Journal Logo

BASIC SCIENCES: Original Investigations

Familial resemblance in ventilatory threshold: the HERITAGE Family Study


Author Information
Medicine and Science in Sports and Exercise: November 2001 - Volume 33 - Issue 11 - p 1832-1840
  • Free


It has been shown that there is familial resemblance for V̇O2max. Bouchard et al. (6), using familial data from the HERITAGE Family Study (HERITAGE), estimated maximal heritability in 429 White sedentary subjects from 86 nuclear families to be at least 50%. Previous studies estimated that familial resemblance, which includes both genetic and shared environmental factors, accounts for 30–67% of V̇O2max(3,4,6,7,10,11,13–17,24).

Although V̇O2max is a measure of maximal aerobic power, most daily activities are performed at submaximal exercise intensities. However, few prior studies (5,9,16,18,19) have looked at the baseline heritability of markers of submaximal fitness. Results from these previous studies have found maximal heritability estimates of 28–48%.

Pérusse et al. (20) investigated the contribution of genetic factors to baseline and response to HERITAGE training in several submaximal performance phenotypes including oxygen consumption (V̇O2) at power outputs (PO) of 50 W, 60% and 80% V̇O2max and PO at 60% and 80% V̇O2max. The maximal heritability of the baseline phenotypes varied from 48–74%, whereas the maximal heritability for the training response phenotypes tended to be lower (23–57%). Studies with monozygotic (MZ) twins exercise trained for periods of 15 or 20 wk on a standardized cycle ergometer found that there were significant within-MZ twin pair resemblances for submaximal measures of V̇O2, suggesting that genetic factors are involved in the trainability of these phenotypes (4).

In the present study, familial resemblance for submaximal aerobic fitness, both in the sedentary state before starting exercise training and in its response to training, are investigated using HERITAGE data. V̇O2 (mL·min−1) at the ventilatory threshold (V̇O2vt) adjusted for body weight is used to estimate submaximal sustainable aerobic fitness. This is justified by the work of many prior investigators who have shown that V̇O2vt and the associated variable of V̇O2 at the lactate threshold (LT) is related to sustainable aerobic work capacity and aerobic fitness. Londeree has summarized many of these studies in a meta-analysis evaluating training at intensities relative to ventilatory threshold (VT) and LT (12). In addition to V̇O2vt, we investigated the familial resemblance of VT as a percentage of V̇O2max (VT%V̇O2max) and the response of V̇O2vt (mL·min−1) to aerobic exercise training. These results represent the first reported estimates of familial influences on baseline measures and the response to training of V̇O2vt and VT%V̇O2max.


Familial resemblance in nuclear two-generation families is estimated using a simple familial correlation model. Since parent-offspring and sibling pairs share half of the same genes, twice the average parent-offspring and sibling correlations represents the maximal familial effect (i.e., maximal heritability). This estimate is considered maximal, since it represents effects attributable to all sources of familial variance, including genetic and shared family environment. On the other hand, the correlation between spouses usually represents effects attributable to familial environments, since spouses share few genes in common, assuming that there is random mating. The familial correlation model was applied to measures of V̇O2vt and VT%V̇O2max for both baseline and the response to training, separately in Black and White families participating in the HERITAGE Family Study.


A total of 339 individuals from 100 White families and 172 individuals from 99 Black families had satisfactory baseline data for both V̇O2vt and V̇O2max. Recruitment of families was derived from extensive media publicity and advertisements at the four participating clinical centers (Arizona State University [later moved to Indiana University], Laval University, University of Minnesota, and the University of Texas at Austin). Several inclusion criteria were used for screening. White families were required to have both biological parents and at least three offspring, whereas Black families could participate in units as small as two first-degree members (i.e., parent-child or two siblings) (5). Parents had to be 65 yr or younger and offspring were between the ages of 17 and 40 yr. With a few exceptions approved by a physician, subjects had a body mass index (BMI) less than 40 kg·m−2. Resting BP did not exceed 159 mm Hg for systolic blood pressure or 99 mm Hg for diastolic blood pressure. In addition, antihypertensive and lipid drug therapy were causes for exclusion. Most importantly, subjects were required to be sedentary at baseline; this was defined as no regular vigorous activity in the previous 6 months. In general, subjects were required to be in good health. Details of subject screening and methodology in HERITAGE have been previously published (5). Approval was obtained from the institutional review boards for the protection of human subjects in research projects at each of the four participating HERITAGE clinical centers, and written informed consent was obtained from each participant.

