Opinion TRENDS in Ecology & Vol.17 No.3 March 2002 115

or mutagens [8–11]. Third, NVP occurs during sickness organogenesis, when the developing embryo is at greatest susceptibility [9] to mutagens. Fourth, the incidence of NVP correlates with [8–11], being largely absent in and embryo quality societies where maize and/or corn are food staples [10,12]. The fifth and most direct evidence is the correlation of NVP with pregnancy outcome: mothers with NVP Scott Forbes experience lower rates of , early pregnancy loss, preterm delivery and low than do mothers without [15–17], a pattern attributed to the prophylactic and are routine features of early pregnancy in humans. But are benefit of food aversions associated with NVP. they adaptive or pathological? Several lines of evidence suggest that they The case for embryo protection therefore appears protect mothers and developing embryos from dietary mutagens and sound. But might there be another explanation? Haig pathogens. Nausea and vomiting in pregnancy (NVP) coincide with the has suggested that pregnancy sickness might arise as a vulnerable period of embryogenesis, are associated with food aversions, and are by-product of maternal–embryo conflict [18,19]. There predictors of positive pregnancy outcomes. Here, I argue that NVP is not directly are numerous potential conflicts between embryo and adaptive, but arises as a byproduct of genetic conflict between mother and mother during pregnancy, and pre- and embryo. The negative correlation between first-trimester spontaneous can be direct results [10,11]. One and NVP is not the result of protection of embryos from environmental mutagens conflict in particular is directly relevant to pregnancy or pathogens, but the result of intrinsic chromosomal defects. These low-quality sickness: who decides whether the pregnancy continues? embryos produce subnormal levels of human chorionic gonadotropin, a For the embryo, and occasionally the mother, it might be involved in pregnancy maintenance, and the probable proximate trigger for NVP. a struggle for life or death (Box 1). Here, the pregnancy hormone HUMAN CHORIONIC GONADOTROPIN (hCG) is Lassitude, sleepiness and loss of appetite signaled the key. Haig describes it as an allocrine hormone [10] onset of pregnancy for the Kwakiutl women of the Pacific produced by one individual to manipulate the coast of Canada. A pregnant woman was careful about endocrine system of another. And hCG does exactly her diet. She would eat dry salmon, fat and berries, but that. Closely related to LUTEINIZING HORMONE (LH), hCG avoided whale meat, squid, and salmon eggs. If she regulates early function. It is secreted had food cravings, her husband attempted to provide by the preimplantation embryo, and subsequently by these foods. And qualmishness and appetite loss were the placental TROPHOBLAST in early pregnancy. Without early signs that the forthcoming child was a girl [1]. hCG, the corpus luteum regresses. Without the corpus For most mothers in most societies, food aversions luteum, production falls. And, without and some nausea and vomiting are routine features progesterone, the pregnancy fails. of early pregnancy [2]. And, for a small fraction, the Maternal endocrine control of gestation is open nausea and vomiting is prolonged and debilitating to subversion by the embryo. Consider an ancestral (: see Glossary), and if left mother and her offspring. She alone regulates early untreated can be fatal [3,4]. But is pregnancy sickness pregnancy with LH. But the embryo has the same beneficial? Might it be an adaptation rather than complement of genes as its mother and placental ? In the late 1970s, Hook and Little access to the maternal bloodstream. Imagine further postulated that nausea and vomiting during a mutation that causes an embryonic gene encoding pregnancy (NVP or ) protected LH production to be turned on, thus adding to the developing embryos from alcohol (a potent dietary maternal supply. Some embryos that would have teratogen) and tobacco smoke [5,6]. Walker et al. [7] been aborted because of low maternal LH production examined NVP in relation to dietary aversions and would now survive. (This neat embryonic trick is cravings, and documented aversions to meat, fish, not exclusive to humans, but is rare in eutherian coffee and fatty foods, and cravings for sour, savory mammals, having arisen on at least three separate and sweet foods, and also for fruits and meat. Profet occasions [18]: in horses, guinea pigs and primates.) [8,9] later argued that NVP shielded embryos from an With a substantial selective advantage, such a array of dietary mutagens and pathogens, particularly mutation would quickly become fixed, triggering an plant toxins. Most recently, Flaxman and Sherman evolutionary arms race. Mothers that now carry suggested that NVP deters the ingestion of spoiled and embryos that they would ‘prefer’ in an evolutionary burned meats, fish and eggs that might affect not only sense to abort (no conscious choice implied) reduce their the developing embryo, but also mothers who are own LH production and/or increase the level of hormone immunocompromised during pregnancy [10,11]. required for pregnancy maintenance. Embryos respond

