Bomb 137Cs in modern honey reveals a regional soil control on pollutant cycling by plants

137Cs is a long-lived (30-year radioactive half-life) fission product dispersed globally by mid-20th century atmospheric nuclear weapons testing. Here we show that vegetation thousands of kilometers from testing sites continues to cycle 137Cs because it mimics potassium, and consequently, bees magnify this radionuclide in honey. There were no atmospheric weapons tests in the eastern United States, but most honey here has detectable 137Cs at >0.03 Bq kg−1, and in the southeastern U.S., activities can be >500 times higher. By measuring honey, we show regional patterns in the biogeochemical cycling of 137Cs and conclude that plants and animals receive disproportionally high exposure to ionizing radiation from 137Cs in low potassium soils. In several cases, the presence of 137Cs more than doubled the ionizing radiation from gamma and x-rays in the honey, indicating that despite its radioactive half-life, the environmental legacy of regional 137Cs pollution can persist for more than six decades.

During the middle of the 20 th century, five countries tested over 500 nuclear weapons in the air, which, taken together released far more ionizing radiation to the atmosphere than any other event or combination of events in human history 1,2 . The majority of these weapons were detonated in just a few locations in the northern hemisphere; the Marshall Islands in the Pacific Ocean (U.S.) and Novaya Zemlya (former U.S.S.R.) hosted over half of the energy yield of all the tests 3 . Many of the air detonations were so powerful that dozens of radioactive fission products (e.g., 137 Cs, 90 Sr, 131 I) were injected into the stratosphere and distributed globally with a residence time of ca. 1 year before deposition primarily by rainfall. In 1963, the Nuclear Test Ban Treaty effectively limited atmospheric testing with the exception of a few dozen smaller yield detonations that occurred primarily in China and Africa in the 1960s and 1970s 1 . The presence of radioactive pollution from nuclear testing is globally ubiquitous, and is detectable on every continent and even in deep ocean trenches 4 . Earth scientists widely apply the "bomb spike" radionuclides as a global chronological maker based on the assumption of strong radionuclide associations with minerals, and there are thousands of published studies using weapons fallout to trace soil transport and for dating sediments deposited in the 1960s 5,6 . The negative consequences of global nuclear fallout to human health are just recently coming into focus 7,8 , but the long-term biogeochemical fate and ecological consequences of radioactive pollution from weapons tests in ecosystems outside the immediate vicinity of test sites has not been carefully evaluated.
While most of the radiation produced by a nuclear weapon detonation decays within the first few days, one of the longest-lived and more abundant fission products is 137 Cs, which has a half-life of 30.2 years. During radioactive decay to stable 137 Ba, 137 Cs atoms emit ionizing radiation, including beta particles with Emax = 0.512 MeV, Ba K x-rays at 0.032 MeV, and a 0.662 MeV gamma ray, and recent research shows that even low amounts of 137 Cs can kill organisms and disrupt essential ecosystem services 9-14 . Cesium is not required for plant growth and functioning 15 . However, because Cs has a similar ionic charge and radius as K, an essential element for plants, 137 Cs is absorbed from the soil via K specific membrane transporters in vegetation 16,17 . This pathway for 137 Cs from soils to plants and into the human diet was predicted decades ago, leading to a government supported radionuclide surveillance program that conducted widespread testing of milk in the late 1950s-1980s 18,19 . However, there is no published research documenting the presence or absence of 137 Cs in eastern U.S. plants or in the U.S. food supply since 1988 20 . Here we present the first measurements of 137 Cs in honey sourced from the eastern U.S. and leverage this with a highresolution dataset of soil potassium which gives us the power to show regional patterns in the biogeochemical cycling of this dangerous radionuclide. We find that soil-plant-pollinator focusing effects can magnify 137 Cs by several orders of magnitude in honey sourced from specific physiographic regions with low soil K.