Exercise training.

After baseline assessments, each subject was trained on a cycle ergometer 3 times·wk−1 for 20 wk. The intensity and/or duration was automatically adjusted for each individual every 2 wk, starting at the heart rate (HR) associated with 55% of baseline V̇O2max for 30 min per session and increasing incrementally until each subject was working at the HR associated with 75% of baseline V̇O2max for 50 min during the last 6 wk. The PO was adjusted automatically to maintain the appropriate HR response during all training sessions by a built-in computer program. Each session was supervised on site and adherence to the protocol was strictly monitored (23).


A battery of measurements was administered to each individual both before (baseline) and after (posttraining) engaging in the 20-wk standardized exercise-training program. Each participant at baseline performed two maximal exercise tests, separated by at least 48 h, and again posttraining using a SensorMedics ergometrics 800S (Yorba Linda, CA) cycle ergometer. Heart rate was monitored by an electrocardiogram. Gas exchange variables (V̇O2, V̇CO2, V̇E, and RER) were recorded as a rolling average of three 20-s intervals. Criteria for reaching V̇O2max included meeting two of the three following criteria: 1) RER > 1.1, 2) plateau in V̇O2 (change of less than 100 mL·min−1 in the last three consecutive 20-s points) or 3) a HR within 10 beats of age-predicted maximal HR (6,23). All subjects met at least two of these criteria both at baseline and at posttraining. Details of the exercise tests have been described elsewhere (5,6,23).

VT was determined by concurrently evaluating graphs of V-slope (change in slope of the V̇O2 vs V̇CO2 plot from less than 1 to greater than 1), ventilatory equivalents (first sustained rise in V̇E/V̇O2 without a concurrent rise in V̇E/V̇CO2), and excess CO2 (first sustained rise in excess CO2 as calculated by the formula (V̇CO22/V̇O2) − V̇CO2) plots. The methods used followed a tight protocol requiring independent investigators to determine VT within 3% of one another for the data to be accepted. Gaskill et al. (8) have previously described details of the combined VT determination method used in this study. The combined method was shown in that study to greatly reduce the standard deviation of the LT − VT difference (70–140% reduction over using individual VT determination methods). Because of the rigid VT determination criteria, VT was adequately determined in only about 60% of the HERITAGE population.

Data adjustments.

Data were adjusted for several covariates before the familial analyses using stepwise (and/or forced) multiple regression. In summary, a variable was regressed on a set of covariates, separately in eight race by sex by generation groups (fathers, mothers, sons, and daughters in each of the White and Black gender samples). In the stepwise regressions, only those terms that were significant at the 5% level were retained. The resulting squared residuals from this mean regression were similarly adjusted for covariate effects in the variance (i.e., heteroscedasticity) using another stepwise regression. The final adjusted variable was the residual from the mean regression, divided by the square root of the predicted score from the second regression. A final standardization step was taken to ensure zero mean and unit variance.

For baseline V̇O2vt (mL·min−1), a two-step regression procedure was used. In the first step, V̇O2vt was adjusted for weight (kg) using a forced regression, and in the second step was further adjusted for a polynomial in age (age, age2, age3) using stepwise regression. The two-step procedure that forced weight into the equation was necessary, since weight was not significant (α > 5%) in one of the race by sex by generation groups (Black fathers). A second baseline V̇O2vt phenotype was constructed similarly to the first, except that in the second regression the stepwise covariates included a polynomial in age, fat mass (FM), and fat-free mass (FFM). Thus, the two analysis variables represent V̇O2vt adjusted for weight and age (B1V̇O2vt), and adjusted for weight, age, FM, and FFM (B2V̇O2vt). With few exceptions, the additional adjustments for FM and FFM had little or no impact on the weight and age analysis and the data are not further presented. VT%V̇O2max was computed as V̇O2vt (mL·min−1) × 100/V̇O2max (mL·min−1) and then adjusted for covariates. Two adjusted variables were constructed for VT%V̇O2max using stepwise regression. The first was adjusted for a polynomial in age (B1%V̇O2max), and the second adjusted for age, FM, and FFM (B2%V̇O2max). With few exceptions, the additional adjustments for FM and FFM had little or no impact on the weight and age analysis, and the data are not further presented except where differences between the two models were noted. The response V̇O2vt was first computed as posttraining (mL·min−1) − baseline (mL·min−1) values, and then adjusted for a polynomial in age and the baseline value (mL·min−1) using stepwise regression (D1V̇O2vt).

