Seasonal trends in sperm quality in Denmark and Florida

Abstract

Background

Spermatogenesis is a temperature-sensitive process, but previous studies on seasonal variation in semen quality have produced conflicting results, often due to differences in sample sizes, methodologies, or local climates. Most studies have reported peak sperm motility in colder months, although findings vary by country. This study investigated whether the quality of semen (defined here by sperm motilities and concentrations) varies seasonally in Denmark and Florida, and whether outdoor ambient temperatures contributed to any seasonal trends.

Design

This retrospective observational study included data from 15,581 men applying to be sperm donors at Cryos International from 2018 to 2024 at two general locations: 10,670 men from four Danish cities (Aarhus, Aalborg, Odense, and Copenhagen) and 4,911 men from Orlando, Florida. These men were 18–45 years old and resided near the collection sites. All ejaculates were analysed within one hour of collection using the same computer-assisted semen analysis (CASA) system and standardised temperature-controlled protocols to quantify ejaculate volume, sperm concentration and the concentrations and counts of progressively motile sperm (grades a and b) in each ejaculate. The effects of outdoor temperatures on sperm motility during the month of collection and two months earlier (representing early spermatogenesis) were also modelled. We used nonlinear statistical methods (GAMs) to accurately analyse and display the month-to-month seasonal variation, while controlling for and illustrating the nonlinear effects of male age, monthly average temperatures, and the longitudinal trends during the study period (2018–2024) on those semen parameters.

Results

Strong and consistent seasonal variation was observed in the concentrations of progressively motile (grades a and b) sperm in both Denmark and Florida. The concentration of rapidly progressive (grade a) sperm was highest in the warmer months (June-July) and lowest in the colder months (December-January) in both countries, despite climatic and demographic differences. There was no evidence of seasonal variation in ejaculate volume or total sperm concentration, indicating that the rate of spermatogenesis did not vary seasonally. The number of progressively motile sperm (TMSC) in an ejaculate also varied seasonally, even after controlling for monthly average outdoor temperature in the contemporaneous and preceding two months, suggesting that other seasonal factors, possibly related to lifestyle or environment, may affect sperm motility.

Conclusions

Semen quality (measured as the concentration and number of progressively motile sperm) exhibits clear and consistent seasonal variation in the male populations living in both a temperate and a subtropical climate. These findings highlight the importance of accounting for seasonality when assessing semen parameters for fertility evaluation and indicate that there is seasonal variation in semen quality, even in warm climates.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12958-026-01537-w.

Introduction

It is well known that many human physiological processes vary across the year, influenced by seasonal variation in environmental factors such as temperature, daylight duration, and lifestyle [1]. While numerous factors such as age, lifestyle, and environmental conditions are known to influence sperm quality, seasonal variation in sperm parameters (e.g., motile sperm concentration and count, semen volume) remains less clearly understood.

Emerging evidence suggests that semen parameters may exhibit seasonal fluctuations. For instance, studies have reported peaks in sperm concentration and motility during specific months, alongside seasonal variation in semen volume [2–5]. These patterns may be influenced by factors such as ambient temperature, which can affect testicular function, or shifts in lifestyle behaviours, including diet and physical activity, which often vary across seasons [6–9].

A comprehensive understanding of seasonal effects on semen quality could have significant clinical and practical implications. It may inform the timing of fertility treatments, optimise assisted reproductive protocols, and provide valuable guidance to couples trying to conceive. Our aim in this study was to investigate seasonal variation in sperm concentration and motility, as well as ejaculate volume, within two large cohorts of men applying to be sperm donors across the different seasons in both subtropical (Florida) and temperate (Denmark) climates. By exploring the underlying trends and contributing factors, our goal was to enhance the understanding of male fertility and improve reproductive outcomes.

Methods

Data collection

The study used retrospective data on men residing in either Denmark or Florida who applied to become sperm donors at Cryos International from 2018 to 2024. These men lived in or near Denmark's four largest cities (Copenhagen, Aarhus, Odense, or Aalborg), or in or near Orlando, Florida, USA. Cryos has facilities in each of these 5 cities where semen samples were collected and analysed. The recruitment process at Cryos required men to be between 18 and 45 years old for them to be invited to provide a semen sample for analysis [10]. As there was no other pre-screening of the men who provided semen samples, there is no reason to believe that these men are not representative of the local populations.