Cs in eastern U.S. honey
Honey is produced by wild and managed pollinators across the world, and, because bees make this product by reducing the water content of flower-derived nectar by nearly 5-fold, environmental contaminants are naturally concentrated in this food 21 . Using low background gamma spectrometry we found detectable 137 Cs ( ≥0.03 Bq 137 Cs kg -1 = 10 5.94 atoms tablespoon -1 ) in 68 of 122 honey samples sourced from North America. Most of these samples came from private small-scale eastern U.S. honey producers where we identified the hive locations at the U.S. county scale (110 samples), for which average soil K concentration and 137 Cs deposition rates are available 22,23 (Fig.1). Atmospheric deposition models and direct soil measurements show that the northeastern United States received slightly higher burdens of 137 Cs from the weapons tests than the southeastern United States 1,24 . Surprisingly, the geographic pattern of 137 Cs in honey is not correlated with this regional gradient of fallout (r 2 <.01; p>0.5). Of 40 honey samples collected from U.S. states north of Virginia, only 12 had detectable 137 Cs, but 36 of 39 honey samples from Florida, Georgia, and South Carolina had detectable 137 Cs (Fig.1). For comparison to the east coast samples, we analyzed 5 honeys from larger commercial operations in the central U.S. where the hives were located primarily on croplands. These honeys were very low in 137 Cs, with 4 undetectable samples and one sample just at the detection limit, and a single sample from Cuba had undetectable 137 Cs. While soils of the eastern U.S. have a relatively narrow range of 1 to 2 kBq 137 Cs m -2 , concentrations in honey from this region spanned nearly 3 orders of magnitude with far higher levels in the southeast (Fig. 1b). We find an inverse relationship between the transfer of 137 Cs from soils to honey and mean soil K concentrations of the county from it was sourced (r 2 = 0.363; p < 0.001; Fig. 2). The southeastern U.S. has relatively old, intensely weathered and leached soils from the warm and wet climate on coastal plain geology, tending to be deficient in phosphorous, K, and other rock-derived nutrients 25 . In contrast, recent glaciation in the northeastern U.S. and freshly exposed bedrock in the Appalachian Highlands maintains a relatively large supply of K 26 . These climate and soil parent material factors create a natural gradient of soil K that subsequently drives a regional control on 137 Cs in honey, but agriculture causes local exceptions. Four of the honeys from FL that were specifically identified as sourced from managed orange groves or pepper crops, where K and N amendments are common were much lower in 137 Cs than FL wildflower honeys, which is consistent with our observations that agricultural honeys from the central U.S. tended to be low in 137 Cs. Physiological differences across plant families also cause variations in Cs uptake 27 , which likely impact our observations, but our hypothesis that soil K is the first-order regional control on the 137 Cs content of honey is well supported by experimental-based research on Cs biogeochemistry 15,17 .

Fig. 2.
The transfer of 137 Cs from soil to honey scales inversely with the average soil K of the county from which it is sourced. Uncertainty on this scaling is dominated by soil K variability within each county, shown here as one standard deviation of the mean.

