Multiple interacting environmental drivers reduce the impact of solar UVR on primary productivity in Mediterranean lakes

Solar radiation, including UVR (280–400 nm), is a key energy source for phytoplankton, and its underwater availability depends on water optical properties and constituents that modulate attenuation. Concurrent global-change drivers—nutrient inputs from runoff and dust deposition, increasing atmospheric CO2, and altered mixing regimes due to extreme weather—also influence phytoplankton. While many studies examine single drivers, fewer quantify interactions among multiple drivers, whose combined effects can be non-additive and context dependent. Lakes, highly productive yet vulnerable ecosystems, are experiencing shifts from clear to turbid conditions. However, no in situ studies have simultaneously tested how UVR effects on natural phytoplankton communities are modified by concurrent CO2, mixing, and nutrients across a transparency gradient. The study aims to determine how multiple interacting drivers modulate primary productivity (P) and how this modulation varies with water transparency (KdPAR). The authors hypothesize that increased CO2 and nutrients under fluctuating light regimes (Mix) will accentuate UVR-induced inhibition of P, with stronger effects in communities from darker environments.

Prior work shows complex, often non-linear interactions among drivers: nutrient inputs can stimulate biomass, elevated CO2 alone may have little effect, yet together they can synergistically increase carbon biomass. UVR effects can be masked or unmasked by nutrient status; enrichment has revealed detrimental UVR impacts on growth and primary production. Interactions are species- and context-dependent, varying with transparency; for instance, UVR×Nut stimulated production in clear Mediterranean lakes but was antagonistic in less transparent Andean lakes. UVR×Mix interactions increased inhibition of primary production and excretion of organic carbon in turbid lakes, with opposite patterns in clear systems. DOM can both attenuate UVR (sheltering organisms) and alter community structure by favoring smaller phytoplankton, influencing acclimation and photoprotection (e.g., MAAs). Despite the importance of light regime shifts, experimental studies probing these effects within multi-driver scenarios remain scarce, highlighting the need addressed by this work.

Study design comprised two components: (1) a literature survey of global lake transparency and (2) an in situ multi-driver factorial experiment across a transparency gradient in Mediterranean lakes. - Literature survey: Compiled 421 KdPAR estimates (1960–2018) from Scopus and unpublished data to characterize global distributions (temperate, tropical, boreal/polar) and provide context for the Mediterranean lakes’ KdPAR. - Study sites: Nine mixed, oligotrophic lakes in Spain spanning a KdPAR gradient of 0.18–0.90 m−1: five high-mountain lakes in Sierra Nevada National Park (siliceous bedrock) and four in Lagunas de Ruidera Natural Park (calcareous substrates with high nitrate). Physical-chemical variables measured included temperature, TP, TN, SiO2, DOC, Chl a, and total biomass; phytoplankton community composition was assessed (diatoms dominant in most lakes; dinoflagellates in two). - Sampling and incubation: Surface water (0.5 m) collected in July 2012, pre-screened (45 µm) to remove large zooplankton. For each lake, water was divided into twelve 2-L PET bottles and incubated overnight at in situ temperature under a 2×2 combination of pCO2 and nutrient treatments: CO2 at ambient (~400 ppm, air-bubbled) or elevated (750 ppm, gas-bubbled to mimic RCP8.5; pH decreased ~0.28 units), and nutrients at ambient or enriched (1 µM P as NaH2PO4 and 29.03 µM inorganic N as NH4NO3; N:P=30, simulating Saharan dust inputs). Total CO2 calculated from alkalinity and pH. - In situ exposure: Following overnight incubation, subsamples were transferred to 50-mL quartz vessels and exposed in the source lakes for 4 h centered at local noon under a 2×2 combination of light quality and mixing treatments: UVR treatment with +UVR (PAR+UVR >280 nm) using uncovered quartz vessels and −UVR (PAR only >400 nm) using UVR-blocking foil; mixing treatment with static exposure at 0.5 m depth or fluctuating exposure using a motorized simulator cycling between surface and 3 m (1 m every 4 min; 10 cycles), approximating in situ mixing speeds. Trays were deployed ~2 m from boats to avoid shading; irradiance for static vs mixed treatments was comparable on average, though lake-specific differences existed. - Measurements: Primary production assessed via 14C-NaHCO3 incorporation in 50-mL samples during in situ exposure (4 h). Physical profiles of UVR (305, 320, 380 nm) and PAR (400–700 nm) using Biospherical radiometers; Kd values determined from ln downwelling irradiance vs depth. Temperature profiles recorded with multiparametric probe. Chemical analyses for TP, TN, Si, DOC followed standard methods. Chl a measured from GF/F filters. Phytoplankton abundance and composition determined by Utermöhl method in Lugol’s-fixed samples. - Experimental design and statistics: For each lake, a 2×2×2×2 full factorial design (UVR × CO2 × Mix × Nut; triplicate) was implemented. Effects on primary productivity (P) were quantified using natural log response ratios (lnRR) for single, double, triple, and four-way interactions (corrected formulation). Differences among lakes and treatments tested by one-way ANOVA with LSD post hoc tests. A five-way ANCOVA evaluated potential modulation by TN:TP, in situ temperature, and mean total irradiance, treating lakes and experimental factors as fixed and the three environmental variables as covariates. Relationships between lnRR (single, double, triple, interactive) and KdPAR were assessed via polynomial regression (chosen over linear based on higher explained variance and lower AIC); model assumptions checked by residual analyses. Global mean irradiances for static vs mixing compared by t test; UVR and PAR exposure comparisons included.