Semen analysis

Semen samples were collected in sterile plastic cups (Sarstedt, Hounisen Laboratory equipment A/S, Skanderborg, Denmark) following a recommended period of 2–5 days of abstinence time. After collection, ejaculates were held at controlled room temperature (21 °C) to liquefy. Ejaculate volume was estimated by weighing the samples (assuming a density of 1.0 g/mL) in accordance with guidelines from the World Health Organisation [11, 12]. Within 30–60 min of production, an aliquot from each sample was analysed using Computer-Assisted Sperm Analysis (CASA) to assess sperm concentration and motility (see Supplementary Material for details of the protocol). Sperm motility was categorised into four grades based on WHO definitions [11, 12]: rapidly progressive (grade a; ≥ 25 µm/s), slowly progressive (grade b; 5– < 25 µm/s), non-progressive (grade c; 0– < 5 µm/s of forward progression), and immotile (grade d; with no active tail movement). All equipment, procedures, and CASA protocols were standardised across all sites.

Datasets

The dataset for each ejaculate included the collection date and city, the man’s age, ejaculate volume (mL), sperm concentration (10⁶/mL), total sperm count (millions), concentrations (10⁶/mL) of rapidly (grade a) and slowly (grade b) progressive sperm, non-progressive (grade c) sperm, and immotile (grade d) sperm. Following WHO definitions of sperm grades [11, 12], TMSC was calculated, as in many previous studies (e.g., [13]), from the concentrations of progressively motile sperm (grades a + b) in an ejaculate multiplied by the ejaculate volume. Monthly outdoor temperatures (averages of daily highs, averages, and lows; Fig. 1) were obtained from the Danish Meteorological Institute (Copenhagen, Denmark) and Climate Data Online (National Centers for Environmental Information). We used those temperatures to test and account for potential ambient temperature effects on sperm quality, as reported in previous studies [14–16]. Although Danish law exempts secondary analysis of anonymised data from ethical approval requirements, ethical clearance was secured from The University of Manchester (ref: 2023–18428-31,578).

Fig. 1.

Monthly average temperatures in the five cities where men provided semen samples for this study

Statistical analyses

We used R version 4.5.2 for all analyses, with the gam function in the mgcv (v. 1.9–4) package for Generalized Additive Models (GAMs) [17, 18]. GAMs were used to predict ejaculate volume, total sperm count, sperm concentration, and sperm motility metrics (rapidly and slowly progressive sperm concentrations, and TMSC), while controlling for and illustrating the effects of the man’s age, monthly average daily temperature, city of donation, and date of semen production (see Supplementary Materials for model structure and other details).

We used GAMs rather than the more traditional Generalized Linear Models (GLMs) to model the seasonal variation in semen parameters as GAMs provide several features for the analysis of annual and seasonal trends. First, they can model nonlinear trends with a more flexible framework (i.e. splines) when modelling complex relationships that are not constrained by strict assumptions about the distributions of residuals. Second, GAMs allow us to incorporate different smoothing functions for each continuous predictor in a model. Finally, like GLMs, GAMs allow us to test and illustrate the contribution of each predictor to the response variables (e.g., semen parameters) while accounting for variation in the other predictors. Thus, unlike previous studies of seasonal variation in semen parameters, we do not assume that the effects of predictors are linear or fit a simple nonlinear function (e.g., quadratic) and we are able to accurately model and display the month-to-month seasonal variation in those parameters. As the P-values associated GAMs are considered to be approximations, we have minimised using the term ‘significant’ and instead refer to the strength of evidence [19] as ‘weak’ (0.05 < P < 0.10), ‘moderate’ (0.01 < P < 0.05), ‘strong’ (0.001 < P < 0.01), or ‘very strong’ (P < 0.001), with P > 0.10 indicating no evidence, given the data.

