Deterioration in nutritional condition with aging could reduce reproductive success but coincides with declines in residual reproductive potential, thus invoking opposing expectations for late-life reproduction. Yet, the mechanisms regulating energy accrual and allocation to reproduction and survival throughout the lifetime of long-lived, iteroparous animals have remained elusive owing to variation in energetic costs across their extended reproductive cycle (from conception to juvenile independence). Using 10 years of repeated measures of both nutrition (i.e., body fat and food availability) and reproductive allocation across the reproductive cycle of 232 free-ranging, adult, female mule deer, we revealed that nutrition is a critical piece in understanding patterns of reproductive senescence and terminal investment. From conception to weaning, age-related patterns of reproduction were influenced by both body fat and environmental conditions. Reproductive senescence was clear across the entire reproductive cycle, although allocation to offspring was partly mediated by nutrition. Terminal investment, however, was most evident towards the end of the annual reproductive cycle and unveiled only when considering nutritional condition and food availability; during years with poor resource availability, older mothers raised larger juveniles (6-month old juveniles). Our work evokes nutrition as a lurking variable in end-of-life reproductive tactics for long-lived animals, while demonstrating the necessity of accounting for energy when considering patterns of reproductive senescence and terminal investment in wild animals.

We used data on age, nutrition, and reproductive investment of female mule deer from December 2013 to December 2022. We captured 70 adult, female mule deer using helicopter net gunning in December 2013. Each spring and autumn following that initial capture event, we recaptured all surviving individuals and captured new individuals to maintain a sample size of 70 animals. At the initial capture event for an animal, we extracted one insiform canine to assess age of individuals using cementum annuli and assigned a unique animal identifier (e.g., “001”). At every capture event, we measured body mass, determined nutritional condition, and fit all animals with a GPS radio-collar programmed to take satellite fixes every five hours from December 2013 to March 2015, every two hours from April 2015 to March 2018 and every one hour from April 2018 to December 2022. Nutritional condition of individuals was determined using developed protocols for mule deer which include measuring subcutaneous rump fat via ultrasonography and body palpation to estimate a body condition score. We calculated percent ingesta-free body fat (hereafter, body fat) of all animals. Each spring, we assessed pregnancy status and fetal number using ultrasonography, and fit all pregnant animals with Vaginal Implant Transmitters.

Beginning in 2015, we monitored all pregnant females each day in spring to identify birth events. Following a VIT expulsion alert, or using movement data to identify parturition events, we located the adult females and assessed if a birth event had occurred. For animals that had given birth, we searched the surrounding area for newborn deer. Upon location, we carefully restrained, collected data on sex, newborn mass (kg), and morphometrics and placed an expandable collar on the animal (ATS VHF 2015; ATS neolink 2016 – 2017; Vectronic Aerospace GPS 2018 – 2022). All newborn animals were assigned a unique animal identifier (e.g., “F001”). Most newborn deer (> 90%) were captured either the day they were born or the day following their birth. Following capture, we monitored survival of all deer captured as newborns throughout the summer. Upon identifying a mortality event, we investigated the mortality as soon as possible to determine the cause of death.

Beginning in 2019, we recaptured all surviving juveniles (i.e., offspring that were 6 months of age) the first December following their birth following the same general protocol of capture and handling adult animals. Following capture of juvenile deer in December, we measured juvenile mass (kg) and determined body fat of all juvenile animals. We did not extract incisiform canines of any juvenile deer because we had known information on birth date and age. Adult animals had the following metrics of reproductive effort: pregnancy status (binary, 0 or 1), fetal number (categorical, 1 - 3), spring body fat (%), mass of newborn (kg), mass of juvenile (kg), survival of offspring, and autumn body fat (%).

Spatial Data

Mule deer in this population migrate each spring and autumn between their seasonal ranges. We delineated the start and end date of both spring and autumn migrations of each individual using net squared displacement in Migration Mapper.  Mule deer in this population have very distinct summer and winter ranges, and net squared displacement is an effective tool at delineating migratory events. Next, we created seasonal home ranges for each individual deer using dynamic Brownian Bridge (dBBs) movement models with 95% contours for both winter and summer ranges. We created winter ranges for each individual using GPS data from 1 December to 28 February of each year. If animals were still completing their autumn migration after 1 December, we removed all GPS locations before their first day following the end of autumn migration. If animals began their spring migration before 28 February, we removed all GPS locations starting on the day they began their spring migration. We created summer home ranges from 1 June to 30 September. If animals had not completed their spring migration by 1 June, we removed all GPS locations until the day after they had completed their spring migration. If animals began their autumn migration before 30 September, we removed all GPS locations beginning on the day they began their autumn migration.

We used remotely sensed data of snow depth available through the Snow Data Assimilation System (SNODAS). We calculated the average snow depth in winter (from December 1 – Feb 28) and extracted the average snow depth during each winter from each home range of individual mule deer using the terra package in program R, version 4.2.2.  

We estimated per capita precipitation on each summer range, which influences fat accumulation of deer over summer. We used remotely sensed data of precipitation with daily estimates of precipitation available through DAYMET. We calculated cumulative precipitation by summing values of each pixel from 1 June to 30 September. We then calculated the average cumulative precipitation on summer range for each individual deer. We divided the cumulative precipitation value by population estimate.