Why age-specific mortality is relevant to welfare biology

Why age-specific mortality is relevant to welfare biology

30 Oct 2019


The documentary series “Our planet” opens with a flamingo chick whose legs have become caked with salt from the mud flats. The young bird can’t keep up with the rest of their flock and is left to die. At the same time, other chicks and healthy adults seem to be living reasonably contented lives, able to find food and overcome other challenges. It seems plausible that an adult flamingo has lived a life characterized more by pleasure than suffering. This chick, though – and some proportion of all flamingos who have been born – never got to experience their best years.

Relating age-specific mortality to welfare

To understand the balance of pleasure and suffering in nature, we need to understand what proportion of animals experience different life fates.1 Life fates describe common patterns of major events in an animal’s life. More technically, a life fate is defined as “a unit that operationally aggregates individuals from the same species based on the similarities of critical life events and hazards befalling them.” For example, in flamingos, one life fate could be dying young, another could be having offspring and dying of old age, and another could be succumbing to disease in early adulthood.

Lifespan is a blunt way of quantifying life fates, because two animals may have very different experiences and yet die at the same age. However, within the same species, the frequency of different lifespans may reflect common challenges associated with specific stages of life. Lifespan can be thought of as a container for the total quantity of positive or negative experiences an individual can have. However, it may be more than that. Because causes of death (e.g. disease, severe weather, predation, starvation) are also causes of suffering, lifespan might also indicate the quality of an individual’s life.

When animals have quick and painless deaths, the harm they suffer is mainly in the deprivation of potential good in the future. However, many animals die in ways that cause them extreme suffering. Moreover, potentially fatal factors can also cause suffering without killing the individual animal.

This is why age-specific mortality patterns can be a useful indicator of the proportion of suffering in a certain animal population. We currently lack data on how welfare varies with age in any wild animal population (cross-disciplinary work comprising contributions from ecology and animal welfare science will shed more light on this question). Nevertheless, for the reasons explained above, it is plausible that average welfare over a given period of time would be proportional to the probability of surviving that same period, relative to other periods of the same length during an animal’s life.

This hypothesis would also imply that the probability that welfare over a given period is net-negative (that is, it contains more suffering than happiness) is inversely proportional to the survival probability. This is because we can expect death to significantly affect an individual’s welfare by causing them to suffer while at the same time reducing their chances of experiencing positive states of wellbeing. This is especially relevant for estimating the welfare of species where most individuals die young.

Basing research on previous findings in animal demography

Ecologists have studied patterns of mortality in many different animal species. A key concept in this context is “life expectancy,” which denotes the average lifespan of an individual born into a given population, as opposed to their theoretical maximum lifespan. So far, research on the life expectancies of different animals has mostly been motivated by human and environmental interests, rather than out of concern for these animals’ quality of life. Nevertheless, the data obtained by such studies can provide valuable information for welfare biology. To be sure, there are individual variations in lifespan, which also magnifies the relevance of differences in quality of life with respect to age. However, learning more about the life expectancies of animals may allow us to make general estimations of what the welfare of such animals tend to be.

Age-specific survival data are often presented through “life tables” that summarize vital statistics of a population. Beginning with a cohort whose members start life together, the life table states, for each interval of age, information such as the number of deaths, the survivors remaining, the rate of mortality, and the expectation of further life. Most recent research involving life tables has used them as a means to create “matrix population models” which use age-specific survival and reproductive rates to predict demographic trends. This report summarizes a set of demographic models for 257 wild animal populations belonging to 126 species, drawn from the COMADRE database, to illustrate the range of life expectancies and age-specific mortality patterns that exist within and among taxonomic groups.

Taxonomic patterns


Patterns of age-specific mortality are too diverse to assign universal classifications to large taxonomic groups. For example, even some insect species have relatively high rates of juvenile survivorship. However, while age-specific patterns vary, animals of certain groups, such as the ray-finned fishes (actinopterygii), do have considerably shorter life expectancies and lower average annual survival rates than others, including birds (aves) and mammals (Figure 1).

Figure 1: The summed frequencies of each life expectancy across the four best-represented taxonomic classes in the dataset. The vast majority of fish populations represented here have life expectancies of less than one year. Reptile populations have the second-shortest life expectancies on average.

Importantly, life expectancy is only an average of the lifespans. Only in five populations out of the 257 considered here do >5% of individuals live past age 8, but this inexorable decline in survivorship with age is a product of diverse age-specific survival probabilities (Figure 2).

Figure 2: A scatterplot of age-specific survival rates through age 12 including all populations in the dataset. Points are jittered about both axes to increase visibility, especially for the growing cluster of points along the zero-survival axi

Figure 2 shows a bifurcation of annual survival probabilities with age across species. By age 12, populations are split between those with annual survival rates of less than 5%  and those with annual survival rates over 80% . The taxonomic makeup of these two groups is not even; mammals are disproportionately represented in the high-survival group (Figure 3a), for example, while insects and fishes are almost exclusively found in the low-survival group (Figure 3b). Bird and reptile populations are represented fairly evenly in both groups.