Because the monthly means of the daily low, average, and high temperatures were highly correlated (r > 0.85 in each pairwise analysis), we included only the monthly means of average daily temperatures (hereafter ‘monthly average temperatures’) in each model (Fig. 1). Exploratory analyses showed that those monthly average temperatures more often resulted in a statistically better-fitting model (based on the Akaike Information Criteria, AIC), given the data. In addition to the contemporaneous monthly average temperatures, we also modelled the monthly average temperatures two months previously to examine the effects of that outdoor temperature during the early stages of spermatogenesis, reported to take 74 days in humans [20].

Results

Study population

The study included semen parameter data from 15,581 men attending the Cryos International sperm banks in Denmark and Florida. A total of 10,670 men were resident in Aarhus, Aalborg, Copenhagen, and Odense in Denmark, and 4,911 men from Orlando in Florida (Table 1). All participants were 18–45 years old.

Table 1.

Descriptive statistics (mean [95%CL] range) for samples analyzed from four cities in Denmark and Orlando, Florida

Variable	Aalborg	Aarhus	Copenhagen	Odense	Orlando
no. of samples	1752	2690	3986	2282	4263
Man’s age (years)	25.4 [25.2, 25.7] 18–45	25.5 [25.2, 25.7] 18–45	27.0 [26.8, 27.2] 18–45	25.3 [25.0, 25.5] 18–45	27.7 [27.5, 27.9] 18–44
Monthly average temperature (°C)	8.99 [8.72, 9.25] –1.2–19.0	9.07 [8.87, 9.28] –0.9–19.4	10.1 [9.96, 10.3] –0.4–21.2	9.58 [9.36. 9.80] –0.6–19.8	23.61 [23.5. 23.7] 14.4–29.9
Ejaculate volume (mL)	3.61 [3.54, 3.69] 0.2–11.0	3.66 [3.60, 3.72] 0.3–13.4	3.69 [3.64, 3.74] 0.1–13.6	3.57 [3.50, 3.63] 0.2–12.3	3.47 [3.42, 3.52] 0.1–14.0
Sperm concentration (× 106/mL)	62.8 [60.5, 65.2] 0.08–377	64.9 [62.7, 67.0] 0.01–496	75.5 [73.5, 77.4] 0.01–485	66.8 [64.6, 69.0] 0.03–372	72.1 [70.5, 73.7] 0.08–465
Sperm count (× 106)	221 [211.7, 230.0] 0.27–1320	227 [218.6, 235.1] 0.02–1988	265 [257.1, 272.0] 0.02–2324	232 [223.3, 240.4] 0.07–1638	246 [238.9, 252.1] 0.1–2607
Progressively motile sperm concentration (× 106/mL)	23.8 [22.6, 25.0] 0–178	29.8 [28.4, 31.1] 0–374	27.2 [26.3, 28.1] 0–238	30.6 [29.3, 31.9] 0–265	43.9 [42.7, 45.2] 0–366
Total Motile Sperm Count, TMSC (× 106)	82.2 [77.8, 86.6] 0–694	103.0 [98.1, 107.7] 0–1129	93.8 [90.6, 97.0] 0–976	105.0 [100.2, 109.9] 0–1167	149.9 [145.1, 154.7] 0–2343

Ejaculate volume and sperm concentration

There was no evidence that either ejaculate volume or sperm concentration varied seasonally in either Denmark (Fig. 2) or Orlando (Supplementary Material Figure S2B,F), suggesting that these parameters remained relatively stable throughout the year. There is, however, very strong evidence from both populations that both parameters varied across the study period (2018–2024) and the men’s ages (Fig. 2, Supplementary Material Figure S2). There is little evidence that either parameter varied in relation to monthly average temperatures in the contemporaneous months (Supplementary Material Table S1), except for a negative effect on ejaculate volume in Orlando.

Fig. 2.