Familial aggregation.

Familial aggregation was tested with an sex-specific familial correlation model using the computer program SEGPATH (21). The program fits the model directly to the family data using maximum likelihood methods under the assumption that the phenotypes within a family jointly follow a multivariate normal distribution. The general model was derived from four types of family members (fathers (f), mothers (m), sons (s), daughters (d)), leading to eight correlations within three familial classes (1 spouse (fm), four parent-offspring (fs, fd, ms, md), and three sibling (ss, dd, sd)). The general model (i.e., all eight correlations estimated) was fit to the data, as were several reduced (alternative) models examining sex and/or generation differences and the significance of the correlations. Reduced hypotheses were tested using the likelihood ratio test (LRT), which is the difference in minus twice the log-likelihoods (−2 ln L) obtained under the general model and a reduced hypothesis. The LRT is distributed as a χ2, with the degrees of freedom equal to the difference in the number of parameters estimated in the two nested models. In addition, Akaike’s information criterion (AIC), which is −2 ln L plus twice the number of estimated parameters, was used to compare nonnested models; the “best” model is the one with the smallest AIC (1).

Several reduced hypotheses tested for sex and generation differences. First, the hypothesis of no sex differences in offspring was tested (fs = fd, ms = md, ss = dd = sd, df = 4). Other models included tests for no sex differences in parents and offspring (fs = fd = ms = md, ss = dd = sd, df = 5), and no sex or generation differences (fs = fd = ms = md = ss = dd = sd, df = 6). In addition, models positing same versus opposite sex correlations (fs = ss = md = dd, fd = ms = sd, df = 5) and sex-specific correlations (fs = ss, md = dd, fd = ms = sd, df = 4) were examined. A maternal or mitochondrial hypothesis (ms = md = ss = dd = sd, df = 4), and an environmental hypothesis consisting of a single correlation (fm = fs = fd = ms = md = ss = dd = sd, df = 7) were also tested. Finally, the hypotheses of no sibling correlations (ss = dd = sd = 0, df = 3), no parent-offspring correlations (fs = fd = ms = md = 0, df = 4), no spouse correlation (fm = 0, df = 1), and no familial resemblance at all (fm = fs = fd = ms = md = ss = dd = sd = 0, df = 8) were tested. Nonrejected hypotheses were combined in order to form the most parsimonious model, from which the familial effects (maximum heritabilities) were derived. Maximal heritability was computed as follows: (rsib + rpo)(1 + rspouse)/[1 + rspouse + (2)(rspouse)(rpo)], where rsib is the average sibling correlation, rpo is the average parent-offspring correlation, and rspouse is the spouse correlation (22). This equation adjusts the familial effect for spouse resemblance, and when the spouse correlation is zero the equation reduces to twice the average sibling and parent-offspring correlation.


Sample statistics and covariate adjustments.

Table 1 gives the sample sizes, means, and standard deviations for each of the measures, separately in eight race by sex by generation groups. Mean group differences were judged by a comparison of standard errors. Race differences were noted for every variable in the offspring (sons and daughters). In the parents, there were race differences for age and mass variables but not V̇O2 variables in the fathers and few race differences in the mothers (age and baseline V̇O2vt). There were both sex and generation differences for most variables in all groups, except for fewer sex differences across Black females (mothers vs daughters).

Table 1
Table 1:
Means and standard deviations (SD) by race and family position.

Because the V̇O2vt was significantly skewed (1.0) and leptokurtotic (1.74), a log-transformation was performed before covariate adjustments, which resulted in a more normally distributed variable (0.29 and −0.002, respectively). Covariate adjustment results are given in Tables 2–4. The adjustment of V̇O2vt (mL·min−1) for weight (Table 2) generally accounts for more variance in Blacks than in Whites except for fathers, where the inverse was seen. Further adjustments for age (Table 3), FFM, and FM (Table 4) generally account for additional variance across offspring (Black and White son and White daughter) groups but not for parents for V̇O2vt. Age is a better predictor of VT%V̇O2max in White parents and Black daughters than is FM or FFM. For the response of V̇O2vt, baseline V̇O2vt levels account for 4–12% of the variance in the White mothers and offspring.

