•UV-B radiation currently represents c. 1.5% of incoming solar radiation. However, significant changes are known to have occurred in the amount of incoming radiation both on recent and on geological timescales. Until now it has not been possible to reconstruct a detailed measure of UV-B radiation beyond c. 150 yr ago.
•Here, we studied the suitability of fossil Pinus spp. pollen to record variations in UV-B flux through time. In view of the large size of the grain and its long fossil history, we hypothesized that this grain could provide a good proxy for recording past variations in UV-B flux.
•Two key objectives were addressed: to determine whether there was, similar to other studied species, a clear relationship between UV-B-absorbing compounds in the sporopollenin of extant pollen and the magnitude of UV-B radiation to which it had been exposed; and to determine whether these compounds could be extracted from a small enough sample size of fossil pollen to make reconstruction of a continuous record through time a realistic prospect.
•Preliminary results indicate the excellent potential of this species for providing a quantitative record of UV-B through time. Using this technique, we present the first record of UV-B flux during the last 9500 yr from a site near Bergen, Norway.
UV-B radiation currently represents c. 1.5% of incoming solar radiation. However, significant changes are known to have occurred in the amount of incoming radiation both on recent and on geological timescales as a result of stratospheric ozone loss, solar activity, Milankovitch orbital oscillations, volcanic events and variations in cloud cover (Rozema et al., 2002b, 2009; Beerling et al., 2007). It has been suggested that the impact of these changes in incoming UV-B flux on terrestrial ecosystems could have affected all aspects of ecosystem functioning and fitness from genes to biomes, including altering the mode and rate of evolution (Cockell, 1999; Visscher et al., 2004; Beerling et al., 2007; Willis et al., 2009).
To date, a detailed measure of UV-B radiation through time has been unobtainable. Instrumental records of incoming UV-B radiation only extend back to the 1920s and are limited in geographical extent (Staehelin et al., 2002), and modelling of past UV-B radiation is extremely difficult because of the unknown spatial variation in cloud and ash cover over time (Shaffer & Cerveny, 2004), which could have had a highly influential role in determining the amount of UV-B reaching the ground.
Some success has been achieved in determining short-term variations (over the past 100 yr) in UV-B radiation through the measurement of phenolic acids, including para-coumaric acid (pCA) contained in the cell wall (sporopollenin of fossil plants) of Lycopodium species (Rozema et al., 2001; Blokker et al., 2006; Lomax et al., 2008). These phenolic acids are ultraviolet-absorbing compounds and are part of the defensive system of plants, and their amount increases with increasing UV-B dose. The presence of these UV-B-absorbing compounds has been also been revealed in subfossil (herbarium) material such as pollen, cuticles and wood (Rozema et al., 2001, 2002b; Blokker et al., 2005, 2006; Watson et al., 2007; Lomax et al., 2008), and in individual fossil samples of Holocene (5–6 kyr) and Palaeogene (c. 55–65 Myr) age (Blokker et al., 2005, 2006; Rozema et al., 2006). It has also been demonstrated that the ultraviolet-absorbing compounds are chemically stable and can survive unaltered for millions of years in the fossil record (Wehling et al., 1989). Such material can thus act as an excellent record of past changes in UV-B radiation and stratospheric ozone (Rozema et al., 2001, 2009).
In glasshouse experiments, it has been demonstrated that the concentration of phenolic acids in sporopollenin varies depending on the species and the amount of UV-B radiation in which they have been grown (Rozema et al., 2001, 2002a,b, 2005). However, field evidence demonstrating a dose–response relationship between the amount of UV-B-absorbing compounds or their ratio and the incoming UV-B radiation is lacking to date (Rozema et al., 2001, 2002a, 2005; Blokker et al., 2005). It has also hithero not been possible to obtain a systematic continuous record of the concentration of phenolic acids through time (time-series data > 100 yr) from one fossil species. This is a consequence, in part, of the scarcity of certain pollen types in the fossil record, but also of the difficulty in extracting small grains from the sedimentary record without using chemicals. Previous research has documented that some of the chemicals normally used to extract fossil pollen grains from sedimentary sequences (e.g. NaOH, HNO3 and acetolysis) significantly alter and reduce the amount of ultraviolet-absorbing compounds (Blokker et al., 2005; Rozema et al., 2009).
