Undermining the Wall of Death

Different fields of science often don't talk to each other, even when coming at the same problem from different angles. A stark example of this can be found in the literature on aging. I'm in the field of evolutionary demography, and aging is one of our central focuses. We ask why and how it happens by studying the demographic patterns of different species under different circumstances. The evolutionary demographic theory of aging is built around the idea that there are alleles that have effects at different ages and natural selection acts on these genes to sculpt the age-specific mortality at different ages. Because dying young (before you've had a chance to reproduce) is more disadvantageous than dying old (after you've already passed on some genes) natural selection acts more strongly to minimize mortality in early adulthood than later adulthood, resulting in a chance of dying that increases in age. There are decades of theory built up around this idea, and the idea is not without merit, but it does assume these age-specific gene effects, generally without bothering to say what the actual genes are or how they influence mortality.

Another field within biology that focuses heavily on understanding aging is biogerontology. Biogerontology focuses on understanding the mechanistic basis of aging at the cellular and molecular level. They describe aging as a process of narrowing of the homeodynamic space, often due to accumulation of damage. Homeodynamic space is a concept related to homeostasis (the tendency of organisms to push their physiological state back to some optimum), but with the recognition that the goal that the individual is pushing towards, and its options for pushing, change over time. For example, as the cells in an organism accumulate mutations, it becomes more dangerous to allow them to continue replicating, because this could spawn a cancer. So the cells are forced to turn down expression of genes that allow for cell replication. But if your cells are replicating less, then you should be more reluctant to allow apoptosis, programmed cell death, because cells that die can't as easily be replaced. But if you've down-regulated the genes involved in apoptosis, this means infected cells will be less likely to kill themselves, so you need to have a stronger inflammation response, so that white blood cells will be brought to areas of infection and kill the infected cells from the outside. But increased inflammation has all sorts of nasty side effects, which themselves need to be compensated for. Note that I am just making this chain up as an example. The point being that the organism, in order to deal with the accumulation of damage, has to adjust various aspects of its physiology, which can cause damage or challenges to the system, which requires further adjustments. The organism gradually loses wiggle room, paints itself into a corner as it were. When this homeodynamic space gets too small, the organism can't respond to whatever insults (internal or external) come along and gets killed.

Reading papers in biogerontology, I am struck by two things. The first is how naive and outdated their evolutionary assumptions tend to be. For example, they still will state that aging is not observed in the wild because no individual lives long enough to grow old in the wild, an opinion that evolutionary biologists began to reject in the 1960s and have now disproved with data from numerous species from plankton to humans and birds to aphids. But I am also struck by how naive they would think our assumptions about age-specific genes are. They state as one of the basic principles of biogerontology that are no genes whose roll it is to cause aging, or which act at a particular age to regulate the chance of death. You will remember I said that such age-specific gene effects, from unspecified genes, are at the center of much of the theory behind evolutionary demography. Yet biogerontologists know such genes not to exist. So our assumptions about the mechanisms are as naive and simplistic as their assumptions regarding the demography.

This lack of communication, with each field basing its thinking on ideas the other has long since rejected, is common in science. There are simply too many journals, papers, conferences, etc., too many fields that may produce important information, for anyone to keep a useful fraction of an eye on most of them. So the lack of communication between fields is to some extent inevitable, but it does have significant consequences.

This is obvious when we introduce the gerontological observation that gene expression is not highly age specific (at least not late in life) to the evolutionary literature on post-reproductive lifespan (PRLS). Much of the study of PRLS has been motivated by the idea that PRLS shouldn't exist unless post-reproductive individuals do something useful for their younger kin. This idea arises from the evolutionary demographic theory of aging I described above. If an individual has reached the age where it can no longer reproduce, the genes it is expressing at that age should be genes that selection doesn't care about at all, because whether she dies at that age has no effect on how many offspring she has. So mutations that kill post-reproductive individuals should accumulate rapidly, unopposed by natural selection. W.D. Hamilton, a preeminent evolutionary theorist of the mid-20th century, wrote in 1966 that “In the absence of complications due to parental care or other altruistic contributions due to post-reproductives, the [mortality] curve should be roughly asymptotic to the age of the ending of reproduction.” By this he means that as the individual approaches the end of her reproductive period, her chance of dying at each instant should approach 100%. This has been dubbed "Hamilton's Wall of Death." Hamilton's work is influential enough, and his basic logic sound enough, that many of my colleagues still believe we should find the Wall of Death. But in fact we can find PRLS in a huge range of organisms where there is no parental care or anything comparable, and the Wall of Death is nowhere to be found. Hamilton's prediction fails because his model is built around high age-specificity of gene expression, which we now know not to exist. Genes which are being expressed at and after the age of reproductive cessation are the same genes being expressed prior to that age, doing the same things they did prior to that age (except of course reproduction) and so they can't just suddenly cause all sorts of lethal effects. This represents a major constraint on the ways selection can shape the pattern of mortality over age, and we evolutionary demographers are just starting to come to terms with the ramifications of this. When I have time to write another longish post, I'll explain how this leads to a major question in evolutionary demography that I have been thinking about but don't yet have any plausible answer to.

