In Angiosperms, perennials typically present much higher levels of inbreeding depression than annuals. Two hypotheses have been proposed to explain this pattern.Because plants do not have a segregated germline, the first hypothesis states that more long-lived species may accumulate more somatic mutations as they grow, which they could then transmit to their offspring and thereby generate higher inbreeding depression. The second hypothesis, which does not contradict the first, stems from the observation that inbreeding depression is expressed across multiple life stages in Angiosperms. It posits that increased inbreeding depression in more long-lived species could be explained by differences in the way mutations affect fitness in these species, through the life stages at which they are expressed. In this study, we investigate the second hypothesis, setting aside somatic mutations accumulation. We combine a physiological growth model and multilocus population genetics approaches to describe a full genotype-to-phenotype-to-fitness map, where the phenotype relates to fitness through biological assumptions. We study the behaviour of mutations affecting growth or survival, and explore their consequences in terms of inbreeding depression and mutation load. Then, we discuss the role deleterious mutations maintained at mutation-selection balance may play in the coevolution between growth and survival strategies.Perennials, which make up the majority of Angiosperms (∼ 70%, Munoz et al., 2016), 1 typically present much higher levels of inbreeding depression than annuals. Indeed, meta-2 analyses found inbreeding depression to span from δ ≈ 0.2 on average in short-lived herba-3 ceous species to δ ≈ 0.5 in long-lived herbaceous species and shrubs, and δ ≈ 0.6 in 4 Under this model, individual size naturally saturates when the energy required to maintain 96 the existing body equals the available energy ( Fig. 2a-2b).
97Genetic assumptions. Mutations are assumed to occur at rate U (per haploid genome) 98 at a large number of loci, which recombine at rate 0 r 1 2 . In three separate models, 99 we consider mutations affecting three different traits. Mutations may affect growth by 100 increasing either their bearer's maintenance cost (c) or production cost (ε), or they may 101 affect its survival. When mutations affect survival, they are assumed to decrease both their 102 6 bearer's probability of being recruited as a juvenile (J) and its adult survival probability 103 (S). The effect of mutations is denoted s, with a dominance coefficient h 1 2 . Loci affect 104 traits multiplicatively, so that for any trait z (z ∈ {c, ε, S}), we havewhere H o (resp. H e ) is the number of homozygous (resp. heterozygous) mutations born 106 by the considered individual.
107Approximation of the expected number of mutations, inbreeding depression 108 and mutation load. We use two approaches to study our model. In the first approach, 109 we make the assumption that selective pressures acting on mutations at various life-stages 110 can be summarised...
