This study was supported, in part, by the Perinatology Research Branch, Division of Intramural Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), National Institutes of Health (NIH); the Department of Health and Human Services (DHHS); and, in part, with federal funds from NICHD, NIH under contract HSN275201300006C (to Dr Romero). Dr Pavlicev was supported by the March of Dimes Prematurity Research Centre collaborative grant 22-FY14-470. Dr Mitteroecker was supported by the Austrian Science Fund (FWF number 29397).

Without cesarean delivery, obstructed labor can result in maternal and fetal injuries or even death given a disproportion in size between the fetus and the maternal birth canal. The precise frequency of obstructed labor is difficult to estimate because of the widespread use of cesarean delivery for indications other than proven cephalopelvic disproportion, but it has been estimated that at least 1 million mothers per year are affected by this disorder worldwide. Why is the fit between the fetus and the maternal pelvis so tight? Why did evolution not lead to a greater safety margin, as in other primates? Here we review current research and suggest new hypotheses on the evolution of human childbirth and pelvic morphology. In 1960, Washburn suggested that this obstetrical dilemma arose because the human pelvis is an evolutionary compromise between two functions, bipedal gait and childbirth. However, recent biomechanical and kinematic studies indicate that pelvic width does not considerably affect the efficiency of bipedal gait and thus is unlikely to have constrained the evolution of a wider birth canal. Instead, bipedalism may have primarily constrained the flexibility of the pubic symphysis during pregnancy, which opens much wider in most mammals with large fetuses than in humans. We argue that the birth canal is mainly constrained by the trade-off between 2 pregnancy-related functions : while a narrow pelvis is disadvantageous for childbirth, it offers better support for the weight exerted by the viscera and the large human fetus during the long gestation period. We discuss the implications of this hypothesis for understanding pelvic floor dysfunction. Furthermore, we propose that selection for a narrow pelvis has also acted in males because of the role of pelvic floor musculature in erectile function. Finally, we review the cliff-edge model of obstetric selection to explain why evolution cannot completely eliminate cephalopelvic disproportion. This model also predicts that the regular application of life-saving cesarean delivery has evolutionarily increased rates of cephalopelvic disproportion already. We address how evolutionary models contribute to understanding and decision making in obstetrics and gynecology as well as in devising health care policies.

Key words : arrest of descent, arrest of dilatation, bipedalism, cephalopelvic disproportion, cesarean delivery, cliff-edge model, encephalization, erectile dysfunction, evolution, failure to progress in labor, fecal incontinence, fetal head, fistula, Homo erectus, mismatch, parturition, pelvic dimensions, pelvic dimorphism, pelvic floor disorder, pelvic inlet, prolapse, starvation during pregnancy, symphysis pubis, urinary incontinence, uterine prolapse, uterine rupture, vaginal prolapse

Cephalopelvic disproportion and obstructed labor

Human childbirth is substantially more difficult than that of most other primate species, owing primarily to the close match between the maternal birth canal and the fetal head1 (Figure 1). As a consequence of this close fit, even a small variation of maternal and fetal dimensions in human populations leads to a considerable rate of cephalopelvic disproportion (CPD) and obstructed labor.

Comparison of cephalopelvic proportions at the level of the pelvic inlet across closely related primates : orangutan (Pongo), chimpanzee (Pan), gorilla (Gorilla), and modern humans (Homo sapiens).

Humans show a very close fit between the pelvic inlet and the fetal cranium. Reproduced, with permission (copyright 2007 Worldwide Fistula Fund), from Wittman and Wall (107).

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 1 : Comparison of cephalopelvic proportions at the pelvic inlet across related primates

Where cesarean delivery is not easily available, particularly in developing regions of the Global South (Africa, south Asia, and Latin America), obstructed and prolonged labor are frequent causes of maternal morbidity and mortality.

Complications can be short term, such as uterine rupture and chorioamnionitis, or long term, such as fistulas and incontinence (2-6). Despite rich epidemiological data on obstructed labor and cesarean delivery, the precise frequency of CPD is difficult to estimate.

Current methods of pelvimetry are unreliable predictors of anatomical disproportion (7). Apart from an arrest of descent, often a sign of CPD (8), cesarean deliveries are frequently performed for other indications, e.g., fetal distress, failed induction, arrest of dilatation, repeat cesarean delivery, and maternal request, some manifesting before an arrest because of CPD (9,10).

With this complexity in mind, reported CPD rates range from 1% to 8% of childbirths across different geographic regions (11) [World Health Organization, 2003 (5)]. Hence, even the most conservative estimate entails about 40,000 affected births in the United States and about 1.3 million worldwide every year.

It is intriguing that childbirth, a process so fundamental to our species’ existence, exhibits such significant complication rates.

