Understanding drivers of vital rates is important for testing life history theory, predicting population dynamics, and guiding conservation. Migratory, long-lived animals are important to life history theory because they present an extreme among trade-offs in vital rates: delayed maturity, low fecundity, variable recruitment rates, and long generation times. Understanding how vital rates respond to environmental variation has been elusive for such species because of difficulties in studying wide-ranging animals over extended periods. Populations that live on elevation gradients can provide tractable study systems because considerable environmental change occurs over small geographic distances. Galapagos tortoises are an iconic example; they are long-lived, migrate seasonally, face multiple anthropogenic threats, and have cryptic early life history stages for which vital rates are unknown. From 2012–2021, we studied the reproductive ecology of two species of Galapagos tortoises along elevation gradients that coincided with substantial change in climate and vegetation productivity. Specifically, we 1) measured physical and reproductive condition of adult females, 2) tracked the movements of 33 adult females using GPS telemetry, and re-located them seasonally to measure condition, 3) recorded nest temperatures, clutch characteristics, and egg survival from 107 nests, and 4) used radio telemetry to re-locate 104 hatchlings to monitor growth, survival, and movements. We also monitored temperature, rainfall, and primary productivity from field and remotely sensed data along the elevation gradient. Adult females were either elevational migrants or year-round lowland residents. Migrants had higher body condition than residents, and body condition was positively correlated with fecundity. Nests occurred in the hottest, driest part of the tortoise’s range, between 6–165m elevation. Clutch size increased with elevation, egg survival declined, and hatchling survival and growth were highest at intermediate elevations. Hatchlings dispersed rapidly between 100–750 m from their nests before becoming sedentary in ranges <0.05 ha. Environmental variability mediated by elevation influenced adult fecundity and juvenile recruitment. Predicted future climates will profoundly impact the relationships between elevation and vital rates of Galapagos tortoises and other species living on elevation gradients. Resilience will be maximized by ensuring connectivity of foraging and reproductive areas within current and possible future elevational ranges of these species.

Study site

The Galapagos Islands straddle the equator in the eastern Pacific ca. 1000km west of Ecuador. This volcanic archipelago consists of some 129 islands, including 13 large islands (over 1km², Fig. 1), the oldest of which are ca. 4M and youngest < 0.5M (Poulakakis et al. 2012). The climate is characterized by a hot wet season from January to May, and a cool dry season for the rest of the year (Trueman and d'Ozouville 2010). However, during the “dry” season, persistent cloud cover results in humid upland conditions on the windward (southern) slopes of the larger islands (Colinvaux 1984). Vegetation patterns are driven by rainfall and substrate, which are largely determined by aspect, elevation, and lava flows.

Our study focused on Santa Cruz Island (Fig. 1), which rises to 860m elevation with a surface area of 986km² (Snell et al. 1996), one of nine islands Giant tortoises are believed to have currently or previously occupied (Caccone et al. 2002). The island hosts the largest human population on the Galapagos Islands, estimated at  >15000 in 2010 (León and Salazar 2012). Most of the moist highland zones have been converted to agriculture and at least 86% of these zones are now degraded by either agriculture or associated invasive plant species (Trueman et al. 2014, Laso et al. 2019).

Tortoises on Santa Cruz occur in two separate areas on the island’s western and eastern flanks (Fig. 1). Western Santa Cruz tortoises, Chelonoidis porteri, are widely distributed from ca. 0–450m elevation along the southwestern flank of Santa Cruz (labelled “El Chato” in Fig. 1), and comprise several thousand individuals (MacFarland et al. 1974, Benitez-Capistros et al. 2018, Tapia A et al. 2021). Eastern Santa Cruz tortoises, C. donfaustoi, occur on the east of the island (labelled “Cerro Fatal” in Fig. 2), between ca. 50–450m elevation (Fig. 2). Both species are of the “domed” morphotype, and they display strong size dimorphism, with females weighing up to ca. 120kg and males occasionally exceeding 260kg (unpub. data). Both species are seasonal partial migrants along the elevation gradients, and females of each species use permanent nesting areas situated at different elevations with different environmental conditions that offer potential to test our hypotheses. Some of our datasets come only from western Santa Cruz tortoises because we began our study in El Chato, before expanding to the eastern Santa Cruz tortoise population.

Environmental data collection

Rainfall and shade temperature across the elevation gradient of each species were quantified via a series of weather stations located at 50m altitude increments from 50–400m in El Chato and 100–400 in Cerro Fatal. At each location, a 100cm x 12cm plastic tube sealed at the bottom was attached vertically to a stake with an unobstructed view of the sky. A layer of engine oil 5cm thick was poured into each tube to prevent evaporation of accumulated rain. At the beginning of each month, the height of the water plus oil column was measured, after which most of the water was syphoned out of the tube. The height of the water column was then re-measured and left to accumulate rainfall during the month. Shade temperature was measured every four hours with iButton thermochrons (iButtons, DS1922L, Maxim Integrated, San Jose, CA) housed in 5mm thick polystyrene cups fitted upside down to the underside of each rainfall accumulators. IButton data were downloaded monthly. To estimate vegetation productivity along the elevation gradient and also in nesting zones, we used the Normalized Difference Vegetation Index (NDVI) obtained from the MODIS (Moderate Resolution Imaging Spectroradiometer) product of the Terra satellite which has been shown to reliably capture variation in vegetation growth, productivity, and resources available to herbivores (Huete et al. 2002, Pettorelli et al. 2005, Mueller and Fagan 2008). MODIS vegetation indices are provided at 250m × 250m resolution every 16 days, yielding 23 composites per year. We calculated pixel values from raw data following methods described by Bastille-Rousseau et al. (2017). Pixel values averaged by month from 0-400m along the elevation gradient within the range of tagged female tortoises were selected and projected onto a digital elevation model of Santa Cruz Island to assign elevation values to each pixel, and then binned at 25m intervals. NDVI within nesting zones and areas used by juveniles was obtained from values of pixels that touched minimum convex polygons formed from hatchling movement data collected in each zone (see below).

