Fucox/Chl a and NPQ are distinct measures, both reflect the capacity of the photosynthetic system to dissipate the excess of absorbed energy, suggesting that, at UE, the loss of this regulatory capacity may have occurred earlier and contributed to mortality, rather than being merely a late symptom of already doomed tissue. The absence of data in several UE combinations, due to total mortality, precludes a direct comparison of pigment plasticity between sites.

In the same vein, the mixed-effects models showed that the temporal trajectory of F v /F m in middle fronds depended on significant site × time (p = 0.004) and irradiance × time (p < 0.0001) interactions: the course of photochemical efficiency responded to specific combinations of factors and not to an isolable effect of each variable, as would be expected if tolerance depends on the environmental history of each population. This contextual dependence aligns with findings from other studies of M. pyrifera, where increasing temperature reduces growth and compromises photosynthetic status and survival, with light availability strongly modulating these effects (Sánchez-Barredo et al., 2020; Bunting et al., 2024, 2025; Umanzor et al., 2020). Equally species- and site-specific responses have been reported in Arctic kelps under simulated marine heatwaves, where Laminaria digitata showed no significant changes whereas Saccharina latissima and Alaria esculenta did exhibit them (Lebrun et al., 2025), supporting the idea that tolerance depends on the condition and history of the organism, and not solely on the magnitude of the stressor. This dependence of M. pyrifera responses on local conditions is consistent with previous studies documenting the role of environmental history and population differentiation (Fernández et al., 2021; Solas et al., 2024).

Tissue-specific differentiation was also evident in the photochemical and pigment responses. Apical fronds combined a marked decline in F v /F m with pronounced pigment adjustments, indicating both sensitivity and plasticity in the most exposed tissue, whereas middle fronds expressed change mainly at the multivariate level, suggesting a coordinated adjustment of pigment balance rather than marked shifts in individual pigments. Sporophylls, in turn, showed a weak F v /F m response and relatively stable pigment concentrations despite high mortality. This decoupling suggests that stress in reproductive tissues may not be fully captured by photophysiological metrics alone, but may instead involve structural deterioration or shifts in resource allocation (Lobban and Harrison, 1994; Graham et al., 2007); consistently, dos Santos et al. (2019) showed that the vegetative and reproductive phases of M. pyrifera differ in pigment composition and photosynthetic capacity. Overall, M. pyrifera cannot be treated as a functionally uniform organism, but rather as a heterogeneous structure in which tissues differ in their baseline condition, plasticity, and tolerance.

One aspect of the design that warrants caution when interpreting the ecophysiological results is salinity, which in this study was preserved at native levels for each population rather than manipulated. Although it differs markedly between the two sites and is a variable strongly modified by glacial melt, it was not evaluated as an experimental factor. More broadly, research on high-latitude kelps remains biased toward temperature and light, with salinity and sediment load receiving comparatively little attention despite their rapid modification by glacial retreat (Farrugia-Drakard et al., 2023). For example, in an Arctic glacial fjord it was observed that S. latissima was able to adjust to, and even benefit from, increased temperature, whereas reduced salinity impaired its performance and triggered marked transcriptomic responses (Li et al., 2020). Nevertheless, pigment content and F v /F m were largely insensitive to salinity changes (Monteiro et al., 2020), suggesting that the photophysiological and pigment variables measured here may be comparatively robust to salinity differences; consequently, their omission probably does not compromise these particular variables, although it does limit extrapolation to other physiological descriptors. Incorporating salinity as an experimental factor nonetheless represents a priority for future work in glacier-influenced systems. In addition, nutrients were not supplemented during the experiment; although partial water renewal with native seawater was performed, progressive nutrient limitation cannot be entirely ruled out as a contributing co-factor towards the end of the 15-day period.