Soil potassium inhibits 137 Cs uptake by vegetation
Soil K depresses 137 Cs uptake by plants and ultimately limits this radionuclide in vegetation via several mechanisms. Experiments using a range of plant species indicate that Cs is not required by vegetation and thus never preferentially absorbed over K, rather, the Cs is taken because its ionic properties are near enough to K to allow mobility through K transport channels [15][16][17] . Increasing available K simply reduces Cs uptake due to the mass action relationships associated with plant-soil cation exchange reactions, thus K-based fertilizers are a proven method to reduce 137 Cs uptake by food crops in heavily contaminated soils 28 . The ammonium ion (NH4 + ) also competes with Cs in these reactions, further explaining why honey sourced from managed agricultural systems tended to be very low in 137 Cs. Kinetic controls on Cs uptake reactions may shift significantly at a threshold of K given that higher absorption rates by vegetation are observed in K limited soil 16 . Our measurements support a threshold model of 137 Cs uptake into plants, given that all 26 honey samples collected from counties with soil K <0.17% had detectable 137 Cs, which included honey from 4 different U.S. States (Fig. 2). Moreover, soil mineralogy likely plays an important role in sequestering 137 Cs from plants. Soils high in K tend to have illite, a family of clay minerals with a strong capacity to absorb or include Cs, thereby reducing its bioavailability 29 . While the mechanisms limiting plant uptake of 137 Cs have been demonstrated experimentally, our analyses of honey show how these processes play out regionally, controlling how plants and pollinators are exposed to widely varying degrees of radioactive pollution because of local soil nutrient status.
The assumption that 137 Cs is strongly fixed to soil and sediments forms the basis for its widespread use as sediment tracing and dating tool by earth scientists 6,30 . Thus, there is an absence of studies documenting the uptake of this fission product by native plants and possible pathways to the food supply in North America. One exception to this is the Pasteurized Milk Network (PMN), started in 1957 by the U.S. Public Health Service. Milk and honey are similar in that each is produced by foraging animals in every U.S. State, but in the 1950s, milk was considered the most likely pathway for nuclear fallout to enter the U.S. food supply. Consequently, a surveillance program sampled milk at central processing plants to monitor the radionuclide content, beginning with stations in five major cities 18 , but expanding with individual state programs to include over 100 stations in the 1960s. With the PMN and individual state programs joining the efforts using the same methodology, over 10,000 measurements of 137 Cs in milk are available, including monthly measurements from nearly every U.S. State for 1960 to 1975 19,31 . These data track the biological uptake of fission products from soil to vegetation to milk in different regions of North America and are valuable to compare with our more recent measurements of 137 Cs in honey.
Nationwide monthly average 137 Cs concentrations in milk peaked in late 1963 at 6 Bq kg -1 and fell sharply to <0.6 Bq kg -1 by 1970 in response to the Nuclear Test-Ban Treaty 19 . Of the 122 honey samples we measured recently, three exceeded 6 Bq kg -1 , and 30 exceeded the 0.5 Bq kg -1 concentration that nationwide average milk remained below after 1970. The highest honey 137 Cs concentration that we measured from Florida in 2018 exceeded any reported monthly milk 137 Cs value between 1958 and 2014, when the program formally ended (Fig. 3), indicating that honey can be highly concentrated in 137 Cs compared with other foods. The geographic pattern of 137 Cs reported by the PMN is consistent with the pattern we find in honey today. Nearly every month between 1960 and 2014, the sampling stations in FL reported the highest 137 Cs concentrations in milk compared with the rest of the United States. The 137 Cs content of milk from FL had a delayed decline compared with NY area milk after 1963, and average honey 137 Cs today in FL is significantly higher than average NY area honey (Fig. 3). Median soil K in FL was about 0.1%, compared with 1.3% in the NY area. Our data taken together with the PMN dataset show that plants growing in K deficient soils of the southeastern U.S. are more prone to absorbing 137 Cs decades after atmospheric deposition, and that this contamination is transferred to foods by animal foragers. To further investigate the widespread biogeochemical cycling of 137 Cs from regional weapons pollution, we analyzed vegetation archives from the Hubbard Brook Experimental Forest (HBEF) and more recent collections to reconstruct 137 Cs in native plants in the eastern U.S. over the last 50 years (Fig. 4). The HBEF one of the world's longest running ecosystem studies and maintains an archive of native foliage collected as early as the 1960s from several different northeastern states. These data show clearly that vegetation across multiple common species in the eastern U.S. has been declining in 137 Cs. At its peak, however, the levels in vegetation during the 1960s through the 1980s were extremely high, at 500 to over 1000 Bq kg -1 , which is orders of magnitude higher than levels known to be dangerous for insects to digest 9,13,14 . 137 Cs in vegetation declines from a median value of 390 Bq kg -1 in the late 1960s to approximately 4 Bq kg -1 in 2019, which is 2 orders of magnitude. The more high-resolution time series of 137 Cs in milk (Fig. 3) shows similar scales of decline, from a 1963 high of 10.7 Bq kg -1 in Tampa, F.L. and 9.8 Bq kg -1 in N.Y. to values very near the 0.1 Bq kg -1 PMN detection limit in both areas by 2010. These data taken together show how the biological uptake of 137 Cs from soils declines significantly over time, as the radionuclide undergoes radioactive decay, but also as the 137 Cs migrates down beneath the active plant rooting depth 5 . If the 137 Cs content of honey followed a similar trend as milk (Fig. 3) and foliage (Fig. 4) then average 137 Cs in honey in the Southeastern U.S. (GA, FL, SC, NC) would have been far over 100 Bq kg -1 in the 1960s-1970s, which even exceeds most national concentration standards for human consumption 32 . Fig. 4. The 137 Cs content of native vegetation (foliage) from the eastern U.S. using archives and more recent collections. Each symbol represents a single plant, and the species and state are given in the legend.