Table 2
Table 2:
Percent of variance accounted for in baseline V̇O2vt by weight.
Table 3
Table 3:
Age adjustments.
Table 4
Table 4:
Age, FM, and FFM adjustments.

Familial aggregation.

The SEGPATH model-fitting results are given in Tables 5 (White sample) and 6 (Black sample). For the age-adjusted V̇O2vt at baseline (B1V̇O2vt) in the White sample, none of the sex or generation tests were significant. For example, there were no sex differences in the offspring (χ24 = 7.08, P = 0.1316) or the offspring and parents (χ25 = 7.45, P = 0.1894), nor were there any generation differences (χ26 = 7.74, P = 0.2580). Same versus opposite sex correlations were not ruled out (χ25 = 6.27, P = 0.2808), nor were sex-specific correlations (χ24 = 4.97, P = 0.2900). Moreover, a single correlation model fit the data (χ27 = 8.73, P = 0.2726), as did the maternal transmission (or mitochondrial) model (χ24 = 5.76, P = 0.2177). Although the sibling (χ23 = 17.38, P = 0.0006) and parent-offspring correlations (χ24 = 30.27, P < 0.0001) were significantly different from zero, the spouse correlation (χ21 = 2.56, P = 0.1093) was not. The hypothesis of no familial effects (model 12) was rejected (χ28 = 46.35, P < 0.0001). The AIC was used to determine the best sex-differences hypothesis, which was model 4 (no sex or generation differences). The combined test of no sex or generation difference and no spouse correlation (models 4 + 9) fit the data by likelihood ratio criterion (χ27 = 9.99, P = 0.1891). However, the AIC for that model (11.99) was larger than that for the single correlation hypothesis (model 10, AIC = 10.73). Therefore, although either model adequately fits the data, the single correlation model provides the “best” fit and was chosen as the most parsimonious.

Table 5
Table 5:
Goodness of fit tests: White sample.
Table 6
Table 6:
Goodness of fit tests: Black sample.

For age-adjusted VT%V̇O2max (B1%V̇O2vt) in Whites, the parent-offspring correlations were significant (χ24 = 10.86, P = 0.0282), although the sibling (χ23 = 5.32, P = 0.1499) and spouse (χ21 = 2.84, P = 0.921) correlations were not. In general, there was a significant familial effect (χ28 = 18.95, P = 0.0256). These separate tests suggest a model for no sex or generation differences, and no sibling and no spouse correlations (models 4 + 7 + 9). Although this hypothesis fit the data by likelihood ratio test (χ27 = 10.91, P = 0.1426, AIC = 12.91), the most parsimonious model by AIC was for a single correlation (AIC = 6.97).

Finally, for the age-baseline V̇O2vt adjusted V̇O2vt response to training (D1V̇O2vt) in Whites, any of the models fit by likelihood ratio (i.e., no hypotheses were rejected). Although several alternative models were tested, the most parsimonious model by AIC was for a single correlation (AIC = 6.59).

The model-fitting results for the Black sample were quite different (Table 6). Because of the small sample sizes, there was not enough power to test for sex differences, so the correlations were tested for significance by familial class (i.e., assuming no sex differences). For the age-adjusted V̇O2vt (B1V̇O2vt), there was significant familial resemblance between sibling pairs but there was no parent-offspring or spouse resemblance. The separate tests suggest a model for no parent-offspring and no spouse correlations (models 3 and 4). However, the best model by likelihood ratio test and by AIC was for a single familial correlation (where sib = parent-offspring) and with no spouse correlation (AIC = 2.31). After V̇O2vt was adjusted for FM and FFM (B2V̇02vt), only the sibling correlations were significant by likelihood ratio. However, the most parsimonious model by AIC was for no spouse correlation and a single correlation model where parent-offspring = siblings (models 4 + 6, AIC = 2.11).

