Quantifying spatial and taxonomic variability in thermal sensitivity is critically important in order to identify vulnerable areas and triage management and conservation efforts in response to climate change. Here we quantified the thermal performance of four dominant habitat forming species of seaweed and seagrass, from four locations across the Mediterranean Sea. We found major differences in thermal performance between species as well as subtle, albeit important differences between populations within species. Interestingly, optimal temperatures most closely resembled local habitat temperatures, whereas upper thermal limits reflected genus-level realised thermal limits. This suggests that thermal performance may have evolved to be optimal under typical local conditions, while upper thermal limits may have been conserved to tolerate rare extreme events. An important exception to this general pattern, was the high thermal tolerance of seagrasses in Cyprus. Cyprus represents the warm distribution edge of many species in the Mediterranean Sea including P. oceanica and yet, displayed the lowest thermal sensitivity due to higher thermal tolerance limits than observed in other locations across the Mediterranean. Observed differences in thermal performance between species and within species provide important context about the potential vulnerability of benthic marine ecosystems to warming and contributes to a general understanding about the underlying drivers of thermal sensitivity.
Major differences in thermal performance between species was the standout finding from this study. This finding is significant, because of the close similarities these species share in their realised thermal distributions. Realised thermal limits are a commonly used metric of thermal sensitivity 7. Yet, despite all species having realised upper thermal limits between 29–30ºC, under experimental conditions thermal limits ranged from 30ºC in P. oceanica to over 40ºC in C. nodosa. In contrast, the global distribution of congeneric species was indicative of differences in upper thermal limits. Posidonia and Cymodocea both have Indo-Pacific distributions outside of the Mediterranean, reflecting the historical connection between the Mediterranean and Indian Ocean prior to the closure of Suez Isthmus, approximately 10 Mya. The contemporary distribution of Posidonia genus is bi-temperate, with all other Posidonia species found in temperate Australia. P. oceanica separated from other Posidonia species around 65 Mya 14. By contrast, with the exception of C. nodosa, all congeneric species of Cymodocea, have a tropical distribution throughout the Indo-Pacific 20. Previous experiments on congeneric species found that Cymodocea serrulata displayed severe loss of biomass at temperatures > 40ºC 21, consistent with estimated upper limits of C. nodosa in the current study. We are not aware of comparable thermal performance experiments on congeneric Posidonia spp. However, following an extreme marine heatwave in 2011, warm-edge populations of P. australis were tolerant of temperatures up to 29ºC 22, 3.8ºC above the long term average 23. The thermal performance of P. pavonica is also consistent with the distribution of Padina spp. globally. Padina spp. is a monophyletic genus distributed worldwide, from the equator to high latitude, cool-temperate regions 15. The broad thermal distribution of Padina spp (-1.8ºC − 31ºC), parallels with the broad thermal niche breadth of P. pavonica observed in experiments, relative to other species. Contrasting patterns in the global distribution and thermal tolerance of congeneric species parallel the differences in the thermal limits of Mediterranean species suggesting that deep evolutionary mechanisms may be contributing to differences in upper thermal limits.
The differences in thermal tolerance between species is also remarkable given the shared selection pressures that have acted on these species. Following the final closure of the Suez Isthmus, the Mediterranean Sea underwent repeated climatic disturbances that fundamentally affected biodiversity in the basin 17,24. At the beginning of the Pleistocene, approximately 2.6 Mya, the Mediterranean Sea underwent rapid cooling, leading to mass extinction of much of its tropical-affiliated biota 13. Glacial and interglacial periods then resulted in alternating waves of cool-affiliated and tropical species colonizing the Mediterranean from the Atlantic, only to become extinct or have their distribution restricted during the subsequent cool or warm cycle. Evidence of this can be seen in the contemporary distribution of many species (e.g. Sparisoma cretense) with disjunct Atlantic-Levantine distributions due range contractions from the western Mediterranean Basin during glaciations. Despite these profound climatic disturbances, P. oceanica, C. nodosa, C. compressa and P. pavonica remain dominant throughout the Mediterranean, suggesting a high degree of plasticity in all four species. Moreover, differences in thermal tolerance between species persist in spite of strong thermal selection pressures over several million years.
