Purpose: We evaluated how constant incubation temperatures affect life-history traits pre-hatching and post-hatching of the six-tubercled Amazon River turtle, Podocnemis sextuberculata.

Methods: We incubated eggs from natural nests at ten semi-constant temperatures between 22.26±1.01°C and 37.37±0.38°C (2013) and at six temperatures between 25.75±0.22°C and 36.17±0.15°C (2016). In 2013, we raised hatchling for 90 days to evaluate effects of temperature on early hatchling growth. We evaluated maternal effects in 2016.

Results: P. sextuberculata displays temperature-dependent sex determination and produces males at colder and females at warmer temperatures (TSD Ia). The estimated pivotal temperature was 33.73 ± 0.15°C and the transitional range of temperatures (TRT) 1.16 ± 0.59°C. Semi-constant temperatures below 26°C and above 38°C were lethal. Intermediate temperatures (32.25°C and 31.5°C, respectively) were optimal for hatching success and produced larger hatchlings that grew slower early in life compared to colder or warmer conditions, which produced smaller hatchlings. Warmer incubation temperatures within the optimal range (28°C-37°C) accelerated embryonic development. In contrast, comparisons of 30, 60 and 90 days-old suggests that warmer incubation temperatures reduced growth and mass gain rates post-hatching, such that incubation temperature effects on body size at emergence disappeared by 3 months of age.

Conclusions: Six-tubercled Amazon River turtles showed the highest pivotal temperature reported for any turtle. The relatively narrow TRT may limit the evolutionary potential of this vulnerable turtle in the face of global warming. Future incubation experiments at a finer scale (33°C-36°C) are warranted to refine the sex-ratio reaction norm. Field studies that monitor natural nests are imperative to evaluate conservation measures and the effect of female-biased illegal hunting and climate change. By providing data about the thermal biology of an understudied lineage of non-model species, our study helps fill gaps in our understanding of the evolution of vertebrate sex determination and its potential adaptive value.

Two semi-constant-incubation-temperature experiments were conducted in 2013 and 2016. Freshly-laid eggs were collected at the Middle Solimoes River within the Mamiraua Sustainable Development Reserve, Central Amazon, Brazil and transported to the laboratory at the Mamiraua Sustainable Development Institute (Tefe, Amazonas State, Brazil). Egg incubation started <24hrs from oviposition. Incubators were made from 120-liter Styrofoam boxes divided into two compartments, each with a 60-watt lamp connected to a digital thermostat (AddTherm, model: A80). Eggs were buried in moist medium-size vermiculite purchased at a local garden store (in a proportion of 1g vermiculite:2g of water following Gallego-Garcia and Paez, 2016) inside plastic containers, along with the thermostat probe. Temperature was recorded hourly with a data logger with (±0.5°C accuracy). The thermostat probe and data logger were placed in the middle of the container. To maintain moisture levels, all boxes with vermiculite were incubated for a week, without eggs, at each temperature prior to the onset of the experiment to determine the rate of water loss per treatment by the change of weight of the box. Then, once the eggs were added and the experiment started, water was added at different intervals, between every 2 days (for warmer treatments with the highest rate water loss) to weekly (for colder treatments with lowest rate of water loss).

In 2013, we collected 100 eggs (10 per clutch), and placed one egg per nest at 10 different target temperatures: 22°C, 23°C, 28°C, 29°C, 30°C, 31°C, 32°C, 33°C, 36°C and 37.5°C. However, the low-tech design of the incubators resulted in thermal regimes that fluctuated with varied amplitude around these target values (Figure S1) (22.86±1.01°C, 23.22±0.65°C, 27.89±0.19°C, 29.14±0.20°C, 29.99±0.23°C, 31.11±0.21°C, 31.7±1.71°C, 33.18±0.40°C, 35.95±0.83°C and 37.37±0.38°C). In 2016, we collected 132 eggs from 11 different nests, and placed two eggs per nest at six different thermal regimes: 25.75±0.22°C, 27.94±0.65°C, 29.83±0.26°C, 31.95±0.29°C, 34.01±0.67°C, 36.17±0.15°C. Because eggs from 22.86°C and 23.22°C did not hatch and lacked signs of embryo development, these temperatures were excluded from all further analyses.

Incubation duration was measured from egg collection to full emergence of hatchlings from the eggshell (emergence). Unhatched eggs were opened at the end of the experiment, and those lacking traces of embryonic development excluded from further analyses, as they were assumed to be infertile or to have suffered embryonic mortality unrelated to temperature (this assumption was based on the fact that statistical tests did not detect any temperature effect on the number of undeveloped eggs: x² = 12, df = 9, p = 0.21) and excluded from further analyses. Straight carapace length (SCL), straight carapace width (SCW), plastron length (PL), plastron femoral width (PFW) and head width were measured on day one after emergence using a 150 mm caliper (Mitutoyo®), and hatchling mass was measured using a digital scale (Pesola®, MS500, precision = 0.1 g).

To investigate the influence of incubation temperature on post-hatching growth, in 2013 we kept the hatchlings for eight days in a plastic box containing moist vermiculite and then raised them individually until day 90. We reared them in open plastic boxes (262x177x147 mm) filled with 40 cm³ of tap water per millimeter of hatchling SCL (adjusted when hatchlings were measured at 1, 8, 15, 30, 45, 60 and 90 days of age), and provided with a rock as dry surface. Boxes were maintained indoors at 23 - 31°C under a ~12h daylight photoperiod and cleaned every two days. Hatchlings were fed commercial turtle food (Alcon® Reptolife: 34% protein, 10% fiber, 15% mineral, 1.6-3% calcium, 0.7% phosphor) ad libitum, starting at 8 days of age. At age 1, 8, 15, 30, 45, 60 and 90 days, we measured and weighed hatchlings.

Hatchlings were euthanized by intravenous (IV) injection of general anesthetics (association of xylazine 2 mg/kg, ketamine 50 mg/kg and acepromazine 0.25 mg/kg), followed by administration of potassium 19.1% chloride (2 mg/kg, IV) after loss of corneal and muscles reflexes, as recommended by Resolution nº1000/2012 of the Brazilian Federal Council of Veterinary Medicine (CFMV, 2012). Gonads were dissected to assess sex by histology, and carcasses were fixed in 10% formalin, stored in 70% ethanol and deposited in the Mamiraua Sustainable Development Institute Reptile Collection in Tefe.

Data files can be opened using any spreadsheet reader, that opens .xls file format. All analyses were performed using R v. 3.5.2