For the age-adjusted VT%V̇O2max (B1%V̇O2max) in Whites, the best model by AIC was for a single correlation (i.e., spouse = parent-offspring = sibling), although none of the correlations were strictly significant by likelihood ratio test. After VT%V̇O2max was adjusted for FM and FFM (B2%V̇O2max), the “best” model was for only a spouse correlation (i.e., parent-offspring = sibling = 0). The most parsimonious model (AIC = 3.14) for the V̇O2vt (mL·kg−1·min−1) response to training (D1V̇O2vt) was for a single correlation (model 5).

The parameter estimates under both the general and the most parsimonious models are given in Tables 7 (White) and 8 (Black). In the White sample, the maximal heritability estimates at baseline were 58% for V̇O2vt and 38% for VT%V̇O2max. Adjustments for FM and FFM were similar. For Whites, the maximal heritability estimate for the response of V̇O2vt to aerobic training was 22%.

Table 7
Table 7:
Parameter estimates (± SE) under general and parsimonious models: White subsample.a
Table 8
Table 8:
Parameter estimates (± SE) under general and parsimonious models: Black subsample.a

In the Black sample, the maximal heritability estimates for baseline V̇O2vt (54%) were similar to those of the White sample regardless of adjustments for FM or FFM. The estimates were similar to those of the White sample for VT%V̇O2max before adjustment for FM and FFM (39%) but not after FM and FFM adjustments, where there was no longer evidence to suggest a nonzero heritability, perhaps because of the small sample size in the Black sample. The response of V̇O2vt to aerobic training in Blacks has a larger estimate of maximal heritability than in the White sample (51%).


Significant familial contributions were found for both baseline V̇O2vt and baseline VT%V̇O2max (V̇O2vt/V̇O2max), as well as for the response to training. Other important findings were that familial factors accounted for more variance in V̇O2vt (over 50%) than in VT%V̇O2max (less than 40%). Also, adjustments for FM and FFM only affected the heritability for VT%V̇O2max in the Black sample. The only Black versus White differences in the familial effects were for VT%V̇O2max after FM and FFM adjustment (greater effect in Whites) and for the V̇O2vt response (greater effect in Blacks).

For baseline V̇O2vt, there was no difference in the familial effect between the White and Black samples, respectively, either before (58% and 54%) or after (47% and 46%) the combined adjustments for FM and FFM. For the response to training in V̇O2vt, there was evidence of significant familial/genetic influences in both races, with the familial influence for the training response in the Black sample (51%) being greater than and about double that of the familial influence in the White sample (22%).

There was a strong familial component for V̇O2vt in the sedentary state. Although the contributions of genetics and familial environment cannot be separately quantified, inspecting the correlation patterns can lead to inferences about their relative contributions to the variation in the phenotype. The significant spouse correlations for baseline V̇O2vt, at least in the White sample, suggest an environmental component as well as a genetic component. Thus, in the sedentary state some individuals will have varying submaximal fitness levels on the basis of genetics and shared familial environment. A β path analysis of the Quebec Family Study sample (17) suggested that nongenetic factors could explain 100% of the transmission of steady-state work rate at a fixed HR of 150 (PWC150) from parents to offspring. In the present study, it is not possible to differentiate genetic versus environmental effects, but it cannot be ruled out that shared environment contributes substantially to the maximal heritability effects though, generally, the parent-offspring and sibling correlations are stronger than the spousal correlations.

These baseline data are similar to submaximal aerobic power results reported by Bouchard et al. (2). In that study, an evaluation of the familial relationship of PWC150, adjusted for body mass, found estimates of the total genetic effect to be about 30–48%. The slightly higher values (47–58%) found in this study may be a result of the controlled heterogeneous sedentary nature of the individuals in the current study or a result of the different variables used to evaluate submaximal fitness. Pérusse et al. (18) and Katzmarzyk et al. (9), using data from the Canadian Fitness Study (11,680 participants from 4,144 nuclear families) estimated general heritability of 28% for PCW150 but, like the current study, could not differentiate between genetic and environmental transmission. The results from the Canadian Fitness Study are similar, though of lesser magnitude, to the results from the current analyses. Most recently, using data for White HERITAGE participants, Pérusse et al. (20) reported that the maximal heritability of several measures of submaximal fitness corrected for weight, age, and sex (V̇O2 at PO of 50 W, 60%V̇O2max, 80%V̇O2max and PO at 60%V̇O2max, 80%V̇O2max) reached values ranging from 48–74%. In the current study, baseline V̇O2vt averaged 55% of V̇O2max and had a maximal, weight-, and age-adjusted heritability for Whites of 58%. This compares favorably with the 50% maximal heritability for 60%V̇O2max determined by Pérusse et al.

