Natural selection acts not on a behavior itself, but on the factors that produce that behavior, and/or the outcome of that behavior. So if we want to have an evolutionary explanation for a behavior, it’s important to understand both what drives it and its consequences. We also need evidence that this behavior (or again, what comes before or after it) has some clear fitness benefit.

The kinds of factors that produce variation in behavior include things like resource availability, group composition, and mating strategies. How much food does the individual have access to, but also what kind? What do they need to do to acquire it and do they need to compete to do so? Who else is in the group and are they related to her? Is she a seasonal breeder or spontaneous ovulator, a promiscuous, monogamous, or polygynous primate?

So there are certainly good reasons to expect many behaviors to have adaptive significance, and for them to improve reproductive success. Yet when it comes to studying humans, too often this evolutionary buttressing collapses, only to be replaced with pop evolution interpretations. These interpretations tend to forget that things we find sexy today may or may not have been preferred in ancestral environments, and that if we cannot understand relationships between behavior and reproductive success without some sort of mechanism, it is little more than storytelling.

Hormones are one surefire way to begin to build a mechanistic link between behavior and reproductive success. Yet I am usually rather ornery about those manuscripts as well. When I complain about articles about hormones and behavior, they usually are based on the following issues:

  • Use of a WEIRD population (Western, Educated, Industrialized, Rich, Democratic)
  • Infrequent sampling through the menstrual cycle, with no effort to determine ovulation
  • No clear mechanism to link the hormone and behavior
  • Evolutionary storytelling that doesn’t link to fitness benefits

So it’s nice to come across a paper* on hormones and behavior that not only manages not to fall into any of the above traps, but surpasses my expectations in terms of the quality of hypothesis testing, statistical analyses, and conclusions.

What they found

Ziomkiewicz et al (2012) found that urban, Polish women with a particular temperament profile – those with high activity and endurance traits but low emotional reactivity – had higher estradiol concentrations and a progesterone profile that rose earlier in the luteal phase compared to women with the opposite profile (low activity and endurance, high reactivity). They used the Formal Characteristics of Behavior – Temperament Inventory developed by Strelau and Zawadzki, and focused on activity, endurance and emotional reactivity because they correlate with personalities that indicate high reproductive success. In their supplementary materials, Ziomkiewicz et al more clearly define these traits:

“Emotional reactivity is the tendency to react intensively in response to emotional stimuli and is expressed by high emotional sensitivity and low emotional endurance. Endurance is the ability to react adequately in situations which require long-lasting or highly stimulative activity or under intensive external stimulation. Activity enables to maintain behaviors with high stimulative value or behaviors that provide high external stimulation” (Ziomkiewicz et al., 2012).

Therefore women with high activity and endurance, but low emotional reactivity – and thus are emotionally stable and extraverted – are called High Ability to Process Stimulation (HAPS) individuals. Those with low activity and endurance but high emotional reactivity – so neurotic and introverted – are called Low Ability to Process Stimulation (LAPS) individuals.

The participants of this study were healthy, urban women from Wrocław, Poland, and they collected urine daily for one menstrual cycle. While there were 116 total participants, the main analyses of the paper compare HAPS and LAPS women, of which there were 26 participants each. This is still a great sample size for a study that has so much analytical richness per participant – for comparison, my own dissertation has half the number of participants (n = 25), and my only other major dataset 46 participants.

Behavior is context-dependent

Ziomkiewicz et al took a life history perspective on the question of whether ovarian hormones play a role in women’s personality. Women who are better at processing stimuli probably have more resources available to them, which makes them better equipped to handle the high resource demands of pregnancy, lactation and childcare. Women who are better at processing stimuli in their environment tend to be more extroverted and sociable, and extraversion and sociability are both highly preferred traits in potential mates, as well as correlated with reproductive success.

So, the authors hypothesized that HAPS women would have hormone profiles that indicate higher fertility compared to LAPS women. This is a far more thoughtful way to develop hypotheses about behavior and ovarian hormones because it comes from a basic principle of understanding that behavior is dependent on context and resource availability. That is, behavior is influenced by underlying genetic variation, and also environmental variation.

Ultimately, what Ziomkiewicz et al are arguing is that the mechanisms that underlie the relationship between extraversion and reproductive success reflect a relationship between these more basic temperamental traits of how well one handles one’s environment and ovarian hormones. So they are developing a model for how behavior is correlated with ovarian hormones in a way that is functionally, biologically significant. I think this is really cool!