Figure 3: Relative representation of the four best-represented taxonomic classes in the dataset in the high and low survival rate clusters distinguished in Figure 2. Birds and reptiles are relatively evenly divided between the two groups, while the vast majority of mammal populations have a >80% annual survival rate by age 12 and ray-finned fish overwhelmingly have a <10% survival probability by that age.

Mortality rates in wild populations are normally determined by a variety of causes (e.g., accidents, diseases, natural predation, human predation, weather), some stochastic, which can make it difficult to detect any mortality process that increases with age. Another obstacle to comparison is that published life tables often begin at different biological stages. Some studies include data about egg and larval mortality, for example, while others omit it or bundle it with the reproductive rate.2 The availability of data across species is further biased by conservation concern and ease of study, both of which favor larger, longer-lived animals. Even though these factors limit the comparability of data between species, they are still illustrative of the diversity of life histories within taxonomic groups.

Specific examples

While massive databases like COMADRE are indispensable for identifying broad trends and examples of unusual age-specific mortality patterns, it is also important to check the specific assumptions and methodologies the demographic models are based on before drawing any conclusions about the welfare of specific taxa. In this section, I will discuss life tables for five particular species with reference to their source methodologies.

Freshwater turtle

Turtles provide a classic example of high early-life mortality paired with long adult life expectancies. It has been studied the age-specific mortality of Australian freshwater turtles, Emydura macquarii and Chelodina expansa (Figure 4).3 These species occupied the same area, but C. expansa was far less abundant and preferred nesting sites away from the water, while E. macquarii nested densely near the shore of the Murray River. Fox predation posed a major threat to eggs and hatchlings of both species, but E. macquarii were especially vulnerable, with an egg/hatchling survival rate of 0.5% compared to 2.2% in C. expansa. Approximately 77% of older juveniles (~1-12 years) of both species survived annually, implying that only around six individuals in ten thousand made it from birth to adulthood. However, adult annual survival rates of ~95% (~13.5 expected years of life) would await them.

Figure 4: Stage-specific annual rates of survival for the two freshwater turtle species, Emydura macquarii and Chelodina expansa. The first-year numbers include egg mortality.


There are comparable estimates of stage-specific survival for American crocodile (Crocodylus acutus) populations in the Everglades National Park of southern Florida, USA (Figure 5).4 The juvenile-to-adult survival rates are broadly similar to those of freshwater turtles. The first-year survival rate appears to be much higher for these crocodiles (although still low at 20%), but omits egg mortality, which is likely to be extremely high. Still, this higher rate may be more relevant to welfare estimates if crocodiles are not sentient during the egg stage.

Figure 5: Stage-specific annual survival rates for the American crocodile (Crocodylus acutus)


Amphibians’ larval and adult stages are much more distinctly specialized than most vertebrates, with the larvae (tadpoles) being adapted for fast growth in an often-transient habitat, into a mobile adult who can find a new place to spawn. They are exquisitely adapted to natural habitat fragmentation. As such, amphibians generally have extremely low rates of survival in their early life stages.

There have been estimated mean annual rates of age-specific survival for the western toad (Anaxyrus boreas), northern red-legged frog (Rana aurora) and common frog (Rana temporaria) by aggregating published data (Figure 6).5 All three species showed losses of around 98% during the first year of life, which includes the embryonic, larval and metamorph stages. In the western toad, most losses (92%) occurred during the metamorph stage, while in the two frog species (Rana), peak mortality occurred earlier, during the larval stage (~96% mortality). Embryo survival was relatively high, at 80-90%. Annual juvenile mortality rates were high, though numerically minor following the first-year losses, and the ratio of juvenile to adult mortality rates varied greatly between species.

Figure 6: Age-specific survival rates for three species of amphibian. The first year encompasses embryonic, larval and metamorph stages.

Density-dependent survival among amphibians

Age-specific mortality has been unusually well studied amongst amphibians because of the importance of age-specific density in regulating their population dynamics. Theory and most studies predict a net-negative effect of population density on survival to adulthood, which could result from competition between juveniles for resources.6 However, the peril of migrating to less populated habitat can sometimes outweigh the penalty of competition. For example, it has been found local population density to be associated with higher adult survival in the great crested newt (Triturus cristatus), in part because individuals in high-density populations were less likely to undertake a dangerous migration to find a mate.7

How population density affects survival during a given life stage can be both a cause and consequence of the dominant cause of mortality (e.g. density-dependent starvation vs density-independent predation).8 At the population level, stage-specific density-dependence can temper fluctuations in juvenile survival. For example, if a storm wipes out 80% of tadpoles, the subsequent reduction in competition for food will allow a greater share of the remaining tadpoles to survive to adulthood than otherwise would have, dampening the effect of the storm on overall recruitment to adulthood. This phenomenon of stage-specific “sensitivity” is of interest to conservationists and welfare biologists alike,9 but has unfortunately justified scientific neglect of some of the most vulnerable groups of animals when their individual survival has little effect on the abundance of their species.

