Artificial light at night alters life history in a nocturnal orb-web spider

Experimental spiders were obtained from eggsacs laid in the laboratory by 18 wild-caughtfemales, collected from sites ranging in light intensity from <0.1 to 40 lux (Skye Instruments Lux Meter, Llandrindod Wells, Wales, UK) in an urban park in Melbourne, Victoria (37.7911 S, 144.9515 E) in February 2016. Spiderlings from these 18 families were reared from hatching at 22 °C under a 12-h day (2000 lux; 12 V cool white LED strip lighting;) and a 12-h night that was either darkness (treatment;= 215; 0–0.06 lux) or dim light at night (treatment;= 235; 20–24.6 lux; 12 V cool white LED strip lighting). We selectedas it sits in the middle of the range of light intensities experienced by spiders in our collection site, and is comparable to other laboratory studies mimicking the effects of street lighting (;). Offspring from each family contributed equally to each of the two light treatments. At 14.95 ± 1.47 (mean ± SE) days after emergence from the egg sac, juvenile spiders were transferred to inverted, transparent plastic cups and maintained until death under standard laboratory conditions (). To maintain humidity, cups were lightly misted with water every two days. Young juveniles (hatching until seventh instar) were provided with three to fiveper week; older juveniles (seventh instar to penultimate instar) were fed three to five house flies () per week; and, adults were fed one juvenile cricket () equal to their body size twice a week. All spiders were provided with equal food appropriate to their size and life stage regardless of treatment group. The amount of food provided to juveniles was based on observations of wild. We did not provide ad libitum food during development, which would have been unrealistic and may have masked the effects of ALAN. Spiders were handfed where possible to minimise the effects of lights on prey capture rates. Prey were rarely left completely uneaten regardless of treatment group but adult spiders often discarded food suggesting that they were likely provided ad libitum food. E. biapicata 0 lux n 20 lux n 20 lux Drosophila melanogaster Musca domestica Acheta domesticus Eriophora Fig. S1 [Durrant et al., 2018] [McLay, Green & Jones, 2017] [Henneken et al., 2015]

Spiders were checked every two to three days and any moults or deaths were noted, to assess the effects of the presence of ALAN on development rates and survivorship. The date of death after maturation was taken as a measure of adult lifespan. The age (in days) at which a spider completed its final moult was defined as its age at maturation. Once spiders reached maturity, we measured their body mass (mg) and the length of the tibia (mm) on the front left leg, and used these to determine total life-time growth.

We assessed whether exposure to ALAN during development affected reproductive success by providing spiders with mating opportunities within their own light treatment groups and measuring the number of offspring produced. Mating pairs were age-matched to ensure senescence (days since maturation) differed as little as possible between treatment groups (mean ± SEM senescence at mating:males = 20.32 ± 2.36; females = 8.14 ± 0.81;males = 18.96 ± 2.56; females = 10.40 ± 2.17). Prior to each mating trial, we allowed the female to build a web in a rectangular Perspex frame (58 × 58 × 15 cm) after ‘sunset’ in the laboratory. Following web construction, we placed the male in the bottom corner of the frame furthest from the hub of the female’s web to allow mating. Following mating, both males and females were returned to their normal housing conditions. If a female failed to mate with a male, she was paired with consecutive different males until she mated. For each mated female, we recorded the time between mating and production of the first eggsac, the total number of eggsacs laid, the number of spiderlings that emerged, and the average mass of individual spiderlings in the first eggsac produced by each female; spiderling mass in subsequent eggsacs were not measured due to time constraints. To measure reproductive costs of ALAN due to physiological disruption, independent of reproductive costs due to shifts in body size, we calculated the number of eggs produced per milligram of female body mass for each mated female. 0 lux 20 lux