Scott Forbes Five lines of evidence support this embryo–maternal by elevating LH production even further. This appears Dept of Biology, protection hypothesis. First, pathogens and mutagens to be exactly what has happened through a series of University of Winnipeg, occur naturally in the diet and some foods have gene duplications [18]. Mothers retaliate, and so it Winnipeg, MB, harmful effects [8–14]. Second, NVP is associated continues, coming to an end when further escalation Canada R3B 2E9. e-mail: with food aversions, and strong and pungent odors becomes too costly for one or both parties. Human [email protected] are often correlated with the presence of pathogens mothers and their embryos just might have reached

http://tree.trends.com 0169-5347/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved. PII: S0169-5347(02)02428-4 116 Opinion TRENDS in Ecology & Evolution Vol.17 No.3 March 2002

Box 1. When does genetic conflict in pregnancy occur?

Flaxman and Sherman [a] argue that if genetic conflict holds it both obligate brood reduction and early spontaneous abortion are should occur late in pregnancy when the resource demands of the prospective adaptations. are greatest, rather than early in pregnancy when nausea Offspring and parent interests, however, can diverge. By and vomiting in pregnancy (NVP) actually occurs. But this producing elevated levels of human chorionic gonadotropin argument fails to recognize the pluralistic nature of genetic conflict (hCG), embryos avert the pregnancy loss that is surprisingly in human pregnancy. Conflict over the division of resources common: between one third and three quarters of human between mother and embryo does occur later in pregnancy, and conceptions are spontaneously aborted [f]. Intriguingly, the manifests as pre-eclampsia and gestational diabetes [b,c]. The most common chromosomal – trisomy 21 conflict that occurs early in gestation, however, is about survival (Down syndrome) – is associated with abnormally high hCG rather than the allocation of current resources. That is, should the production [b,f], further implicating hCG in a system that avoids pregnancy be terminated or continued? According to the theory maternal spontaneous abortion. of parent–offspring conflict [d], the mother should hold a more References conservative view about such decisions than should the offspring a Flaxman, S.M. and Sherman, P.W. (2000) Morning sickness: a mechanism (from an evolutionary rather than a conscious perspective). Here, for protecting mother and embryo. Q. Rev. Biol. 75, 113–148 a striking parallel occurs between early spontaneous abortion in b Haig, D. (1993) Genetic conflicts in human pregnancy. Q. Rev. Biol. 68, 495–532 humans and obligate brood reduction in birds, where a maximal c Haig, D. (1996) Altercation of generations: genetic conflicts of pregnancy. benefit is obtained from early brood reduction before any Am. J. Reprod. Immunol. 35, 226–232 proximate food limitation occurs [e]. When, for whatever reason, d Mock, D.W. and Parker, G.A. (1997) The Evolution of Sibling Rivalry, Oxford the circumstances for continued pregnancy are, from a mother’s University Press e Forbes, L.S. and Ydenberg, R.C. (1992) Sibling rivalry in a variable perspective, adverse (e.g. significant food shortfall), she could environment. Theor. Popul. Biol. 41, 335–360 potentially avoid the major costs by early spontaneous abortion f Forbes, L.S. (1997) The evolutionary biology of spontaneous abortion in before substantial resources are committed to the embryo. Thus, humans. Trends Ecol. Evol. 12, 446–450