Drawing conclusions relevant to welfare

For an individual animal to have had a “life worth living,” they must have experienced enough pleasure during their life to compensate for a potentially painful death.10 For animals who are able to live out most of their full lifespans, this seems plausible, but the vast majority of animals who are born survive to experience only a small fraction of their potential lives. There may simply not be enough time in their lives to experience enough happiness to outweigh the pain of death. As animals live to older ages, they are likely to experience stressors including disease, vulnerability to predators, and competition for food and mates. Even if they survive these, such factors would also be connected to states of suffering, as they have been shown to lead to chronic stress and poor physical condition.11

Variation among individuals in their circumstances of birth can set them up for different experiences in later life, including variable mating success, lifespan, and cause of death.12 This variation within cohorts of animals can be captured under the paradigm of “life fates,” mentioned above, which uses additional categories of life history data to more holistically group animals with presumably similar experiences.

The field of welfare biology is at a very early stage, and little dedicated work from the life sciences has been invested up until recently. While progress is still limited by the lack of empirical studies of wild animal welfare, knowledge of age-specific mortality patterns will be essential for contextualizing the anticipated data in order to understand what being a member of any species is like on average.


1 Alonso, W. J. & Schuck-Paim, C. (2017) “Life-fates: Meaningful categories to estimate animal suffering in the wild”, Animal Ethics, p. 3 [accessed on 4 September 2019].

2 See for instance Davis, A. J.; Hooten, M. B.; Phillips, M. L. & Doherty, P. F., Jr. (2014) “An integrated modeling approach to estimating Gunnison sage‐grouse population dynamics: Combining index and demographic data”, Ecology and Evolution, 4, pp. 4247-4257 [accessed on 22 September 2019].

3 Spencer, R. J. & Thompson, M. B. (2005) “Experimental analysis of the impact of foxes on freshwater turtle populations”, Conservation Biology, 19, pp. 845-854.

4 Richards, P. M. (2003) “Evaluating the relative effects of life history stages in the conservation of the American Crocodile (Crocodylus acutus) in Florida”, Florida Scientist, 66, pp. 273-286.

5 Biek, R.; Funk, W. C.; Maxell, B. A. & Mills, L. S. (2002) “What is missing in amphibian decline research: Insights from ecological sensitivity analysis”, Conservation Biology, 16, pp. 728-734.

6 See for instance Berven, K. A. (1990) “Factors affecting population fluctuations in larval and adult stages of the wood frog (Rana sylvatica)”, Ecology, 71, pp. 1599-1608; Vonesh, J. R. & De la Cruz, O. (2002) “Complex life cycles and density dependence: Assessing the contribution of egg mortality to amphibian declines”, Oecologia, 133, pp. 325-333. See also Lande, R.; Engen, S.; Sæther, B. E.; Filli, F.; Matthysen, E. & Weimerskirch, H. (2002) “Estimating density dependence from population time series using demographic theory and life-history data”, The American Naturalist, 159, pp. 321-337.

7 Cayuela, H.; Schmidt, B. R.; Weinbach, A.; Besnard, A. & Joly, P. (2019) “Multiple density‐dependent processes shape the dynamics of a spatially structured amphibian population”, Journal of Animal Ecology, 88, pp. 164-177 [accessed on 1 October 2019].

8 Leão, S. M.; Pianka, E. R. & Pelegrin, N. (2018) “Is there evidence for population regulation in amphibians and reptiles?”, Journal of Herpetology, 52, pp. 28-33.

9 Unger, S. D.; Sutton, T. M. & Williams, R. N. (2013) “Projected population persistence of eastern hellbenders (Cryptobranchus alleganiensis alleganiensis) using a stage-structured life-history model and population viability analysis”, Journal for Nature Conservation, 21, pp. 423-432.

10 See for instance Scherer, L.; Tomasik, B.; Rueda, O. & Pfister, S. (2018) “Framework for integrating animal welfare into life cycle sustainability assessment”, The International Journal of Life Cycle Assessment, 23, pp. 1476-1490 [accessed on 13 September 2019]. See also Farm Animal Welfare Council (2009) Farm animal welfare in Great Britain: Past, present and future, London: Farm Animal Welfare Council [accessed on 28 September 2019].

11 Clinchy, M.; Sheriff, M. J. & Zanette, L. Y. (2013) “Predator‐induced stress and the ecology of fear”, Functional Ecology, 27, pp. 56-65 [accessed on 4 September 2019]. Bateson, M.; Emmerson, M.; Ergün, G.; Monaghan, P. & Nettle, D. (2015) “Opposite effects of early-life competition and developmental telomere attrition on cognitive biases in juvenile European starlings”, PLOS ONE, 10 (7) [accessed on 14 October 2019].

12 Fay, R.; Barbraud, C.; Delord, K. & Weimerskirch, H. (2018) “From early life to senescence: Individual heterogeneity in a long‐lived seabird”, Ecological Monographs, 88, pp. 60-73 [accessed on 8 September 2019].