Measured testosterone changes precede later fertility declines in three documented cases. New hormone, behavior and conception data constrain the intermediate steps of BERM.
Research update · 9 September 2026
Finnish population samples, the regional US MMAS cohort and Israeli clinical data document lower testosterone before a later national TFR decline episode. This supports temporal precedence in these cases. The effect size, exact lag and cause remain separate calibration questions.
Lower T: observation-window end
2002
Later TFR peak
2010
TFR decline begins
2011
Age-group and birth-cohort comparisons in samples collected in 1972, 1977 and 2002. The preliminary report appeared in 2006 and was archived in 2008; the journal article followed in 2013.
Lower T: observation-window end
1997
Later TFR peak
2007
TFR decline begins
2008
Table 3: 500 → 444 ng/dL in 1987–89 versus 1995–97, at comparable median ages of 65 and 64. Regional, older men compared with a later national fertility trend.
Lower T: observation-window end
2015
Later TFR peak
2017
TFR decline begins
2018
All 30 age-specific means for ages 20–49 are lower in 2013–15 than in 2006–09. Values digitized from Figure 1A; clinical samples, four observation windows.
The spaces between these dates are calendar gaps between observation windows, not estimates of a common biological lag. Age groups within one window are not independent time points.
A later decline episode is a post-2000 local TFR peak followed by at least three annual decreases and a lower value five years later. The three dates also agree across alternative fertility sources. This descriptive comparison used previously seen data; it is not a prospective forecast. UN WPP 2024i
Download the timing assessment, including qualified and non-supporting casesThe wider assessment retains differences: Denmark’s total-T result weakens after BMI adjustment; South Africa’s included follow-up shows no preceding total-T decline. These three cases concern later decline episodes, not the beginning of every country’s historical fertility transition.
The new collection contains 71 published component estimates and summaries, not 71 independent studies. NHANES adds individual hormone and behavior records. Overlapping publications remain grouped by study family.
6,638 measured T · 4,040 linked to sexual frequency
Men aged 18–69 in three repeated cross-sectional surveys. In our exploratory weighted analysis at ages 20–49 (n = 2,982), the odds ratio for at least weekly vaginal or anal sex was 0.99 per +100 ng/dL (95% CI 0.93–1.07), adjusted for age, BMI, survey cycle and examination time. No clear uniform slope; this endpoint is not fertile-window intercourse.
+0.47 sexual-activity events/day at 12 months
Randomized treatment versus placebo in 1,161 men aged 45–80 with low T, low libido and cardiovascular disease or risk increased the composite activity measure (95% CI 0.11–0.83). Events include more than intercourse. Finkelstein’s experiments also separate testosterone and estradiol contributions. Neither result supplies a natural population T slope.
Weeks to months in suppression/recovery studies
Serum T and intratesticular T are different states: testosterone treatment can suppress gonadotropins and intratesticular T. In a pooled recovery analysis of 1,549 men, the median time to recover 20 million sperm/mL was 3.4 months (95% CI 3.2–3.5). These experiments do not support a fixed 10–15-year tissue-response lag.
First sperm: 7 months · pregnancy: 21 months
Medians in a retrospective gonadotropin-treatment cohort of 35 men with hypogonadotropic hypogonadism and azoospermia (IQRs 5–13 and 18–30 months). Sperm production and time to pregnancy are distinct transitions. This is not a universal delay for healthy populations.
18.8% versus 27.5%; adjusted OR 0.65
In an observational male-T analysis within a fertility-treatment trial, live births occurred in 21/112 low-T and 184/669 other couples. The adjusted 95% CI was 0.38–1.12. The direction is compatible with the proposed link, but uncertain; male testosterone was not randomized.
Baseline T predicts fatherhood; T then falls
Among 624 men followed for 4.5 years, higher baseline waking T predicted subsequent partnered fatherhood; new fathers then showed lower T. Prospective baseline measurements must be separated from hormonal changes after parenthood.
Hormones can affect behavior and gamete production through different responses and delays. The probability of conception per cycle conditions each transition on earlier stages:
A: an at-risk cycle; P: partner contact; X: sperm exposure in the fertile window; H: hormone history; Z: relationship context, intentions and contraception; Q: semen quality; F: female reproductive state.
Births follow conceptions through gestational survival and delay. TFR is aggregated from age-specific live-birth rates. Correlated hormone, behavior and semen effects must be integrated jointly; four assumed 20% reductions can count shared effects repeatedly.
Candidate response shapes include a regularized spline, a saturating curve, and a nearly flat population-range response with a stronger deficiency effect. Fit their shape and state-dependent delays, then compare predictions on countries or periods excluded from fitting.
