Temperature‐induced multi‐species cohort effects in sympatric snakes

Abstract In reptiles, reproductive maturity is often determined by size rather than age. Consequently, growth early in life may influence population dynamics through effects on generation time and survival to reproduction. Because reproductive phenology and pre‐ and post‐natal growth are temperature dependent, environmental conditions may induce multi‐species cohort effects on body size in sympatric reptiles. I present evidence of this using 10 years of neonatal size data for three sympatric viviparous snakes, Dekay's Brown snakes (Storeria dekayi), Red‐bellied Snakes (S. occipitomaculata), and Common Garter snakes (Thamnophis sirtalis). End‐of‐season neonatal size varied in parallel across species such that snout–vent length was 36%–61% greater and mass was 65%–223% greater in years when gestating females could achieve higher April–May (vs. June–July or August–September) operative temperatures. Thus, temperature had a larger impact during follicular enlargement and ovulation than during gestation or post‐natal growth. Multi‐species cohort effects like these may affect population dynamics and the magnitude of these effects may increase with climate change.


| INTRODUC TI ON
For many reptiles, more rapid growth results in earlier maturity (Bronikowski & Arnold, 1999;Ford & Seigel, 1994;Frazer et al., 1993;Gibson & Hamilton, 1984). This means that growth early in life can influence population dynamics through effects on generation time and survival to reproduction (Cole, 1954;Gibbons et al., 1981;Oli & Dobson, 2003). Rapid neonatal growth can produce a "silverspoon effect" in which individuals that grow quickly early in life also experience higher growth rates later in life (Baron et al., 2010;Le Henanff et al., 2013;Madsen & Shine, 2000). Pre-natal events can also influence post-natal growth While et al., 2009). For example, in Meadow Vipers (Vipera ursinii ursinii), earlier parturition is associated with greater offspring mass and body condition and faster post-natal growth (Baron et al., 2010). Environmental temperature can influence both the timing of parturition (Blanchard & Blanchard, 1940;Cadby et al., 2010;Wapstra et al., 2010) and the rate of post-natal growth (Adolph & Porter, 1996;Peterson et al., 1993), potentially inducing cohort effects with long-term impacts on population dynamics (Beckerman et al., 2003;Lindstrom & Kokko, 2002;Wittmer et al., 2007). If environmental temperature has similar effects on multiple sympatric species, multispecies cohort effects may result.
Here, I provide evidence that pre-natal thermal conditions have parallel effects on end-of-season neonatal size in wild populations of three sympatric viviparous snakes, the Red-bellied Snake (Storeria occipitomaculata), Dekay's Brown snake (S. dekayi), and the Common Garter snake (Thamnophis sirtalis). All three are colubrid snakes in the subfamily Natricinae (Pyron et al., 2013), are widely distributed and locally abundant across eastern (and western, T. sirtalis) North American (Powell et al., 2016), and have similar reproductive phenology. Mating mostly occurs in spring, followed by follicular enlargement and ovulation (Noble, 1937). At my study site in Illinois, enlarged follicles are first detectable by palpation in April and May.
Gestation spans several months and parturition, as indicated by the appearance of neonates and post-partum females, commences in late July or early August. Neonates lack yolk reserves at birth (Mack et al., 2017) but begin feeding soon after parturition and grow rapidly until cold weather brings about the cessation of aboveground activity (late September-mid-October). Body size differs markedly among species, with S. occipitomaculata ranging from 67 to 284 mm snout-vent length (SVL) and 0.4-15.8 g, S. dekayi ranging from 76 to 378 mm SVL and 0.4-32.4 g, and T. sirtalis ranging from 115 to 780 mm SVL and 0.9-277.6 g at my study site. Diet also differs among species with S. occipitomaculata consuming almost exclusively slugs, S. dekayi consuming slugs, snails, and earthworms, and T. sirtalis consuming earthworms, amphibians, rodents, and birds (Virgin & King, 2019; personal observation).

| ME THODS
I conducted a capture-mark-recapture study of S. occipitomaculata, S. dekayi,and T. sirtalis at Potawatomi Woods Forest Preserve in northern DeKalb County,Illinois (42.4051 N,