this point. The cost is pregnancy sickness, a mild sickness, given that it is more common when the fetus inconvenience for most mothers (but see [20]), but in a is female than when it is male [26]. minority of cases, hyperemesis gravidarum results. Establishing a direct link between hCG and pregnancy sickness has been difficult, partly because The link between hCG and pregnancy sickness hCG exists in multiple forms, and hyperemesis Is there a hormonal link to NVP, as would be expected gravidarum might be linked to abnormal hCG under the genetic conflict described above? The balance (Box 2). High serum concentrations of of evidence suggests that there is (Box 2), but whether one particularly potent variety, for example, are hCG is the direct antecedent of NVP remains an open characteristic of HYDATIDIFORM MOLES that are associated question. hCG might be an integral part of the with excess thyroid stimulation and high levels of endocrine pathway that underlies NVP. It has multiple hyperemesis gravidarum [27–29]. Embryos with effects, including stimulating the corpus luteum to trisomy 21 (Down syndrome) also produce produce progesterone and ESTRADIOL, and the thyroid to exceptionally high levels of hCG [30], and high levels produce throxine (Box 3). Alternatively, its production of NVP [2,31]. Although hCG is clearly involved in this could reflect placental vigor and could be correlated maternal–embryo conflict, its exact role is still unclear. with another placental product that underpins NVP. This distinction notwithstanding, low hCG production Is pregnancy sickness adaptive? is associated with early pregnancy loss [21,22], and It could be argued that the mild NVP that induces high hCG levels are often, although not always, linked food aversions is adaptive, but as symptoms become to the occurrence of hyperemesis gravidarum (Box 2). progressively more severe, the plausibility of such Twin produce high levels of hCG and high an adaptation wanes. Hyperemesis gravidarum, levels of hyperemesis gravidarum [23,24]. Similarly, in particular, is obviously maladaptive. Can it be hCG levels in VANISHING TWIN pregnancies are lower accommodated within an adaptive framework? Some than in normal twin gestations [25]. And there is an variability is expected for nearly every trait and intriguing link to fetal gender: female babies are more hyperemesis gravidarum might represent the extreme likely to produce qualmishness in pregnancy [20]. end of a continuum. But even normal levels of NVP Among all Swedish births between 1987 and 1995 might bear fitness costs. In subsistence populations, (over one million in all), there was an overall such costs are evident because marginally nourished male:female ratio of 51.4:48.6. However, in mothers mothers with NVP experienced later nutritional admitted to hospital with hyperemesis gravidarum in and adverse pregnancy outcomes [20]. These effects the first trimester, the ratio was reversed: 44.3:55.7. are largely masked in modern societies with access Severe pregnancy sickness was much more likely when to virtually unlimited nutrition, and thus their a mother was carrying a girl. This effect again directly importance might be underestimated by proponents indicates hCG as the proximate cause of pregnancy of the embryo–maternal protection hypothesis.

http://tree.trends.com Opinion TRENDS in Ecology & Evolution Vol.17 No.3 March 2002 117

Box 2. hCG and pregnancy sickness

Ever since its discovery, human chorionic gonadotropin (hCG) demonstrate a relationship between hCG and hyperemesis has been suggested to be the proximate cause of pregnancy gravidarum [b,c]: one was based on very small samples [c], and sickness [a]. However, the link between hCG and pregnancy the other [b] did not clearly define hyperemesis gravidarum; sickness has been obscured by two studies that failed to differences in also existed between emetic and nonemetic patients. But the balance of evidence does suggest a strong and direct link between hCG and nausea and vomiting in 70 75 pregnancy (NVP). Most obviously, the incidence of pregnancy sickness mirrors the increase in hCG levels that begins two to 70 four weeks after conception, peaking at six to eight weeks and 60 disappearing by 14 weeks (Fig. I: red line, data for hCG taken from 65 [c]; blue line, data of the incidence of NVP taken from [d]). hCG

) excretion [a,e] and serum concentrations of hCG are higher in

Ð1 60 50 pregnant women with hyperemesis gravidarum than in normal

Women with nausea (%) pregnant women [f]. More recently, O’Connor et al., in a large- 55 scale hormonal and epidemiological study of women in rural 40 Bangladesh [g], established a strong association between 50 elevated levels of hCG and nausea and vomiting in pregnancy. References 45 a Schoeneck, F.J. (1942) Gonadotrophic hormone concentration in emesis 30 gravidarum. Am. J. Obstet. Gynecol. 43, 308–311 40 b Depue, R.H. et al. (1987) Hyperemesis gravidarum in relation to estradiol

level in maternal serum (IU l level levels, pregnancy outcome, and other maternal factors: a seroepidemiologic β study. Am. J. Obstet. Gynecol. 156, 1137–1141 20 35 c France, J.T. et al. (1996) Serum concentrations of human chorionic hCG gonadotrophin and immunoreactive inhibin in early pregnancy and recurrent 30 : a longitudinal study. Aust. NZ J. Obstet. Gynaecol. 36, 325–330 10 d Tierson, F.D. et al. (1986) Nausea and vomiting of pregnancy and association with pregnancy outcome. Am. J. Obstet. Gynecol. 155, 1017–1022 25 e Fairweather, D.V.I. and Loraine, J.A. (1962) Urinary excretion of human chorionic gonadotropin in patients with hyperemesis gravidarum. Br. Med. J. 3, 666–669 0 20 f Goodwin, T.M. et al. (1992) The role of chorionic gonadotropin in transient 45678910111213141516 of hyperemesis gravidarum. J. Clin. Endocrinol. Metab. Week of gestation 75, 1333–1337 g O’Connor, K.A. et al. (1999) Reproductive and pregnancy-related Fig. I TRENDS in Ecology & Evolution sickness in a prospective study of Bangladeshi women. Am. J. Phys. Anthropol. 108, 172

