Tree rings reveal extent of exposure to ionizing radiation in Scots pine Pinus sylvestris
- First Online:
- Cite this article as:
- Mousseau, T.A., Welch, S.M., Chizhevsky, I. et al. Trees (2013) 27: 1443. doi:10.1007/s00468-013-0891-z
- 715 Downloads
Tree growth has been hypothesized to provide a reliable indicator of the state of the external environment. Elevated levels of background ionizing radiation may impair growth trajectories of trees by reducing the annual growth. Such effects of radiation may depend on the individual phenotype and interact with other environmental factors such as temperature and drought. We used standardized growth rates of 105 Scots pine Pinus sylvestris located near Chernobyl, Ukraine, varying in the level of background radiation by almost a factor 700. Mean growth rate was severely depressed and more variable in 1987–1989 and several other subsequent years, following the nuclear accident in April 1986 compared to the situation before 1986. The higher frequency of years with poor growth after 1986 was not caused by elevated temperature, drought or their interactions with background radiation. Elevated temperatures suppressed individual growth rates in particular years. Finally, the negative effects of radioactive contaminants were particularly pronounced in smaller trees. These findings suggest that radiation has suppressed growth rates of pines in Chernobyl, and that radiation interacts with other environmental factors and phenotypic traits of plants to influence their growth trajectories in complex ways.
KeywordsChernobylGrowthInteraction between stressorsIonizing radiationTree heightTree rings
On 26 April 1986, one of the Chernobyl nuclear power plant reactors exploded and a nuclear fire burned for 10 days releasing between 9.35 × 103 and 1.25 × 104 PetaBq of radionuclides into the atmosphere. By contrast, the Three Mile accident in PA, USA on 27 March 1979 released just 0.0005 PetaBq. Although many of these radionuclides either dissipated or decayed within hours, days or weeks [e.g., iodine-131 (131I)], cesium-137 (137Cs) still persists today in the environment even hundreds of kilometers from Chernobyl (Shestopalov 1996; Zakharov and Krysanov 1996; Møller and Mousseau 2006; Yablokov et al. 2009). Likewise, strontium-90 (90Sr) and plutonium-239 (e.g., 239Pu) isotopes are common within the Chernobyl Exclusion Zone and in areas in Russia and Belarus. Given the 30, 29 and 24,000 years half-life for Cs-137, Sr-90 and Pu-239, respectively, these contaminants are likely to be of significance in the health of plants and animals including humans for many years to come. This accident provides a unique, but relatively unexploited opportunity to study the effects of ionizing radiation under field conditions. Many recent studies have documented the large effects of radiation on the abundance and reproductive success of different taxa (e.g., Møller and Mousseau 2007a, b, 2008, 2009). In particular, the accident provides a unique opportunity to conduct common garden or reciprocal transplant experiments (e.g., Kovalchuk et al. 2000) that will allow tests of environmental and genetic effects of radiation, and hence tests of adaptation to radiation.
Trees are particularly suitable for studies of the negative impact of environmental conditions (such as ionizing radiation) on growth, because growth parameters of the same individual are stored permanently in tree rings and these can readily be compared between years with and without exposure to the environmental condition, while simultaneously using trees in control areas as untreated controls (e.g., Ceulemans and Mousseau 1994; Cherubini et al. 2003; Ufar 2007). Previous studies of the effects of radiation on tree growth are few and the scope of these studies is very limited. Based on subjective and qualitative assessments, Arkhipov et al. (1994) suggested that doses of less than 0.1 Gy did not cause any immediate visible external damage to trees, although internal damage was not quantified. Although many studies have indicated that radioactive fallout from nuclear tests or nuclear bombs can be traced in tree rings, this effect may vary among radioactive isotopes due to differences in migration among isotopes across the tree trunk (e.g., Kagawa et al. 2002). Schmitt et al. (2000) showed for a very small sample of Pinus sylvestris pine trees grown near Chernobyl that xylem formation decreased during 1987–1989, but had recovered by 1990. Ionizing radiation had no direct effect on the cambium in 1986, but affected the differentiation of xylem mother cells in that year. Schmitt et al. (2000) concluded that reduced wood formation was due to massive loss of needles in 1986 rather than uptake of radiobiologically active elements, although this conclusion was based on only three trees felled at 1, 8 and 20 km from Chernobyl. Woodwell and Miller (1963) reported that pitch pine Pinus rigida exposed to chronic levels of radiation for several years in the 1950s at the Brookhaven facility reduced the width of growth rings, with reductions being the greatest at the base of the tree. This reduction was affected by the size of the crown and climate, with trees with large crowns showing small effects, and such effects increasing in years with greater climate perturbations. Again, these results were based on just a few trees making it difficult to draw general conclusions.