In Denmark, there is no evidence that either ejaculate volume or sperm concentration varied seasonally. In A and E, means ± 95%CLs are shown for raw data pooled across all years. In B-D and F–H, predicted trends ± 95%CLs from GAM models are shown (with approximate P-values): by month (B,F), by the longitudinal trend over the study period (C,G), and by the man’s age (D,H), in each case controlling for the other two variables in the model as well as City. Tick marks above the x-axes (rugs) indicate the values for which there is data. See Supplementary Material Figure S2 for similar results for Orlando

Rapidly progressive sperm (grade a)

In both Denmark and Orlando, there is very strong evidence that the concentration of rapidly progressive sperm varied seasonally (Fig. 3B,F; Supplementary Material Figure S3), with the predicted means from the statistical models (Table 2) being lowest in winter (December-January) and highest in summer (June-July). In both countries, there is very strong (Denmark) to moderate (Orlando) evidence that rapidly progressive sperm concentration was positively related to monthly average temperature (Table 3) such that the highest mean concentrations predicted by the model shifted to spring (March–May) and the lowest means to autumn (September–November; Table 2, Fig. 4C,G). The predicted seasonal patterns in the two countries, controlling for contemporaneous monthly average temperatures, were virtually identical (Fig. 4C,G).

Fig. 3.

Variation in the concentration of rapidly progressive sperm in Denmark and Orlando. In A and E, the monthly means ± 95%CL of raw data pooled across all years are shown. In B-D and F–H, predicted trends ± 95%CLs from GAM models are shown (with approximate P-values): by month (B,F), by the longitudinal trend over the study period (C,G), and by the man’s age (D,H), in each case controlling for the other two variables (and for City in the model for Denmark). See Supplementary Material for model details

Table 2.

Predicted values [95%CL] for semen parameters in the months when they were highest and lowest

	DENMARK				ORLANDO
Semen parameter	Lowest		Highest		Lowest		Highest
	month	Predicted	month	Predicted	month	Predicted	month	Predicted
Rapidly progressive sperm concentration (millions/mL)	Dec-Jan	6.20 [5.71, 6.73]	Jun	8.00 [7.32, 8.74]	Dec-Jan	16.68 [15.50, 17.96]	Jun	18.74 [17.37, 20.22]
Slowly progressive sperm concentration (millions/mL)	Jul	20.10 [18.60, 21.72]	Oct	22.09 [20.50, 23.80]	Dec-Jan	25.68 [23.36, 28.23]	Jun	28.35 [25.73, 31.24]
Non-progressive + immotile sperm concentration (millions/mL)	Jul	47.30 [44.68, 50.08]	Nov	52.75 [49.94, 55.71]	May	24.07 [22.37, 25.90]	Nov	25.62 [23.85, 27.52]
Progressively motile sperm concentration (millions/mL)	Dec-Feb	28.37 [26.63, 30.23]	Jun-Jul	28.54 [26.77, 30.42]	Nov	44.12 [40.89, 47.60]	Jun	48.68 [45.08, 52.57]
Total Motile Sperm Count (TMSC, millions)	Dec-Jan	96.72 [89.76. 104.23]	Jun	100.94 [93.47, 109.01]	Nov	153.22 [143.06, 164.09]	May	165.12 [153.77. 177.32]

Predicted values of semen parameters from the Denmark and Orlando samples from GAMs that control for variation in the man’s age, and the longitudinal trends across the study period (2018–2024), as well as City in the models for Denmark

Table 3.

Effects of monthly average temperatures on sperm parameters thought to positively influence on fertilty

	Denmark		Orlando
To predict:	no lag	lag 2 months	no lag	lag 2 months
Rapidly progressive sperm concentration (millions/mL)	1.00 [54855.9] < 0.0001	1.00 [54857.3] 0.07	1.00 [35916.9] 0.04	1.01 [35917.3] 0.37
Slowly progressive sperm concentration (millions/mL)	1.01 [86035.8] 0.68	4.84 [86034.6] 0.32	1.01 [42778.9] 0.13	4.45 [42771.2] 0.04
Progressively motile sperm concentration (millions/mL)	1.30 [92035.2] 0.18	1.00 [92034.3] 0.17	1.00 [46942.3] 0.10	4.28 [46936.8] 0.06
Total Motile Sperm Count (TMSC; millions)	1.26 [118018.8] 0.05	4.25 [118017.1] 0.28	1.03 [58812.1] 0.57	4.34 [58807.8] 0.12

Effects (edf [AIC] P) shown here are from statistical models (GAMs) to predict semen parameters from contemporaneous monthly average temperatures (no lag) compared to those temperatures from two months previously (lag 2 months), at the beginning of spermatogenesis. The model with the lowest AIC in each pair (no lag vs lag 2 months for each parameter and country) is the best-fitting model. Only the model in each pair where AIC is lower by > 2 are statistically better fits (bold text) given the data. An edf close to 1.00 indicates that the effect of temperature is linear. P-values are approximate, testing for the effect of temperature on the variable of interest while controlling for the man’s age and longitudinal trends across the study period (2018–2024), as well as for City in the models for Denmark

Fig. 4.