Human anatomical traits have been subject to natural selection for millions of years and are therefore often considered to be the best fit available for a given function. Hence, evolutionary anthropologists have long asked the questions :

Why is the human fetus so tightly matched to the maternal birth canal and thus so prone to birth complications ; why is there not a greater safety margin, as in most other primate species (Figure 1) ?

These long-standing evolutionary puzzles, which are of immediate relevance to obstetrics, gynecology, and public health, have received renewed attention in recent years. Here we review current theoretical and empirical research on the evolution of the human pelvic form and childbirth. Because several recent reviews highlighted the manifold social, cultural, psychological, and legal components underlying obstructed labor and surgical delivery (12-19), we focus on the biological aspects and their evolution since the split of the human lineage from the great apes.

The conflicting effects of bipedalism and encephalization on the pelvic architecture Bipedalism (see Glossary) evolved in the human lineage 5 million to 7 million years (myr) ago and coincided with a major remodeling of the pelvis.

Approximately 2 myr ago it was followed by the massive increase in brain size in the fully bipedal genus Homo, accompanied by an increase in the size of the fetal head (20).

Studying pelvic evolution throughout time is difficult for several reasons. First, the pelvis is prone to fragmentation and thus poorly represented in the paleontological record (21). Second, the pelvis is a highly integrated structure, in which selection on any one part results in changes to many others to preserve anatomical and functional integrity (22,23).

Finally, the human pelvis encountered multiple episodes of different selection pressures, each one leaving traces in the shape of the pelvis that are contingent on previous changes ; this challenges the separate reconstruction of these selective forces (24).

Despite these limitations, there is no doubt that the evolution of bipedalism coincided with major anatomical restructurings of the primate pelvis, as can be seen by comparing pelves of early bipedal hominids, such as australopithecines, to human-like apes (Figure 2).

The phylogenetic tree (left) shows the split between humans and their closest living relatives of the genus Pan (bonobo and chimpanzee). Bipedalism arose early in the human lineage and is common to the 3 genera, Ardipithecus, Australopithecus, and Homo, of which relatively good fossil remains have been studied.

Overall body size, adult brain size, and neonatal body mass increased in the human lineage (middle panel). Relative size and shape of the skull reflect the encephalization in genus Homo throughout the last 2 myr. The pelvic birth canal increased in size in the human lineage and shifted to a gynecoid shape. Partially redrawn and combined from Pontzer (25,108) and Gruss and Schmitt (25,108) myr, million years.

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 2 : Timeline of human evolution and obstetrically relevant traits.

These modifications include the shortening and widening of the illium, the alignment of the sacrum and the pubic symphysis in a dorsoventral plane, the broadening of the sacrum, the development of more prominent ischial spines, and a reduction of the distance between the hip joints and the vertebral column.

Many of these pelvic changes relate to the muscular requirements for efficient upright locomotion and balance of the upright body as well as for support of the viscera. Pelvic changes during the transition to bipedalism also sculpted a very specific birth canal. In australopithecines, the birth canal is mediolaterally broad both at the level of the inlet and at lower levels (25). Obstetric aspects likely played a minor role in the evolution of the australopithecine pelvis because the skull was relatively small (24).

Several myr after bipedalism evolved, the pelvis encountered a new evolutionary challenge in the genus Homo : the gradual increase of relative brain size (encephalization, see Glossary) and, hence, of fetal size. Evolutionary effects of encephalization on the pelvis usually are inferred from comparisons of modern humans to early representatives of the genus Homo, mostly Homo erectus.

Even though the cranium of H. erectus is already substantially larger than that of Australopithecus, pelvic differences between australopithecines and H. erectus are likely confounded by co-occurring ecological and behavioral changes that are independent of encephalization, such as changes in body size, ranging behavior, and thermoregulation (24). It is currently unknown when the increase in brain size started to affect pelvic form, but it is clear that this process continued after the last common ancestor (see Glossary) with H. erectus.

Relative to modern humans, H. erectus has flaring illiac blades and a mediolaterally broad (platypelloid) birth canal (Figure 2). The prevailing view is that birth-related pelvic changes within the human lineage accounted for the anterio-posterior broadening of the birth canal. Compared with H. erectus, the modern human pelvis is mediolaterally narrower, but this narrowing is relatively recent and appears to have occurred after the adaptation to changed obstetric demands. Finally, modern humans also became taller, which adds another source of strain to the pelvis by increasing the requirements for locomotory musculature and support of the inner organs.

Despite the common evolutionary history and functional demands, pelvic shape varies considerably within and across modern human populations (for the classic types, see Caldwell and Moloy (26,27 ; Figure 3). Studying this variation, along with its functional consequences and complications, can help to determine the manifold roles of the pelvis in locomotion, pelvic floor support, and childbirth (28), as detailed below.

