Phytoplankton primary productivity in aquatic ecosystems is fundamentally driven by solar radiation, including ultraviolet radiation (UVR). Water transparency, a key factor influencing the penetration and spectral composition of underwater light, is increasingly affected by global change factors such as increased rainfall, runoff, and dust deposition. These changes, along with rising atmospheric CO2 and altered mixing regimes, create a complex interplay of environmental drivers that unpredictably impact phytoplankton communities and lake functioning. This research addresses the critical knowledge gap concerning how these interacting drivers affect primary productivity, focusing particularly on the potential moderating role of decreased water transparency. The central hypothesis is that increases in CO2 and nutrients under fluctuating light regimes will exacerbate the inhibitory effect of UVR on primary productivity, especially in darker environments. Understanding these interactions is crucial for predicting the future of lake ecosystems and their role in the carbon cycle, given their high productivity and vulnerability to environmental shifts. Previous research predominantly focused on the individual effects of these drivers, overlooking the importance of their combined influence, which often results in non-additive, synergistic, or antagonistic outcomes. This study aims to quantify the combined effects of UVR, CO2, nutrients, and mixing regimes on phytoplankton primary productivity in Mediterranean lakes across a gradient of water transparency, using an *in situ* experimental approach.

Numerous studies have explored the individual impacts of global change drivers on aquatic ecosystems. For instance, increased nutrient inputs are known to stimulate phytoplankton biomass, while elevated CO2 concentrations may have less pronounced or even contrasting effects. However, the combined impact of these drivers often deviates from simple additive models, exhibiting synergistic or antagonistic interactions. Previous research has shown the synergistic effects of nutrient enrichment and UVR on phytoplankton, highlighting the complex responses and species-specific adaptations involved. Furthermore, the interaction between UVR and mixing regimes also exhibits context-dependent effects, with synergistic inhibition of primary productivity and increased organic carbon excretion in turbid lakes, but opposing trends in clearer waters. Despite the ecological significance of lakes and the increasing prevalence of turbid conditions, *in situ* experimental studies that concurrently examine the impacts of UVR, CO2, nutrients, and mixing regimes under varying light environments remain scarce. This lack of comprehensive research makes it difficult to accurately assess the collective influence of these environmental stressors on lake primary productivity.

This study utilized a full factorial experimental design to assess the individual and interactive effects of UVR, CO2, nutrient concentration, and mixing regimes on primary productivity (P) in nine Mediterranean lakes exhibiting a gradient of water transparency (KdPAR). A literature review was conducted to determine global KdPAR values for comparison. The study area comprised lakes from the Sierra Nevada and Lagunas de Ruidera Natural Parks, differing in altitude, nutrient levels and phytoplankton community composition (diatoms and dinoflagellates were dominant). Surface water samples were collected and subsequently divided into 12 2-L PET bottles. These were incubated overnight under two pCO2 levels (ambient and 750 ppm, simulating RCP 4.5 scenario) and two nutrient concentrations (ambient and enriched). After incubation, samples were further divided into 50-mL quartz vessels and incubated *in situ* for 4 hours centered at local noon, under two light qualities (+UVR and -UVR, achieved using UV-filter foil) and two mixing regimes (static and fluctuating, mimicking in-situ conditions using a custom-built mixing device). Primary productivity (Pc) was measured using 14C incorporation. Data analysis involved one-way ANOVA and ANCOVA to test for significant differences among lakes and treatments, with polynomial regression analyses used to assess relationships between lnRR (natural logarithm response ratios) and KdPAR. This approach aimed at evaluating the significance of single, double, triple, and four-level interactions under different light conditions. Statistical assumptions (normality, homoscedasticity) were checked prior to analysis. Data on physical and chemical variables (temperature, nutrient concentrations, chlorophyll a, biomass) were also obtained for each lake.

The study revealed that UVR was the dominant driver affecting primary productivity (P), exerting an effect 3 to 60 times stronger than the other drivers. UVR also demonstrated the largest difference in driver magnitude across treatments. Double and triple interactions among drivers showed two response patterns: a weaker inhibitory effect of individual drivers (particularly UVR and CO2) and an antagonistic effect in approximately 50% of double and triple interactions. The four-factor interaction (UVR × CO2 × Nut × Mix) exerted a synergistic, inhibitory effect on P, most prominently in lakes with KdPAR > 0.7 m-1. However, the magnitude of this four-factor interaction was 40-60% lower than the effect of UVR alone or two- and three-level interactions. The study found no significant effects of TN:TP ratios, in situ temperature, or mean solar irradiance on the response of P to the drivers and their interactions. The relationship between lnRR and KdPAR revealed that UVR and Nut had increasing and decreasing inhibitory effects, respectively, on P with increasing KdPAR. By contrast, CO2 and Mix had no significant effects along the KdPAR gradient. Double and triple interactions showed a unimodal response pattern, shifting from slight antagonism to synergism with increasing KdPAR. Overall, the inhibitory effect of single drivers decreased as the number of interacting drivers increased, suggesting mitigation effects. The mean lnRR for single drivers was -0.23, while lnRR for double and triple interactions were 0.04 and 0.03, respectively. The lnRR for the interactive effect was 0.19, indicating a reduction in the inhibitory effect with increasing driver complexity.

The findings highlight the importance of water transparency (KdPAR) as a key predictor of the combined effects of environmental drivers on primary productivity in Mediterranean lakes. Although the results are specific to Mediterranean lakes, they demonstrate that increasing environmental complexity diminishes the magnitude of individual driver effects, particularly UVR. These community-level results extend the previous model-based observations suggesting that the overall biotic response to multiple drivers depends heavily on the response to the single most influential driver. While the study examined acute physiological responses, the results are ecologically relevant due to the use of natural communities already adapted to various in situ conditions and the simulation of future environmental scenarios. The stronger inhibitory UVR effect in darker environments is consistent with previous research and can be attributed to lower levels of photoprotective compounds in dominant phytoplankton groups (diatoms and dinoflagellates). The reduced primary productivity under the four-factor interaction in darker waters could be a result of increased productivity efficiency as a compensatory mechanism for reduced light, or possibly due to community tolerance resulting from chronic exposure and adaptation to UVR. Phytoplankton communities from darker environments might possess a competitive advantage when multiple stressors are present due to higher photosynthetic energy-conversion efficiency compared to those in clearer waters. The results are not biased by nutrient ratios, temperature, or solar irradiance, as these factors did not significantly affect Pc.

This research provides evidence of significant changes in planktonic community structure when lakes shift toward turbid conditions. Low water transparency under multiple interacting drivers can reduce near-surface primary productivity by approximately 40%, although this negative impact is 40-60% less severe than that of UVR alone. Given the study's small spatial and temporal scales, future research should focus on longer-term ecosystem-level studies to more accurately quantify the impacts of global change and inform management strategies. Further work could investigate the role of acclimation and adaptation in mitigating these multiple-stressor effects.

The study's short-term *in situ* experimental design did not allow for complete community acclimation or adaptation to the simulated environmental conditions. The findings are specifically based on Mediterranean lakes, and their generalizability to other lake types might be limited. The choice of specific nutrient enrichment levels and mixing velocities was based on previously established methods and in-situ observations, potentially affecting the interpretation of interactions.