Discussion
The immune phenotype of marmots varies with their age. While the leukocyte concentration remains stable over the course of an individual life, the lymphocyte count decreases, and the neutrophil count increases. In Mammals, lymphocytes and neutrophils make up the majority (80%) of the leukocytes (Jain, 1993). Lymphocytes play a central role in acquired immunity, being involved in immunoglobulin and memory cell production and in the modulation of immune defence (Jain, 1993; Roitt et al., 2001; Vandervalk & Herman, 1987). Involved in the innate immune response, neutrophils are the primary phagocytic leukocytes, and circulating phagocytes proliferate in response to infections, inflammation and stress (Jain, 1993).
A decrease in lymphocytes, together with an increase in neutrophils (Cheynel et al., 2017; Kirk et al., 2010), and more broadly, a decrease in the acquired immune system combined with an increase (or upkeep) in the innate immune system with age, was observed in various vertebrate species (reviewed in Peters et al., 2019). A decrease in the acquired immune system with age is often interpreted as a consequence of the thymus involution (Hakim & Gress, 2007). The observed increase in neutrophil count does not necessarily mean a higher performance of the innate immune system with age. Indeed, the phagocytic ability of neutrophils could decrease with age (Gomez et al., 2008) and increasing their number could be an adaptive compensatory mechanism. However, this increase in neutrophil count could also be a compensation for a decrease in the efficacy of the acquired immunity. A remodeling of the immune system could indeed occur due to changes in the selective pressures when getting old. Given the lower probability to encounter new pathogens at old ages, downregulating the acquired immune system could be adaptive (Fulop et al., 2018). Immunosenescence should not be considered as a unidirectional deterioration, and would probably be better described by taking into account remodeling and reshaping of the immune functions with age (Fulop et al., 2018).
We observed less lymphocytes for marmot males than for females. Various hypotheses such as sex-differences in allocation strategy, intra-sexual competition (Metcalf & Graham, 2018; Sheldon & Verhulst, 1996) or inhibition of the immune system by some steroid hormones were often suggested to induce differences between males and females (Gubbels Bupp et al., 2018; Klein & Flanagan, 2016; Taneja, 2018). However, we did not observe sex-specific differences in the variation of the immune phenotype with age. So far, studies of sex-specific patterns of immunosenescence remain equivocal: some suggested sex differences (e.g. (Gubbels Bupp et al., 2018; Tidière et al., 2020; van Lieshout et al., 2020; Bichet et al., submitted ), while others did not (e.g. Brooks & Garratt, 2017; Cheynel et al., 2017; Kelly et al., 2018; Peters et al., 2019).  For instance, van Lieshout et al. (2020) found a decrease in the proportion of lymphocytes with age in male badgers (Meles meles ), but not in females. The authors argued that this result could be explained by the high testosterone levels observed in male badgers, due to their polygynandrous mating system (Buesching et al., 2009), contrary to monogamous species (Sugianto et al., 2019) such as the Alpine marmot (Allainé, 2000; A. Cohas et al., 2006).
In our study, individuals with fewer lymphocytes but more neutrophils were more likely to die. This result was further confirmed by a significant selective disappearance of individuals with this phenotype. Innate cellular response (involving neutrophils) is considered as costly in terms of energy and autoimmune damage (Lee, 2006). Individuals with neutrophil-oriented response may be unable to mount an appropriate immune response against challenges encountered at old ages (Froy et al., 2019), and/or may have an excessive cost to this response and die.
Our study also illustrates the importance of longitudinal analyses and the use of appropriate statistical tools to avoid misleading conclusions regarding immunosenescence (Peters et al., 2019; van de Pol & Wright, 2009). At the population level, our analyses revealed quadratic age effects on immune parameters, probably due to a combination of variations in the strength of selective disappearance with age and of within-individual variations (Figure 1). Our current knowledge derived from cross-sectional studies thus has to be taken with caution (Peters et al., 2019). So far, only three studies investigated longitudinal variations with age in the immune functions (Beirne et al., 2016; Andrea L. Graham et al., 2010; Schneeberger et al., 2014) which is clearly not enough to understand senescence in a complex system like immunity. For instance, in the Greater Sac-Winged Bat (Saccopteryx bilineata ), it was found that the number of leukocytes decreased with age, both within and between individuals, while the immunoglobulin G concentration was higher in older individuals, but did not vary within individuals, and the bacterial killing capacity of the plasma did not vary with age, at both levels (Schneeberger et al., 2014). These variations with age also impacted the short-term survival probability (Schneeberger et al., 2014). More longitudinal studies, like the present one, are highly necessary to properly understand the patterns and consequences of immunosenescence for wild individuals and populations.