One pollen type that is large and ubiquitous and has a long fossil record is Pinus spp. We hypothesized that Pinus would be a good taxon to provide a continuous fossil record for the following reasons: it produces large, distinctive pollen grains (typically 100–200 μm in diameter) that can be easily manipulated under a stereomicroscope; it is a highly prolific pollen producer, making collection of the requisite sample size (50–100 grains per measurement) relatively easy, without the need to use strong chemicals for pollen extraction from fossil sediment (which can alter the aromatic signal); it is common in fossil sequences throughout Europe and North America and has a long continuous fossil record, extending back at least 200 Myr (Willis & McElwain, 2002); and it is fairly intolerant of shading and as a result Pinus trees generally grow in sunny stands, exposed to direct solar UV-B light.
Here, we report the findings of a study to determine the potential of the UV-B-absorbing compound para-coumaric acid (pCA) in Pinus spp. pollen as a proxy for recording variations in UV-B radiation through time. Two key objectives are addressed: to determine whether there is a clear relationship between UV-B-absorbing compounds in the sporopollenin of extant pollen and the magnitude of UV-B radiation in which it is grown; and to determine whether these compounds are apparent and can be extracted from a small enough sample size of fossil pollen to make reconstruction of a continuous record through time a realistic prospect. To address the first objective, we undertook measurements of the UV-B-absorbing compound pCA from extant Pinus spp. populations growing in a latitudinal range from 70 to 35°N, where a correspondent change in UV-B is evident. To address the second objective, we extracted fossil Pinus sylvestris pollen from a sedimentary sequence from a small lake near Bergen, Norway, spanning the last 9500 yr before present (cal yr BP) and compared the fossil estimates for the amount of changes in UV-B radiation to modelled estimates for the same interval in time.
Our approach was designed to test the following hypotheses:
• that the concentration of UV-B-absorbing compounds in extant sporopollenin of Pinus spp. pollen can be related to spatial variation in UV-B flux;
• that the temporal differences during the Holocene in UV-B flux resulting from differences in solar flux can be quantified using the UV-B-absorbing compounds in fossil Pinus spp.
Materials and Methods
Extant pollen of Pinus spp. (mainly Pinus sylvestris L.) was collected from 18 populations located in forests, botanical gardens, and parks covering the full latitudinal gradient in Europe from northern Norway (69°37′N) to Crete (35°16′N) during spring–early summer of 2008 and 2009 (Fig. 1; Table 1). For southern Europe (Crete and Athens), pollen of Pinus pinaster Aiton was sampled as no P. sylvestris was found. To extend the latitudinal transect, pollen of Pinus canariensis C. Sm. was additionally sampled from Tenerife, Canary Islands (31°39′N).
Table 1. Location details (latitude and longitude in degrees and minutes and as decimal degrees) of sites for collection of extant Pinus spp. pollen
Athens, National Garden
Athens, Botanical Garden
At each site, stalks containing pollen were collected from approximately five pine trees situated in small stands in full sunlight. Each of the trees constituted a sample. In order to collect the mature pollen, a paper envelope was placed around the pollen stalk. The envelope was shaken to collect as much of the loose mature pollen as possible, and then the stalk was cut and placed in the same envelope. Each envelope was sealed at all edges with sticky tape to prevent leakage and then sealed in another paper envelope. Envelopes containing pollen stalks from each tree were stored separately in order to avoid cross-contamination. Trees growing at the edge of a stand were preferentially chosen for sampling in order to reduce the possible effect of shading from other trees. The distance between trees was between 5 and 10 m. All populations sampled grew at low to mid elevations (< 600 m asl).
Fossil P. sylvestris pollen was extracted from a 14C dated sedimentary sequence from a small lake at Fjell, near Bergen, Norway (Gardstjorna, 60°19′N, 5°4′E), spanning the last 10 500 yr (A. E. Bjune & H. J. B. Birks, unpublished). This sequence was chosen because it was known to contain large quantities of P. sylvestris pollen. A new methodology for extracting the fossil pollen was devised which involved washing a small amount (> 10 mg) of sediment in distilled water and extracting the individual pollen grains using a micromanipulator in combination with a Nikon eclipse ME600 microscope.
