Hot mothers, cool eggs: nest-site selection by egg-guarding spiders accommodates conflicting thermal optima


Correspondence author. E-mail:


1. In oviparous species providing maternal care, the choice of nest site is crucial for the survival of both the eggs and the mother. Most embryos only develop successfully within a narrow range of incubation conditions, which may differ from the mother’s own requirements.

2. How, then, do nest-attending mothers select sites that provide suitable conditions for embryonic development, without compromising their own viability?

3. We investigated nest-site selection in flat-rock spiders, Hemicloea major, a species that guards fixed egg sacs in a thermally challenging environment (under sun-exposed rocks). Females glue egg sacs beneath rocks during late spring and guard their eggs during summer, when temperatures beneath rocks often exceed 50 °C.

4. Our field surveys show that spiders laid eggs beneath rocks that were larger and thinner, and thus hotter, than were most available rocks. However, the egg sacs almost invariably were glued to the coolest sites on the substrate beneath a rock, rather than to the (hotter, by about 9 °C) underside of the rock.

5. By affixing their egg sacs to the coolest locations beneath the hottest rocks, females ensured that their developing offspring experienced moderate temperatures and avoided lethal extremes and, simultaneously, gave themselves access to much hotter areas (that enhance their feeding and growth rates) under the same rock. This strategy allows mobile adult spiders to actively select higher temperatures than can be tolerated by their embryos, while remaining close enough to their eggs for effective nest guarding.


Animals of many taxa provide direct care to eggs or offspring, thereby increasing survival rates of these vulnerable life stages (Clutton-Brock 1991; Huang & Wang 2009). However, parental care can be costly to the parent, for example, by increasing energy expenditure or delaying future reproduction (Clutton-Brock 1991). In order for the benefits of increased offspring survival to outweigh the associated costs, nest-guarding parents should be under strong selection to choose nesting microhabitats that satisfy the requirements of both the parent and offspring. Thermal conditions in the nest-site may present a challenge in this respect, because the thermal preferences of free-living organisms may be very different from the requirements of their eggs (Bursell 1974). Commonly, embryos can develop successfully only under a limited range of thermal conditions, often narrower than the thermal tolerance ranges of conspecific adults (Bursell 1974; Shine 1987; Li & Jackson 1996; Andrews, Qualls & Rose 1997; Deeming 2004; Hanna & Cobb 2006). The problem is relatively minor in oviparous taxa without parental care, because the female needs to visit the oviposition site only briefly before returning to her usual habitat (e.g. Doody et al. 2006; Pike, Webb & Andrews 2011a). Maternal costs are greater if the habitat used by females does not provide adequate thermal conditions for the eggs, necessitating migration to specific nesting areas that provide suitable conditions (Bonnet, Naulleau & Shine 1999; Angilletta, Sears & Pringle 2009). The costs are greater still in viviparous taxa and in egg-guarding oviparous species. The prolonged physical association between mothers and their offspring means that to facilitate embryonic development, females may have to accept thermal conditions different from those that they would usually select (e.g. Beuchat 1986; Beuchat & Ellner 1987; Webb, Christian & Shine 2006). This constraint holds true regardless of whether the eggs are guarded at a fixed site or carried around on the female’s body (e.g. wolf spiders; Lycosidae). How do reproducing females balance these conflicting thermal requirements, such that they minimize their own costs while optimizing incubation conditions for their offspring?

Flat-rock spiders (Hemicloea major; Gnaphosidae) shelter and nest beneath rocks on sandstone outcrops in New South Wales, Australia (Child 1977; Brunet 1996; Main 2000; Fig. 1a–c). These nocturnal, dorsoventrally flattened spiders select thin sun-exposed rocks for shelter, and the consequently high temperatures in these sites confer fitness benefits to the spiders by increasing developmental rates (Goldsbrough, Hochuli & Shine 2004), increasing strike speed, and decreasing prey handling time (both tested at 5 °C intervals from 20 to 40 °C; van den Berg 2009). These sit-and-wait predators rarely leave their selected rock and capture prey encountered beneath these rocks (Brunet 1996; Forster & Forster 1999; van den Berg 2009). Temperatures under rocks fluctuate seasonally, and many ectotherms abandon these microhabitats during summer, when temperatures can exceed 40 °C (Webb & Shine 1997, 1998; Webb, Pringle & Shine 2004; Pike, Webb & Shine 2010). Flat-rock spiders have a high thermal tolerance that increases seasonally (in laboratory experiments, the critical thermal maximum, CTmax, ranged from 48·3 °C in spring to 49·6 °C in summer; van den Berg 2009), enabling them to shelter beneath rocks year round. Because flat-rock spiders nest beneath these rocks during late spring, and the egg incubation period lasts through to summer (Goldsbrough, Hochuli & Shine 2004), developing embryos potentially are exposed to lethally high temperatures. Female flat-rock spiders glue opaque disc-shaped egg sacs to rock substrates and do not move them during incubation (Fig. 1a–c; Child 1977; Brunet 1996). The female guards the egg sac throughout incubation, and thus, the nest site must accommodate the thermal requirements of both the female and her offspring. Here, we investigate whether female spiders choose nest sites non-randomly and, if so, whether microscale thermal variation beneath individual rocks influences oviposition site choice by females.