Statistical analyses were carried out using R version 3.4.2 (). We used linear mixed models to determine the effects of light treatment (fixed factor), sex (fixed factor) and family (random factor) on age at maturation, length of the intermoult period, and the number of recorded moults. We used a generalised linear mixed model (with an assumed logit link) to compare the proportion of juveniles in each treatment group surviving to maturity, with family as a random factor. We used linear mixed models to determine the effects of light treatment (fixed factor), sex (fixed factor) and family (random factor) on body mass, and tibia length in adults. In the case of a significant treatment × sex interaction, we used post hoc comparisons with a Bonferroni correction. Due to the strong correlation between treatment group and age at maturation, relationships between age at maturation and body mass, tibia length, or number of juvenile moults were analysed separately for each treatment group. The marginal () and conditional ()values are presented for these analyses. Survival curves (for entire lifespan and separately for adult longevity only) for each treatment group were built using a Kaplan–Meier survival analysis (lifespan or adult longevity as response, light treatment as fixed factor, family as random factor; sex was a fixed factor for the analysis of adult longevity). Curves were compared using a log-rank test (survival package in R;). Linear mixed models were used to test the effects of light treatment (fixed factor) and family (random factor) on the number of eggsacs produced, the number of spiderlings per eggsac, the time to the first eggsac, and the mass of spiderlings in the first eggsac. [R Core Team, 2017] [Therneau, 2015] R R R m c 2 2 2

The probability that a juvenile reached maturity was comparable for the two light treatments (72 of 215spiderlings; 64 of 235spiderlings; GLMM: χ= 2.54, d= 1,= 0.11). However,spiders matured significantly earlier () and required fewer moults to reach maturity () compared withspiders, for both males and females. The intermoult interval (days) was comparable forandspiders (). There was a positive relationship between time to maturation and the recorded number of moults (= 109.1,< 0.0001;). Overall, the total number of days survived (from emergence from the egg to death) byspiders was fewer thanspiders (log-rank test: χ= 10.90, d= 1,< 0.001;) but adult survival was comparable for the two light treatments (χ= 0.001, d= 1,= 0.97). Regardless of lighting treatment, males matured earlier than females () and required fewer moults to reach maturity (). The intermoult interval (days) was shorter for females compared to males () and adult females lived longer than males (χ= 41.85, d= 1,< 0.0001;). 0 lux 20 lux f P 20 lux 0 lux 20 lux 0 lux F P 20 lux 0 lux f P f P f P 2 2 2 2 Table 1A Table 1B Table 1C Fig. 1 Fig. 2 Table 1A Table 1B Table 1C Fig. 3 (1,131)

There was a significant interaction between light treatment and sex in their effect on body mass at maturation (). Post hoc analyses revealed that female body mass was significantly greater than male body mass in thegroup (estimate ± SE: 169.88 ± 19.05;= 8.92,< 0.0001), but not in thetreatment group (estimate ± SE: 36.32 ± 19.98;= 1.82,= 0.29). Additionally,spiders were significantly smaller thanspiders, with a bigger effect in females (males: estimate ± SE: 158.40 ± 18.14;= 8.73,< 0.0001; females: estimate ± SE: 291.97 ± 20.70;= 14.11,< 0.0001). There was also a significant treatment–sex interaction for tibia length (). Post hoc analyses demonstrated that male tibia length was greater than female tibia length forspiders (estimate ± SE: 0.66 ± 0.17;= 3.91,< 0.001) but the sexes were comparable inspiders (estimate ± SE: 0.18 ± 10.16;= 1.10,= 1.00). Tibia length was significantly greater incompared tospiders in both males (estimate ± SE: 1.51 ± 0.15;= 9.79,< 0.0001) and females (estimate ± SE: 2.00 ± 0.18;= 11.37,< 0.0001). There was a positive relationship between time to maturation and adult body mass (= 262.35,< 0.0001;) and adult tibial length (= 40.55,< 0.0001;). Table 1D Table 1E Fig. 4 Fig. 5 0 lux t P 20 lux t P 20 lux 0 lux t P t P 20 lux t P 0 lux t P 0 lux 20 lux t P t P F P F P 130.6 124.3 131.4 124.9 124.7 130.2 131.2 125.4 (1,131) (1,131)