While species-level differences in thermal performance were most conspicuous, more subtle differences in thermal performance were also observable between some populations. Most notably was the thermal performance of P. oceanica in Cyprus, which did not display a decline in growth at higher temperatures as observed in other populations. This finding contrasts with the prevailing paradigm that thermal sensitivity will be highest in warm edge populations. Warm edge populations, by definition face the highest habitat temperatures of any population across a species range, which if thermal limits are conserved between populations, mean they also have the highest sensitivity to warming. In the case of P. oceanica we observed a u-shaped pattern in thermal safety margins, with central populations living closest to their upper thermal limit and cool and warm range edge displaying the highest thermal safety margins.
This finding provides context for the paradigm that P. oceanica is highly sensitive to warming 3,18. This idea emerged primarily through studies conducted in the western Mediterranean (e.g. Mallorca) where temperatures above 28ºC have driven severe declines in shoot density of P. oceanica 3. Our findings support this previous work insofar as thermal limits of P. oceanica were estimated to be 30ºC, very close to the maximum temperatures that previously led to population declines. However, our findings also help to explain how P. oceanica can thrive at its warm distribution limit in Cyprus where average daily summer temperatures are already 29ºC and peak daily summer temperatures frequently exceed 30ºC.
With respect to metabolic rates, warming resulted in a reduction of net production in all species except for C. nodosa. A decrease of NP with warming indicates that climate change could result in a change to heterotrophic metabolism, becoming a CO2 source and an oxygen sink, further aggravating global warming. A net reduction of NP owing to a steeper increase in respiration rates than in gross primary production has been predicted by the Metabolic Theory of Ecology 25 owing to the fact that activation energies for autotrophs are half that of heterotrophs 26,27. Previous studies have shown a higher increase of respiration than primary production for planktonic communities in response to warming 26–31, whereas for benthic-dominated communities this prediction has been questioned 32. Our results demonstrate that macrophyte species can also increase their respiration rates faster than primary production with warming, resulting in heterotrophic metabolic rates at high temperatures.
Finally, experimental artefacts need to be considered in regard to their potential contribution to the variable thermal performance results. Thermal performance under experimental conditions is contingent on the experimental design, acclimation rates and seasonality, among other factors 33. As such, thermal performance measured under experimental conditions needs to be considered as a relative, not absolute estimate of the ecological reality that might occur in a natural ecosystem. Nevertheless, by using standardised methodologies of collection, transportation, acclimation and experimentation in the same experimental system, with identical conditions between experiments, methodological variation was minimised between experiments.
Seasonal timing was one potential source of variation between experiments. While all experiments took place around summer, timing of experiments ranged between June – September, due to logistical constraints. For seaweeds, poor condition of C. compressa from Crete at the time of collection (July) was likely due to seasonal senescence and resulted in it being removed from the analysis. For P. oceanica, it is unlikely that seasonality was responsible for the poor growth relationship to temperature in Cyprus. Growth rates of P. oceanica from Cyprus were the highest recorded in any experiments, suggesting that the poor growth relationship was not an artefact of poor performance. Performance of seagrasses may, however, have been influenced by experimental conditions as a result of their clonal growth physiology. Seagrasses are clonal plants connected by rhizomes and are dependent on the connectivity between ramets for nutrient translocation 34. The consistently negative NP observed across temperatures for seagrasses may be indicative that these species are less suited to growing in fragments under experimental conditions than seaweeds which source their nutrients from the water column.
Our findings reveal the thermal performance of four dominant habitat forming species across their geographical range in the Mediterranean Sea. Between species differences in optimal and upper thermal limits were pronounced despite all species sharing similar geographical distributions. Within species thermal performance was relatively conserved between populations across a 6ºC climate gradient in the Mediterranean, with the clear exception of P. oceanica at its warm range edge. Our findings highlight nuanced inter and intra-specific patterns in thermal performance that would be overlooked through a reliance on realised distributions or measurements from a single population. Contemporary patterns in thermal performance for Mediterranean macrophytes potentially reflect deep evolutionary legacies of species and local differences in the historical and contemporary marine climates. Population-specific patterns in thermal performance have important implications for the distribution and abundance and conservation of Mediterranean benthic habitats in response to climate change.