The classic pelvic types with respect to the shape of the pelvic canal (109). The predominant type is gynecoid. Reproduced from Caldwell-Moloy (Proceedings Roy Soc Medicine) (26).

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 3 : Classic pelvic types with respect to the shape of the pelvic canal.

The concept of an obstetrical dilemma

In 1960, Sherwood L. Washburn (29) introduced the term “obstetrical dilemma” in a review of the effect of tool use on human evolution. Washburn proposed that bipedalism, while freeing hands for tool use, also resulted in the selection for a larger brain given the increasing tool use. Eventually, this led to conflicting selection pressures on the pelvis : concurrent with the evolution of bipedalism, the size of the birth canal had been reduced relative to that of other primates, but the subsequent increase in brain size required a large birth canal.

Such an opposition of selective forces is able to bring net evolutionary change in a trait to a halt, hence the term obstetrical dilemma. Washburn (29) proposed that advantages gained by increased pelvic width are outweighed by biomechanical disadvantages for bipedal gait. For example, individuals with a wide pelvis would have an easier labor and delivery process ; however, they would be less successful in gathering food and providing for offspring.

Moreover, he reemphasized an idea suggested earlier by Portmann, that the problem of large fetuses was partially ameliorated in human evolution by giving birth at an earlier, more altricial, developmental stage, at which cranial size is smaller (30,31).

Whereas primates, in general, are born at an advanced (precocious) developmental stage relative to, for example, mice, this trend is secondarily reversed in the human lineage (secondary altriciality, see Glossary). Hence, human neonates are more developed than those of truly altricial animals but less developed than those of other primates.

The proposal of Washburn (29) has been highly influential and gained wide acceptance in the anthropological community.

The details of this idea, however, have undergone further scrutiny. For simplicity, we will refer to generally wide/broad and narrow pelves in the discussion in the following text, referring to a composite of multiple characteristics, including the width of the birth canal, that make the female pelvis more or less suitable for childbirth (see Figure 4). More specific terms will be used when referring to detailed assessments of pelvic shape.

Clearly, a broader birth canal is beneficial for childbirth. This female specific selection for childbirth is manifested in the strong sexual dimorphism of pelvic anatomy, specifically in the size and shape of the pelvic outlet (Figure 4).

Average male and female pelves (2 middle columns) in anterior, superior, and lateral views along with 5-fold linear extrapolations of the sex differences in pelvic shape (leftmost and rightmost columns). In females the pelvic canal is more spacious, the illiac blades are shorter and reach further laterally, the subpubic angle is broader, and the sacral bone is shorter and more outward projecting than in males. Reproduced with permission, from Fischer and Mitteroecker (32).

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 4 : Pelvic sexual dimorphism.

In fact, the pelvis is the most dimorphic element of the human skeleton [(eg, Fischer and Mitteroecker (32)].

The source of selection for a narrow pelvis, however, is less clear. Washburn (29) proposed that further widening of the pelvis would impede bipedal gait, but recent biomechanical studies did not support this hypothesis (Figure 5).

The results of Warrener and colleagues,34 who tested Washburn’s hypothesis that pelvic width negatively affects gait efficiency. Mass-specific locomotor costs (net volume of oxygen consumed during exercise) during walking and running on a treadmill were measured for 8 males and 7 females.

Results showed no significant difference between the sexes despite wider hips in women than in men. These results suggest that bipedal gait is not the source of selection for a narrow pelvis. Reproduced from Warrener et al (34).

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 5 : Results of Warrener et al, who tested Washburn’s hypothesis.

Warrener and colleagues (33,34) showed that females and males do not differ in efficiency in either walking or running on a treadmill (Figure 5). The kinematic study of Whitcome et al (35) showed that during walking at higher speed, females translate more of the pelvic rotation into strides than males, which leads to similar efficiency of locomotion despite wider hips. Thus, instead of decreasing efficiency, a wide pelvis contributes to stride length and, particularly in individuals with shorter legs, may even be beneficial for walking (36).

These findings do not contradict the assumption that human pelvic architecture evolved in response to bipedal gait ; they show only that, in modern humans, pelvic width does not considerably affect walking efficiency.

This implies that walking efficiency is unlikely to be the cause for the evolutionary retention of our modern narrow pelvis ; a broader pelvis, which would clearly ease childbirth, must have other functional disadvantages than only locomotion.

The importance of the pelvic floor In quadruped mammals and primates, the pelvic floor musculature functions primarily for moving the tail and as sphincter (37) whereas in upright humans, the muscles and fasciae of the pelvic diaphragm create a horizontal pelvic floor that supports the abdominopelvic organs. It has thus been suggested that maintaining a relatively small pelvic outlet enhances this support function (37).