Effect of contemporary monthly average temperatures on the concentration of rapidly progressive sperm and TMSC. In A-B and E–F, the effects of temperature are shown whereas (C-D) and (G-H) show the resulting seasonal variation in those variables when controlling for the variation in temperature. Each pair of graphs (AC, BD, EG, and FH) shows the predicted trend lines ± 95%CL from GAMs that also control for the longitudinal trend over the study period and the man’s age (and for City in the model for Denmark). The dotted blue line in (C) is the seasonal variation in Orlando from (G) with the scale on the right-hand axis

Slowly progressive sperm (grade b)

There is moderate (Denmark) to strong (Orlando) evidence that slowly progressive sperm concentration also varied seasonally (Supplementary Material Figure S4). In Denmark, the concentration of slowly progressive sperm was highest in spring (April) and fall (October) and lowest in summer (July), whereas in Orlando those concentrations were highest in early summer (June) and lowest in winter (December-January; Table 2, Supplementary Material Figure S4). There is only weak evidence that slowly progressive sperm concentrations in both locations varied seasonally (Supplementary Material Figure S4) when controlling for the non-significant effects of average temperatures in the contemporary months (Table 3).

Progressively motile (grades a + b) sperm concentration

In Orlando, there was very strong evidence for seasonal variation in the concentrations of progressively motile sperm (grades a + b) with the highest concentrations in early summer (June-July) and the lowest in winter (November-January; Table 2); there was no evidence to support a seasonal pattern in Denmark (Supplementary Material Figure S5). Controlling for monthly average temperature in the contemporaneous month provided only weak evidence for a seasonal pattern of progressively motile sperm concentration in Orlando and none in Denmark (Supplementary Material Figure S5).

Non-progressive (grade c) and Immotile (grade d) sperm concentrations

There is very strong (Denmark) to moderate (Orlando) evidence that the concentration of non-progressive and immotile sperm (grades c + d) also varied seasonally (Supplementary Material Figure S6), with the highest values in early winter (November) and lowest in early-to-mid summer (May–July; Table 2). Controlling for contemporaneous monthly average temperature, there was still moderate evidence for seasonal variation in the concentrations of non-progressive and immotile sperm at both locations (Supplementary Material Figure S6).

Total Motile Sperm Count (TMSC)

There is weak (Denmark) to moderate (Orlando) evidence that Total Motile Sperm Count (TMSC; grades a + b sperm) varied seasonally (Fig. 5, Supplementary Material Figure S7), with high and low values in the same seasons (May–June and November-January, respectively) in both countries (Table 2). Controlling for contemporaneous monthly average temperature, there was still weak evidence for seasonal variation in TMSC in Orlando but not in Denmark (Fig. 4).

Fig. 5.

Variation in Total Motile Sperm Count (TMSC) in Denmark and Orlando. In A and E, the monthly means ± 95%CL of raw data pooled across all years are shown. In B-D and F–H, predicted trends ± 95%CL from GAM models are shown (with approximate P-values): by month (B,F), by the longitudinal trend over the study period (C,G), and by the man’s age (D,H), in each case controlling for the other two variables (and for City in the model for Denmark)

In both locations, there was very strong evidence for a longitudinal trend in TMSC across the study period (Fig. 5C,G) and the effect of the man’s age (Fig. 5D,H) on TMSC. The effect of age is similar in both locations with TMSC being highest between ages 25 and 35. The longitudinal trends across the study period, however, are remarkably different between the two countries with TMSC in Denmark declining from 2019–2022 (as also reported in ref. 21) and in Orlando increasing from 2018–2024. TMSC was positively related to contemporaneous monthly average temperatures but the relationship for Orlando was not statistically significant. Controlling for contemporaneous monthly mean temperature, there was still weak evidence for a seasonal pattern in Orlando (Fig. 4G) but none in Denmark (Fig. 4C).