Evidence against genetic conflict? The link between diet and NVP would seem NVP is absent in certain pregnancies, and certain inconsistent with genetic conflict. Maize and/or corn- societies. Is this evidence against genetic conflict? If based diets are correlated with an absence of NVP vigorous embryos do produce NVP and poor quality [10,12] but, as the cliché goes, correlation is not embryos do not, then should not pregnancies without causation. It could simply be that mothers who rely NVP be expected to invariably end in early pregnancy on only a single food staple are undernourished loss? This argument is too simplistic, because NVP (reducing the incidence of NVP; see [33]) relative to is obviously multifactorial in origin. Female mothers with access to multiple food staples. Or it embryos, for example, produce more NVP than do could be that the correlation itself is spurious: neither males [32]; smoking decreases the incidence of the original analysis by Minturn and Weiher nor that emetic pregnancies [33], and allergies, gallbladder by Flaxman and Sherman controlled for phylogeny as disease [34] and twins increase it (Box 2). Heavier has become de rigueur in comparative analyses. The mothers are more prone to NVP [33], and even the latter authors acknowledge this problem, discuss it size and position of the corpus luteum can influence at length, and conclude that the lack of independent its severity [35]. Embryo quality is thus only one of contrasts is unlikely to affect the outcome of the many possible causes and the resulting relationship result. Perhaps, but three of the seven societies between NVP and embryo quality will vary. lacking NVP, for example, were North American Moreover, to argue that an absence of NVP is aboriginal. For many, this would represent a serious pathological [9] is almost certainly wrong. It could problem of pseudoreplication. A further possibility, as well be that the mother-to-be is carrying a healthy Flaxman and Sherman note, is that genetic conflict male singleton fetus, that she is slim, allergy free, and embryo–maternal protection are not mutually smokes, and has a small corpus luteum in the exclusive, but rather represent different levels of left ovary. causation, a notion worthy of further exploration.

http://tree.trends.com 118 Opinion TRENDS in Ecology & Evolution Vol.17 No.3 March 2002

Box 3. The role of hCG during early pregnancy

Human chorionic gonadotropin (hCG) is linked to series of eutherian mammals, luteinizing hormone released by the anterior hormonal pathways during early pregnancy, two of which are pituitary stimulates this progesterone production: in humans, hCG of direct interest (Fig. I). First, it plays a key role in pregnancy triggers this progesterone production by the corpus luteum. maintenance in humans. hCG is secreted by the pre-implantation The exact link, if any, between hCG and nausea and vomiting in embryo and by the SYNCYTIOTROPHOBLAST (see Box Glossary) of the pregnancy, is still unclear. The second key pathway involves the after implantation, and is detectable in maternal thyroid. hCG is structurally similar to thyroid-stimulating hormone only one day after implantation occurs [a]. Embryonic hCG (TSH) and stimulates the thyroid in early pregnancy. This effect is production appears to have supplanted the role of the maternal more potent with some forms of hCG than with others [c], and pituitary in early pregnancy [b]. Maintenance of the uterine excess stimulation results in transient elevation of thyroid ENDOMETRIUM in humans requires progesterone production from hormone levels and hyperemesis gravidarum, a syndrome the corpus luteum for the first eight weeks of pregnancy. In other termed gestational thyrotoxicosis [d,e]. hCG also stimulates estradiol production by the corpus luteum [f], and this might Nausea and vomiting provide an alternative pathway to causing nausea and vomiting ? [d]. hCG also stimulates the production of the hormones estrone, ? 17α-hydroxyprogesterone, and relaxin by the corpus luteum.