Anatomical features support this suggestion : the ischial spines, which severely narrow the birth canal in modern humans, serve as attachments of the muscles and fasciae of the pelvic floor (Figure 6). These spines are considerably smaller and located more dorsally in quadrupeds, in which the pelvic floor is a vertical structure, and the weight of the fetus rests predominantly on the backbone (37). Similarly, the sacral bone protrudes anteriorly into the birth canal in modern humans, whereas it is much shorter and posteriorly oriented in apes.

A, In the transition from quadruped to biped, the weight of the inner organs came to lie on the pelvic floor. The comparison of pelves between a quadruped gorilla and a bipedal human shows an increase in the size of pelvic floor-supporting structures : the ischial spines and the sacrum. B, The comparison between the H. erectus and modern human pelves shows a further enlargement of the ischial spines, concomitantly with an increase in the mass and cranial size of the fetus. Reproduced from Huxley (110). (A) ; and Gruss and Schmitt (25), with permission (B).

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 6 : Evolution of pelvic floor support.

After the evolution of bipedalism, the importance of pelvic support further intensified with the increase of fetal weight in the genus Homo. Indeed, the ischial spines are less prominent in early hominoids (such as H. erectus, Figure 6) than they are in the modern human pelvis (37), and the modern narrow pelvis likely provides more support for the pelvic floor than the broader archaic one.

This evidence suggests that a broad pelvis is more suitable for childbirth but less suitable for carrying the heavy fetus throughout a long pregnancy.

Evolutionarily, this situation intensifies the obstetric dilemma because both carrying and delivering are part of the same process: pregnancy (Figure 7).

According to this model, advantages and disadvantages of a broad pelvis do not pertain to different functions, such as pregnancy and walking. Rather, they both pertain to pregnancy: while a broad pelvis is advantageous for childbirth, it is disadvantageous for supporting the large fetus. We propose that this disadvantage limits the broadening of the pelvis.

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 7 : The model of obstetric selection trade-off.

If gait was the major selective force for a narrow pelvis, walking efficiency might have been compensated for, independently, by changes in leg musculature or behavior. But if pelvic floor support was the crucial driver for a narrow pelvis, such a compensation would not be possible : an evolutionary change that affects one function, either carrying or delivering the fetus, would always affect the other function, too.

Pelvic floor disorders and the obstetrical dilemma Pelvic floor disorders (PFDs) are common complications associated with pregnancy in general and obstructed labor in particular. The pelvic floor is a complex structure comprised of ligaments, muscles, and fasciae, arranged in several layers (38). The term pelvic floor disorders refers to a wide range of complications, affecting different parts of pelvic floor architecture and reflecting disruptions of distinct functions (39).

One group of PFDs arises at birth as a direct consequence of injuries to the muscles of the lower part of the pelvic floor, such as fistulas, uterine rupture, and injury to the sphincter muscles.40

These injuries are frequent outcomes of prolonged labor as a result of CPD and thus reflect disadvantages of a narrow birth canal, specifically for passing the large fetus during parturition.

According to the pelvic floor hypothesis, another group of disorders is expected to originate from strain to the muscles and connective tissue while carrying the heavy fetus. This strain is more likely to result in complications such as late-onset weakness of ligaments, associated with pelvic floor prolapse and incontinence, as a long-term consequence of multiple pregnancies (although acute injury can also cause these disorders). The strain, a consequence of carrying, is exacerbated when the bony pelvis offers little support and the pelvic floor is weak. Thus, both a wide pelvis during carrying the fetus as well as a narrow pelvis during birthing can result in the same general group of PFDs.

Studies addressing the association between PFD and pelvic anatomy often pool different PFDs to obtain sufficient sample size, but this challenges the detection of associations between specific complications and specific pelvic features. Nonetheless, several studies detected associations of PFD with both a narrow and a wide bony pelvis. For example, in a study of 59 women with PFD and 39 controls, Handa et al (41) found that the transverse diameter of the pelvic inlet is greater in women with pelvic floor disorders than in those without them. Interestingly, they also showed that prolapse is associated with a narrow obstetrical conjugate (the shortest pelvic diameter of the birth canal, see Glossary), presumably because of a predisposition to injury. Similarly, in a study on 34 patients and 34 controls, Sze et al (42) found that women with vaginal prolapse have a wider transverse diameter of the pelvic inlet than the control group. Two other large studies reported associations of a broad birth canal with urinary incontinence.

Stav and co-authors (26) found that women with wide transverse and anterio-posterior inlet diameters, as well as with a wider pelvic outlet diameter are more frequently affected by incontinence than women with a narrower inlet.