Effects of temperatures during spermatogenesis

To examine the relationship between ambient outdoor temperatures during spermatogenesis, we also modelled the effects of monthly average temperatures two months previously on sperm parameters (Table 3). For the concentrations of progressively motile sperm, there was weak (Denmark) and very strong (Orlando) evidence for a seasonal pattern when controlling for average temperatures two months previously (Supplementary Material Figure S8c), although there was only weak (Orlando) or no (Denmark) evidence for an effect of those temperatures (Table 3). There was also moderate evidence for a seasonal pattern in TMSC in both locations when controlling for those temperatures (Supplementary Material Figure S9), with the highest values for TMSC in summer and the lowest in winter, even though there was no evidence for the effects of those temperatures (Table 3).

Discussion

Our analyses of semen parameters in a large number of men between 2018 and 2024 in Florida and Denmark show clear seasonal patterns in sperm quality, as measured by the concentrations (Figs. 3B,F; Supplementary Material Figures S3-S5) and numbers of progressively motile sperm (TMSC; Fig. 5B,F; Supplementary Material Figure S7) in ejaculates. Whilst other sperm parameters (e.g., morphologies, viability, DNA fragmentation) might have some influence on sperm quality that affect male fertility, to date there is the strongest evidence only for the effects of the numbers and concentrations of progressively motile sperm on the spontaneous ongoing pregnancy rate [12]. We recognise that sperm motility is more often measured in studies of fertility than those other parameters and that further research that includes those other potential measures of sperm quality is warranted.

The seasonal pattern in the concentration of rapidly progressive sperm is supported by the strongest evidence (Fig. 3B,F), and is almost identical in both Denmark and Florida, being highest during the summer months and lowest in winter (Figs. 3B,F, 4C,G). The similarities between these male populations persist in spite of large climatic differences and likely differences in ethnicities, lifestyles, and exposure to environmental contaminants that might influence sperm quality. These populations also differed in the year-to-year patterns in sperm quality (e.g., Figs. 3C,G and 5C,G; Supplementary Material Figures S3-S9), as well as in the effects of outdoor monthly average temperatures (Table 3, Fig. 4A,B,E,F; Supplementary Material Figures S4-S6, S8-S9).

By constructing Generalized Additive Models, we have been able to accurately illustrate the month-to-month seasonal patterns in semen parameters (Figs. 2B,F, 3B,F, 4C,D,G,H, 5B,F) while accounting for the nonlinear effects of other variables (man’s age, monthly average temperature, longitudinal trends) on those parameters. This has the advantage of allowing meaningful comparisons between populations in two different climatic environments, revealing that those seasonal patterns, in Denmark and Florida, are very similar (Table 2). In addition, these statistical models have also revealed similar nonlinear variation in the effects of male age (Figs. 3D,H, 5D,H) and monthly average temperature (Fig. 4A,B,E,F) on semen parameters in these two populations, but very different year-to-year variation (Figs. 3C,G, 5C,G) in those parameters. The patterns in Denmark of year-to-year changes in rapidly progressive sperm concentration (Fig. 3C, Supplementary Material Figure S3) and TMSC (Fig. 5C) corroborate our previous finding of declines in those parameters from 2017–2023 [21]. The clear differences from those year-to-year patterns in Florida (Figs. 3G, 5G) deserve further study.

As there was no evidence for seasonal variation in ejaculate volume or sperm concentration in either location (Fig. 2B,F; Supplementary Material Figure S2), the rate of spermatogenesis does not appear to have varied seasonally. However, the concentrations of progressive motility of spermatozoa produced did (Fig. 3B,F; Supplementary Material Figure S4). Whilst there were strong seasonal patterns in the concentrations of progressively motile sperm, the number of those sperm in ejaculates (TMSC) showed less pronounced seasonal patterns (Fig. 5B,F). This suggests seasonal variation in the relative numbers of rapidly (grade a) and slowly (grade b) progressive sperm, but further work will be needed to investigate the nature of that apparent trade-off.