Female tortoise movements, physical and reproductive condition, and the timing of nesting

Between 2010–2021, custom-made GPS telemetry tags (e-obs GmbH, Munich, Germany) were fitted to a sample of adult female tortoises from both species (Blake et al. 2013, Blake et al. 2015, Bastille-Rousseau et al. 2016a, Bastille-Rousseau et al. 2016b). Tags were glued to the front of the carapace using non-toxic plumber’s epoxy (Fix-It Stick Epoxy Putty, Oatey, Cleveland, Ohio, USA). Tags collected fixes every hour and stored the data on a memory card that was periodically downloaded in the field using a wireless modem.

Four times per year (Nov-Dec, Mar, Jun-Jul, Sep) between 2013–2015, we attempted to find all tagged female tortoises to determine their body condition and reproductive status. During these periods we also searched for non-tagged females dispersed along the elevation gradient and recorded the same metrics. For each adult female encountered we recorded mass to the nearest 0.5kg and curved carapace length to the nearest 0.5cm with a tape measure. Body Condition Index (BCI) was calculated as 

BCI = m/L^2.98

where m= mass, and L=curved carapace length. The power coefficient was calculated based on the best fit for a large sample of length/mass measurements collected by Galapagos National Park Service (GNPS) rangers (Blake et al. 2015). We used a portable ultrasound machine (E.I. Medical Imaging Portable Ultrasound Ibex®) with a curvilinear 3.8 MHz transducer probe to scan adult female tortoises for the presence of eggs and developing follicles following methods of Robeck et al. (1990) and Casares et al. (1997). Complete counts of follicles and eggs were not possible, so we recorded presence/absence of follicles or eggs.

Monitoring of nesting zones and identification of nests

Studies of tortoise nesting were focused mostly in the three known nesting zones in El Chato, with supporting data from two zones in Cerro Fatal (Fig. 1). Nesting zones are distinguished by their open soil, evidence of existing and old nests and signs of intensive use by tortoises, and these concentration sites are well known to Galapagos National Park rangers. Mean elevation in the three El Chato nesting zones was 13m, 58m and 107m in the lower, middle, and upper zones respectively, and 90m and 165m for the lower and upper zones in Cerro Fatal. Nesting zones are separated by extensive areas of lava rock unsuitable for nesting.

Between 2013–2016 from July to October, we searched for freshly made nests in the three El Chato nesting zones (n=98), and in 2016 for the two Cerro Fatal zones (n=12). We identified freshly constructed nests by signs of recent digging and the presence of moisture on the nest cap (tortoises often urinate and defecate in and on the nest site). On finding a fresh nest, we dug carefully under the hardened nest cap and removed it. We then removed all eggs, maintaining their orientation and marking each egg with a unique number ID. We recorded breaks and cracks in each egg, weighed them to the nearest gram and, with calipers, measured the diameter to the nearest millimeter. Eggs were then carefully put back into the nest in the same position in which they had been found. For a sample of 41 nests, we estimated the center of the egg chamber where we placed iButton thermochrons (temperature loggers, DS1922L, Maxim Integrated Products, San Jose, CA.) as we replaced the eggs. iButtons recorded temperature data every four hours. As we replaced the eggs back in their original location, we placed soil around each egg to secure it in its original configuration and packed surface soil to replicate the cap as closely as possible.

It is the policy of the Galapagos National Park Service (GNPS) to protect tortoise nests in designated areas from feral pigs with wire mesh placed over nests and to open nests at the predicted end of incubation to extract hatchlings. This reduces the risk of entombment, but hatchlings are liberated before they would naturally emerge. This precluded us from reporting on natural incubation times and emergence dates, as well as mortality caused by entombment. Following GNP regulations, when incubation was estimated to be complete, we reopened the sample of nests. Hatchlings had either already exited the nest or had hatched but not emerged. When present, hatchings were measured to the nearest millimeter (linear length as opposed to CCL used for adults), weighed to the nearest gram, visually inspected for abnormalities, and then released at the nest. To quantify growth, survival, and movements, we selected a sample of hatchlings from randomly chosen nests to be fitted with VHF tags (RI-2B, Holohil Ltd. Carp, Ontario, Canada). Tags weighed 5g each; only tortoises that weighed >60g were fitted with transmitters. The same attachment method as for adults was used, except the tags were placed at the rear of the carapace only. A survival analysis revealed no effect of transmitter mass relative to initial body mass on hatchling mortality (Table S10d). Thereafter, we identified the location of tagged hatchlings approximately bi-weekly, recording location obtained from a handheld GPS unit (Garmin GPSMAP 64, Olathe, KS., USA.), weight, and length. Hatchlings found dead were usually in an advanced state of decay and cause of death could not be determined.  

All animal handling procedures followed the guidelines of the GNPS, the Max Planck Institute of Animal Behavior, and IACUC protocol #121202 of the State University of New York, College of Environmental Science and Forestry.