Moreover, Berger et al (43) reported that stress urinary incontinence is associated with a wide subpubic angle and a wide pelvic outlet (large interspinous and intertuberous diameters) (43,44). Because of the relatively small size of the effect observed, larger samples may be critical given that some of the smaller studies found no differences in bony dimensions (45,46). But even weak statistical associations between pelvic shape and pelvic floor function, which may be of limited clinical relevance today, can drive evolution : any consistent association between a morphological trait and survival rate or reproductive success imposes a selective pressure on this trait.

To understand the conflicting selection pressures acting on pregnancy, it will be important for future studies to distinguish between the different disorders and their associations with specific pelvic traits. Such studies may help to predict the risk of certain PFDs based on pelvic anatomy and to develop preventive strategies for women with a wider pelvis, such as strengthening specific aspects of the pelvic floor, introducing external support for carrying, or encouraging behavioral changes that ameliorate unnecessary strain.

The human pubic symphysis is particularly inflexible

Softening and widening of the pubic symphysis as well as relaxation of the sacroiliac joints during pregnancy widens the birth canal and thus ease childbirth ; it may appear as a human specific adaptation to the tight fetopelvic fit.

However, in several other mammals with relatively large neonates, flexibility of the pubic symphysis is much larger than in humans (47). In guinea pigs, for instance, the mean diameter of the fetal head is 20 mm, whereas the pelvic canal in early pregnancy is just 11mm wide. To accommodate the fetal head, the pubic bones separate up to 23 mm during late pregnancy in response to estrogen and relaxin, and a ligament appears in the middle of the joint (48,49). This flexibility is crucial because calcification of the symphysis, which occurs if the first pregnancy is delayed, frequently leads to CPD in guinea pigs (50).

The emergence of an interpubic gap, bridged by a flexible ligament, is also a prerequisite for successful delivery in other species, including mice, bats, deer mice, and macaques (49,51). In mice, for instance, the gap measures 4e10 mm at delivery (52), whereas in humans the mean increase of the interpubic gap is only 3 mm (53,54). In other mammals, by contrast (eg, in many odd- and even-toed ungulates and in big cats), the pubic symphysis is completely fused by ossification (47,49).

Why is the flexibility of the pubic symphysis so variable across mammals, and why is it so inflexible in humans, thus contributive to difficulty in childbirth ?

A widened pubic symphysis increases the birth canal and thus the vulnerability of the pelvic floor. Most species with a wide pubic gap are either very small and hence experience little pressure on the pelvic floor and engage in postures that reduce force on the pelvis and the pelvic floor (eg, roosting head down in bats), or they reduce the pressure of the viscera and the fetus on the pelvic floor by living in water. An ossified symphysis, by contrast, allows for a higher net force applied by the muscles to the rest of the body and thus may facilitate more energetically efficient locomotion (47). Apart from many fast running species, pubic flexibility is also strongly reduced in other large-bodied bipedal species, such as kangaroos.

In humans, a wide symphysis during pregnancy and birth is associated with severe pelvic girdle pain (55,56), common among athletes and patients with traumatic pelvic injuries (53,57). It is aggravated by weight bearing and associated with difficulty in walking (58). Evolutionarily, the larger birth canal that would result from increased symphysial flexibility apparently was outweighed by the increased risk of injuries to the pelvis and the pelvic floor. Grunstra et al (47) even hypothesized that bipedal locomotion in humans has not primarily constrained pelvic width, as classically suggested, but the flexibility of the symphysis.

Selection for a narrow pelvis may act via males : a role in erectile function

Both the obstetric demands and the additional strain exerted on the pelvic floor by the heavy fetus are specific to females. The evolutionary trade-off thus differs between the sexes and gave rise to the sexual dimorphism in pelvic morphology and pelvic canal size observable today. Similar dimorphism patterns are also seen in other primate species with a large fetal head relative to the size of the pelvic canal (cephalopelvic index, see Glossary), such as in lar gibbons, rhesus macaques, and squirrel monkeys, suggesting that pelvic dimorphism is, at least in part, a consequence of fetal size (59,60).

It is important to consider the specific conditions required for sexual dimorphism to arise in the evolution of traits that are present in both sexes. Divergent evolution (see Glossary) of such traits requires opposing selection pressures in the 2 sexes ; selection acting in only 1 sex is not sufficient because the developmental genes and processes underlying a trait, in this case the pelvis, are mostly the same whether they find themselves in a male or a female organism (genetic correlation between the sexes, see Glossary). This means that selection for a broader pelvis in females must have been counteracted by selection for a narrow pelvis in males. If there is no selection in males against the female adaptation, selection in females will also change, at least to a great extent, male shape, and no dimorphism will arise.