Thyroid Estradiol Estrone References a Yen, S.C. et al. (1999) Reproductive Endocrinology (4th edn), W.B. Saunders 17α-hydroxyprogesterone b Haig, D. (1993) Genetic conflicts in human pregnancy. Q. Rev. Biol. 68, 495–532 Relaxin c Yoshimura, M. et al. (1994) Thyrotropic activity of basic isoelectric forms of human chorionic gonadotropin extracted from hydatidiform mole tissues. Corpus J. Clin. Endocrinol. Metab. 78, 862–866 luteum d Goodwin T.M. et al. (1992) The role of chorionic gonadotropin in transient hCG hyperthyroidism of hyperemesis gravidarum. J. Clin. Endocrinol. Metab. 75, 1333–1337 e Kimura, M. et al. (1993) Gestational thyrotoxicosis and hyperemesis Progesterone gravidarum: possible role of hCG with higher stimulating activity. Clin. Endocrinol. 38, 345–350 f Tulchinsky, D. and Hobel, C.J. (1973) Plasma human chorionic gonadotropin, estrone, estradiol, estriol, progesterone and 17α-hydroxyprogesterone in human pregnancy. III. Early normal pregnancy. Am. J. Obstet. Gynecol. 117, 884–890 Uterine endometrium Box Glossary Placenta Endometrium: lining of the uterine wall. Syncytiotrophoblast: cell mass of the developing embryo that initially invades the Fig. I Embryo TRENDS in Ecology & Evolution maternal endometrium.

Pregnancy sickness and spontaneous abortion: embryos to be destined to be aborted, as advocates of post hoc ergo propter hoc? embryo protection have suggested. Rather, low-quality Mothers with NVP are less likely to lose embryos to embryos cause spontaneous abortion. Advocates of spontaneous abortion than are mothers without. the embryo protection hypothesis have reversed the Advocates of the embryo protection hypothesis argue that direction of causality. The inverse correlation between this is due to the prophylactic benefit of food aversions NVP and early pregnancy loss is explained more afforded by pregnancy sickness. This argument is almost parsimoniously as signaling embryo quality. Thus, it certainly wrong. Fifty to 70% of first-trimester recognized seems that mothers can indeed be a ‘little bit’pregnant. spontaneous are chromosomally abnormal Such pregnancies are associated with low levels of [36,37]. In perhaps the most detailed study [38], hCG, fail to produce NVP and usually end in early where spontaneous abortion was documented with spontaneous abortion. This idea of course is not new. ultrasonography, 77% of the karyotyped abortuses were Both Stein and Susser [40] and Haig have previously chromosomally defective. The actual rate of genetic defect made similar arguments, and clinicians even use in first trimester spontaneous abortions is probably early hCG levels as indicators of embryo–fetal health. higher, perhaps considerably so, as very early, occult With this pillar of the embryo–maternal protection spontaneous abortions are usually unnoticed (reviewed argument removed, support for NVP as an adaptation in [39]) and less obvious genetic defects are not begins to crumble. Could it be that NVP is a red detected by karyotyping technology. Thus, most herring in the study of maternal dietary adaptations first-trimester spontaneous abortions (when most to pregnancy? The advocates for embryo–maternal spontaneous abortions occur) are unrelated to diet, but protection have gathered an impressive array of rather stem from intrinsic genetic causes. An absence of evidence that human mothers alter diet during NVP during the first trimester does not cause low-quality pregnancy, and only the unreasonably skeptical

http://tree.trends.com Opinion TRENDS in Ecology & Evolution Vol.17 No.3 March 2002 119

