Cohort effects and proximal versus distal metrics for lifetime reproductive success in Richardson’s ground squirrels

Lifetime reproductive success (LRS) in the form of number of descendants is a commonly used measure of individual fitness, but the life stage at which descendants are counted varies among studies. Conceptual and logistical trade-offs exist along the gradient of proximal-to-distal LRS metrics. Although proximal metrics, such as number of offspring weaned, are logistically easier to collect than distal metrics, such as number of reproductively mature offspring or grandoffspring, they may be of little evolutionary significance if stochastic events heavily influence the realized number of descendants. We use a 25-year demographic data set based on 954 adult female Richardson's ground squirrels (Urocitellus richardsonii) from 22 annual cohorts to characterize and compare 6 metrics for LRS: lifetime production of litters, numbers of weanlings, weanling daughters, adult daughters, weanling matrilineal granddaughters, and adult matrilineal granddaughters. Most adult females weaned only 0 (21%), 1 (47%), or 2 (22%) lifetime litters. All LRS distributions were right skewed, with 53% and 77% of adult female Richardson's ground squirrels having no adult female matrilineal descendants after 1 and 2 generations, respectively. Survival of daughters and mothers covaried with calendar year, and LRS was strongly influenced by the calendar year in which females recruited into the breeding population as yearlings. Catastrophic flooding in 2005 killed nearly all descendants from the 2004 and 2005 cohorts. Daughter survival to adulthood explained more variance in lifetime production of adult daughters and granddaughters than number, size, or sex ratio of litters. Overall, to have a >= 50% chance of 1 adult granddaughter, a female needed to produce 2 litters, 7 weanling daughters, or 2 adult daughters. All distal (response) versus proximal (predictor) LRS metrics had significantly positive relationships, but variability increased with each distal step in the response variable. Our long-term study highlights sampling issues with LRS studies, variability within and among cohorts, and the role of stochastic events in uncoupling a mammal's reproductive effort from realized number of descendants.