W) between
April 2009 and October 2018. Fieldwork was focused in a wet sedge meadow and adjacent old field (approximately 5 ha; Figure 1). To facilitate snake detection, I placed 33-41 artificial cover objects (used rubber conveyor belt measuring ca. 60 × 90 × 1 cm) 15-20 m apart in an irregular grid. I checked artificial cover objects approximately weekly and captured snakes by hand. I classified snakes by species and sex and measured snout-vent length (SVL) using a cloth tape and mass using an electronic balance (Fitch, 1987). Snakes were individually marked by clipping ventral scales (using 3.5-× magnification) and released where captured, usually within 10 min.
I identified neonates (animals captured prior to their first hibernation) as a distinct age class by plotting SVL against day of year (DOY) separately for each year and species (an example is shown in  Operative temperatures were estimated separately for three periods corresponding to follicular enlargement and ovulation (April-May), gestation (June-July), and post-natal growth (August-September).
For April-May and June-July, I set animal mass to the mean mass of gravid females at my study site (S. occipitomaculata = 9.7 g, S. dekayi = 17.5 g, and T. sirtalis = 76.8 g); for August-September, I set animal mass to the mean mass of neonates at my study site (S. occipitomaculata = 1.0 g, S. dekayi = 1.5 g, and T. sirtalis = 3.5 g; see Appendix 1 for other model settings). For each species, year, and period, I computed the number of hours that body temperature exceeded 25°C, the temperature at which natricinae digestive rate, crawling speed, oxygen consumption, and tongue flick rate reach ca. 50% of their maxima and above which oxygen consumption increases rapidly from baseline (Stevenson et al., 1985). At Long Point, Ontario, a site similar in latitude and elevation to my study site, more than 90% of active T. sirtalis body temperatures exceeded 25°C ( Figure 1 in Gibson & Falls, 1979). The thermal biology of Storeria is less well known but body temperatures in excess of 25°C occurred throughout the active season at a site in northwestern PA (Gray, 2014) and locomotor performance increased from 10 to 20 to 30°C (Gerald & Claussen, 2007). I used analysis of covariance with species as a factor to test whether end-of-season SVL or mass covaried with hours >25°C in April-May, June-July, or August-September. I first tested for a significant factor-by-covariate interaction to determine if the slope of the relationship between end-of-season SVL or mass and hours >25°C differed among species. When no such interaction was detected (Results), main effects were tested in follow-up analyses of covariance with the factor-by-covariate interaction omitted.
I generated estimates of effect size (partial η 2 ; the proportion of variation in end-of-season SVL or mass explained by hours >25°C after removing variation attributable to species; Richardson, 2011) to assess the magnitude of each period's influence. For comparison, I computed mean April-May, June-July, and August-September air temperatures from daily temperature data downloaded from https:// prism.orego nstate.edu/. IBM SPSS Statistics Version 26 (Armonk, New York) was used for analysis with α = .05.

| DISCUSS ION
Sympatric S. dekayi, S. occipitomaculata, and T. sirtalis showed parallel patterns of variation in neonatal size across 10 years such that end-of-season SVL was 36%-61% greater and end-of-season mass was 65%-223% greater in years with maximal size relative to years with minimal size. Furthermore, end-of-season SVL and mass were associated with the amount of time that gravid females could achieve April-May body temperatures >25°C. This result suggests that the rate follicular enlargement and timing of ovulation had especially large impacts on neonate size. Variation in the amount of time that females could achieve June-July body temperatures >25°C (gestation and parturition) or the amount of time that neonates could maintain August-September temperatures >25°C (post-natal growth) had less impact. Of these three periods, the amount of time that snakes could achieve body temperatures >25°C was least for

April-May (averaging ca. 260 h vs. 660 h in June-July and 490 h in
August-September) and had the largest among-year coefficient of variation (23%-27% depending on species vs. 9%-11% in June-July and 8%-9% in August-September). Behavioral thermoregulation may allow gestating females to achieve their preferred body temperatures more easily in June-July when ambient temperatures are high than in April-May when ambient temperatures are lower (Huey et al., 1989;Peterson, 1987). Possibly, the small size of neonates limits their thermoregulatory ability during August-September (Bittner et al., 2002). Alternatively, the thermal dependence of physiological processes in neonates may differ from that of adults as suggested by the significant association of end-of-season SVL and mass with August-September hours >20°C but not >25°C.

Experimental manipulations of environmental temperatures
in semi-natural enclosures could provide more rigorous tests of Year Body temperature hours >25°C
In addition, the larger size attained by neonates in warm years may result in increased survival (Jayne & Bennett, 1990) independent of age at reproductive maturity. Consequently, the cohort effects described here may generate temporal variation in population abundance, density, and size structure much like patterns of geographic variation attributed to differences in activity season (Adolph & Porter, 1996). Given the degree of dietary overlap among snake species at my study site, especially between S. dekayi and S. occipitomaculata (Virgin & King, 2019), temporal variation in abundance and density may affect competitive interactions among snake species and have top-down and bottom-up effects on their prey and predators. Additional data on the degree to which cohort effects persist beyond the neonatal life stage and the extent to which reproductive maturity is size-versus age-dependent (Bronikowski & Arnold, 1999) would aid in evaluating their impact on population dynamics.
Cohort effects like those observed here are not unusual, having been documented in a wide range of plant and animal taxa (Lindstrom & Kokko, 2002 and citations therein). What is unusual, although not unexpected, is the occurrence of parallel cohort effects across multiple sympatric species. Because of their physiological dependence on environmental temperature (Huey, 1982), ectothermic vertebrates are likely candidates for exhibiting multi-species cohort effects but similar patterns are anticipated in other taxa as a consequence of different shared environmental drivers (e.g., water availability in plants, Streng et al., 1986; beech masting in rodents, Wittmer et al., 2007;fire in grassland birds, Powell, 2006).
Although analyses of cohort effects on single-species population dynamics have been fruitful (Beckerman et al., 2003;Le Galliard et al., 2010;Lindstrom & Kokko, 2002;Wittmer et al., 2007), multi-species cohort effects, with their potential impacts on competitive and predator-prey interactions, warrant further study (Huss et al., 2013). The more frequent occurrence of extreme weather events (IPCC, 2014) may result in even larger cohort effects than those observed here (Cadby et al., 2010;Lourdais et al., 2004). Equally interesting are situations where weather or other environmental drivers have contrasting effects on sympatric species due to differing ecological traits (e.g., Ma et al., 2018). For example, a hot year might have negative effects on diurnal or open-habitat species, but positive effects on nocturnal or shade-dwelling species, as has been suggested in the context of climate change (Huey et al., 2012;Paaijmans et al., 2013). The fact that cohort effects can arise from pre-natal or preovulatory environmental conditions has the potential to magnify the impact of climate change on demography and life history.