For VT%V̇O2max, the pattern of results was quite similar to those for V̇O2vt, except that the magnitude of the effect was lower (38% and 39% for Whites and Blacks, respectively) than for V̇O2vt. However, in the Black sample, the familial effects appeared to be entirely explained by FM and FFM. In an analysis of the HERITAGE data, we have found that V̇O2vt declines at a slower average rate of −0.113 mL·min−1·kg−1·yr−1 between ages 17 and 65 in our study population, than does V̇O2max (−0.465 mL·min−1·kg−1·yr−1). These differences in the age-related loss in submaximal versus maximal cardiorespiratory fitness account for 43% of the variance in VT%V̇O2max with age. Furthermore, the V̇O2vt heritability data presented in the present study and the heritability in HERITAGE of V̇O2max reported by Bouchard et al. (6) probably contribute significantly to the reported familial component of VT%V̇O2max, at least in the White sample.

Finally, the maximal heritability in the response to training of V̇O2vt in the current study is similar to results found by others. Pérusse et al. (22) found maximal heritability in the response to HERITAGE training in several submaximal phenotypes (V̇O2 at PO of 50 W, 60%V̇O2max, 80%V̇O2max and PO at 60%V̇O2max, 80%V̇O2max) to range from 23–57% in White families. The intensity of training studied by Pérusse et al. that best corresponds to the intensity of exercise at VT for this population is the training response for V̇O2 at 60%V̇O2max. The maximal heritability in White families of 23% for the 60%V̇O2max phenotype compares favorably with maximal heritability of 22% for White families in the current study.

The evidence for a genetic influence in the V̇O2vt response to a standardized aerobic training program suggests that there are genetic factors, as well as the shared exercise program, influencing the training response. Gaskill et al. (8) previously reported that the absolute intensity of exercise training during the final 6 wk of the HERITAGE exercise program was responsible for about 28% and 35% (Black and White families, respectively) of the V̇O2vt aerobic exercise training response. Combining the familial and training intensity data would suggest that a maximum of 50% of the V̇O2vt response variability in Whites and 90% in the Black sample could be accounted for by these two variables.

In summary, V̇O2vt has similar maximal heritabilities of 58% and 54% for the White and Black samples, respectively, in the sedentary state after adjustments for age and body mass. In contrast, the V̇O2vt response to training is characterized by apparently divergent familial components (22% and 51%) in Whites and Blacks, respectively.

The HERITAGE Family study is supported by the National Heart, Lung, and Blood Institute through the following grants: HL45670 (C. Bouchard, PI), HL47323 (A. S. Leon, PI), HL47317 (D. C. Rao, PI), HL47327 (J. S. Skinner, PI), and HL47321 (J. H. Wilmore, PI). A. S. Leon is supported in part by the Henry L. Taylor Professorship in Exercise Science and Health Enhancement. C. Bouchard is also partially supported by the George A. Bray Chair in Nutrition.

Address for correspondence: Steven E. Gaskill, Ph.D., Department of Health and Human Performance, The University of Montana, 112 McGill Hall, Missoula, MT 59812; E-mail: [email protected]