The objectives of our study were to investigate to what extent radiation from Chernobyl affected means and variances in the growth rate of trees. We relied on comparisons of growth rate before and after 1986 in the same individual trees, using the additional design feature of including trees that ranged in exposure from normal background radiation of 0.05 μSv/h to 34.5 μSv/h, or an increase in level of radiation by almost a factor 690 compared to normal background radiation. This sampling design allowed us to assess the change in mean and variance in growth rate before and after 1986 for a large number of trees. Scots pines were not often found in areas of contamination above 30 μSv/h, having been killed by exposure during the disaster with little or no recruitment in the more highly contaminated areas since that time.
All previous studies of the effects of radiation on growth rate suffered from problems of small sample sizes, and we avoided this problem by studying more than 100 trees across 12 different sites. A second objective of our study was to investigate the effects of alpha and beta radiation compared to gamma radiation on trees differing in size, assuming that small trees would be more susceptible to the negative impact of radiation (Woodwell and Miller 1963) because they could only extract water and nutrients from the topmost part of the soil where most of the radioactive material resided (Shestopalov 1996). A third objective was to investigate variation in annual mean growth rates due to radiation and other environmental factors, because radiation is not the only possible environmental stressor and the effects of radiation could interact with other stressors. In fact, we expected that radiation would have an effect on growth rate, but that this effect would be exacerbated by the impact of other environmental factors such as high temperatures and drought in the sandy soils of Chernobyl. We used Scots pine as a study organism because it is common and widespread in the region, but also because previous studies have shown that these pines are more susceptible to the negative impact of radiation than many other species of trees (Arkhipov et al. 1994; Kal’chenko and Fedotov 2001; Kal’chenko et al. 1993a, b; Kovalchuk et al. 2003; Rubanovich and Kal’chenko 1994; Shevchenko et al. 1996).
Study sites and choice of study specimens
Tree cores and phenotype of trees
We extracted tree cores at a height of 1.5 m using a 7 mm increment borer (Suunto). We measured the height of trees at an accuracy of 0.5 m using a Nikon hypsometer. The diameter of the tree at 1.5 m was measured at an accuracy of 1 cm with a DBH tape. Tree cores were subsequently removed and stored for later treatment and measurement at the laboratory.
Quantifying tree growth rates
We used standard dendrochronology methods (Cook and Kairiukstis 1990) and digital imaging software to measure the annual growth rings and quantify tree growth rates before and after the 1986 Chernobyl accident. Tree increment cores were properly oriented and mounted in grooved wooden boards. Increment cores were then sanded with successively finer grades of abrasive paper until a polished surface was obtained and individual cell walls were clearly visible under a 20× dissecting microscope. Cores were then scanned at a 1,200 dpi resolution on a flatbed scanner (Canon, Canoscan 8800f). To ensure suitable detail to be captured in the digital images, a dissecting microscope was used to compare each core to its scanned image. The digital imaging software, (CooRecorder 7, Cybis Elektronik and Data AB, http://www.cybis, Larsson 2008a), was used to register coordinates for the annual growth ring boundaries in the scanned images. The distances between annual ring boundary coordinates were recorded at an accuracy of 0.1 mm and used to calculate raw annual growth rates. We used the software program CDendro (Larsson 2008b) to standardize the annual growth rate by applying the Baillie and Pilcher normalization and negative exponential detrending procedure that uses a 2-year running average to calculate the proportion of growth attributable to the current year relative to the previous year, followed by the use of a negative exponential regression of within-tree variation in growth to remove (i.e., detrend) the effects of systematic changes in growth as the tree ages. Standardized data were used to control the effects of tree age and site-specific conditions (e.g., soil type) on annual tree ring width, allowing comparisons among differently aged trees and sampled sites. Thus, we derived standardized growth rates and detrended growth rates as a second standardized measure of tree growth rate. All annual growth rings were dated to year based on the time of collection. All measurements were made blindly with respect to the level of background radiation for the trees.