Progress!

The first draft of the abstract of a first chapter of my thesis! None of this will survive the editing process.

Human females have the unusual life-history trait of frequently surviving well past their reproductively fertile period. While a variety of adaptive hypotheses have been proposed to explain this trait, some authors argue that post-reproductive lifespan (PRL) is a phylogenetically widespread trait, requiring no special adaptive explanation for humans. Still others have argued that PRL is the result of cultural and physiological traits, not adaptive evolution. We suggest that the continued confusion on this front arises from two primary sources, the treatment of non-alternative hypotheses as mutually exclusive, and the use of PRL, an inconsistently calculated and theoretically ill-suited parameter. Given the drawbacks of PRL as a comparative measure, a variety of more useful and comparable measures of post-reproductive survival (PRS) can be calculated using data in the form of standard demographic life tables. Using life tables from 20 human populations, 78 non-human primate populations and two non-primate species, we present a set of measures of PRS which allow for direct comparability between populations and to evolutionary null hypotheses. We find strong support for the uniqueness of the scale of human PRS, for the widespread presence of PRS in primates and for the influence of culture in extending PRS.

Very rough section of a very rough draft.

Today between 11 and midnight I wrote a very rough section of a section of a rough draft of one of the several papers that will go into my thesis. Feels like progress. I don't expect to have much time for blogging over the next few months, but I will try to post bits like this that are potentially interesting, and that show what efforts keep me from having time for blogging.

The Measurement of Post-Reproductive Lifespan
Advancement in the study of PRLS has been hampered by differences over terminology, the use of a wide range of non-comparable measures and the failure to put measures of the scale of PRLS in the context of the time scales on which the organisms live.

Some authors have used the term "post-fertile" rather than "post-reproductive" arguing that anything that an organism does that increases her genetic representation in future generations is a form of reproduction, and that "post-reproductive" is therefore an inaccurate term to apply to post-fertile individuals who are still caring for their young (REFS). Indeed Hamilton (1966) argues that, "if the organism practises parental care 'birth' should be considered to occur... at the age at which the offspring becomes independent." While not disputing the biology behind this argument, we feel that the term "post-reproductive" is deeply enough ensconced in the literature on this topic that the use of alternative terminology to convey the same concept may tend to muddy communication. For this reason we use the term "post-reproductive" to refer to life after direct reproduction (fertility), excluding indirect reproduction (care of young and indirect fitness benefits).
Beyond semantic disagreements, so many methods have been used to calculate the scale of PRLS that efforts at comparisons across species and studies have been few and confusing. For example, XXXX and ZZZZ (REF) present a table of PRSL for 12 primate species, all given in units of years, but calculated in six different ways. Disagreements exist as to how to define the end of reproduction, how to determine the end of survival, and which individuals to include. The measures vary because the type of data used vary, and the interests of the authors vary, figuratively leading to comparisons of the shelf-life of apples to the refrigerator hardiness of oranges. The effect of sample size on these estimates is generally not addressed.
Even when these drawbacks are not found, authors generally fail to correct for the overall longevity of the species in question. One should expect a species that lives 100 years to, on the average, experience more years of PRLS than a species that lives 20 years. Without a denominator related to the time scale of the organism's life history, the numerator of PRLS is fairly uninformative.

In this study, we use a type of data that allow for broad comparability: age specific mortality and fertility figures as calculated in standard demographic methodology. Because the form of the data is highly standardized, the same measures can be calculated across taxa, for males and females, and in a wide range of environments. The use of data sources as information rich as are age specific mortality and fertility tables allows for the use of multiple measures which illuminate different aspects of PRLS, but which need not be falsely compared to each other, because we can calculate every measure for each population for which these data are fully available. Furthermore, the use of age-specific demographic tables allow us to put our measures of PRLS in the context of the reproductive and actuarial longevity of the organisms, allowing for meaningful comparisons between populations with very different lifespans.