The exploratory analysis of the associated microbiota adds a complementary dimension to this framework. Among the samples that survived until the end of the experiment, the structure of the bacterial communities was determined primarily by site of origin, with a minor contribution from temperature and irradiance, and no effect of frond type. Thus, the environmental control associated with each locality exerted a stronger influence on the epiphytic microbiota than tissue identity, even though host ecophysiology did differ among frond types. This result is consistent with Rodríguez (2026), who identified environmental context as the main axis of microbial differentiation in natural populations of M. pyrifera from these same sites, and aligns with the broader idea that kelp biochemistry and physiological condition are shaped by the local environment (Qiu et al., 2019; Soto et al., 2024; Niedzwiedz et al., 2025).

Beyond this geographic organization, UE communities showed a greater dominance of Alteromonadales, whereas those of PC were more heterogeneous, with larger contributions of Flavobacteriales and Rhodobacterales. Despite these taxonomic differences, the functional profiles predicted by FAPROTAX were similar between sites, predominantly related to chemoheterotrophy, indicating a possible shared metabolic background. Beyond this common background, the ecological role of the dominant orders detected suggests a plausible, hypothesis-generating relationship with host ecophysiology. Alteromonadales, which frequently exceeded 50% of relative abundance at both sites, has been linked with carbon remineralization, the use of algal exudates, vitamin provision, and antagonistic functions against opportunistic taxa (Sawabe et al., 1998; Wang et al., 2008; Thomas et al., 2021), functions that could provide nutritional support and protection to the host.

This microbial reorganization coincides temporally with the ecophysiological changes described in the previous sections. At UE, the temperature increase to 10 °C coincided with a relative reduction of Rhodobacterales (a group that includes recognized producers of cobalamin (vitamin B12), an essential cofactor of algal metabolism; Croft et al., 2005; Dogs et al., 2017) and with a greater homogenization of the community toward Alteromonadales. This pattern of homogenization associated with the temperature increase is consistent with the phenomenon of dysbiosis described in the surface microbiome of M. pyrifera by Minich et al. (2018), and in the kelp Nereocystis luetkeana, where experimental warming was observed to reduce taxa considered beneficial to the host while favoring the proliferation of previously low-abundance lineages (Molnar et al., 2025). The functional relevance of this type of disruption has been documented experimentally in gametophytes of the kelp E. radiata, where microbiome disruption under thermal stress drastically reduced survival (Veenhof et al., 2025), supporting the coupling between thermal stress, dysbiosis, and host performance observed in this study. It is worth noting that in the 10 °C-I5 combination, the growth rate of apical fronds reached its highest value, reinforcing the idea, already discussed, that apparent growth does not necessarily capture the underlying physiological and microbial deterioration. Although this correspondence is descriptive and cannot be interpreted as causal, it is consistent with the hypothesis that host physiological deterioration under its least favorable thermal condition coincides with a simplification of its associated bacterial community, potentially including the loss of lineages with nutritional support functions such as vitamin B12 provision.

At PC, in contrast, the same temperature increase was not associated with generalized mortality in apical and middle fronds, since the most marked mortality at PC occurred under high irradiance, in apical fronds at 6 °C-I50 and sporophylls at 10 °C-I50, whereas growth became negative specifically under 10 °C-I50. On the contrary, the temperature increase coincided with an increase in taxonomic diversity (a greater contribution of Flavobacteriales and Rhodobacterales) and with relatively higher Fucox/Chl a values than at UE under equivalent thermal conditions. This suggests that, at the site with greater baseline photoacclimatory plasticity, the temperature-associated microbial change could reflect a functional adjustment of the holobiont, possibly linked to the production of exudates or mucilage (English et al., 2025), rather than a sign of deterioration. At UE, where the thermal tolerance margin is narrower, the same environmental change coincided with signs of stress both at the host level and in its associated microbiota. Taken together, these functional attributions (both the shared metabolic background inferred by FAPROTAX and the specific roles attributed to Alteromonadales and Rhodobacterales) should be regarded as hypotheses consistent with the taxonomic patterns and not as evidence of mechanism, since in this study the functional assignments are based on taxonomic inference and not on direct measurements of gene expression or metabolic activity.