- UVR dominance: UVR was the dominant driver of primary productivity responses; its individual effect size was on average 3–60 times larger than that of CO2, Mix, or Nut, and exhibited the greatest magnitude differences among treatments. - General inhibition: In 6 of 9 lakes, all four drivers individually reduced P (Pe rates 0.02–0.80 h−1), with Nut and Mix less inhibitory than UVR or CO2 (|lnRR| for Nut and Mix < 0.52). - Interactions mitigate inhibition: Many double and triple interactions showed reduced inhibition relative to single-driver effects, and about half of these interactions were antagonistic on P (i.e., effects less negative than expected). - Four-way interaction: The UVR×CO2×Mix×Nut combination synergistically reduced P in darker lakes (e.g., Río Seco Superior), but the magnitude (minimum lnRRinteractive < −0.7) was 40–60% lower than the strongest two-way (minimum lnRR ≈ −1.5), three-way (minimum lnRR ≈ −1.2), and single UVR (minimum lnRR ≈ −2) effects. - Transparency dependence: Along the KdPAR gradient (0.18–0.90 m−1), UVR inhibition of P increased with KdPAR, while the inhibitory effect of Nut decreased with KdPAR; CO2 and Mix showed no significant trend with KdPAR. Double and triple interactions (UVR×Nut; UVR×CO2×Mix; UVR×CO2×Nut; UVR×Mix×Nut) exhibited unimodal shifts from slight antagonism to synergism as KdPAR increased. The four-way lnRRinteractive similarly shifted from antagonistic to synergistic over the KdPAR gradient. - Averaged effects: Mean lnRR values across lakes indicated diminishing inhibition with increasing driver count: single −0.23; double 0.04; triple 0.03; four-way 0.19. - Covariates: No significant effects of TN:TP ratios, in situ temperature, or mean solar irradiance on P responses to UVR, CO2, Mix, Nut, or their interactions.

Water transparency (KdPAR) emerged as a key predictor of how multiple drivers affect phytoplankton primary productivity in Mediterranean lakes. Increasing environmental complexity (more interacting drivers) tended to reduce the magnitude of inhibitory effects, especially those driven by UVR, supporting the concept that community responses to multiple drivers are strongly governed by the dominant single driver. The strongest UVR inhibition occurred in darker waters, consistent with prior observations from turbid systems. Likely mechanisms include lower photoprotective capacity (e.g., fewer MAAs) in communities chronically sheltered by DOM, and physiological downregulation (e.g., Calvin cycle/RuBisCO, excess reducing power not dissipated via NPQ), diverting carbon and energy toward stress responses and storage compounds. Despite synergy at high KdPAR, the four-driver scenario produced smaller inhibitory magnitudes than simpler interactions or UVR alone, potentially due to increased photosynthetic efficiency in low-photon environments, stress-induced community tolerance, and the capacity of small, variable systems to buffer multiple stressors. Communities from darker environments may gain a competitive advantage under interacting drivers, and reductions in maximal photosynthetic efficiency under multiple drivers may be less severe in darker near-surface waters than in clear waters. Results were robust to variation in nutrient ratios, temperature, and irradiance within the tested ranges.

Low water transparency under multiple interacting drivers can synergistically reduce near-surface primary productivity by about 40% in Mediterranean lakes, yet the magnitude of this negative impact is 40–60% lower than when UVR acts alone. UVR is the dominant inhibitory driver, with its effects intensifying as waters darken, while increasing numbers of interacting drivers tend to temper its impact. These findings highlight the predictive value of water transparency for multi-driver responses and suggest that communities in darker environments may be less susceptible to compounded reductions in photosynthetic efficiency. Future research should extend to longer temporal scales and whole-ecosystem studies to refine projections of global change impacts and guide management strategies.

- Short-term, acute in situ exposures (4 h at local noon) capture immediate physiological responses and do not allow acclimation or adaptation, limiting inference on longer-term dynamics. - Spatial scope restricted to nine Mediterranean lakes; generalization to other regions and lake types should be cautious. - Exact matching of UVR and PAR exposures between static and mixing treatments across all lakes was not fully achievable due to lake-specific attenuation differences. - Community compositions differed among lakes, potentially influencing species-specific responses, although statistical controls suggested covariates (TN:TP, temperature, irradiance) did not significantly modulate outcomes within the experimental ranges.