The authors have declared that no competing interests exist.
Altitude is one of the most demanding environmental pressures for human populations. Highlanders from Asia, America and Africa have been shown to exhibit different biological adaptations, but Oceanian populations remain understudied [Woolcock et al., 1972; Cotes et al., 1974; Senn et al., 2010]. We tested the hypothesis that highlanders phenotypically differ from lowlanders in Papua New Guinea, as a result of inhabiting the highest mountains in Oceania for at least 20,000 years.
We collected data for 13 different phenotypes related to altitude for 162 Papua New Guineans living at high altitude (Mont Wilhelm, 2,300–2,700 m above sea level (a.s.l.) and low altitude (Daru, <100m a.s.l.). Multilinear regressions were performed to detect differences between highlanders and lowlanders for phenotypic measurements related to body proportions, pulmonary function, and the circulatory system.
Six phenotypes were significantly different between Papua New Guinean highlanders and lowlanders. Highlanders show shorter height (p-value = 0.001), smaller waist circumference (p-value = 0.002), larger Forced Vital Capacity (FVC) (p-value = 0.008), larger maximal (p-value = 3.20e -4) and minimal chest depth (p-value = 2.37e -5) and higher haemoglobin concentration (p-value = 3.36e -4).
Our study reports specific phenotypes in Papua New Guinean highlanders potentially related to altitude adaptation. Similar to other human groups adapted to high altitude, the evolutionary history of Papua New Guineans appears to have also followed an adaptive biological strategy for altitude.
New Guinea was settled 49,000 years ago (kya) [
Altitude is one of the strongest environmental stresses for human populations [
Despite these detrimental conditions, some human populations have permanently settled in high-altitude locations. For thousands of years, Tibetans, Andeans, and Ethiopian Amharas have lived at high altitude, which has led to the emergence of specialised cultures and an adapted biology [
Despite the complex nature of phenotypic traits usually related to high-altitude adaptations [
Previous studies on the health of PNG populations have shown that highlander groups have some specific phenotypic variations potentially linked to altitude: an increase in ventilatory lung function and haemoglobin concentration [
In this study, we tested the hypothesis that PNG highlanders have developed phenotypic traits known to be adapted to altitude, as a result of inhabiting the highest mountains in Oceania for at least 20 kya. We collected data on 13 phenotypic traits related to altitude for 162 PNG individuals living at high and low altitude locations. We explored the variability of these phenotypes to: (i) identify phenotypic traits specific to PNG highlanders, and (ii) compare this variation to other high-altitude human groups.
This study was approved by the Medical Research Advisory Committee of Papua New Guinea under research ethics clearance MRAC 16.21 and the French Ethics Committees (Committees of Protection of Persons IE-2015-837 (1)). Permission to conduct research in Papua New Guinea was granted by the National Research Institute (visa n°99902292358) with full support from the School of Humanities and Social Sciences, University of Papua New Guinea.
All data were collected from unrelated healthy adults (42.19 years old [95% CI 39.66–44.72]) after they had signed informed consent forms. None of the women participating in the study were pregnant. A presentation of the project was shown to all participants, followed by a discussion with each person to ensure that they had understood the project. Participants completed a questionnaire to determine language affiliation(s), current residence, date and place of birth, and a short genealogy of up to two generations to establish regional ancestry. We excluded individuals who had one or more parents or grandparents from another region (e.g. province) and only kept individuals with a full regional ancestry over at least two generations. Phenotypic data were collected from a total of 162 individuals by the same investigators using the same equipment between November 2016 and August 2019 in two different places (
Map: Sébastien Plutniak/QGIS 3.14.16. Projection: EPSG 4326. DOI:
We measured phenotypic traits reported as adaptive in other highlander groups worldwide [