Box 4. Where next? Glossary

The present evidence for embryo–maternal protection is almost entirely Corpus luteum: a structure formed from an egg follicle after circumstantial: experimental tests are needed badly. An obvious avenue ovulation; it produces progesterone necessary for the maintenance of pregnancy. for future research would be to link diet choice to pregnancy outcome in Estradiol: a sex steroid hormone produced in the ovaries. animal systems. Rats, with catholic diets, might be suitable models for Human chorionic gonadotropin (hCG): a pregnancy hormone use in cafeteria-style experiments. Are the foods that carry teratogens or produced by the embryo and placenta. pathogens avoided during pregnancy and, most importantly, does such Hydatidiform mole: masses of fetal tissue that arise from abnormal fertilization and result in abnormal placental growth, diet choice reduce the incidence of birth defects or spontaneous abortions? often without an embryo/fetus; most moles are diploid with both Although humans are not appropriate experimental subjects, further sets of chromosomes paternally derived. observational work would yield useful insights. Brown et al.’s study [a] of diet Hyperemesis gravidarum: an extreme form of pregnancy sickness choice in human mothers is a first small step in this direction. They examined characterized by persistent nausea and vomiting resulting in imbalance; it is sometimes defined simply as pregnancy whether pregnant mothers with and without nausea differed with respect to the sickness requiring hospitalization. Often extends beyond the first consumption of vegetables that Profet identified as potentially harmful, and trimester of pregnancy, and can be fatal if untreated. whether there was any link to pregnancy outcome (there was not). However, Luteinizing hormone: produced by the anterior pituitary; stimulates the gonads to produce sex hormones. this study did not compare diet before and during pregnancy, involved Progesterone: female sex hormone produced by the corpus domesticated plants with relatively low levels of phytotoxins [b], and did not luteum that maintains pregnancy. examine the role of meats and fish that Flaxman and Sherman subsequently Trophoblast: outer layer of the developing embryo that makes identified as especially hazardous to pregnant mothers and their embryos [c,d]. direct contact with the maternal tissues. Vanishing twin: the early loss of one or both embryos from twin An obvious test of the genetic conflict hypothesis would be to conceptions; a phenomenon that only began to be detected with administer human chorionic gonadotropin (hCG) to mammals without the advent of ultrasound technology. chorionic gonadotropins. For example, hCG is biologically active in mice and rats because of its close structural relationship to luteinizing hormone. Such an experiment would effectively simulate the origin of chorionic could doubt that configured this gonadotropins, and would be expected to reduce the incidence of adaptively. But whether the nausea and vomiting spontaneous abortion by co-opting the maternal regulation of pregnancy. associated with the first trimester of pregnancy are

References adaptive is much less clear. Perhaps the cascade of a Brown, J.E. et al. (1997) Profet, profits, and proof: do nausea and vomiting of early hormonal events that arises from genetic conflict pregnancy protect women from ‘harmful’vegetables? Am. J. Obstet. Gynecol. 176, 179–181 between embryo and mother exaggerates already b Profet, M. (1995) Pregnancy Sickness: Using Your Body’s Natural Defenses to Protect existing, and indeed adaptive changes to diet during Your Baby-to-Be, Addison-Wesley pregnancy. If so, NVPmight serve as a useful probe for c Flaxman, S.M. and Sherman, P.W. (2000) Morning sickness: A mechanism for protecting mother and embryo. Q. Rev. Biol. 75, 113–148 the further study of these mechanisms (Box 4). But NVP d Sherman, P.W. and Flaxman, S.M. (2001) Protecting ourselves from food. Am. Sci. 89, 142–151 itself might prove ultimately to be maladaptive sequelae of the early struggle between mother and embryo.

