Biological clocks as age estimation markers in animals: a systematic review and meta‐analysis

Abstract

Various biological attributes associated with individual fitness in animals change predictably over the lifespan of an organism. Therefore, the study of animal ecology and the work of conservationists frequently relies upon the ability to assign animals to functionally relevant age classes to model population fitness. Several approaches have been applied to determining individual age and, while these methods have proved useful, they are not without limitations and often lack standardisation or are only applicable to specific species. For these reasons, scientists have explored the potential use of biological clocks towards creating a universal age‐determination method. Two biological clocks, tooth layer annulation and otolith layering have found universal appeal. Both methods are highly invasive and most appropriate for post‐mortem age‐at‐death estimation. More recently, attributes of cellular ageing previously explored in humans have been adapted to studying ageing in animals for the use of less‐invasive molecular methods for determining age. Here, we review two such methods, assessment of methylation and telomere length, describing (i) what they are, (ii) how they change with age, and providing (iii) a summary and meta‐analysis of studies that have explored their utility in animal age determination. We found that both attributes have been studied across multiple vertebrate classes, however, telomere studies were used before methylation studies and telomere length has been modelled in nearly twice as many studies. Telomere length studies included in the review often related changes to stress responses and illustrated that telomere length is sensitive to environmental and social stressors and, in the absence of repair mechanisms such as telomerase or alternative lengthening modes, lacks the ability to recover. Methylation studies, however, while also detecting sensitivity to stressors and toxins, illustrated the ability to recover from such stresses after a period of accelerated ageing, likely due to constitutive expression or reactivation of repair enzymes such as DNA methyl transferases. We also found that both studied attributes have parentally heritable features, but the mode of inheritance differs among taxa and may relate to heterogamy. Our meta‐analysis included more than 40 species in common for methylation and telomere length, although both analyses included at least 60 age‐estimation models. We found that methylation outperforms telomere length in terms of predictive power evidenced from effect sizes (more than double that observed for telomeres) and smaller prediction intervals. Both methods produced age correlation models using similar sample sizes and were able to classify individuals into young, middle, or old age classes with high accuracy. Our review and meta‐analysis illustrate that both methods are well suited to studying age in animals and do not suffer significantly from variation due to differences in the lifespan of the species, genome size, karyotype, or tissue type but rather that quantitative method, patterns of inheritance, and environmental factors should be the main considerations. Thus, provided that complex factors affecting the measured trait can be accounted for, both methylation and telomere length are promising targets to develop as biomarkers for age determination in animals.

Telomere length is not a useful tool for chronological age estimation in animals

“…[14] The individual variation in TL, TL change and maintenance is consequently very large, which is mainly because of differences in individual experiences (e.g., reproduction, activity level, growth [15,16] ) and exposure to various environmental factors and stressors (e.g., disease, food availability, abiotic conditions [16][17][18][19] ), but also due to biological differences (e.g., sex [20,21] ) and genetic [22][23][24][25] and epigenetic variation. [26,27] Accordingly, Le Clercq et al [1] found that most of the variation (69%) in TL within species could not be explained by age. The great individual heterogeneity in telomere dynamics is what makes the area interesting to a broad range of fields from ecophysiology, conservation biology, and evolutionary ecology to life-history theory, [28][29][30] but TL does not predictably track time as a clock.…”

“…[33,[45][46][47] In some species, [48] but not in others, [47] telomere shortening is faster in early life than later in life, which is not surprising given the effects of higher levels of cell proliferation in early life on TL. [49,50] Curiously, Le Clercq et al [1] suggested that telomeres do not shorten in somatic cells in juveniles, but there is no evidence for their general claim that "telomere length remains stable in most tissues throughout childhood from infancy to early adulthood." In humans, a rapid decline in TL is observed in infants, [51][52][53][54] which is similar to results from, for example, baboons (Papio hamadryas [55] ), Soay sheep (Ovis aries [56] ) and some bird species.…”

“…Telomeres are not currently among the targets being developed as biomarkers for chronological age determination in wild animals. Fortynine of the 69 studies included in the meta-analysis by Le Clercq et al [1] did not consider using telomere dynamics to estimate age (see Table S1 in the Supporting Information). Twelve studies concluded that telomeres should not be used for age estimation in their focal species, four studies suggested it may be used given future developments, and four studies suggested that TL could be used for age estimation thus sharing the conclusions of Le Clercq et al [1] Of the latter studies, Hatase et al [87] paradoxically found no correlation between TL and age in loggerhead sea turtle (Caretta caretta); Haussmann et al [69] found that TL could classify common terns (Sterna hirundo) into three age classes with 80% accuracy; Cherdsukjai et al [88] found that TL could assess whether dugongs (Dugong dugon) were below 20 years, after which there was no TL change; and Pauli et al [89] used TL together with information on body size, sex, population density, and age class (juvenile or adult) to assign martens (Martes spp.)…”

“…[31] Several studies have demonstrated that phylogenetic nonindependence cannot be ignored when analyzing telomere dynamics across species. [94][95][96][97] Since birds appear to be overrepresented in telomere studies, Le Clercq et al [1] applied an unconventional method of randomly excluding most of the bird studies from their primary meta-analysis and performing a separate analysis of birds "to avoid phylogenetic bias." Randomly excluding studies from a systematic meta-analysis is problematic, and there may still be phylogenetic signals in telomere traits within birds, [94,98] as well as within other classes included.…”

“…Furthermore, qPCR has lower and more variable repeatability compared to TRF, [96] which may lead to methodological effects across studies as seen in recent meta-analyses. [1,16,47] Both TRF, qPCR, and FISH may underestimate the abundance of the shortest telomeres, [112] which may be more relevant to the aging process than the mean TL. [108] These can be better resolved using, for example, the (low-throughput) approaches of single telomere length analysis (STELA), [113] the telomere shortest length assay (TeSLA) [112] or long-read sequencing.…”