Plastic response in Pinus spp., determining the temporal window of response and species-level variation of UV-B absorbing compounds to short-term variation in UV-B radiation

Abstract

The total amount of ultraviolet radiation (UV-B wavelength = 280–315 nm) reaching the Earth’ surface has probably experienced large changes throughout Earth’s history. An untested hypothesis is that past variations in UV-B flux have had a significant impact on the tempo and mode of speciation and extinctions. To address the impacts of variability in exposure to surface UV-B radiation in the geological past, a proxy is required. A pollen-based UV-B proxy, using a newly-developed thermally assisted hydrolysis and methylation pyrolysis-gas chromatography and mass spectrometry method (THM py- GC/MS), is proposed as a promising new proxy to reconstruct changes in past UV-B. However, measurements from py-GC/MS can have a high variance and, as a consequence, low analytical precision along with reproducibility issues over long analytical periods. Despite these challenges, previous work using pollen has established i) a general dose-response relationship between UV-B radiation and UV-B absorbing compounds, ii) a relationship between mean long-term UV-B and UV-B absorbing compounds across Europe and iii) a successful palaeoecological reconstruction of past UV-B flux across the Holocene using the THM py-GC/MS method. Pollen development in Pinus spp. occurs one to two months before dehiscence. During this period the pollen grains are coated with UV-B absorbing compounds, and these compounds are produced to protect themselves from harmful UV-B radiation. In light of this, we investigated whether Pinus spp. pollen displays a plastic response to shortterm changes in UV-B radiation by altering the production of UV-B absorbing compounds. We then further investigated whether this response could be pinpointed to a more accurate time period, e.g. weeks before dehiscence. Since Pinus species remain difficult to identify using traditional light microscopic methods, but have notably different pollen sizes, we wanted to investigate whether different Pinus species exhibit species-level variations in their production of UV-B absorbing compounds. First, in order to improve analytical precision of the THM py-GC/MS method, we modified the method by adding an internal standard of Nonadecanoic acid to calculate the p-Coumaric acid ratio (pCA). We developed a protocol adapted to analyse pollen from Pinus spp. by testing the effect of number of grains on pCA ratio and signal-tonoise ratio. In addition, we performed various tests on different sample preparations, machine settings and cleaning procedures. In order to answer whether Pinus spp. show a plastic response to UV-B, we conducted two experiments; (i) we artificially shaded one branch on ten trees with a 90% shading cloth one month before flowering. During dehiscence we collected pollen from the shaded and sun-exposed branches. (ii) We collected pollen from the same trees of several Pinus spp. in two consecutive years with natural variation in UV-B. These pollen samples were also used to investigate specieslevel variation between years. In order to investigate the timing of the response we collected Pinus pollen from 13 arboreta along a geographical gradient across Europe. We analysed our results using a Bayesian hierarchical model, which enabled us to take into account uncertainties in machine performance and pollen picking precision, which is a novel approach in this research area. Our results include two major improvements of the THM py-GC/MS method; we introduced an internal standard which increased the analytical precision of the UV-B absorbing compounds by 50%, and by using calibration solutions we were able to detect and correct for drift in machine precision and between batches and column cuts. Number of pollen grains has a strong linear correlation with pCA ratio, and the highest signalto- noise ratio was at 250 grains. We did not detect Ferulic acid which is another UV-B absorbing compound found in e.g. Alnus spp. pollen. In the shading experiment the shaded pollen produced 21% less UV-B absorbing compounds than the sun-exposed pollen. In Geneva the UV-B radiation in April-May 2014 was 1034 J m2 higher compared to April-May 2013, while there were minor differences in UV-B exposure between the two previous growing seasons. The pollen collected in 2014 consisting of multiple species contained 24% more UV-B absorbing compounds than pollen collected in 2013. The species from Geneva also displayed species-level variations. This was still apparent after applying corrections for a difference in biovolume. Lastly, the results from pollen collected across Europe show UV-B radiation during the last two weeks before dehiscence to be the best predictor variable of UV-B absorbing compound production. The last two weeks prior to dehiscence had the lowest deviance information criterion (DIC) values and highest slope coefficients. Our findings in both experiments demonstrate a plastic response of UV-B absorbing compounds to short-term changes in UV-B radiation. This is further supported in the broad scale study across Europe on timing of the response, which indicates that the amount of UV-B absorbing compounds is determined by the UV-B during the last two weeks before pollen dehiscence. During the last two weeks in pollen development, the tapetal cells produce the peritapetal membrane containing UV-B absorbing compounds which covers the pollen grain. Even though our pollen size corrections did not fully remove species-level variation, we expect measuring mean actual pollen size of a sample would remove these differences. Our findings have implications for the usage of Pinus pollen and UV-B absorbing compounds as a proxy for past changes in UV-B flux; First our findings show that Pinus spp. exhibit a plastic short-term response to UV-B which suggests that the signal will reflect changes in local UV-B at the given site. Second, this short term response will provide a seasonal (spring-time) UV-B signal in palaeoecological reconstructions of changes in past UV-B. A typical sample from the pollen record represents anywhere between 5 and 20 years, and any between-year variation in pollen season UV-B is therefore averaged out, giving us confidence that we can observe long-term trends. Finally, the species-level variation observed in our study suggests that UV-B reconstructions may be more difficult in sediment cores sampled in an area with known occurrence of several species of Pinus. We expect that applying size corrections to pCA ratios could improve the accuracy of the reconstruction of past UV-B, especially if specific size-correction factors for a given site can be established. The results from this thesis are novel and contribute important implications for the THM py-GC/MC method and the use of a Pinus spp. pollen-based UV-B proxy. It opens the door for future investigations into the drivers of change of pollen-chemistry.