The fact that pelvic sexual dimorphism has nevertheless evolved implies that there has been selection opposing the widening of the pelvis in males.

Studying the male pelvis may thus provide insights into the advantages of a narrower pelvis independent of disadvantages because of birthing complications (25). Two relevant selection regimens are plausible : males can either share the same selective pressure for a narrow pelvis with females (as was suggested for gait or for visceral support) but lack the pressure for widening. Alternatively, selection in males may be for a male-specific function, perhaps also indirectly affecting the pelvis in females via a correlated effect.

If pelvic width is associated with pelvic floor strength in males, as is the case in females, male-specific pelvic floor disorders may offer interesting insights.

In addition to general effects such as pelvic pain and sphincter dysfunction, one of the disorders associated with a weak pelvic floor in males is erectile dysfunction (61). The pelvic floor contributes to erectile function by the muscles that form the penile basis, in particular the ischiocavernosus and bulbocavernosus muscles (Figure 8).

The 2 muscles shown in red, bulbospongiosus and ischiocavernosus, are involved in the female pelvic floor and crucial in maintaining the erection in males. Reproduced from https://www.qrsthai.com/erectile-dysfunction.

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 8 : Comparison of the male and female pelvic floors.

These muscles act as part of the complex involuntary muscular interplay, establishing and maintaining the pressure during erection and ejaculation.

The stability of these muscles may have become particularly important for erection in the human lineage because human males have evolved relatively large penises without a penile bone, a situation uncommon among primates (62).

A summary of US epidemiological studies, including the large Massachusetts Male Aging Study (63-65) reports severe erectile dysfunction in 10% of men aged 40-70 years, with another 25% experiencing intermittent erectile difficulties, and 17% with minimal difficulties (http://www.bumc.bu.edu/sexualmedicine/physicianinformation/epidemiology-of-ed). The prevalence of erectile dysfunction increases steeply with age, but even 5%-10% of men younger than 40 years are affected.

Vascular diseases and aging are the most important proximate factors underlying erectile dysfunction, but it is unclear to what extent they are conditional on pelvic floor strength, which has long been associated with male and female reproductive functioning (66,67). Recent studies reported that improving pelvic floor strength considerably alleviated erectile problems in a substantial percentage of men (61), even though direct estimates of how frequently a weak pelvic floor is the main cause of erectile dysfunction are, to our knowledge, lacking. For example, a study of 55 men showed large improvement of erectile function after 6 months of exercise of the pelvic floor musculature, with 40% regaining normal erectile function and a further 35.5% improving.68 Large randomized trials similarly revealed considerable positive effects of exercise on erectile dysfunction (69-71).

In short, indirect evidence indicates that the strength of the pelvic floor affects erectile function and thus, likely, male reproduction. Furthermore, the fact that the evolutionary widening of the female pelvis did not spill over to widen the male pelvis implies a selective advantage of a narrow pelvis in men. The missing information to link these 2 lines of evidence is whether pelvic width is indeed associated with pelvic floor strength in males. Data in women suggest that this is the case, but we are not aware of studies that have addressed this question in men.

One way to address this question would be an individual-based study examining the association between pelvic measurements and erectile dysfunction.

Such data may be readily available to urologists. An alternative but less direct way would be a comparison of sex specific pelvic morphology with the prevalence of erectile dysfunction across different populations. Literature suggests that interpopulation variation in pelvic shape and erectile dysfunction prevalence is large (72-76) ; a study combining these data, accounting for confounding factors such as overall health status, may be able to address the influence of pelvic morphology on erectile dysfunction.

The cliff-edge model

Despite the multiple sources of selection for a narrow pelvis in females and males, it still appears puzzling that a population’s fitness loss resulting from maternal and fetal mortality has been evolutionarily outweighed by the presumably small advantages for locomotion and pelvic floor stability. The recently developed cliff-edge model of obstetrical selection (77-79) revealed why natural selection, however strong, is not able to fully optimize the match between fetal and maternal dimensions, thus avoiding CPD.

The cliff-edge model is based on an idealized variable D (disproportion), which represents the difference between the size of the fetus and that of the maternal birth canal. When the fetus is smaller than the birth canal and encounters no obstruction, D is negative, whereas when the size of the fetus exceeds that of the birth canal, thus leading to obstructed labor, D is positive (Figure 9A). To study the evolutionary dynamics of fetopelvic disproportion, as expressed by D, an evolutionary fitness value is assigned to each value of D (y- and x-axes in Figure 9A, respectively).