A large-scale analysis of 21,715 semen samples from southern China reported peak progressive motility in February and March, with a subsequent decline in the summer [16]. Similar seasonal trends have been found in other studies, where sperm motility was highest in the colder months and decreased in the warmer months [22–25]. In contrast, a retrospective study conducted in Italy reported that progressive sperm motility was highest during the summer [26], similar to what we report here. The differences among these studies suggest that regional lifestyle and environmental factors, such as differences in temperature, humidity, or social behaviours may influence how sperm motility varies seasonally [27–29].

In our study, there was relatively little evidence for a relationship between sperm motility and outdoor temperatures in either the contemporaneous month or two months previously (Table 3). With respect to average temperatures in the contemporaneous month, there was evidence only for an effect on rapidly progressive sperm concentration (Fig. 4, Table 3) in Denmark (strong) and Orlando (moderate), and moderate evidence for an effect on TMSC in Denmark (Table 3). As spermatogenesis is reported to take 74 days, some effect of ambient temperatures on sperm quality during spermatogenesis might be expected [20]. However, there was only weak to moderate evidence for an effect of average temperature two months prior to ejaculation on the concentrations progressively motile sperm (grades a + b) and slowly progressive (grade b) sperm and only in Florida (Table 3).

While those statistical effects of outdoor average temperatures in our study are intriguing, we hesitate to suggest a causal relationship for two reasons. First, it has been well established that climate, and particularly environmental temperatures, influence many human activities [30]. Thus, it might be lifestyle that is associated with outdoor temperatures and not the temperatures themselves that influence the concentration of progressively motile sperm. Second, the temperature variables that we tested are simple monthly mean values rather than more fine-grained weather measures, like daily wind chill and precipitation and the variability of weather, that could also influence both spermatogenesis and activities. The potential effects of temperature on sperm motility deserve further study as our measures of ambient temperature in this study were rather coarse grained as we did not know which temperatures each man experienced when samples were provided nor during the entire period of spermatogenesis. Nor did we have any information on the health, lifestyles, and activities of individual men that might have had an effect on sperm parameters. It is known or expected, for example, that a man’s duration of sexual abstinence, BMI, diet, sleep, and occupational exposures to toxins will affect sperm motility and concentration (see below). Indeed, the seasonal pattern in rapidly progressive sperm concentration remained when we accounted for average temperature either in the contemporaneous month (Fig. 4A,E) or two months earlier (Supplementary Material Figure S9), suggesting that something other than temperature also affected the seasonal variation in sperm motility, or that our measures of temperature experienced by the men were not fine-grained enough.

To the best of our knowledge, only two recent studies have studied a large and longitudinal dataset to examine both seasonal variation and the effects of temperature on semen quality, both from sperm banks in China [4, 5]. Both studies support the idea that temperature plays a key role. The first studied 41,689 semen samples from 10,802 men from Wuhan in central China [4] found significant effects of ambient outdoor temperatures on semen quality while accounting for a measure of ambient pollution level as well as the subject’s age, body mass index (BMI), smoking and alcohol use, education level, and recent sexual activity. All measures of semen quality except ejaculate volume decreased markedly as daily mean temperature, averaged over the 90 days prior to semen collection, increased from 13 °C to 30 °C. Importantly, some semen parameters (total motile sperm number and total sperm number) were correlated with temperatures 70–90 days before ejaculation, at the start of spermatogenesis. While the sample sizes and mean age of the men (28.3 years) in that study were similar to ours (mean age = 26.0 in Denmark, 27.7 in Orlando), the weather in Wuhan is different, with relatively cold, damp winters and hot, rainy summers, usually characterised as a humid subtropical monsoon climate.

The second studied 44,564 semen samples from 11,050 men from Guandong province in southern China [5], also found significant effects of outdoor temperatures on semen quality, while accounting for the same additional predictors in their models. In contrast to the findings from the Wuhan study, all measures of semen quality except ejaculate volume decreased markedly as daily mean temperature, averaged over the 90 days prior to semen collection, increased from 10 °C to 30 °C. Again, measures of sperm motility tended not to decline at ambient temperatures above 20 °C, supporting the idea of a threshold effect. Climate might be partly responsible for the differences from the Wuhan study as Guandong province is further south, where the climate is different, with short, mild, relatively dry winters and long, hot, wet summers [5].