The studied phenotypes correspond to three categories: body proportions, pulmonary function and circulatory system. Measurements were explained to participants in detail. Participants were at rest during data collection, and without shoes. Measurements were taken by a member of the same group of researchers who had been trained in the standardised protocols. Sample sizes were not necessarily the same for each phenotypic measurement (
Height was measured with a rigid ruler following the standard protocol [
Minimal and maximal chest depth was measured with an anthropometer (Lafayette Large anthropometer, Holtain Ltd, Harpenden, UK), after maximal expiration and maximal inspiration, respectively, at the height of the fifth thoracic vertebra [
Haemoglobin concentration was measured using a portable haemoglobin analyser (Diaspect TM and haemoglobin cuvettes, EKF, France) [
The mean of each phenotypic measurement in PNG highlanders and lowlanders was compared with a Mann-Whitney U test [
The analysed phenotypes clustered into five groups of correlated phenotypes (
This study describes the phenotypic differences between highlanders and lowlanders in PNG (
PNG lowlanders | PNG highlanders | Mann-Whitney U test | |||
---|---|---|---|---|---|
N = 86 | N = 70 | ||||
Mean [95% CI] | Mean [95% CI] | raw data |
Residuals age, sex |
Residuals age, sex, height |
|
- | |||||
Weight (kg) | 68.46 | 64.82 | 0.137 | 0.224 | 0.873 |
[65.86–71.15] | [63.21–66.53] | ||||
BMI (m2/kg) | 24.04 | 24.00 | 0.661 | 0.337 | - |
[23.19–24.97] | [23.33–24.76] | ||||
Forced expiratory volume after | 2.72 | 3.31 | 0.008 |
0.177§ | 0.124 |
1 second (FEV1)(L) | [2.45–3.00] | [3.01–3.61] | |||
Peak expiratory flow (PEF) (L/min) | 6.09 | 6.32 | 0.747§ | 0.638§ | 0.962 |
[5.28–6.92] | [5.39–7.34] | ||||
- | |||||
- | |||||
Systolic pressure (mmHg) | 131.76 | 123.94 | 0.071 | 0.245 | - |
[126.14–137.23] | [117.48–129.93] | ||||
Diastolic pressure (mmHg) | 85.29 | 81.11 | 0.156 | 0.233 | - |
[82.33–88.26] | [77.21–84.99] | ||||
Heart rate (bpm) | 67.77 | 64.38 | 0.042 | 0.057 | - |
[65.49–70.09] | [61.67–67.2] |
Significant results are in bold.
a: p-value of the comparison between PNG lowlanders and PNG highlanders for raw measurements.
b: p-value of the comparison between PNG lowlanders and PNG highlanders for age and sex residuals.
c: p-value of the comparison between PNG lowlanders and PNG highlanders for age, sex and height residuals.
*: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01).
§: data corrected to avoid ties.
**: People suffering from anaemia were removed following the World Health Organization (WHO) standard cut-offs (exclusion of non-pregnant lowlander women < 12g/dl, lowlander men <13g/dl, exclusion of non-pregnant highlander women < 13.3g/dl, highlander men <14.3g/dl) [
PNG highlanders have significantly shorter statures (mean: 164.6 cm [95% CI 163.21–166.03]) than PNG lowlanders (mean: 168.8 cm [95% CI 166.84–170.72]) even when corrected for age and sex (p-value = 0.001) (
Phenotypes | Papuans | Andeans | Tibetans | Ethiopians |
---|---|---|---|---|
Height | HL < LL | HL < LL [ |
HL < LL [ |
HL < LL [ |
Weight | HL = LL | HL < LL [ |
HL < LL [ |
HL < LL [ |
BMI | HL = LL | HL < LL[ |
HL < LL [ |
HL < LL [ |
Waist circumference | HL < LL | HL < LL [ |
HL < LL [ |
NA |
Chest size | HL > LL | HL > LL [ |
HL > LL [ |
HL > LL [ |
FVC | HL > LL | HL > LL [ |
HL > LL [ |
HL > LL [ |
FEV1 | HL = LL | HL > LL [ |
HL > LL [ |
HL > LL [ |
PEF | HL = LL | HL > LL [ |
NA | NA |
Haemoglobin concentration | HL > LL | HL > LL [ |
HL ≈ LL [ |
HL = LL [ |
Systole | HL = LL | HL < LL [ |
HL < LL [ |
HL > LL [ |
Diastole | HL = LL | HL <LL [ |
HL < LL[ |
HL < LL[ |
Heart rate | HL = LL | HL = LL [ |
HL = LL [ |
HL < LL [ |
HL = highlanders LL = lowlanders*.
*:In order to condense the main comparisons of a diverse literature in this table, we used the word “lowlanders” to refer to the control group living at low altitude that studies compare with Andean, Tibetan or Ethiopian highlanders. Nonetheless, lowlander control groups may be heterogenous and vary between studies. Tibetan highlanders are traditionally compared to Han Chinese [
Waist circumference is also significantly smaller in PNG highlanders (mean: 83.54 cm [95% CI 81.99–85.17]) than in PNG lowlanders (mean: 91.77 cm [95% CI 89.44–94.16]), even when corrected for age, sex and height (p-value = 0.002) (
Surprisingly, there is no significant difference in weight and BMI between PNG highlanders and lowlanders (p-value>0.05;
Cultural adaptation has certainly influenced body proportions especially weight, in PNG highlanders, but our data clearly suggest that these traits may be generally adapted to altitude in PNG populations.