Measurement of UV-B-absorbing compound pCA in extant and fossil Pinus spp. pollen
A known number of pollen grains (c. 50) for each sample (extant and fossil) were counted under a binocular microscope and transferred into quartz pyrolysis tubes (diameter of 0.9 mm) with a drop of demineralized water using an Eppendorf pipette. The samples were then placed in a drying cupboard to allow water to evaporate and subsequently wetted with 2–5 μl of Tetramethylammonium (TMAH) reagent (Blokker et al., 2005). They were left at room temperature for 30 min, and then incubated at 70°C for 2 h. Samples were further left at room temperature overnight, as we found that results were more consistent and the signal was clearer with overnight storage, probably because of a more thorough absorbance of the reagent into the pollen grains.
Each sample then underwent thermally assisted hydrolysis and methylation–gas chromatography–mass spectrometry (THM-py GC-MS) to determine the concentration of pCA contained in the sporopollenin. The micropyrolysis methodology applied in this study follows the protocol of Blokker et al. (2005), with slight modifications. Previous studies have indicated that this is the most suitable method for the quantitative analysis of UV-absorbing compounds where only a small amount of analyte is available (Blokker et al., 2006).
The tubes were manually placed into the pyrolysis filament and combusted at 750°C at maximum ramp speed for 1 min in a CDS Pyroprobe 2000 coupled with a CDS 1500 valved interface, which was in turn coupled with an Agilent 6890 gas HP5-MS column and an Agilent 5973 mass spectrometer. The samples were run in a split mode (ratio 20 : 1), which appears to show the best speed transfer of the compounds and to reduce the memory effect as a result of the long residence time of the compounds when the next sample is run. The GC oven temperature was programmed to run for 32 min (6 min starting at 40°C) to 130°C at 15°C min−1, then to 250°C at 8°C min−1 and for 1.5 min to 320°C at 20°C min−1. Helium at a column gas flow rate of 1.2 ml min−1 was used as a carrier gas. The mass spectrometer was run in selective ion monitoring (SIM) mode to detect pCA (molecular ions m/z 192). The retention time for pCA was 19.2 min. To detect the full chemical spectrum of pollen of Pinus spp. one sample was also run in full-scan mode.
Current and past UV-B climate data
A measure of the annual incoming UV-B radiation (J m−2) occurring presently at each pollen collection site was obtained using the satellite-derived surface UV-B dose corrected for cloudiness and ozone (available on request from Jean Verdebout, Institute for Environment and Sustainability, European Commission, Joint Research Centre, Ispra, Italy). This data set includes estimates of UV dose over a period of 20 yr on a half-hourly basis (1984–2003) and has a high geographical resolution of 0.05° (Verdebout, 2004).
To obtain an estimation of past UV-B flux and to provide a comparison with the fossil record obtained, model simulations were obtained using the Hadley Centre climate model (HadCM3) (methodology described in Singarayer & Valdes, 2010). Equilibrium snapshots of solar flux (W m−2) for the Bergen region (60–62.5°N and 7.5–11.23°E) at a 1000-yr temporal resolution were obtained for the time interval 0–10 000 calendar years before present (cal yr BP). In the model set-up there are no changes in solar activity or ozone production. All changes in downward shortwave flux are therefore assumed to relate to changes in climate (i.e. cloudiness) which result from changing the model boundary conditions of the Earth orbital configuration, atmospheric greenhouse gas (carbon dioxide, methane and nitrous oxide) concentrations, and global ice-cover. This model output therefore only describes one of several factors that may have influenced UV-B flux during the Holocene and the values should be treated as a preliminary, minimum estimate, and demonstrating only the broad temporal trends.