To quantify measurement errors, we randomly chose 18 cores and scanned each twice, and each image was digitized twice. Repeatability R between scans was 0.99 (SE = 0.00), F = 2,129.80, d.f. = 450, 1,349, r2 = 0.999, P < 0.0001, and repeatability between digitizing events was 0.99 (SE = 0.00), F = 2,117.79, d.f. = 452, 1,347, r2 = 0.999, P < 0.0001.
Measuring background radiation levels
We measured radiation levels in the field and cross-validated these with measurements by the Ministry of Emergencies, at ground level at each tree using a handheld dosimeter (Model: Inspector, SE International, Inc., Summertown, TN, USA). We measured levels two to three times at each site and averaged the measurements. Such data have previously been validated with correlation against data from governmental measurements at ground level published by Shestopalov (1996), estimated as the midpoint of the ranges published. These analyses showed a high degree of consistency between the two methods (Møller and Mousseau 2007a). Radiation levels vary greatly at a local scale due to heterogeneity in deposition of radioactive material after the Chernobyl accident (Fig. 1; Shestopalov 1996).
A subsample of 44 trees was used to estimate radionuclide activity within cores using gamma spectrometry. All measurements were conducted using a Berkeley Nucleonics SAM940 radionuclide identifier system equipped with a 10 × 10 cm NaI detector housed within a “cave” comprising about 1,200 kg of lead shielding. This amount of shielding resulted in a very low background radiation level of about 7.5 gcps, thus permitting very high sensitivity. System calibration was accomplished using a 0.204 microCi source. Counts were recorded for the known source to produce a calibration factor used to convert the counts from unknown samples to Ci. The source was approximately the same size and shape as the samples and was placed in the same configuration for reading as the samples, thus ensuring the same geometry. Cores were usually counted between 2 and 12 h, depending on the expected activity levels, with higher activities requiring shorter counting times for measurement. Twenty-cm-long cores were cut into four 5-cm pieces to fit atop the scintillation detector. The average 20-cm3 core had a dry mass of 0.81 g. A subsample of 12 cores was counted twice to determine the repeatability of measurements. Activity measurements on these cores was highly repeatable [linear regression: activity (2) = 0.034 + 1.056 * activity (1); F = 7.39, d.f. = 1, 10, P < 0.0001, r2 = 0.92]. The outer 5-cm pieces for 13 cores were counted individually to assess whether radionuclides were evenly distributed throughout the core or concentrated toward the outer regions of the tree trunk. Gamma spectra generated for each sample were inspected and compared to a background reference spectrum. 137Cs activity was estimated by integrating the activity above the background threshold at 662 keV, the energy level for 137Cs photon decay products. Following counting, cores were weighed on a Mettler electronic balance and individual mass was used to standardize radioactivity across samples for most samples. Small pieces of some cores were destroyed during the measurement process and a mass based on the average for all cores was used to estimate mass for these damaged samples.
We used the E-OBS gridded dataset (version 2.0) maintained by the European Climate Assessment & Dataset (ECA&D) for studying temperature and rainfall (Haylock et al. 2008). We calculated mean temperature and rainfall for the main growing season April–August for each year.
The background radiation level was log10-transformed. We developed statistical models to assess the relationship between standardized growth rate and radiation, as implemented in the statistical software JMP (SAS 2012). We tested for differences in the mean growth rate using Welch ANOVA, while simultaneously testing for the significance of a difference in variance among years using Bartlett’s test.
We tested for differences in means and variances in growth rates in relation to the level of background radiation using a repeated-measures design that allowed for tests among periods. Means and standard deviations were recorded for the five 5-year periods 1981–1985, 1986–1990, 1991–1995, 1996–2000 and 2001–2005.
We developed three generalized linear models. First, we developed a model of the relationship between the mean standardized tree growth rate in 1981–1985 and 1986–1990, using a repeated-measures design to partition the effects of radiation and period. We used a similar model for standard deviation in standardized tree growth rate for the two periods. Second, we used a model of the difference in standardized growth rate between 1981–1985 and 1986–1990 as the response variable, and radiation, height and their interaction as predictors. Third, we developed a full-factorial model of standardized growth rate as the response variable, and background radiation, temperature, rainfall, period (1950–1985 vs. 1986–2008) and their two-way interactions and the only three-way interaction that approached statistical significance as predictors.
We calculated the mean residual standard growth rate for different years and used these data in a logistic regression model with suppressed growth (1—suppressed growth less than −0.10 for mean standardized growth rate, 0—for normal growth) as the response variable, and period (0—before 1986, 1—1986 or later), temperature, precipitation and their two-way interactions as predictors.