There was no effect of light treatment on the likelihood of a given pair mating (= 1.34,= 0.25). The number of eggsacs laid per female was comparable across light treatments (), butfemales produced fewer spiderlings overall (). On average,females produced 63.96% fewer spiderlings per eggsac (main effect: χ= 218.35, d= 1,< 0.0001;). The number of spiderlings per eggsac declined with eggsac number (main effect: χ= 30.07, d= 1,< 0.0001), doing so faster infemales (interaction: χ= 8.42, d= 1,< 0.004;). The egg to mass ratio (the ratio of the number of eggs per eggsac to female body mass at maturity) was not significantly different between treatment groups (= 2.11 ± 0.12;= 1.84 ± 0.16 eggs per mg of female;= 1.33,= 0.19). The time between mating and laying the first eggsac was comparable between treatment groups (mean ± SEM days:= 28.54 ± 2.54;= 22.92 ± 2.15;= 2.83,= 0.1). There was no significant effect of light treatment on the mass of individual spiderlings in the first eggsac (). F P 20 lux 20 lux f P f P 0 lux f P 0 lux 20 lux t P 0 lux 20 lux F P (1,79) 45 (1,49) Table 2A Table 2B Table 2C Fig. 6 Table 2D 2 2 2

Seasonal shifts in temperature and daily photoperiod regulate patterns of growth and development in the majority of animals, through underlying effects on circadian and circannual rhythms (;;;). Spiders exposed to ALAN, which may emulate the longer photoperiod normally associated with spring and summer, progressed through fewer juvenile moults and matured earlier. This is consistent with similar observations of extended photoperiods reducing instar number and stimulating earlier maturation in the house spiderKoch 1841 (). Similarly, ALAN exposure shortened seasonal diapause and caused earlier maturation in other arthropod species (;), and overwintering as early instar juveniles has been documented for several spider species (). One possible explanation is that spiders exposed to ALAN may skip several winter moults as the nocturnal photic cues they experience mimic spring and summer photoperiods. In our study, we did not observe an overwintering period during which spiders did not moult for either treatment group. Variation in the number of moults required to reach maturity in spiders has been attributed to the photoperiod experienced (). Spiders such asmay exhibit this plasticity if the ultimate moult is related to when their sexual organs are sufficiently developed, rather than moult number. Hence, perception of an extended photoperiod (indicating the longer days typical of the reproductive season) would stimulate faster sexual development and earlier maturation, potentially driven by a fundamental shift in the biological clocks of the spiders. It is possible that moulting occurs during natural overwintering, but these moults are likely associated with only small increases in body mass. This shift in the rate of development may be related to a documented physiological effect of ALAN: its suppression of the nocturnal synthesis of melatonin (). Melatonin is a highly evolutionarily conserved molecule () that is linked to circadian and circannual rhythms and appears to have a regulatory function for moulting in arthropods (). However, the functions of melatonin in spiders remain largely untested. [Adkisson, 1966] [Gaston et al., 2013] [Fonken & Nelson, 2014] [Navara & Nelson, 2007] [Miyashita, 1987] [Shah et al., 2011] [Van Geffen et al., 2014] [Schaefer, 1987] [Miyashita, 1987] [Jones et al., 2015] [Vivien-Roels & Pévet, 1993] [Girish, Swetha & Reddy, 2015] Achaearanea tepidariorum Eriophora

Earlier maturation has potential consequences for body size, as a shorter development time affords less time to accumulate body mass. Here,spiders were smaller at maturity, largely due to their earlier maturation. However, when maturation time was statistically controlled spiders (particularly females) reared under light at night were also smaller thanspiders, suggesting that ALAN may directly affect growth rate in this species. This contrasts with a previous study in the cabbage moth which found no evidence that exposure to ALAN affected rates of body mass growth (), but such effects have received little attention across other invertebrate taxa. The length of the intermoult period in spiders is largely determined by rates of energy accumulation and increases in body mass (). ALAN may disrupt the body mass threshold for moulting and maturation to occur, resulting in lower increases in body mass and therefore a smaller body size at maturation. Further, artificial light potentially impaired their ability to consume food, or to convert food into body mass. Food consumption was kept as similar as possible between the two treatment groups, so a physiological disruption is the most likely explanation. The biological clocks that are regulated by melatonin in turn modulate feeding behaviours through effects on hormones that regulate hunger and satiety (;). These systems can also affect the physiological processing of food (;;). Hence, the impacts of ALAN on melatonin can cascade to fundamentally alter development and growth. To our knowledge, the effects of ALAN on these hormones have not been explicitly tested in arthropods, but analogues of some of these hormones have comparable physiological roles in insects (). 20 lux 0 lux [Van Geffen et al., 2014] [Nijhout, 2003] [Challet, 2015] [Fonken & Nelson, 2014] [Challet, 2015] [Fonken & Nelson, 2014] [Fonken et al., 2010] [Perić-Mataruga et al., 2009]