To assess how present-day UV-B flux is related to the spatial variation in pCA abundance of the Pinus spp. pollen collected from different latitudes, regression analysis was performed. Because of the nested structure of the data (several trees within a site), the linear mixed effect model with sites as the random factor and UV-B radiation as the fixed factor was evaluated following the procedure of Zuur et al. (2009). Comparing a model, either with or, without the random factor, with the fixed factor in the model showed that the inclusion of the random factor did not improve the model (L. ratio = 0.007, P =0.93). In addition, including the random factor gave very similar P-values for the fixed factor and did not alter our conclusions compared to using a simpler linear regression model. The residuals of the models were also inspected for any spatial trends by plotting the residuals in geographic space and constructing a semivariogram. These analyses did not reveal any spatial dependence in the residuals. All these preliminary analyses suggest that linear regression is adequate for analysing our data and simple linear regressions were therefore performed to evaluate the effect of UV-B on the pCA content. In the regression analysis, the average pCA abundance for each site was used as the response variable and the measured UV-B radiation was used as the predictor variable. A locally weighted scatterplot smoother (LOESS) (span = 750 yr) was used to summarize the temporal trend of pCA abundance in the fossil P. sylvestris from Fjell.
Details of site name and geographical location are provided in Table 1. These sites span a latitudinal range between c. 70°N and 31.7°N and the annual UV-B flux varied between c. 29 910 and 7650 J m−2, respectively.
In total, 75 extant pollen samples and 44 fossil Pinus spp. pollen samples (∼50 grains in each sample) were analysed for pCA using THM-py GC-MS. Results indicate that all samples contained measureable quantities of pCA. The relative abundance of pCA per sample was calculated by dividing the total area of the peak measured for the pCA compound using THM-py GC-MS by the number of grains contained in the sample.
Results show that mean values of pCA abundance in extant pollen of Pinus spp. increase from northern to southern Europe (Fig. 2). Populations growing in southern Europe (Crete, Athens and Valencia) and the Canary Islands have the highest values, whereas those from Norway have the lowest. The estimated surface UV-B dose (satellite data corrected for cloudiness and ozone) has a high positive correlation with latitude (r =0.89, P <0.0001). Linear regression analyses reveal that the pCA content in Pinus pollen is strongly correlated with the satellite-estimated annual surface UV-B dose (Fig. 3).
Results for the fossil pollen extracted from the sedimentary sequence collected from near Bergen (60.35°N, 5.25°E) indicate values of pCA abundance that are predominantly within the range of those collected for the present-day latitudinal range (Fig. 4). Overall, the time-series obtained indicates an interval of higher UV-B flux and variability between 3000 and 7000 cal yr BP and higher flux in the early Holocene.
Model simulations of solar influx during the Holocene over the Bergen region (Fig. 5a) indicate a negative correlation with the amount of cloud cover over the region. The surface downward solar influx is slightly larger in the early- (c. 9000 cal yr BP) and mid-Holocene (between 4000 and 7000 cal yr BP) than in the late-Holocene and in the pre-industrial period. The model results also suggest that the inter-annual variability is larger in the early Holocene than in the late Holocene where the inter-annual variability is quantified as the standard deviation of the mean. The LOESS-smoothed data interpolated at a similar time-span to the model data (span = 750 yr) indicate similar trends to the modelled data (Fig. 5b).
The results of our study demonstrate that pollen from Pinus spp. contains measurable amounts of the ultraviolet-absorbing compound pCA in sporopollenin. Significantly, we found that only a relatively small sample size (c. 50 grains) was required using THM-py GC-MS analysis to detect this compound. Previous work on pollen of Alnus glutinosa showed a minimum detection limit of c. 60 grains for pCA and that THM-py GC-MS gives the best results with c. 100–250 grains (Blokker et al., 2005). The larger size of Pinus spp. pollen grains (c. 2 times bigger than those of Alnus glutinosa) probably accounts for the smaller sample size needed, and this, combined with ease of extraction, makes obtaining the requisite sample size from most fossil sedimentary sequences a realistic possibility.