Dying for Sex

One of the projects I am working on currently is an analysis, using data from primates, of what life history factors are correlated with sex-biased longevity. To put that plainly, I want to know why in some species the females live longer, in some the males live longer, and in some they live equally long. One of the dozens of hypotheses out there explaining why females live longer, in species where they do, is the 'risky male behavior' hypothesis' which says that males don't live as long because they take risks while out looking for sexual partners.

There has been limited support for this hypothesis, and most of the others, because so many hypotheses make the same predictions that one can rarely conclude that a particular factor is at play unless one ignores all the other possibilities (which seems to be the standard practice.)

This paper from Proc.Roy.Soc.B. takes an interesting new tack, looking not at whether males that are shorter lived than their mates are taking more risks, but rather at whether their short-livedness can be explained by increased mortality during the season of risk taking.

Here is the abstract:

Abstract

Male excess mortality is widespread among mammals and frequently interpreted as a cost of sexually selected traits that enhance male reproductive success. Sex differences in the propensity to engage in risky behaviours are often invoked to explain the sex gap in survival. Here, we aim to isolate and quantify the survival consequences of two potentially risky male behavioural strategies in a small sexually monomorphic primate, the grey mouse lemur Microcebus murinus: (i) most females hibernate during a large part of the austral winter, whereas most males remain active and (ii) during the brief annual mating season males roam widely in search of receptive females. Using a 10-year capture–mark–recapture dataset from a population of M. murinus in Kirindy Forest, western Madagascar, we statistically modelled sex-specific seasonal survival probabilities. Surprisingly, we did not find any evidence for direct survival benefits of hibernation—winter survival did not differ between males and females. By contrast, during the breeding season males survived less well than females (sex gap: 16%). Consistent with the ‘risky male behaviour’ hypothesis, the period for lowered male survival was restricted to the short mating season. Thus, sex differences in survival in a promiscuous mammal can be substantial even in the absence of sexual dimorphism.

Residual life

Most of us take for granted that we will live for a long time after we stop having kids (for those of us who will have kids). The standard picture for Americans these days is to stop at 40 or so, hope to live to twice that age. And this doesn't seem unrealistic. But when viewed evolutionarily, it seems a bit bizarre. Most organisms do not have a post-reproductive lifespan, human females very clearly do. I've previously written about the most common hypothesis to explain this fact, is the "Grandmother Hypothesis."

On Tuesday I was giving a talk on my research to Berkeley's Primate Research Group, including my experimental examination of the Grandmother Hypothesis. After the talk, I got a lot of good feedback, including an interesting question. How certain are we that females of other species of primates wouldn't live as long post-menopause as humans if they lived as cushy lives as we do?

The only answer I could give them is that I haven't seen any data suggesting otherwise. But then, on my way home, it occurred to me that I already have probably the world's best data set for answering exactly that question.

Primates in well run zoos tend to greatly outlive their wild cousins. Medical care, reliable food supplies, no predators and few pathogens. Not to say the life of a captive primate is perfect, or that there isn't significant variation in the quality of care, but for many species maximum longevity in captivity is much greater than in the wild. And it just so happens that I have life tables, including age specific reproductive rates and mortality rates, for 120 species of primates. These come from ISIS data, meaning data from relatively well run zoos, and I will need permission from ISIS to use them in this way, but I doubt they will have any major objections.

The idea of writing a paper based on data I already have is exciting to me. Usually I spend years between having an idea and having assembled the data to address it. I have almost all the data I need to address this question safely on several computers. I'll get the hang of this science thing yet.

Why no "grandfather effect"?

Much of my time at present is consumed by setting up an experimental test of what is known as the grandmother hypothesis. The grandmother hypothesis, in short, is the best guess we have as to why the females of humans and a few other species can live well past the age of reproductive cessation. In most species, and indeed in human males, there is no significant post-reproductive lifespan. Individuals are physiologically capable of reproducing for as long as they live. But the females of humans, a few other primates, a couple of cetations and maybe elephants go through menopause and then can live a significant portion of their lifespan after that. Our closest relatives, the chimps and bonobos, apparently go through menopause at the same age as our females, but live at most a few years after that.