A more detailed investigation of the relationship between radioactivity contained within the tree cores and annual growth rates was performed by calculating the mean difference in absolute growth (mm per year) for the 5 years after the disaster versus the growth for the 5 years before 1986 and relating this to the level of radioactivity measured within each tree core (Fig. 5b). To minimize the potential effects of tree size on changes in growth, only large trees older than 10 years (mean = 22.6 years) at the time of the accident in 1986 were included.
We evaluated the magnitude of associations using effect sizes estimated as Pearson product–moment correlation coefficients. Cohen (1988) proposed explicit criteria for evaluation of small (Pearson r = 0.10, explaining 1 % of the variance), intermediate (9 % of the variance) and large effects (25 % of the variance).
We measured 3,758 tree rings from 105 Scots pines. Annual growth increments were on average 2.93 mm (SE = 0.03, median 2.55 mm, range 0.12–13.49 mm). Residual growth rates were on average −0.042 (SE = 0.004, median = −0.014, range −2.46 to 0.71). Detrended standardized annual growth (det_bail) was on average −0.038 (SE = 0.004, median = −0.010, range 0.707 to −2.456). These two standardized growth rates were strongly positively correlated (Pearson r = 0.999), and, therefore, we used det_bail in all subsequent analyses (these are listed as standardized growth rates in the rest of the paper). Individual annual standardized growth rates had a lognormal distribution (KSL test, D = 0.00, P = 0.15). Trees were 7–25 m tall, mean 15.6 m (SE = 0.4, median = 15.8 m), had a diameter of 19–49 cm, mean 29.3 cm (SE = 0.52, median = 29 cm), and were 9–94 years old, mean 42.1 years (SE = 0.3, median = 42 years).
Radiation activity levels for Scots pine trees in the vicinity of the Chernobyl nuclear power plant
Background radiation (μSv/h)
Full core (Bq/kg)
Outer 5 cm (Bq/kg)
Repeated-measures analysis of variance of mean and standard deviation in residual standardized growth rate in relation to background radiation level, period and their interaction for 105 Scots pine trees
Source of variation
R × P
R × P
Difference in mean residual standardized growth rate of Scots pine trees between 1986–1990 and 1981–1985 in relation to the level of background radiation, tree height and their interaction
Sum of squares
R × H
Relationship between residual standardized growth rate of Scots pine trees between 1950 and 2008 in relation to the level of background radiation, temperature and rainfall during April–August, and whether the data were obtained before or after 1986
Sum of squares
Before/after 1986 (B)
T × P
T × R
T × B
P × R
P × B
R × B
T × P × B
The main findings of this study of growth rate of Scots pine were that reduced growth increments and increased variance in the size of growth increments were associated with elevated levels of background radiation. The magnitude of the radiation effect depended on the size of trees, because small trees showed disproportionately reduced growth when exposed to radiation compared to large trees. Other environmental stressors (e.g., temperature and low annual precipitation) interacted with radiation to reduce growth. While our data showed a 3-year continuous effect of radiation on growth suppression, our observations indicated greater variation in annual growth after 1986 and suppressed growth in 1992, 1996, 2003 and 2006. Indeed, standardized growth rate was significantly related to background radiation in 8 years, with a significant mean effect across all years, suggesting a continual mean effect of the Chernobyl accident that extends to the present. This effect size of an intermediate magnitude of 0.30 is similar to the mean effect size of 0.28 found across all meta-analyses in the biological sciences including meta-analyses of the effects of CO2 on plants (Møller and Jennions 2002). This conclusion is supported by the fact that these exceptionally poor years in terms of growth are not associated with elevated temperatures or drought, nor could we document any interaction between weather and radiation.
The degree of suppression of the mean level of growth during 1986–1990 (post-Chernobyl) compared to 1981–1985 (pre-Chernobyl) of individual trees was caused by radiation interacting with tree height, with radiation effects being disproportionately greater in small compared to large trees. There are several possible interpretations. First, more than 90 % of all radioactive material is located in the topmost 20 cm of the soil. Short trees having shallow root systems that do not extend deep into the soil may extract more radionuclides than tall trees with deep root systems. Second, growth rate effects may be more readily discerned in small trees given the larger absolute growth rates in short trees (Koch et al. 2004). Third, this may be caused by an interaction related to differential effects of radiation on mycorrhizae, which may significantly influence radionuclide uptake (Dighton et al. 2008).