A, The x-axis represents the variable D, which captures the difference between cranial size and pelvic size. This value is negative when the fetal cranium is smaller than the pelvic canal (D < 0) and positive when fetal size exceeds the size of the pelvic canal, leading to CPD. The y-axis on the left represents individual fitness. As long as the birth canal can encompass the fetus (D < 0), individual fitness (blue curve) increases linearly with D (both a larger fetus and a narrower pelvis lead to higher fitness). For D > 0, fitness is zero given CPD. The black curve represents the pdf of D, that is, the distribution of pregnancies across different values of D. Evolution by natural selection maximizes the average fitness in a population, which leads to an evolutionary optimum that entails an average value of D (gray dashed line) less than the individual optimum (ie, on the left side of the cliff edge) and a fraction of individuals with CPD (the red area beyond the cliff edge, at D > 0). This evolutionary stable state trades off the high fitness of the pregnancies with D < 0 (large fetuses that can fit through the birth canal) against the fraction of pregnancies with CPD. Modified from Mitteroecker et al (77). B, Cesarean delivery removes the cliff at the fitness threshold at D = 0, thus leading to a further evolutionary increase of D and the resulting CPD rate. Note that only cesarean deliveries performed as a result actual disproportion have this effect. Modified from Mitteroecker et al (77).

CPD, cephalopelvic disproportion ; D, disproportion ; pdf, probability density function.

Pavlicev. New ideas about the obstetrical dilemma. Am J Obstet Gynecol 2020.

Figure 9 : The cliff-edge model.

In evolutionary theory, fitness refers to the average number of offspring of individuals with a particular genotype or phenotype (in this case D). The offspring number is affected both by survival rate and fecundity.

What do we know about survival and fecundity with respect to childbirth that enables us to attribute fitness to values of D ?

Clearly, without cesarean delivery, individuals with D > 0 do not survive, so their fitness equals zero.

Epidemiological data document that larger neonates have higher survival rates (80,81) and thus higher fitness than smaller neonates. Furthermore, mothers with a wide pelvis have higher rates of PFD, which can lead to serious infections and social ostracization, as is common in sub-Saharan Africa (40,82,83). These relationships translate into the blue line in Figure 9A : fitness increases approximately linearly as a function of D, as long as the fetus is smaller than the birth canal. When fetal size exceeds the size of the birth canal (D > 0), survival and thus fitness drop sharply in the absence of medical care.

This highly asymmetric fitness curve resembles a cliff edge (hence the name of the model).

The association of fitness with D imposes natural selection on the corresponding fetal and maternal dimensions.

Only heritable traits can evolve by natural selection, and indeed, a number of studies reported considerable heritability (see Glossary) for fetal size and pelvic dimensions (84-87). Hence, the disproportion D is expected to respond to selection and increase in the population. In other words, the population becomes enriched for those trait values that confer high fitness so that the trait distribution gradually approximates the fitness curve. In the case of our trait D, one would expect many pregnancies to be just below 0, that is, with fetuses that are as large as possible while still being able to be born.

Yet here lies the inherent limitation of evolving biological systems : many phenotypic traits in a population, including fetal size and pelvic dimensions, have approximately symmetrical (normal) distributions and thus cannot match the asymmetrical fitness curve specific of obstetrics. Symmetrical distributions arise because most traits are affected by many genes with small additive effects along with symmetrically distributed environmental influences.

This mismatch limits the adaptive evolution of cephalopelvic fit : evolution by natural selection maximizes average fitness in the population (ie, the average number of offspring per individual), which results in an average value of D below the value with maximal individual fitness (at D = 0), and a certain proportion of cases exhibiting CPD (D > 0 ; Figure 9A). The persistent rates of CPD in human populations therefore do not suggest that natural selection has not worked but rather that evolution is limited by the inherent biology of human reproductive traits.

Given the cliff-edged fitness function, the model also shows that only weak selection toward a greater D is necessary to explain the observed rates of CPD (83).

In other words, against the common intuition, only weak fitness advantages for large newborns, or for a narrow female pelvis, or both, are necessary to evolve the relatively high CPD rates.

The heritability of fetal and pelvic dimensions implies that women born by cesarean delivery because of CPD are more likely to experience CPD when giving birth to their own offspring, compared to women who were born without CPD.

Large multigenerational studies on well documented populations (Scandinavia, United States) have found extensive evidence for inheritance of CPD (88-91), primarily via the maternal genome.

Despite different time periods, geographical regions, and methodical details, these studies report relatively homogenous odds ratios for CPD ranging from 1.55 to 2.06, which likely are underestimates (nonanatomical reasons for obstructed labor are not or only weakly heritable but may be misdiagnosed as CPD). The expected odds ratio inferred from the cliff-edge model is at 2.7779 (using an estimated heritability of 0.5 for D), a value remarkably close to the empirical observations, which supports the cliff-edge model and its assumptions.

