Solar activity at birth predicted infant survival and women's fertility in historical Norway

Abstract

Ultraviolet radiation (UVR) can suppress essential molecular and cellular mechanisms during early development in living organisms and variations in solar activity during early development may thus influence their health and reproduction. Although the ultimate consequences of UVR on aquatic organisms in early life are well known, similar studies on terrestrial vertebrates, including humans, have remained limited. Using data on temporal variation in sunspot numbers and individual-based demographic data (N = 8662 births) from Norway between 1676 and 1878, while controlling for maternal effects, socioeconomic status, cohort and ecology, we show that solar activity (total solar irradiance) at birth decreased the probability of survival to adulthood for both men and women. On average, the lifespans of individuals born in a solar maximum period were 5.2 years shorter than those born in a solar minimum period. In addition, fertility and lifetime reproductive success (LRS) were reduced among low-status women born in years with high solar activity. The proximate explanation for the relationship between solar activity and infant mortality may be an effect of folate degradation during pregnancy caused by UVR. Our results suggest that solar activity at birth may have consequences for human lifetime performance both within and between generations.

1. Introduction

Environmental factors during the early development of an organism can have downstream effects on the phenotypic quality and reproductive performance of that organism [1]. Several long-term studies on a wide variety of species [2], including humans [3–5], have revealed that the environment an organism is exposed to early in life may influence adult life-history traits, such as survival, fertility and lifetime reproductive success (LRS). Individuals may differ in their sensitivity to stressors during early development, which can be influenced by gender and life stage. First, it is generally accepted that males are more vulnerable to environmental stressors than females [6], and second, such effects may vary at different life stages, with greater vulnerability appearing during early development [6].

Exposure to high levels of ultraviolet radiation (UVR) is one such type of environmental stressor that can affect later survival and reproductive performance [7,8]. Levels of UVR vary with solar activity [9], latitude/altitude [10] and photoperiod [11]. The detrimental effects of high UVR exposure during development are unclear but may act via multiple molecular (degradation of folate and DNA damage) and cellular (membrane damage) mechanisms [10,12–14] in the developing organism. Such effects may lead to detrimental consequences later in life [15]. However, organisms can exhibit differential defences against UVR damage, including behaviour (avoidance), accumulation of photo-protective compounds (pigmentation, e.g. melanin and carotenoids) as well as cellular defence mechanisms (DNA-repair and antioxidants) [16,17], and specific genotypes [11]. Moreover, the presence of such UVR defence mechanisms is often costly and indicates that UVR represents a potential environmental factor in life-history evolution [13]. Indeed, a number of aquatic studies indicate that ambient UVR exposure during early development affects life-history traits later in life [7,8]. However, to our knowledge, such research with respect to terrestrial vertebrates is lacking (but see [18]). A few studies have focused on how periods of high solar activity (as a proxy for UVR) during gestation may adversely affect humans through fetal loss [19,20] and through the onset of diseases in adult life [21–23]. Furthermore, some studies have investigated how solar activity during embryonic development predicts lifespan. Whereas one study found that individuals born during years with high solar activity experienced a reduced lifespan [24], another study found there to be no effect [25]. While certain studies suggest stronger long-term effects in men than in women with respect to both lifespan and disease owing to early life exposure to high UVR [9,24], no study has yet considered the effect on survival across the complete lifespan or captured changes in effects across age and gender. This approach makes it possible to identify the specific phases where selection occurs as such information would help determine the causal mechanism responsible for long-term consequences of high UVR during development. However, thus far, no human study has explored the fitness consequences of this association in terms of both survival and reproductive performance.

Herein, we use an individual-based dataset from two human populations in Norway that spans 160 years. The data include information regarding natural fertility and mortality and enable us to test the hypothesis that high solar activity during early development affects human life-history traits (survival, fertility and LRS) and that these associations are modified by gender and age. Specifically, we examine whether the solar activity at birth is (i) related to the probability of survival from birth to the age of 20 years; (ii) related to fertility; and (iii) related to reproductive success (LRS). We analyse each gender separately as UVR has been found to affect men and women differently [9,24]. In contrast to data used in previous studies, the longitudinal individual-based data enable us to control for factors known to affect survival and fitness, such as socioeconomic status, age at first reproduction and ecology [26].