A long and dangerous legacy from the bombs
While the concentrations of 137 Cs we report in honey today are well below the 50-100 Bq kg -1 dietary threshold level of concern observed by many countries 32 , and not evidently dangerous for human consumption, the widespread residual radiation from 0.03 to 19.1 Bq 137 Cs kg -1 is surprising given that 2 half-lives have elapsed since most of the bomb production of 137 Cs. All life on earth has naturally occurring sources of alpha, beta, and gamma radiation. The dominant naturally occurring source of gamma-radiation in plants and most foods is 40 K, which we measured concurrently with 137 Cs. In 34 cases, ionizing radiation sourced from the 137 Cs accounted for more than 20% of the xray and gamma radiation from the honey, and, in 13 cases (ca. 11%), 137 Cs contamination more than doubled the total the x-ray and gamma-ray radiation of the sample. Research emerging in the last decade from two different continents shows that small amounts of 137 Cs pollution can be lethal to pollinating insects and cause measurable damage to surrounding ecosystems 10,14 . Given that increased exposure to ionizing radiation comes with increased risk for DNA damage in plants, insects, and humans, the magnification of 137 Cs in honey should be carefully considered as a major past and current threat to honeybee health 33,34 .
Experiments with the Pale Blue Grass Butterfly, Zizeeria maha indicate a linear dose-dependent response in deformations and mortality when the insects eat vegetation ranging from 2.5 to 48 Bq Cs kg -1 in Japan 13,14 . A significant fraction of honey samples collected from the southeastern United States (16 out of 49 south of VA) fell in this range in 2018-2019, and decades ago it must have been far higher. It is important to note that because radiocesium emits penetrating beta particles and gamma rays, honeybee exposure to increased ionizing radiation comes from both consuming polluted honey and from residing in a hive in close proximity to 10 to 20 kg of radioactive honey for most of the day. Populations of pollinating butterflies and bumblebees across a 137 Cs pollution gradient near the Chernobyl Nuclear Power Plant in the Ukraine are significantly reduced at even the lowest incremental exposure increases between 0.01 and 0.1 S hr -1 . As the pollinator abundance declines around Chernobyl, significant impacts on ecosystem functioning have been quantified through reductions in fruit production and tree recruitment 10 . These studies taken together indicate that when radioactive pollution doubles the ionizing radiation dose to pollinating insects, as we find that 137 Cs in today's honey does in low potassium regions, significant impacts on insect health and declines in ecosystem functioning are predicted.
Several long-standing international agreements on nuclear nonproliferation and arms control have been dismantled in recent years, and some leaders have begun to put increased emphasis on the role of nuclear weapons in military strategies 35 . These developments may lead to renewed weapons testing activities by the nine nuclear capable nations and could encourage other nations to begin testing programs of their own. Our study indicates that vegetation far from a nuclear detonation site can cycle one of the most abundant and dangerous fission products for more than six decades, and in regions of low soil potassium, honeybees get disproportionally high exposure to ionizing radiation from this pollution source. Given that pollinating insects provide vital services to the world's ecosystem and are essential in maintaining global food security 36 , this environmental threat alongside threats to human health should be carefully weighed as policies that control nuclear detonations from testing or military scenarios are formed.

Materials and Methods
Analytical Conditions 122 honey samples were procured from beekeepers as raw, pure and unfiltered. Honey samples were analyzed for radionuclides by directly photon counting 50 to 150 grams of sample using shielded high resolution (peak full-width half-maximum < 1.5 keV at 662 keV) intrinsic germanium gamma spectrometers for >150,000 seconds. Detector efficiency for Canberra 5030 Broad Energy Detectors is determined by counting a certified liquid 137 Cs source (Eckert & Ziegler) and high purity KCl (Fisher Scientific) for the different counting geometries. We also scanned each honey spectrum for 134 Cs, a strong gamma emitter at 605 keV and 796 keV, which given its short half-life of 2.06 years is a tracer of modern nuclear reactor leakage. Given that there was never a statistically significant photopeak at either 605 keV or 796 keV for any honey sample we conclude that the 137 Cs we measure here is predominately legacy nuclear pollution associated with the weapons testing era. Concentrations of 137 Cs and 40 K along with two-sigma uncertainties based on counting statistics and baseline subtraction are given in the Supplementary Data Table. Soil Potassium and Cs-137 Data In 110 of the 122 samples we identified the locations of the hives at the scale of the U.S. county or counties. Cs-137 fallout depositional fluxes to soils have been calculated from 1953-1972 based on precipitation records and measurements of 90 Sr in air and precipitation and match soil records reasonably well 3,5,8 . To determine county-averaged soil potassium, we use a high-resolution soil potassium, dataset determined by airborne radiometric surveys described in detail by the U.S. Geological Survey 9 . For plotting data and statistical tests, "undetected" 137 Cs samples were assigned a value of ½ the detection limit, = 0.015 Bq 137 Cs kg -1 .

Data Availability
All analytical results for 137 Cs and 40 K activities in the 122 honey samples are given along with 2sigma analytical uncertainties in Table S1. Soil potassium concentrations and 137 Cs deposition for each county is given, along with the standard deviation of the soil county potassium data.