Future dynamics of cephalopelvic disproportion

Based on the cliff-edge model, we predicted that CPD rates would increase, and may continue to increase, as a consequence of cesarean delivery (Figure 9B). The regular and safe use of cesarean delivery in cases of CPD removes the fitness threshold at D = 0 ; if the fetus is larger than the birth canal, fitness no longer drops to zero ; it may even continue to increase with increasing D. Thus, cesarean delivery has changed the selective pressure on maternal and fetal dimensions. As a consequence, evolutionary dynamics will move the population distribution toward greater D, leading to increasing rates of CPD (Figure 9B). In other words, the evolved equilibrium between the opposed selective forces of childbirth, pelvic floor stability, and infant survival has been disrupted by surgical delivery, giving rise to a new evolutionary trend.

Assuming an average heritability of 0.5 for fetal and maternal traits, as well as a duration of 2 generations since cesarean deliveries have been regularly and safely performed, we estimated that the original frequency of CPD would increase by about 10%-20% (depending on assumptions about male fitness and the genetic correlation between sexes), which is approximately half a percentage point of CPD incidence. Note, however, that only cesarean deliveries with CPD are contributing to this effect, not those performed for other indications (9,92).

This rough estimate is based on several assumptions (such as the representation of fetopelvic disproportion by a single quantity). Nonetheless, it shows that evolution by natural selection, triggered through medical intervention, can affect human anatomy over decades and not only within thousands or millions of years (93). Yet the model does not specify whether the increased rate of CPD results primarily from a change in fetal or maternal dimensions.

Historical data on pelvic dimensions, to our knowledge, are not available, and changes in neonatal dimensions are mainly driven by changes in nutrition and obstetric practices (78,92). In fact, the average neonatal weight has been decreasing in several industrialized countries because of shorter gestational length and a higher frequency of preterm birth, often because of medically indicated preterm birth, coupled with a higher survival rate because of the advances in neonatal care.

It is worthwhile to note that Berg-Lekas et al (90) reported an increase of primiparous dystocia from 2.2% to 12.5% between the 1950s and 1990s, an estimate that is probably attributed, at least in part, to changes in the reporting of this diagnosis and in short-term changes of economic development and nutritional status (such as post-war recovery). In this case, women developed during a period of poor nourishment, resulting in smaller body size, whereas when they became pregnant, their fetuses developed in an environment without nutritional deprivation, which fostered fetal growth.

Such environmental mismatch between maternal and fetal development adds to the increased frequency of cephalopelvic disproportion in countries recovering from famine or experiencing rapid socioeconomic development (94).

The evolutionary pressures that once increased the frequency of cephalopelvic disproportion have now been reduced by increased access to neonatal care and treatment of pelvic floor disorders, promotion of the survival of neonates, and the higher rate of preterm neonates with a lower birthweight and of women with a weak pelvic floor. These factors, in turn, reduce selection toward larger fetuses and narrower pelves. Fetal size is further constrained by maternal factors other than birth canal size, such as metabolic capacity (95) that are not removed by cesarean delivery and will set limits on an evolutionary increase in cephalopelvic disproportion.

What are the benefits for obstetrics and gynecology in understanding the evolutionary history of human traits and the selective pressures that have shaped them?

Evolutionary explanations for frequent obstetrical challenges (obstructed or prolonged labor, etc) can facilitate research on medical disorders, including the development of models for their prediction and strategies for prevention (96-99). For example, the idea that distinct pelvic disorders can result from a pelvis that is either too broad for some functions or too narrow for others can refine research questions and help to predict and interpret research outcomes.

Contingency on pelvic architecture may also explain some of the variation associated with mutations affecting connective tissue and associated PFDs. Furthermore, evolutionary understanding can guide targeted interventions for women who need support for carrying or who are particularly prone to injury at childbirth.

Finally, the insight that the male pelvis is likely subject to selective pressures for decreased width raises questions about which functions mediate this selection.

We discussed erectile function in this article, but other reasons may exist. Studying the male pelvis and its relationship with other disorders or traits is a promising area of investigation for understanding the conundrum of human childbirth and pelvic architecture.

Specific evolutionary models, such as the cliff-edge model, allow for predictions of how regular medical treatment may affect ongoing evolutionary dynamics (93). The evidence presented herein shows that medical interventions can sometimes lead to evolutionary changes that can take place in several generations, rather than across millennia.

Importantly, the reduction of natural selection by medical treatment does not necessarily reduce evolutionary change ; indeed, the disruption of evolutionary equilibriums can even induce new biological trends.

Clearly the goal of evolutionary reasoning is neither to criticize medical interventions nor to impose naturalistic ideals on modern society but to understand the transition of complex biosocial interactions and to inform public health and research policies.