A major cost of exposure to ALAN is an increase in mortality (;). Spiders exposed to ALAN exhibited higher lifetime mortality rates but, as ALAN also stimulated earlier maturation, the number of spiders that reached maturity and adult longevity were both comparable across the two groups. This increase in mortality may have been driven by increased oxidative stress, as is suggested for other species (). Alternatively, it may have resulted from their accelerated development, as resources were diverted from growth and maintenance into reproductive development (). In contrast to previous studies (), there was no effect of ALAN on adult longevity. This can be explained by a trade-off between the deleterious effects of aging and the benefits of accruing more body reserves. Spiders in thetreatment aged more prior to maturing, so their adult lifespan would be shortened; however, they also had greater body reserves, prolonging their adult lifespan. The high mortality we observed in both treatment groups may have been a result of the unnatural diet our spiders experienced, particularly as their diet lacked moths. However, this unnatural diet affected both treatment groups equally, so it is unlikely that it would dramatically alter our conclusions. [Eisenbeis & Hassel, 2000] [Longcore & Rich, 2004] [Jones et al., 2015] [Boggs, 1992] [McLay, Green & Jones, 2017] 0 lux

The impact of ALAN on the viability of urban spider populations depends on its effects on the reproductive fitness of individual spiders in a population. Thespiders produced the same number of eggsacs but considerably fewer spiderlings compared tospiders. The reduction in fecundity forfemales was largely explained by ALAN induced differences in body mass, as similarly sized females produced comparable numbers of eggs regardless of light treatment. The lack of strong size-independent effects of ALAN on fecundity is contrary to previous results in(). Nonetheless, our results suggest that a smaller body size due to ALAN exposure will lead to a reduction in reproductive fitness. There was no difference between treatment groups in adult longevity, so both groups had an equal amount of time for egg production. However, under natural conditions, predation and declining winter temperatures are major sources of mortality, so earlier maturation may extend the breeding period (;). Hence, the ALAN-induced trade-off between maturing earlier or maturing larger also implies a trade-off between a longer breeding period or producing more spiderlings per eggsac for a shorter period. 20 lux 0 lux 20 lux Drosophila [McLay, Green & Jones, 2017] [Dominoni, Quetting & Partecke, 2013] [Schneider & Elgar, 2002]

In this study, we tested the effects of light treatment with both groups provided with the same amount of food, whereas spiders that build their webs near lights are likely to receive considerable foraging benefits (;). If greater food intake increases reproductive output (;), the foraging benefits associated with artificial lights have the potential to at least partially compensate for the reproductive costs of this trade-off (). This benefit may be tempered by observations of LED lights failing to attract moths—25 webs over three nights failed to catch moths, despite the observed presence of moths in the habitat (NJW, under review)—and decreasing moth capture rates in other orb-web spiders (). Hence, the potential for foraging benefits to mask developmental costs depends on how ALAN affects the types of prey available to insectivores foraging around lights. We found no effect of light treatment on the mass of individual spiderlings. This has important potential fitness implications, as offspring size is related to offspring performance, including starvation tolerance and the ability to capture prey (). However, intergenerational effects may not be evident without further measurements of performance and development in subsequent generations. [Adams, 2000] [Craig, 1987] [Kreiter & Wise, 2001] [Reed, Nicholas & Stratton, 2007] [Lowe, Wilder & Hochuli, 2014] [Yuen & Bonebrake, 2017] [Walker, Rypstra & Marshall, 2003]