Model boundary: Lindgren-derived geometry remains the physical premise. These studies supply empirical downstream biology; BERM proposes their conditional composition. The L2 operator’s gauge, physical scale, tissue kernels and sign, and the human hormone-to-TFR transfer remain open. This update does not calibrate an EMF cause. FieldState may supply physical observations only.
Calcium · redox · hormone production
Qin’s Leydig-cell field experiments connect calcium-related signalling, redox and testosterone. Component studies locate the next steps: CaMKI cooperates with NUR77 to regulate StAR, and a cholesterol analogue can bypass a transport defect. Autophagy also supplies the cholesterol needed for normal steroid production.
Hormone production, blood concentration, receptor response and reproductive success remain distinct observations along this route.
Explore the shared mechanism and its studiesThe timing result concerns specified later TFR decline episodes. Total T does not measure free or intratesticular hormone or receptor action. Hormone interventions constrain individual components; the full population transfer and environmental attribution remain open.
The data explorer includes two published NHANES periods (1988–1991 and 1999–2004), with fully adjusted testosterone means and confidence intervals. These estimates account for age, race/ethnicity, body size, smoking and alcohol; they describe an adjusted comparison rather than an annual population trajectory.
Explore the US hormone and health data →Hormone concentration and hormone action are different observables. BERM now follows the entire serial chain, so a stable total-T value cannot by itself rule out altered binding, delivery, receptor function or downstream signal use.
Reading a hormone measurement
Total testosterone counts free and protein-bound hormone together. Tissue action also depends on the receiving system.
Total T includes all three forms shown here. The symbols illustrate forms, not their proportions. T marks the hormone; protein outlines are schematic.
Availability, receptor function and downstream signal use are separate steps. The testicular local environment is also distinct from the sampled circulation.
| Total testosterone | The combined concentration in the sampled blood; it does not directly measure receptor function. |
|---|---|
| Free testosterone | The unbound fraction, measured or estimated with a stated method and binding assumptions. |
| Tissue response | A separate functional outcome. It cannot be read directly from either blood concentration alone. |
SHBG and albumin change free-hormone availability at the same total concentration; intratesticular T is a separate compartment.
AR or ZIP9 abundance, affinity and post-receptor gain can change tissue action without a proportional serum total-T change.
Evidence boundary: SHBG/free-T physiology and Sertoli-cell AR necessity are established components. A 2605 MHz rat/Sertoli study directly implicated ZIP9 but did not show short-term sperm impairment; a randomized acute MRI study found no testosterone or SHBG change in 24 men. Human chronic EMF→androgen-use calibration remains open. Narinx et al. (2022)i · De et al. (2004)i · Yu et al. (2023)i · Møllerløkken et al. (2012)i
Santi et al. 2025 introduced a differential diagnostic based on simultaneous hormone trends:
Reduced central stimulation is compatible with this pattern; simultaneous local steroidogenic limitations can still be present.
Compensatory LH is compatible with limited testicular output. Calcium, redox, clock and substrate supply locate potential mechanisms; the hormone pattern alone does not identify their trigger.
T↓ with low LH is compatible with reduced central stimulation
BERM includes both central regulation and local Leydig-cell capacity. T↓ with low or inappropriately normal LH is compatible with reduced central drive; it does not exclude concurrent calcium, redox, clock or cholesterol-supply limitations in the testis. A serum pair does not identify the environmental trigger or hormone use at the target.
The following historical comparison describes candidate exposure patterns. The integrated BERM mechanism allows central and local effects for either exposure class; the rows are hypotheses to assess with matched measurements.
| EDC | EMF | |
|---|---|---|
| LH response | Central/local state dependent | Central/local state dependent |
| Dose geography | Tracks chemical industry and agriculture | Tracks electrification and wireless density |
| Cross-species pattern | Aquatic species near discharge sites | Gradient across all domestication levels |
| Temporal onset | Post-1960 (mass plastics) | Post-1920 (electrification); accelerating post-1990 (wireless) |
The integrated mechanism adds a local EMF–steroidogenesis research branch to the central route. Chemical and physical perturbations can converge on shared calcium/redox and hormone-production machinery. Their contribution is separated by protocol, measured intermediate and timing; LH–T alone cannot assign it. SHBG, free T and receptor response remain further stages.
Seven species/population groups arranged by estimated cumulative EMF exposure show a dose-response relationship with reproductive decline:
r = 0.84, p = 0.017, n = 7 species groups
Ecological correlation across species with heterogeneous decline measurements and EMF burden estimates. The species differ in body size, lifespan, generation time, and confounders. Consistent with but not proof of dose-response. This applies equally to conventional explanations.
Derived prediction · L* level
This section describes predictions derived from the BERM framework that have not yet been directly tested. They are presented as testable hypotheses, not established findings.
Next test: compare state-dependent hormone response curves and lag distributions on countries or periods withheld from fitting. Keep temporal precedence, predictive improvement and exposure attribution as separate tests.
See predictions →