The mean growth rate of Scots pines from Chernobyl varied significantly among years, with a highly significant difference in variance among years. High levels of variation in growth were much more pronounced during 1986–2009 following the accident in 1986 than during 1914–1985. Suppression of growth occurred during three subsequent years 1987–1989, and during 1992, 1996, 2003 and 2006, and this suppression was associated with radiation, but not significantly with temperature, precipitation or their interactions with radiation. Increasing variance in growth can be a consequence of increasing age and stem diameter (e.g., Fritts 1976; Carrer and Urbinati 2004) although this explanation is difficult to reconcile with a sudden change in mean and variance in standardized tree growth in 1986. Schmitt et al. (2000) reported suppressed xylem growth in Scots pine during 1987–1989, but not during 1986, when the accident happened in spring. The probability of having three consecutive years of suppressed growth was significantly less than expected randomly, suggesting that the run of 3 years of suppressed growth during 1987–1989 was exceptional. Given that radiation effects were not observed in all years after 1986, we can conclude that other stressors interacted with radiation to suppress growth. We hypothesize that the annual weather patterns interacted synergistically with spatially congruent patterns of site conditions (e.g., the sandy soils of Chernobyl) and levels of Cs-137 deposition at the landscape scale to produce significant growth suppression. This hypothesized interaction is supported by our observations of greater variation in annual growth after 1986 and suppressed growth in 1992, 1996, 2003 and 2006, when both spatial and temporal patterns of environmental stressors matched to produce effects on growth. Repeated-measures analyses of variance confirmed these conclusions by demonstrating that means were reduced and variances increased during some but not all periods, that these effects were related to radiation, and that radiation effects differed among periods. Thus, our data show a 3-year continuous effect and ongoing effects for more than 20 years, which likely extends to the present.
There are uncertainties concerning the nature of Chernobyl effects on tree growth including the mechanisms underlying the effects of radionuclides on tree growth. The effects on somatic and germline mutation rates are well documented (Arkhipov et al. 1994; Kal’chenko and Fedotov 2001; Kal’chenko et al. 1993a, b). The physiological mechanisms of uptake of mixtures of radionuclides by plants remain poorly known, as are the relative effects of external exposure to radiation versus internal exposure to radioactive heavy metals that are carried through the xylem to growing tissues. Our observation of a dose-dependent response provides support for the hypothesis that radionuclides are in large part responsible for our results. By using a sampling regime that included a relatively large number of samples with a wide spatial distribution with respect to contamination level (Fig. 1.), we were able to test our hypothesis. In other words, depressed growth was observed in trees exposed to even small levels of Chernobyl-derived radioactive contaminants exceeding the natural background level by a factor ten independently of distance from the reactor site and this is most parsimoniously explained by a direct effect of radioactive contaminants on growth.
Although we only investigated the effects of radiation on growth rate, pines were clearly also affected in terms of composition of the wood (Fig. 2). Changes in quality and quantity of wood may not only have important implications for decomposition and use as a construction material, but also for forest fires that are known to be a significant threat by redistribution of radionuclides to inhabited areas even far outside the Chernobyl Exclusion Zone (Kashparov et al. 2000; Yoschenko et al. 2006a, b). The very high activity levels of Cs-137 in tissues of trees in highly contaminated areas reported here further emphasize the need for investigation concerning the potential impacts of forest fires on the dispersal of radionuclides in populated regions.
In conclusion, we have demonstrated a landscape-scale effect of severely reduced growth in Scots pine during three consecutive years following the nuclear disaster at Chernobyl, and recurrently in several subsequent years when environmental stressors were spatially and temporally congruent. Given that significant levels of radionuclides were dispersed across 200,000 km2 in Europe as a consequence of the Chernobyl disaster, these findings suggest that there may be ecosystem-scale impacts on productivity that have not previously been suggested.
We are grateful for logistic help during our visits to Ukraine and Belarus from M. Bondarkov and A. Litvinchuk. We also thank L. Dobbs for assistance with radio-dosimetry at USC. We received funding from the University of South Carolina School of the Environment, Bill Murray and the Samuel Freeman Charitable Trust, the National Science Foundation, NATO, the Fulbright Program, CRDF and the National Geographic Society to conduct our research. We acknowledge the E-OBS dataset from the EU-FP6 project ENSEMBLES (http://www.ensembles-eu.org) and the data providers in the ECA&D project (http://eca.knmi.nl).