2. Material and methods

(a). Study system and data collection

We used demographic data collected from Norwegian church records of two different populations in mid-Norway that included 9062 individuals born between 1676 and 1878 [26,27]. The two populations were located at similar latitudes (63°N) and experienced similar climatic conditions [28,29]. However, the Smøla population represented a coastal island population at low altitude (0 to 70 m.a.s.l), whereas the Soknedal population represented an inland population at a somewhat higher altitude (200 to 600 m.a.s.l) [30,31]. Both populations comprised two distinct socioeconomic groups and each individual belonged to either a low (poor) or high (wealthy) status group [26,27]. The chief demographic characteristics of both populations during the study period can be described as strictly monogamous, where both men and women married late and experienced moderate infant mortality and natural fertility [32].

Solar activity, measured as the number of sunspots (SS) observed on the solar surface, varies in an 11-year cycle, with 8 years of low activity (solar minimum—SS_min) followed by 3 years of high activity (solar maximum—SS_max) [24]. The mean annual numbers of sunspots were downloaded from the Oceanic and Atmospheric Administration (NOAA), US Department of Commerce website (http://solarscience.msfc.nasa.gov/SunspotCycle.shtml). First, to capture UVR effects on early development, we divided all individuals into two groups based on whether they were born during a solar maximum or solar minimum period [9]. However, we categorized all children born between 1800 and 1825 as part of the solar minimum period as the sunspot peaks during this period were most of the time below (mean: 20.1, range: 0–47.5) the mean number of sunspots for the entire study period (mean: 46.5, range: 0–154.4) [33]. This resulted in 1041 boys and 965 girls born in a solar maximum period (mean: 96.9, range: 47.7–154.4) and 3362 boys and 3294 girls born during a solar minimum period (mean: 34.0, range: 0–118.1). Second, we investigated the effects of UVR during early life on the probability of survival to adulthood (age 20). The age of each individual was calculated from their birth year and death year because neither birth month nor death month was available.

Of all the children (from both groups), 2006 (22.1%) died before the age of 20 years. Third, we analysed potential effects of solar activity during early life on adult survival and fitness by restricting our sample to individuals who survived to adulthood, got married and sired at least one child. Consequently, we excluded individuals who were either unmarried (925 individuals) or emigrated (1422 individuals). We also excluded individuals with incomplete data (1375 individuals), thus leaving full-life-history information on 1498 women and 1623 men.

We used two different measures to capture the variation in fitness: fertility (number of children born) and LRS (number of children surviving to the age of 20 years) (see electronic supplementary material, table S1, for descriptive data, including life expectancy at age 20). In all analyses, each gender was analysed separately. We controlled for confounding effects of socioeconomic status, ecology, age at first reproduction and year of birth, all of which have been found to be associated with variations in survival, fertility and LRS in these populations [26,27]. LRS may be a better estimate of fitness in these populations as high infant mortality can inflate any measure of fitness based on fertility. Thus LRS is a more suitable measure of fitness in this study [34].

(b). Statistical analysis

We used generalized linear mixed models (GLMM) to control for cohort effects and used year of birth as a random factor. The generalized mixed models were analysed with the statistical software R (R Core Team 2014) using the package ‘lmerTest’ [35]. The analyses were run separately for each gender. We always included the interaction between solar period (maximum/minimum) and socioeconomic status and the interaction between solar period and population. Non-significant interactions were sequentially excluded from the models, while significant interactions were further explored by rerunning the analyses for each group separately (i.e. low and high status, Smøla and Soknedal populations, respectively).

For the analyses of child survival rates, we fitted a binomial model of survival to the age of 20. In the survival analyses, we also included identity of the mother as a random factor in addition to year of birth to control for maternal effects. This was not possible, however, in the fitness analyses as doing so would require full information on four generations and hence diminish the sample size. Nonetheless, as including mother ID in the survival analysis only produced a minor change in the results, we assumed that it did not have a large effect on the fitness analysis. For the analyses of fertility (number of children born), we fitted the data to a Poisson distribution. For the analyses of LRS, we fitted a ‘cbind’ binomial GLMM with number of children born reaching the age of 20 as the nominator and number of children born not reaching the age of 20 as the denominator. We also analysed survival using Kaplan–Meier survival plots and Cox proportional hazard models controlling for socioeconomic status and population using the package ‘survival’ [36] in R.