In addition to the clear seasonal patterns, there was evidence for variation in sperm motility across years and the men’s ages in both countries. In Denmark, sperm motility declined precipitously from 2019–2022 as reported in Lassen et al. (2024) using the same dataset but a less comprehensive analysis [21]. That decline appears to have stopped by 2023 and starting to increase in 2024 (Figs. 3C, 5C), as expected if the decline was due to changes in lifestyle or health during the COVID pandemic that began in late 2019. In Orlando, overall sperm quality (as measured by TMSC) increased gradually from 2018 to 2024 (Fig. 5G). We have no explanation for this strongly supported trend, and it deserves further study.

In both countries there is very strong evidence for sperm quality changing with the age of the men (Figs. 3D,H and 5D,H). Sperm motility was lowest at both ends of the age spectrum (< 25 and > 40 years) and peaked during their thirties. Indeed, there is enough uncertainty in the predicted values to suggest that men between ages 25 and 40 have little variation in sperm quality, another pattern that deserves further study. Lower sperm quality before the age of 25 (Figs. 3D,H, 5D,H) could be explained by increased sexual activity or reduced abstinence duration which we did not quantify, whereas decreasing sperm quality is expected as a normal result of aging [31, 32].

One striking feature of the samples that we analysed is the large amount of variation in sperm parameters. Within the 315 samples from Copenhagen in March 2024, for example, the mean concentration of rapidly progressive sperm was 6.93 millions/mL with the range from 0–72 millions/mL and a standard deviation of 9.84. Thus, the coefficient of variation (CV) for this sample was 142%. To put this in perspective, the CV for most phenotypic traits is usually < 10% for traits influenced by natural selection and 10–20% for those influenced by sexual selection [33, 34]. The reasons for this extreme variation are unknown but are undoubtedly due to genetic variation, lifestyle (diet, exercise, smoking and alcohol consumption), recent sexual activity and men’s ages. Despite the extreme variation in the traits that we measured we were able to find evidence for a large amount of seasonal variation in mean traits in part by accounting for variation in age, year and month, but also because we had large sample sizes over several years.

Conclusions

The results of this study indicate a clear seasonal pattern in sperm quality—as measured by the concentration and number of progressively motile sperm—that is almost identical in both Denmark, where the climate is temperate with cold winters and warm summers, and Florida, where the climate is warm-to-hot all year round (Fig. 1). We also found some limited evidence for relationships between those sperm quality parameters and monthly average temperatures both in the contemporaneous month and two months previously when spermatogenesis began, particularly in the concentration of rapidly progressive motile sperm. Most importantly, however, when controlling for those temperature, the seasonal patterns persisted, suggesting that other factors may contribute to these seasonal trends.

Authors’ contributions

All authors took part in conceptualising the study, as well as editing drafts of the manuscript. EL and RM drafted the paper, and RM undertook the data analysis. All authors read and approved the final manuscript.

Funding

The study received no external funding.

Data availability

The data used for all analyses in this study are publicly available at ref. 35. Supplementary Material is available at *Reproductive Biology and Endocrinology* online.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

E.L. and A.-B.S. are employees of Cryos International. AAP is also a member of the Cryos External Scientific Advisory Committee, he also undertakes consultancy for Carrot Fertility, and in the last two years has delivered educational lectures for IBSA Institut Biochemique SA, and Mealis Group but all monies were paid to the University of Manchester. He is also the co-chair of the UKNEQAS Reproductive Sciences Advisory Committee, is a member of the Advisory Boards for the Progress Educational Trust (Charity Number 1139856) and the Science Media Centre (Charity Number 1140827) and Patron of the Fertility Alliance (Charity Number 1206323 (all unpaid). He is a member of the Guidelines Development Groups for the National Institute for Health and Care Excellence, and the World Health Organisation (again all unpaid). None of the authors were directly involved in the collection or physical analysis of semen samples.