Perturbations in growth trajectory due to early diet affect age‐related deterioration in performance

Summary Fluctuations in early developmental conditions can cause changes in growth trajectories that subsequently affect the adult phenotype. Here, we investigated whether compensatory growth has long‐term consequences for patterns of senescence. Using three‐spined sticklebacks (Gasterosteus aculeatus), we show that a brief period of dietary manipulation in early life affected skeletal growth rate not only during the manipulation itself, but also during a subsequent compensatory phase when fish caught up in size with controls. However, this growth acceleration influenced swimming endurance and its decline over the course of the breeding season, with a faster decline in fish that had undergone faster growth compensation. Similarly, accelerated growth led to a more pronounced reduction in the breeding period (as indicated by the duration of sexual ornamentation) over the following two breeding seasons, suggesting faster reproductive senescence. Parallel experiments showed a heightened effect of accelerated growth on these age‐related declines in performance if the fish were under greater time stress to complete their compensation prior to the breeding season. Compensatory growth led to a reduction in median life span of 12% compared to steadily growing controls. While life span was independent of the eventual adult size attained, it was negatively correlated with the age‐related decline in swimming endurance and sexual ornamentation. These results, complementary to those found when growth trajectories were altered by temperature rather than dietary manipulations, show that the costs of accelerated growth can last well beyond the time over which growth rates differ and are affected by the time available until an approaching life‐history event such as reproduction.

The duration for which males maintained red throat coloration was longer in the first breeding season than in the second, and longer in the Winter experiment than in the Spring (Table S2 and Fig. S3). Dietary treatment affected the duration of red throat coloration (i.e. R males were red for a shorter period of time). While there was no overall effect of photoperiod, there was an interaction between age and photoperiod (the decline in redness with age being much less pronounced in the delayed photoperiod, Table S2, Fig. S3). There was no effect of compensatory growth rate (F 1, 51.23 =0.055, P=0.816), but a male's length at the end of the period of dietary manipulation (= manipulated fish length) was positively related to the length of time he remained red and there was also an interaction between age and manipulated fish length (Table S2), the effect of fish size being more pronounced in the first breeding season.
There was a significant difference in the rate of nest building between the first and the second breeding season, and between the Winter and Spring experiment (Table S3). Males completed nests within 3.3±0.2 days of receiving nest material in their first breeding season (Winter: 3.4±0.3 days and Spring: 3.1±0.2 days) but took longer (4.0±0.3 days) in their second breeding season (Winter: 4.0±0.3 days and Spring: 3.9±0.6 days), and males from the Winter experiment took longer than those from the Spring experiment. While there were no overall effects of diet (F 1, 53.08 =0.71, P=0.405) or photoperiod (F 1, 57.71 =0.02, P=0.891) on the rate of nest building, there was an interaction between season and diet (Table S3): R males took longer than C males to complete nests in the Spring experiment whereas there was less of an effect of diet treatment (after controlling for growth rate -see below) in the Winter experiment (Fig. S4).
There was a negative effect of manipulated fish length on duration, plus a significant interaction between age and manipulated fish length (Table S3): the larger the male at the end of the period of dietary manipulation, the shorter the time he took to build a nest. Compensatory growth rate negatively affected the rate of nest building and there was a significant interaction between season and compensatory growth rate (Table S3): the faster the compensatory growth rate, the longer the time needed to build a nest, particularly in the Spring experiment.

REPRODUCTIVE INVESTMENT IN FEMALES
A total of 24 and 25 females (out of 29 and 35 that were alive at the time) spawned during the first breeding season in the Winter and Spring experiments respectively, but only 9 females in Growth perturbation and age-related deterioration S4 the Winter experiment and 7 in the Spring experiment spawned in the second season (out of 23 and 31 that were still alive at that time). Given the low numbers of females spawning in the second season, the analysis of reproductive investment is based primarily on the first breeding season, and analysis of individual egg mass and number of eggs per clutch is only based on the first clutch since the number of clutches varied between females (mean (±standard deviation) number of clutches per female in the first season = 1.26+0.65). The mean mass per egg from the 1st clutch of each female was significantly heavier in the Winter experiment (3.3±0.1 mg) than in the Spring (2.4±0.2 mg; Table S4). While there was no effect of compensatory growth (F 1,31.72 =2.54, P=0.121), the mass of an egg was related to a female's length at the time of spawning (Table S4), with larger fish producing heavier eggs. Dietary treatment also affected egg mass (with R females of a given size producing lighter eggs, Table S4) whereas there was no effect of photoperiod treatment (F 1,21.76 =0.53, P=0.475). Egg mass was affected by interactions between season and diet and between diet and length at time of spawning (Table S4): the effect of diet was strongest in the Spring experiment, and females from the R group showed less of an effect of fish size on egg size ( Fig. S5a and c).
The number of eggs in the first clutch was not significantly different between the Winter (63.6±2.8) and Spring experiments (52.2±5.5) whereas there was an effect of dietary treatment (Table S4), with R fish producing fewer eggs than C fish ( Fig. S5b and d). Females from the delayed photoperiod group spawned more eggs than those under an ambient photoperiod (Table   S4, Fig. S5b and d). As with egg size, there was no effect on clutch size of compensatory growth (F 1, 32.25 =0.55, P=0.465) but a positive effect of the female's length at time of spawning (Table   S4 and Fig. S5b and d). The interaction between season and photoperiod significantly affected the number of eggs, with delayed photoperiod fish in the Winter experiment spawning more eggs (Table S4).
The relative investment in the first breeding season (defined as (the total number of eggs a female produced in Period 3) / (combined number of eggs she produced in Periods 3 and 5)) was analysed to understand how growth trajectories influenced the investment by the female over time. There were significant differences between the Winter and Spring experiments in the proportion that the first season's eggs made up of the total egg production in the two years (Table S5), with females from the Spring experiment showing a greater relative investment in the first season (Fig. S6). While there was no effect of diet (Table S5)

S5
P=0.919), there was a significant interaction between season and diet (Table S5): R females in the Spring experiment invested relatively less in egg production in the second breeding season than did the corresponding females in the Winter experiment (Fig. S6). While there was no effect of length at time of first spawning (F 1, 39 =3.68, P=0.062), compensatory growth rate positively affected the proportion of eggs produced in the first year (Table S5).

MAXIMUM LIFESPAN
The maximum lifespan (defined as the age at which 90% of the population had died) was similar to the median lifespan (Table S6) Table S6).  Table S1. Description of experimental manipulations. Note that during Period 1 Restricted (R) fish were fed a restricted diet (2% of body mass) while Control (C) fish were fed ad libitum.

Supplemental references
After Period 1, all fish were fed ad libitum. Temperature was held at 10°C during Periods 1, 2 and 4, but was increased to 14°C during the breeding periods in 2008 and 2009 (Period 3 and 5).
These manipulations were conducted on separate fish in the Winter and Spring experiment.    experiments. Data plotted as means ± SE. See Table S2 for statistical analysis.  Table S3 for statistical analysis.