Measurements of chest depth during respiration significantly differ between PNG lowlanders and highlanders when corrected for age, sex and height (p-value
We tested pulmonary function of PNG highlanders and lowlanders. PNG highlanders show a significant higher forced vital capacity (FVC) than PNG lowlanders (mean highlanders: 4.09 L [95% CI 3.78–4.40]; mean lowlanders: 3.32 L [95% CI 3.02–3.63]) even after correction for age, sex and height (p-value = 0.008) (
We also detected a difference in forced expiratory volumes (FEV1) between PNG highlanders (mean: 3.31L [95% CI 3.01–3.61]) and lowlanders (mean: 2.72L [95% CI 2.45–3.00]), although this was not significant after correction (
Haemoglobin concentration is significantly different between PNG highlanders and lowlanders, even after correction for age and sex (p-value = 3.51e -4). PNG highlanders have a higher mean haemoglobin concentration (mean: 14.45 g/dl [95% CI 13.84–15.06]) compared with PNG lowlanders (mean: 12.63 g/dl [95% CI 12.07–13.20]) (
We measured three phenotypic traits related to heart activity: heart rate, diastolic and systolic blood pressure. Heart rate is not significantly different between PNG highlanders (mean: 67.77 bpm [95% CI 65.49–70.09]) and lowlanders (mean: 64.38 bpm [95% CI 61.67–67.2]) (
There is no significant difference in diastolic and systolic blood pressures between PNG highlanders and lowlanders (
We found significant differences among PNG highlanders for phenotypes related to body proportions, pulmonary function and circulatory system (
Our study raises a major question on the adaptive processes to altitude in Oceania. Our data might be influenced by phenotypic changes caused by long-term stays at high altitude independent of ancestry (i.e., acclimatization), due to adaptation to environmental pressures other than hypoxia (e.g., cold), population differentiation following genetic drift, or lifestyle differences (e.g., diet, training state). However, they could also reveal heritable, genetically coded traits (i.e., selection) as has been observed in other intermediate populations from the Andean plateaus and Ethiopian highlands [
This latter hypothesis is particularly interesting since PNG populations have settled in the highlands for at least 20,000 years [
Many of these traits were also observed in other high-altitude populations worldwide, suggesting that altitude acts on the same biological pathways, independent of latitude (
Another hypothesis would be an adaptative introgression from Denisovan knowing that Papuans carry the highest proportion of Denisovan introgressions and that low haemoglobin concentration at high altitude in Tibetans is potentially inherited from Denisovan [
Our study shows that high altitude is a major factor contributing to the phenotypic variation observed in PNG. PNG highlanders tend to have a shorter stature, a larger chest depth and pulmonary volume, and a higher haemoglobin concentration compared with PNG lowlanders. The phenotypic differences observed in this study between lowlanders and highlanders, may be considered as preliminary data indicating mild positive selection in response to mild selective pressure for hypoxia. This hypothesis will request further physiological and genetic studies to be confirmed. The long-isolated history of PNG populations may have favoured the independent emergence of biological traits adapted to their environment. The driving force is probably multifactorial, including cultural, environmental, and genetic aspects. Although human biology is plastic, allowing individuals to temporarily adapt to new ecological contexts, altitude is a major selective pressure on the genomes of human populations. Given their unique genetic diversity, genomic data from PNG highlanders have the potential to shed new light on an exceptional chapter of the history of human biological adaptation.
Spearman correlation was used. Only significant correlations (p-val <0.05) are showed. Selected 5 groups of correlated phenotypes are enclosed in red boxes.
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for raw measurements.*: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01)
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for raw measurements. *: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01). §: data corrected to avoid ties.
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for raw measurements. *: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01). §: data corrected to avoid ties. **: People suffering from anaemia were removed from these violin plots, following the World Health Organization (WHO) standard cut-offs (exclusion of non-pregnant lowlander women < 12g/dl, lowlander men <13g/dl, exclusion of non-pregnant highlander women < 13.3g/dl, highlander men <14.3g/dl))[
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for age and sex residuals. *: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01)
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for age and sex residuals. *: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01). §: data corrected to avoid ties.
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for age and sex residuals. *: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01). §: data corrected to avoid ties. **: People suffering from anaemia were removed from these violin plots, following the World Health Organization (WHO) standard cut-offs (exclusion of non-pregnant lowlander women < 12g/dl, lowlander men <13g/dl, exclusion of non-pregnant highlander women < 13.3g/dl, highlander men <14.3g/dl))[
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for age and sex residuals. *: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01).
(TIF)
P-value are given for Mann-Whitney U test comparison between PNG lowlanders and PNG highlanders for age and sex residuals. *: Significant Mann-Whitney U test with Bonferroni correction for 5 multiple tests (adjusted p-value = 0.01).
(TIF)
(PDF)
(XLSX)
We thank Kylie Suseki, Roxanne Tsang, Teppsy Beni, (UPNG) Jimmy Onu and Willie Pomat (P