1. Akaike, H. A new look at the statistical model identification. IEEE Trans. Automat. Control 19: 716–723, 1974.
2. Bouchard, C., G. Lortie, J. A. Simoneau, C. Leblanc, G. Theriault, and A. Tremblay. Submaximal power output in adopted and biological siblings. Ann. Hum. Biol. 11: 303–309, 1984.
3. Bouchard, C., R. Lesage, G. Lortie, et al. Aerobic performance in brothers, dizygotic and monozygotic twins. Med. Sci. Sports Exerc. 18: 639–646, 1986.
4. Bouchard, C., F. T. Dionne, A. Simoneau, and M. R. Boulay. Genetics of aerobic and anaerobic performances. Exerc. Sport Sci. Rev. 20: 27–58, 1992.
5. Bouchard, C., A. S. Leon, D. C. Rao, J. S. Skinner, J. H. Wilmore, and J. Gagnon. The HERITAGE Family Study: aims, design and measurement protocol. Med. Sci. Sports Exerc. 276: 721–729, 1995.
6. Bouchard, C., E. W. Daw, T. Rice, et al. Familial resemblance for V̇O2max in the sedentary state: the HERITAGE family study. Med. Sci. Sports Exerc. 30: 252–258, 1998.
7. Fagard, R., E. Bielen, and A. Amery. Heritability of aerobic power and anaerobic energy generation during exercise. J. Appl. Physiol. 70: 352–362, 1991.
8. Gaskill, S. E., A. J. Walker, R. A. Serfass, et al. Changes in ventilatory threshold with exercise training in a sedentary population: the HERITAGE family study. Int. J. Sports Med. 22: 1–7, 2001.
9. Katzmarzyk, P. T., L. Pérusse, D. C. Rao, and C. Bouchard. Familial risk ratios for high and low physical fitness levels in the Canadian population. Med. Sci. Sports Exerc. 33: 614–619, 2000.
10. Klissouras, V., F. Pirnay, and J. M. Petit. Adaptation to maximal effort: genetics and age. J. Appl. Physiol. 35: 288–293, 1973.
11. Lesage, R. J., A. Simoneau, J. Jobin, J. Leblanc, and C. Bouchard. Familial resemblance in maximal HR, blood lactate and aerobic power. Hum. Hered. 34: 182–189, 1985.
12. Londeree, B. R. Effect of training on lactate/ventilatory thresholds: a meta-analysis. Med. Sci. Sports Exerc. 29: 837–843, 1985.
13. Lortie, G., C. Bouchard, C. Leblanc, et al. Familial similarity in aerobic power. Hum. Biol. 54: 801–812, 1982.
14. Maes, H. H., G. P. Beunen, R. F. Vlietinck, et al. Inheritance of physical fitness in 10-yr-old twins and their parents. Med. Sci. Sports Exerc. 28: 1479–1491, 1996.
15. Montoye, H. J., and R. Gayle. Familial relationships in maximal oxygen uptake. Hum. Biol. 50: 241–249, 1978.
16. Pérusse, L., G. Lortie, C. Leblanc, A. Tremblay, G. Theriault, and C. Bouchard. Genetic and environmental sources of variation in physical fitness. Ann. Hum. Biol. 14: 425–434, 1987.
17. Pérusse, L., C. Leblanc, A. Tremblay, et al. Familial aggregation in physical fitness, coronary heart disease (CHD) risk factors and pulmonary function measurements. Prev. Med. 16: 607–615, 1987.
18. Pérusse, L., C. Leblanc, and C. Bouchard. Familial resemblance in lifestyle components: results from the Canada fitness survey. Can. J. Public Health 79: 201–205, 1988.
19. Pérusse, L., C. Leblanc, and C. Bouchard. Inter-generation transmission of physical fitness in the Canadian population. Can. J. Sport Sci. 13: 8–14, 1988.
20. Pérusse, L., G. Gagnon, M. A. Province, et al. Familial aggregation of submaximal aerobic performance in the HERITAGE Family Study. Med. Sci. Sports Exerc. 33: 597–604, 2001.
21. Province, M. A., and D. C. Rao. General purpose model and a computer program for combined segregation and path analysis (SEGPATH): automatically creating computer programs from symbolic language model specifications. Genet. Epidemiol. 12: 203–219, 1995.
22. Rice, T., J. P. Despres, E. W. Daw, et al. Familial resemblance for abdominal visceral fat: The HERITAGE Family Study. Int. J. Obes. 21: 1024–1031, 1997.
23. Skinner, J. S., K. W. Wilmore, J. B. Krasnoff, et al. Adaptation to a standardized training program and changes in fitness in a large heterogeneous population: the HERITAGE Family Study. Med. Sci. Sports Exerc. 32: 157–161, 2000.
24. Sundet, J. M., P. Magnus, and K. Tambs. The heritability of maximal aerobic power: a study of Norwegian Twins. Scand. J. Med. Sci. Sports 4: 181–185, 1994.


© 2001 Lippincott Williams & Wilkins, Inc.