3. Results

(a). Effects of early solar activity on child survival

Of all children, 2006 (22.1%) died before the age of 20 years. Mortality was very high (8.2% of live births) in the first year of life and then declined in the second (3.3%) and third (1.7%) years. It then remained stable over the next 17 years (0.3–0.8%). There were no significant interactions between solar activity and status or population for either girls or boys (GLMM, results not shown). Models with main effects only revealed a strong significant effect of solar activity on survival for girls and a tendency for boys (table 1 and figure 1). Kaplan–Meier plots and Cox proportional hazard models corroborated the results in the GLMMs and showed a strong significant effect of solar activity on survival for girls and a strong tendency for boys (electronic supplementary material, table S2 and figure S1). The analyses also indicated that children from the coastal population, irrespective of sex, had lower survival probability than those from the inland population and that low-status boys had lower survival probability than high-status boys (table 1).

Table 1.

GLMM on the probability of survival to adulthood (age 20) in relation to solar activity at birth, status and population for each sex. Survival was fitted to the model using a binomial distribution. The interactions among solar activity and status and population, respectively, were not statistically significant for either sex and hence were removed from the analyses.

	estimate ± s.e.	Z	p
(a) girls (N = 4259 individuals, 1476 mothers, 200 years)
intercept	1.08 ± 0.11	9.49	<0.001
SSmin	0.27 ± 0.11	2.52	0.012
statuslow	–0.03 ± 0.09	–0.33	0.740
populationinland	0.21 ± 0.08	2.53	0.011
(b) boys (N = 4403 individuals, 1526 mothers, 198 years)
intercept	1.05 ± 0.11	9.19	<0.001
SSmin	0.18 ± 0.10	1.71	0.087
statuslow	0.21 ± 0.10	2.15	0.032
populationinland	0.19 ± 0.09	2.11	0.035

Figure 1.

Probability (mean ± s.e.) of survival to adulthood in relation to solar activity for boys and girls. **p < 0.1 and *p < 0.05. Dark grey bars denote SS_max, light grey bars denote SS_min.

(b). Effects of solar activity on fertility

The mean number of children born across all women and men in the dataset was 4.8 (s.d. = 2.9, range 1–16). For women, there was a significant interaction between solar activity at birth and status (GLMM, SS × status: F_1,1843 = 0.13, p = 0.036), but there was not a significant interaction between solar activity at birth and population ID (GLMM, SS × population: F_1,1843 = −0.02, p = 0.68). Therefore, we reanalysed the data for each status group separately and found that high solar activity significantly reduced fertility in low-status but not high-status women (table 2 and figure 2). For men, none of the interactions were statistically significant (GLLM, SS × status: F_1,1718 = −0.07, p = 0.27; SS × population: F_1,1718 = 0.01, p = 0.49), and the analysis of the main factors revealed no significant effect of solar activity on fertility (table 2).

Table 2.

GLMM indicating the number of children born (fertility) in relation to solar activity at birth, status, population and age at marriage. Number of children was fitted to the model using a Poisson distribution. As there was a significant interaction between solar activity and status for women, separate models for each status group were analysed.

	estimate ± s.e.	Z	p
(a1) women, high-status (N = 1222 individuals, 202 years)
intercept	3.21 ± 0.08	41.30	<0.001
SSmin	0.00 ± 0.04	0.00	0.990
populationinland	0.00 ± 0.03	0.00	0.980
age at marriage	–0.06 ± 0.00	–22.00	<0.001
(a2) women, low-status (N = 621 individuals, 162 years)
intercept	3.03 ± 0.12	25.65	<0.001
SSmin	0.14 ± 0.06	2.21	0.027
populationinland	–0.01 ± 0.05	–0.16	0.871
age at marriage	–0.06 ± 0.00	–15.97	<0.001
(b) men (N = 1718 individuals, 195 years)
intercept	2.62 ± 0.07	36.40	<0.001
SSmin	0.03 ± 0.04	0.70	0.490
statuslow	–0.20 ± 0.02	–7.90	<0.001
populationinland	0.00 ± 0.02	0.20	0.840
age at marriage	–0.03 ± 0.00	–15.20	<0.001

Figure 2.

Fertility (mean ± s.e.) in relation to solar activity for men and for women in relation to socioeconomic status. *p < 0.05. Dark grey bars denote SS_max, light grey bars denote SS_min.