The content of pCA was reproducible in the Pinus spp. pollen subsamples of the same population growing under the same UV dose, although there were some small discrepancies in the measured pCA abundance between subsamples of the same population, as can be seen from the error bars in Fig. 3. These are probably the result of sporopollenin structure. Previous studies on the chemical structure and pattern of sporopollenin have indicated that they are generally complicated by the complex structure of the sporopollenin polymer, which contains diverse intermolecular linkages (Osthoff & Wiermann, 1987; Wehling et al., 1989; Collinson et al., 1993; Hemsley et al., 1996, 2004; Thom et al., 1998). This complex structure has been previously assumed to result in the division of sporopollenin into different aliphatic and aromatic fragments during pyrolysis (Blokker et al., 2005; Rozema et al., 2009).
We also determined that the content of pCA in extant Pinus spp. pollen demonstrates a strong positive relationship with estimated spatial variations with surface UV-B radiation (Fig. 3). Thus with a change of 75% in the UV-B radiation, there is a corresponding increase of 60.8% in pCA between our most northerly (70°N) and southerly populations (31°7′N). We found a higher pCA content in Pinus spp. pollen at locations with an elevated UV-B dose. Conversely, lower pCA abundance was apparent at locations with a lower UV-B dose. Interestingly, some sites located at more southerly locations in Norway (e.g. Oslo) showed a lower content of pCA than sites situated at more northern locations in the same general climatic region of Norway (Fig. 3). This inverse trend strongly coincides with a geographical pattern in estimated UV-B radiation, and may be attributable to a more variable ozone column at higher latitudes (Lomax et al., 2008).
In reconstructing a fossil record, results from this study demonstrate that a similar sample size (c. 50 grains) is required to obtain measurable quantities of pCA from fossil Pinus spp. pollen as from extant pollen. Values obtained from the fossil samples extracted from the sedimentary sequence at Gardstjorna, near Bergen, Norway are within the range apparent in the present-day latitudinal gradient (Fig. 4). Results also show that the pCA abundances apparent at high latitude during the mid-Holocene climatic optimum are closer to those currently experienced at much more southerly locations, that is, in Crete (Figs 2, 3). However, there is considerable variability in this signature and some of the lower values obtained (e.g. at c. 2000 and 7500 cal yr BP) indicate intervals of time during the Holocene when UV-B flux at Bergen was similar to values presently recorded at higher latitudes. The trend obtained for variations in UV-B flux over the past 9500 yr also indicates a close similarity to that obtained from model output for solar flux over the same interval in time (Fig. 5a,b). Work is currently underway to analyse this trend in more detail and obtain a longer temporal sequence.
Measurement of the ultraviolet compound pCA contained in fossil Pinus spp. pollen holds great potential as a technique for recording variations in UV-B flux through time. As a consequence of the large size of the grain and its abundance in fossil records, we have demonstrated that it is possible to extract pollen grains easily from sedimentary sequences in the absence of chemicals. It is also possible to obtain a requisite sample size for measuring pCA in the sporopollenin using THM-py GC-MS (c. 50 grains) with ease. Similar to other studies, the results indicate that the abundance of pCA in Pinus spp. pollen grains is influenced by the amount of UV-B radiation that the plant grows in, with a higher abundance of pCA apparent in pollen grains that develop in regions with a stronger UV-B dose. Consequently, this method can be successfully applied to fossil records with a limited amount of biological material.
Reconstruction of UV-B flux at a site near Bergen over the past 9500 yr using this technique demonstrates clear variations through time. The resulting trend in pCA obtained from the Pinus spp. sporopollenin follows a similar pattern to that obtained for modelling solar flux over the same interval in time. This demonstrates the potential of this new proxy to deliver, for the first time, a quantitative measure of variation in UV-B flux through time; data that are important to many avenues of climatological and biological research.
We are grateful to the following contributors of pine pollen samples: Keith Bennett (Northern Ireland), Donatella Magri (Italy), Tina Badal (Spain), Maja Andric (Slovenia) and Lea de Nascimento (Tenerife). K.J.W. gratefully acknowledges the following bodies for funding of this pilot study: the Natural Environmental Research Council, UK (grant NE/G010730/1) and the Oxford University John Fell Fund. This is publication number A334 from the Bjerknes Centre for Climate Research (AEB).