Evolutionarily, this makes sense. If one is no longer increasing one's lifetime reproductive success, staying alive offers no obvious selective advantage. No point in investing in physiologies and structures that will last 100 years if one is only going to reproduce for 50 years. Better to put those resources into having more kids now.

But under a certain set of circumstances, reproduction does not end with, or shortly after, childbirth. If your young aren't really able to take care of themselves for a decade or two, you aren't done reproducing until they don't need you any more. In most hunter gatherer societies, the survival rate of five year olds whose mothers die is quite low. So for human women, having a kid in the last several years of life was likely a waste of time.

Worse, the kid who didn't make it took time and resources that could have been put into other kids, and childbirth, particularly late in life, is dangerous. Plus, elder human females are important for helping their daughters raise their own young, and learn how to do so. It has been shown that young mothers in several societies have a higher success rate raising kids if their mothers are around. The women who stopped having kids and focussed on the kids and grandkids they already had, and avoided the risk of late life childbirth, are thought to have ended up getting more of their genes into future generations than women who kept giving birth as long as they lived. If so, and if this variation in life history was heritable, which seems likely, this differential reproductive success would inevitably lead to a population with more and more women stopping early and fewer and fewer giving birth late in life. This is, we think, why we ended up with this "grandmother effect" of women living well past reproductive age.

The benefit of having a grandmother around seems to be restricted to maternal grandmothers. And this observation, that paternal grandmothers don't seem to make as much of a difference (at least in the societies studied) to the survival of their grandkids, points to at least two possible reasons why we don't see a "grandfather effect" to go along with this "grandmother effect."

First, in most societies, at least those studied in this context, males are providing less in the way of vital care. So if a women has a son who has kids, perhaps she is less involved in care, or in teaching how to care, because her son is not as involved as his mate in that care, and the daughter-in-law is not nearly as likely to look for advice and help from her husband's mother than her own mother. And perhaps this same logic applies to grandfathers on both sides. If they are not who the primary caregiver can go to for help and advice, the advantage of having them around to help is smaller.

Second, paternal grandmothers are less certain of which is really their genetic grandchild. If a woman gives birth to a daughter, and watches that daughter give birth to babies, she can be very confident that those are her descendants. If a woman gives birth to a son, and then watches that husband's mate give birth, there is a significant chance (and we have the genetic data to substantiate this) that the baby was fathered by some other man, and those babies aren't her genetic kin. So investing in them heavily may not be doing her any good. This argument is doubly true for grandfathers. The daughter who is giving birth may not even be his. A couple of generations removed, and who can be sure?

A final reason males may not have evolved to have a post reproductive period comes back to that risk in late life childbirth. Men don't give birth, so the risk to late life survival posed by late life reproduction may be greatly reduced, or completely absent. Without that trade-off, why not keep on breeding as long as possible?

A test of the idea that the voice recognition software works poorly for technical topics.

I will now train ViaVoice to understand my speech and know some of the technical words I am likely to use in my writing.

Senescence. I'm going to talk about senescence. Senescence has been defined many ways. In the biological literature definitions of senescence fall into 4 main categories. Physiological senescence, cellular and tissue senescence, actuarial senescence and reproductive senescence. These last 2 can be combined into the measure I will refer to as Hamiltonian senescence.
When most people think about senescence they think of physiological senescence. Hardening of the arteries graying of the hair gradually increasing dementia and a wide range of other traits are easily observable signs of physiological senescence. Physiological senescence can be defined as age related deleterious changes in the anatomy, biochemistry or physical functions an organism.
This is distinct from tissue senescence leaf senescence cellular senescence and other adaptive degradations of disposable subsets of an organism. In this meaning of senescence, the senescence of the whole organism is not being discussed, the breakdown, recycling or discarding of a particular piece of the organism. In a fitness context, senescence refers to the decrease in mean remaining individual fitness with age. This can be broken down into reproductive senescence, the decrease in mean reproductive output by age and actuarial senescence the decrease in mean survivorship by age. The definition of reproductive senescence can be further refined to include both direct reproduction (i.e. fertility) and indirect reduction ( i.e. aid and transfers of resources to related individuals).
as Hamilton argues, these means should be calculated across all individuals born into a particular population or cohort, rather than simply those individuals surviving to the age in question.

Much of the biological literature suffers from the conflation of these various definitions both with each other and with the observed variables used as proxies for senescence. Different authors studying the same organisms or even the same datasets have come to very different conclusion regarding the senescence those organisms because of confusion regarding the definition and measure being used.