These results can be placed in the context of the ongoing environmental changes in the MSAE. Glaciers along the Andes, including the Patagonian Icefields adjacent to the study area, have experienced sustained mass loss over the past two decades in response to regional warming (Dussaillant et al., 2019), and continued glacial melt is expected to increase freshwater input, reinforcing the stratification and salinity gradients that characterize these coastal systems (Bahamonde et al., 2022). The results obtained do not allow quantitative predictions for M. pyrifera populations, given the single-site design and the short experimental duration; nevertheless, they offer a mechanistic basis for anticipating the likely direction of their response. Because tolerance to temperature and irradiance was contingent on the environmental history of the population and on tissue identity, the vulnerability of these populations is unlikely to be uniform: tissues acclimated to the low-light and low-temperature conditions of glacier-influenced sites could be particularly susceptible if the light or thermal regime shifts beyond the range to which they are acclimated, as suggested by the higher mortality of UE tissues under the most demanding experimental combinations. This is consistent with global evidence that warming has strong negative effects on kelps across life-cycle stages and physiological levels, intensifying with the magnitude and duration of exposure (Roethler et al., 2025).

By jointly assessing host ecophysiology and the associated epiphytic microbiota in M. pyrifera from contrasting glacier- and ocean-influenced sites, this study contributes to filling several knowledge gaps for the species in the sub-Antarctic region. First, it extends to an experimental context the functional differentiation among frond types previously documented mainly in natural populations and along depth gradients (Marambio et al., 2017, 2023; Coral-Santacruz et al., 2024), showing that frond types differ not only in baseline condition but also in their tolerance and plasticity when exposed to controlled changes in temperature and irradiance. Second, it provides an integrated, holobiont-oriented characterization in which host ecophysiology and epiphytic microbiota are examined simultaneously; this approach remains scarce for M. pyrifera and, to our knowledge, has not been previously applied to glacier-influenced sub-Antarctic populations. Third, it identifies the site of origin as a shared organizing axis of both host and microbial responses, providing a framework and testable hypotheses for future studies that explicitly evaluate host–microbiota interactions under environmental change through balanced designs and direct measurements of microbial activity.

This study addressed three questions, each linked to an explicit hypothesis stated in the Introduction. The first hypothesis, that the glacier-influenced population (UE) would show a low-light photoacclimation signature, was supported: UE individuals exhibited higher F v /F m and distinct pigment profiles than those from PC. The second, that frond types would differ in tolerance and plasticity, was also supported, though with nuances: middle fronds were the most resistant, apical fronds the most photochemically sensitive yet the most plastic, and sporophylls the most vulnerable overall, a vulnerability not always captured by F v /F m. The higher mortality of UE tissues under the most demanding conditions was consistent with a mismatch between the environmental history of the population and the imposed conditions, rather than with a non-specific stress response. The third hypothesis, that the epiphytic microbiota would be structured primarily by site, was likewise supported: site was the main axis of community structure, with no significant effect of frond type once other factors were accounted for, and only minor contributions of temperature and irradiance.

The convergence of host ecophysiology and epiphytic microbiota along the site axis is consistent with a holobiont perspective, in which local environmental context acts as a common organizer of the responses of both the host and its associated bacterial community, a relationship that remains to be evaluated directly in future studies. These findings should nonetheless be interpreted in light of three main limitations. First, the use of a single site per environmental condition prevents separating the effect of environment type from that of the specific locality. Second, because the two sites were sampled in different years (austral summers of 2022 and 2023), interannual variability cannot be separated from site identity, and the design does not allow natural variability to be characterized across spatial or temporal scales. Third, differential mortality conditioned the sampling of the microbiota. Even so, the study underscores the need to incorporate both population-level environmental context and within-thallus functional heterogeneity when assessing the vulnerability and acclimation capacity of sub-Antarctic kelp forests under ongoing environmental change.