(c). Effects of early solar activity on lifetime reproductive success

The mean LRS in the dataset was 3.6 (s.d. = 2.4, range 0–14) children. There were no significant interactions between solar activity at birth and status or population for either sex (data not shown). The GLMM with main effects showed a tendency for high solar activity to reduce LRS among women but not among men (table 3 and figure 3). Furthermore, while socioeconomic status showed no effect on LRS, the inland population (Soknedal) had significantly higher LRS than the coastal population (Smøla). Age at marriage tended to be important in men with older age at marriage increasing LRS (table 3).

Table 3.

GLMM indicating the number of children reaching maturity (LRS) in relation to solar activity at birth, status, population and age at marriage. The data were fitted to the model using a ‘cbind’ binomial distribution with the number of children born reaching the age of 20 as the nominator and the number of children born not reaching the age of 20 as the denominator. The interactions among solar activity and status and population, respectively, were not statistically significant for either sex and hence were removed from the analyses.

	estimate ± s.e.	Z	p
(a) women (N = 1498 individuals, 195 years)
intercept	0.45 ± 0.18	2.50	0.013
SSmin	0.16 ± 0.09	1.70	0.088
statuslow	0.01 ± 0.06	0.24	0.810
populationinland	0.60 ± 0.06	10.39	<0.001
age at marriage	0.01 ± 0.01	0.82	0.410
(b) men (N = 1623 individuals, 194 years)
intercept	0.50 ± 0.18	2.77	0.006
SSmin	0.05 ± 0.10	0.47	0.640
statuslow	0.08 ± 0.061	1.25	0.210
populationinland	0.53 ± 0.057	9.34	<0.001
age at marriage	0.01 ± 0.006	1.80	0.072

Figure 3.

LRS (mean ± s.e.) in relation to solar activity for men and women. *p < 0.1. Black bars denote SS_max, grey bars denote SS_min.

4. Discussion

Previous studies have found that different environmental conditions experienced early in life may influence adult life-history traits [2–5]. Furthermore, solar activity during early development has been found to predict the onset of diseases later in life [21–23] that may reduce the lifespan by 1.5 years ([24], but see [25]). Such effects are most likely stronger in men [9,24] than in women. Herein, we show for the first time that not only infant survival and thus lifespan but also fertility is statistically associated with solar activity at birth in these two populations. Our findings support the hypothesis that high solar activity during early life affects subsequent survival and fitness. In addition, we tested whether these associations were modified by gender, and contrary to our predictions, the associations were stronger in females than in males.

Individuals born in years with high solar activity had a lower probability of surviving to adulthood than those born in years with low solar activity. On average, the lifespans of women and men born in a solar maximum period were 5.1 and 5.3 years shorter, respectively, than women and men born in a solar minimum period, a result that is consistent with previous findings [24]. Importantly, we have taken this finding one step further by showing that early exposure to solar activity affects childhood survival, particularly during infancy. One proximate explanation for the relationship between solar activity and infant mortality may be an effect of folate degradation (vitamin B) caused by UVR [11,37,38]. Folate is needed for DNA synthesis and for the maintenance of the epigenome [39] and is thus essential for the development of healthy and fecund individuals during gestation [13]. Folate deficiency during pregnancy is associated with higher morbidity and mortality (reviewed in [40]), and hence, increased folate degradation in solar maximum periods could result in folate deficiency in pregnant women and, consequently, fetal loss [19,20] as well as reduced subsequent survival of children [13]. Another candidate explanation may be the selection for specific genotypes associated with folate loss and vitamin D biosynthesis at the time of conception or early pregnancy in relation to solar activity. While folate is UV labile, vitamin D biosynthesis is UV dependent. Previous studies have found that photoperiod can influence both vitamin D receptor and nuclear folate gene variants via differential embryo survival [11,19]. Importantly, the genotypes that influence embryo survival are also associated with late-life clinical phenotypes [11,19]. Thus, the selection of these vitamin D and folate-related gene variants could partly explain the association between mortality and solar activity found in this study.