References 11 Sherman, P.W. and Flaxman, S.M. (2001) Protecting 22 Homan, G. et al. (2000) Human chorionic gonadotropin 1 Boas, F. (1966) Kwakiutl Ethnography, University ourselves from food. Am. Sci. 89, 142–151 as a predictor of outcome in assisted reproductive of Chicago Press 12 Minturn, L. and Weiher, A.W. (1984) The technology pregnancies. Fertil. Steril. 73, 270–274 2 Fairweather, D.V.I. (1968) Nausea and vomiting in influence of diet on morning sickness: a cross- 23 Grun, J.P. et al. (1997) The thyrotrophic role of pregnancy. Am. J. Obstet. Gynecol. 102, 135–175 cultural study. Med. Anthropol. 8, 71–75 human chorionic gonadotrophin (hCG) in the 3 Abell, T.L and Riely, C.A. (1994) Hyperemesis 13 Ames, B. et al. (1990) Nature’s chemicals and early stages of twin (versus single) pregnancies. gravidarum. Gastroenterol. Clin. North Am. synthetic chemicals: comparative toxicology. Proc. Clin. Endocrinol. 46, 719–725 21, 835–849 Natl. Acad. Sci. U. S. A. 87, 7782–7786 24 Kauppila, A. et al. (1975) Twin pregnancy. A 4 Broussard, C.N. and Richter, J.E. (1998) Nausea 14 Schardein, J.L. (1996) Naturally occurring clinical study of 355 cases. Acta Obstet. Gynecol. and vomiting of pregnancy. Gastroenterol. Clin. teratogens. J. Toxicol. Toxin Rev. 15, 369–391 Scand. (Suppl.) 44, 5–12 North Am. 27, 123–151 15 Weigel, R.M. and Weigel, M.M. (1989) Nausea and 25 Kelly, M.P. et al. (1991) Human chorionic 5 Hook, E.B. (1978) Dietary cravings and aversions vomiting of early pregnancy and pregnancy gonadotropin rise in normal and vanishing twin during pregnancy. Am. J. Clin. Nutr. 31, 1355–1362 outcome: a meta-analytical review. Br. J. Obstet. pregnancies. Fertil. Steril. 56, 221–224 6 Little, R.E. and Hook, E.B. (1979) Maternal Gynaecol. 96, 1312–1318 26 Danzer, H. et al. (1980) Maternal serum human alcohol and tobacco consumption and their 16 Brandes, J.M. (1967) First trimester nausea and chorionic gonadotropin concentrations and fetal association with nausea and vomiting during vomiting as related to outcome of pregnancy. sex prediction. Fertil. Steril. 34, 336–340 pregnancy. Acta Obstet. Gynecol. Scand. 58, 15–17 Obstet. Gynecol. 30, 427–431 27 Bruun, T. and Kristoffersen, K. (1978) Thyroid 7 Walker, A.R.P. et al. (1985) Nausea and vomiting 17 Medalie, J.H. (1957) Relationship between function during pregnancy with special reference and dietary cravings and aversions during nausea and/or vomiting in early pregnancy and to hydatidiform mole and hyperemesis. Acta pregnancy in South African women. Br. J. Obstet. abortion. Lancet 2, 117–119 Endocrinol. 88, 383–389 Gynaecol. 92, 484–489 18 Haig, D. (1993) Genetic conflicts in human 28 Egwuatu, V.E. and Ozumba, B.C. (1989) 8 Profet, M. (1992) Pregnancy sickness as pregnancy. Q. Rev. Biol. 68, 495–532 Observations on in Enugu, adaptation: A deterrent to maternal ingestion of 19 Haig, D. (1996) Altercation of generations: genetic Nigeria. Int. J. Gynaecol. Obstet. 29, 219–225 teratogens. In The Adapted Mind (Barkow, J. conflicts of pregnancy. Am. J. Reprod. Immunol. 29 Glick, M.M. and Dick, E.L. (1999) Molar et al. eds), pp. 327–365, Oxford University Press 35, 226–232 pregnancy presenting with hyperemesis 9 Profet, M. (1995) Pregnancy Sickness: Using Your 20 Pike, I.L. (2000) The nutritional consequences of gravidarum. J. Am. Osteopath. Assoc. 99, 162–164 Body’s Natural Defenses to Protect Your pregnancy sickness: a critique of a hypothesis. 30 Cuckle, H. (2000) Biochemical screening for Down Baby-to-Be, Addison-Wesley Hum. Nat. 11, 207–232 syndrome. Eur. J. Obstet. Gynecol. Reprod. Biol. 10 Flaxman, S.M. and Sherman, P.W. (2000) 21 O’Connor, J.F. et al. (1998) Differential urinary 92, 97–101 Morning sickness: A mechanism for protecting gonadotrophin profiles in early pregnancy and 31 Källén, B. (1987) Hyperemesis during mother and embryo. Q. Rev. Biol. 75, 113–148 early pregnancy loss. Prenat. Diagn. 18, 1232–1240 pregnancy and delivery outcome: a registry

http://tree.trends.com 120 Opinion TRENDS in Ecology & Evolution Vol.17 No.3 March 2002

study. Eur. J. Obstet. Gynecol. Reprod. Biol. 35 Samsioe, G. et al. (1986) Does position and size of 38 Guerneri, S. et al. (1987) Prevalence and distribution 26, 291–302 corpus luteum have any effect on nausea of of chromosome abnormalities in a sample of first 32 Askling, J. et al. (1999) Sickness in pregnancy and pregnancy? Acta Obstet. Gynecol. Scand. 65, 427–429 trimester internal abortions. Hum. Reprod. 2, 735–739 sex of child. Lancet 354, 2053 36 Boue, J. et al. (1975) Retrospective and prospective 39 Forbes, L.S. (1997) The evolutionary biology of 33 Klebanoff, M.A. et al. (1985) Epidemiology of vomiting epidemiological studies of 1500 karyotyped spontaneous abortion in humans. Trends Ecol. in early pregnancy. Obstet. Gynecol. 66, 612–616 spontaneous human abortions. 12, 11–26 Evol. 12, 446–450 34 Gadsby, R. et al. (1997) Pregnancy nausea related 37 Hassold, T.J. et al. (1980) A cytogenetic study of 40 Stein, Z. and Susser, M. (1991) Miscarriage, to women’s obstetric and personal histories. 1000 spontaneous abortions. Ann. Hum. Genet. , and the epiphenomena of pregnancy: the Gynecol. Obstet. Invest. 43, 108–111 44, 151–178 causal model. Epidemiology 2, 163–167