Within a given population, if you have higher estradiol concentrations than the norm, you likely are able to allocate more energy both to ovulating a high-quality egg and getting your endometrium to grow nice and thick. If you also have high progesterone concentrations, it means you have a nice, big corpus luteum (that’s the yellow body left behind by the ovulated egg, again indicating it was probably high quality). Further, high progesterone indicates you can not only decidualize your endometrium, which means give it all the nooks and crannies and growth factors that make it a nice place to implant, but you can maintain the endometrium and a potential trophoblast until the placenta can make its own progesterone. This increases the chance for implantation and pregnancy but may also reduce the risk of early pregnancy loss.

The reason I make the point that all of this matters in terms of intrapopulation variation, is that different lifestyles and subsistence behaviors help produce populational norms for hormone concentrations. What is high for a rural Polish woman, for instance, is not the same as high for an urban Polish woman, and is probably not the same as high for an urban American woman. Not only that, but since these behaviors are context and resource-dependent, then the contexts and resources available to a given population, and how much they vary based on socioeconomic status and many other factors, are also important.

Physiology is context-dependent

This paper is cool for behavior people, but it’s also cool for reproductive physiology nerds like me. I mean, look at these graphs!

Ziomkiewicz et al 2012 Figure 1 showing HAPS women have higher estradiol and an earlier progesterone curve than LAPS women.

Profiles of urinary estrogen (a) and progesterone (b) during the menstrual cycle in HAPS (black circle) and LAPS (white square) women.

These are very likely biologically significant differences, at least in the estradiol curves. The authors did find that the day dependency of progesterone was different in the two groups, and you can see that while the overall values aren’t that different, that the HAPS women have higher values in the early part of the luteal phase. I would like to see these values aligned by ovulation so we can better see them up against the implantation window, which is 6-12 days after ovulation. Aligning there rather than menses might give us some better insight into whether this difference is biologically meaningful. If progesterone is higher in HAPS women in the early part of the window, that would definitely be interesting. And of course, it would be interesting to one day also run these analyses alongside other physiological variables like follicle waves or endometrial thickness.

Meeting the criteria for a good behavior paper

As I mentioned before, the following issues make me cranky:

  • Use of a WEIRD population
  • Infrequent sampling through the menstrual cycle, with no effort to determine ovulation
  • No clear mechanism to link the hormone and behavior
  • Evolutionary storytelling that doesn’t link to fitness benefits for the behavior

Ziomkiewicz et al sampled daily throughout the menstrual cycle, provided a clear and coherent mechanistic link between hormones and behavior, and provided an interesting life history perspective on why temperament and ovarian hormones would be linked. The only criterion they don’t quite meet is in the use of an urban, Polish population, yet I still think this is different enough from the American and UK populations that are more often sampled that they pass here as well. From having done fieldwork in Poland over three summers, the activity patterns and diet composition is still pretty different in urban areas there versus in the US.

One final issue that Ziomkiewicz et al handle well is in discussing limitations and alternative hypotheses. They mention that energy availability can impact temperament, and so the relationship they found between temperament and ovarian hormones might only be the association between energy availability and ovarian hormones that has already been found countless times. However, they controlled for body fat, which partially addresses energy availability and the authors have previously shown influences estradiol.

Ziomkiewicz et al 2008 Figure 1 demonstrating that women with very high and very low body fat have lower estradiol concentrations

From Ziomkiewicz et al 2008. Women with very low and high body fat have lower estradiol in this sample.

Jasienska and Ellison 1998 Figure 3 showing the impact of energy expenditure on progesterone

From Jasienska and Ellison 1998, demonstrating that higher energy expenditure is correlated with lower progesterone concentrations in this sample of rural, Polish women.

At the same time, differences in energy availability using other measures like energy expenditure and intake tend to affect progesterone and luteal activity first, as you can see in the figure above, yet that was definitely the lesser finding in this study. So I really think the authors have something here.

Of course I am biased (remember to see*), but I do think Ziomkiewicz and colleagues have put together a thoughtful and important contribution to our understanding of the ways in which behavior and hormones influence each other.

*Full disclosure: I know several of the authors on this paper and two (Ziomkiewicz and Jasienska) are collaborators. However I knew nothing of this project or manuscript until I saw the final, published form.


Jasienska G, Ellison PT. 1998. Physical work causes suppression of ovarian function in women. Proceedings of the Royal Society of London Series B 265(1408):1847-1851.

Ziomkiewicz A, Ellison PT, Lipson SF, Thune I, Jasienska G. 2008. Body fat, energy balance and estradiol levels: a study based on hormonal profiles from complete menstrual cycles. Hum Reprod 23(11):2555-2563.

Ziomkiewicz A, Wichary S, Bochenek D, Pawlowski B, Jasienska G. 2012. Temperament and ovarian reproductive hormones in women: Evidence from a study during the entire menstrual cycle. Hormones and Behavior 61(4):535-540.