Next, we tested whether solar activity at birth affected reproductive success. Our results revealed that low-status but not high-status mothers born in a solar maximum period had reduced fertility whereas high-status mothers did not, and that all mothers born in a solar maximum period had reduced LRS, although the latter did not reach statistical significance. Our findings suggest that maternal exposure to solar activity during gestation can affect the fitness of female children. The effect of socioeconomic status on the relationship between solar activity and fertility suggests that high-status pregnant women were better able to avoid the adverse effects of high solar activity. One possible explanation for this is that poor pregnant women were exposed to higher doses of UVR than rich women because low-status women were outdoor workers, whereas pregnant high-status women could spend more time indoors [30,31]. In addition, the quality of diet was better among the high-status women than it was for low-status women [26]. High-quality food, in contrast to low-quality food, is rich in antioxidants and carotenoids as well as coenzymes that are involved in both the protection of DNA and the reparation of damaged DNA. Consequently, individuals on a low-quality diet are expected to be more susceptible to UV-induced DNA damage [41]. However, the mechanism linking the exposure of pregnant women and DNA damage in the fetus is not entirely clear and requires further research.

A competing explanation for the relationships found in this study is that fluctuations in sunspot numbers also reflect other aspects of the environment other than the levels of UVR, such as periods with heavy workloads for pregnant women or famines, which could influence early development. These explanations, however, are not likely. First, pregnant women with heavy workloads tend to experience reduced fetal growth and abortion [42]; however, in our data, we did not find any differences in the total number of children born in the populations between SS_max (56.1) and SS_min (56.6) years. Second, the nutritional status of a pregnant woman is important for the development of the fetus and its subsequent survival and reproductive success [6]. For example, infant mortality increases during years of famines [43]. However, SS_max years did not follow famine years in these populations [30,31]. Why, then, is the fitness of women more sensitive to high solar activity during early life than is the fitness of men? First, solar activity could differentially affect the abortion rate of male and female fetuses. If only the most robust male fetuses survive, while all female fetuses survive, male fertility could be less affected. Thus, a male-biased sex ratio at birth would reflect such a situation. However, this explanation is not likely because, in our data, the sex ratio at birth in SS_max years (1.08) was not significantly less male biased than in SS_min years (1.02) (χ² = 1.27, d.f. = 1, p = 0.26), and the tendency is that more males are born in SS_max years. A more likely explanation for the observed gender differences is that the development of the reproductive organs is more costly in females than in males, and thus more at risk for developmental abnormalities when exposed to a stressor during organogenesis [44]. Concurrently, as the developmental period of the reproductive organs in the two sexes is different, testicular and ovarian organs most likely experience different hazards of in utero exposure [44] to, for example, UVR.

Our results contradict one study that did not find any reduction in the lifespan of individuals born in years of high solar activity when using a population-based dataset [25]. One reason for this discrepancy may be that the individual-based data have important advantages over population-based data. Our data allowed us to control for both socioeconomic status and ecology, factors affecting child survival in these populations [45]. Because status and ecology may conflate the effects of early life solar activity on lifespan, we avoid any probability of a type II error in our tests. Furthermore, our populations may be more susceptible to UVR damage with their high latitudinal position and less protective skin pigments. Variations in skin pigmentation (skin type) may be adaptive partly because melanized skin reduces the UVR-induced damage [13,46,47]. Natural populations at lower latitudes are thus better adapted to the detrimental effects of UVR on health and fitness. Future studies are required to test the variability/repeatability of our findings across latitudes and skin types. For example, among pregnant Caucasian females living near the equator, we might expect increased abortion rates and morbidity.

Our study also has some limitations. First, we lack exact timing of birth—those born early or late in the 3-year period of SS_max may not have intense UVR during the entire gestation. Furthermore, we cannot fully distinguish between pre- and postnatal exposure to UVR. Nonetheless, as the gestation period among mammals is the most vulnerable developmental stage, we assume that the effects evidenced in this study are owing to high UVR during development in utero as such periods lasted for, on average, 3 years. Second, individuals who emigrated were not included, which could bias the results; however, it is unlikely that these individuals differ with respect to fertility, LRS and lifespans from those who were included in the data.

To conclude, this study is the first to emphasize the importance of UVR in early life and life history in human populations. UVR is a global stressor with potential ecological impacts and the future levels of UVR are expected to increase owing to climate change and variation in atmospheric ozone [48]. Our results are thus highly relevant and contribute to a deeper knowledge of the consequences of UVR for animal life history, including human health.

Data accessibility

The datasets supporting this article may be requested by contacting G.R.S. (gine.skjarvo@bio.ntnu.no).