systematic biology community has resoundingly Typology versus accepted phylogenetic systematics (cladistics) as the means of determining evolutionary relationships, and if an analysis does not use this method, the hypothesis transformation in the is not generally considered adequately tested. Phylogenetic systematics has been the standard method for determining evolutionary relationships origin of birds for over two decades [4,5]. Cladistics groups organisms only by new features that are identified in the descendants of a common ancestor; therefore, the Kevin Padian and John R. Horner order of evolution that cladistics establishes is not based on overall similarities, degree of ecological differentiation, or ideas about adaptive value or The questions of bird ancestry and the evolution of typically ‘avian’ features appear necessity. Repeated independent cladistic analyses all never-ending. This is true in the press but not in the scientific trenches, where conclude that the closest relatives of birds comprise a the methods of comparative biology are productively used to settle many major small group of theropod dinosaurs (velociraptorines, questions, including the origin of birds. Opponents of the view that birds evolved dromaeosaurines and troodontids) [6–10]. No from dinosaurs tend to use typological characterizations of ‘reptiles’ and ‘birds’; cladistic analysis has produced a different result. although they accept evolution, their approaches do not use the transformational And no opponent of the bird–theropod hypothesis approaches of phylogenetic systematics, against which hypotheses of evolutionary has done a full cladistic analysis of the question. change in function, physiology and behavior should be tested. ‘Typologists’ also Opponents to the cladistic view rely on other kinds tend to depend on knowledge of how evolutionary processes must work, rather of knowledge. The theropod dinosaurs in question than on comparing independent patterns of evidence. Both typological and were too large, too late in time, could not climb trees, transformational approaches can be evolutionary, but the utility of typology is lacked postulated ‘key features,’ could not pass limited because it stresses taxonomic gaps rather than mosaic transitions. through an allegedly necessary gliding phase, or were physiologically incapable of performing birdlike If the pages of many science journals, weeklies and functions [1,11,12]. These are all propositions that newspapers are to be believed, several long-standing have been answered on their own terms, whether and perennially interesting questions in dinosaur functional, stratigraphic, or metabolic [10]; but the biology remain endlessly controversial, if not important point is that none was based on any completely intractable. Did birds evolve from evidence of relationship, so they do not really test dinosaurs? Were dinosaurs warm-blooded? Did they the question of bird origins [10]. No alternative take care of their young? Surprisingly, much of the hypothesis has withstood cladistic testing; and, in debate over these questions, however engaging to the fact, there have not been any specific alternative press, is rooted less in evidence than in methodology. hypotheses for >20 years. No other method of It reduces to a disagreement about the appropriate phylogenetic analysis has been proposed and argued methods and philosophical approaches to to supplant cladistics, which is why the field, as a evolutionary questions, and to the persistence of a whole, remains unconvinced by these objections. Kevin Padian* pre-evolutionary philosophy that still invests much Dept of Integrative thinking about evolutionary problems. Typology versus transformationism Biology and Museum of The philosophical problem at the root of this Paleontology, University of California, Berkeley, Are birds dinosaurs? (And how do we know?) methodological impasse is an old one (Fig. 1). Typology CA 94720-3140, USA. Not all workers agree that birds evolved from began as a pre-evolutionary idea that treats all *e-mail: kpadian@ Mesozoic theropod dinosaurs [1,2]. But to most organisms in a named group as if they share the same socrates.berkeley.edu scientists in the relevant fields, these objections do physiological and structural features [13]. It reinforces John R. Horner not test or falsify the hypothesis that birds evolved the perceived gaps between groups and makes it harder Museum of the Rockies, from dinosaurs because they are not based on to distinguish transitional features and forms. Typology Montana State University, Bozeman, MT 94718, testable hypotheses and ignore standard methods of is associated with the works of Linnaeus, which divided USA. approaching questions of relationship [3]. Why? The living organisms into discrete groups, innocent both

http://tree.trends.com 0169-5347/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved. PII: S0169-5347(02)02409-0