Figure 1.

 (a) Female flat-rock spider Hemicloea major (body length, c. 2.5 cm) guarding an egg sac; the two egg sacs on the left have hatched (evident by exit holes left by emerging offspring), whereas the one on the far right contains eggs. (b) A typical loose surface rock used by spiders, (c) the underside of the same rock showing a spider egg sac (arrow) and thermal images taken at 1300 h, shown for (d) the underside of the rock and (e) the substrate with egg sac visible (arrow). In (d) and (e), rock temperatures range from 20 to 57 °C, with darker colours indicating cooler temperatures. The substrate photograph shows an obvious outline of where the rock sat, because the previously shaded substrate is much cooler than the underside of the rock.

Materials and methods

We studied microhabitat use and nest-site selection of Hemicloea on Monkey Gum plateau in south-eastern New South Wales, Australia (35°S, 150°E; 355 m a.s.l). The plateau and surrounding valley are dominated by closed canopy eucalypt forest, except for bare rock outcrops adjacent to elevated cliffs. Our study area comprised 75 rock outcrops (each c. 100 m2 in size) containing 971 individually numbered rocks (for details see Pike, Webb & Shine 2011b). Prior to our study, we recorded the length, width and maximum thickness (to nearest mm) of each loose surface rock, and removed all spider egg sacs. We also took hemispherical photographs above a representative subset of these rocks (n = 762), which were used to calculate canopy cover (% openness) and solar radiation transmitted through the canopy (mols m−2 per day, both calculated for the spider reproductive season using Gap Light Analyzer; Frazer, Canham & Lertzman 2000). These two attributes directly influence the thermal regimes beneath rocks (Pringle, Webb & Shine 2003).

We sampled rocks monthly from May 2007 to November 2009 (n = 31 months, spanning three autumn–winter–spring periods and two summers) by turning each rock and recording the presence of spiders and freshly laid egg sacs. We recorded whether egg sacs were affixed to the underside of the rock or to the substrate beneath the rock. We classified each rock as used: (i) by spiders only, (ii) by nesting spiders (i.e. a spider and an egg sac were present; note that because female spiders guard their egg sacs during incubation, we never observed egg sacs containing incubating eggs without a female spider present), or (iii) not used (hereafter ‘unused’). We compared rock attributes among rock types using manova with rock area, rock thickness, canopy openness, and transmitted solar radiation as the dependent variables. Fisher’s PLSD pairwise comparisons were used to test for differences between individual rock types for each attribute.

To evaluate broad-scale patterns of rock selection based on temperature, we quantified thermal regimes beneath rocks used by spiders only (n = 25), used by spiders as nest-sites (n = 30), or not used (n = 77) using miniature thermal dataloggers (Thermochron iButtons; accurate to ±0·5 °C). Dataloggers were sandwiched between the substrate and underside of rocks (and in contact with both of these surfaces) and recorded average temperatures at 1·5-h interval from October to November 2009 (n = 52 days), when eggs were incubating. We randomly chose one clear, sunny day for analysis and used a repeated-measures anova to compare thermal regimes among rock types using time as the repeated measure and temperature as the dependent variable. We then calculated the daily maximum and mean temperatures throughout this period and used separate repeated-measures anovas to compare daily maximum and mean temperatures among rock types using day as the repeated measure and temperature as the dependent variable.

We used a thermal imaging camera (ThermaCAM S65; FLIR Systems Inc., Boston, Massachusetts, USA) to quantify thermal variation beneath 17 individual spider nest rocks. On 10 December 2009 (when eggs were incubating), we took thermal images of the underside of rocks and the substrate beneath the rocks at six time intervals throughout the day, beginning at 0700 h and ending at 1500 h. This period encompassed the time when rocks were warming in the morning and began cooling in the afternoon (the experiment was terminated when it began raining, which decreased rock temperatures). This allowed us to generate thermal profiles for the underside of the rock, the substrate and the egg sac. We set the emissivity of the camera for rough sandstone surfaces (ε = 0·935). Thermal images were taken within 5 s of lifting each rock. We replaced all rocks in their exact locations to ensure the consistency of subsequent temperature readings. We downloaded images and analysed them using special-purpose software (ThermaCAM Research Pro 2.9; FLIR Systems Inc.). For each photo (640 × 480 pixels), we manually drew an outline around the bottom of the rock or the location where the rock sat and determined the minimum, maximum, mean and standard deviation of temperatures within that area (to 0·1 °C). We then added a spot metre over the egg sac and recorded this value as the temperature of the egg sac.

We tested whether each thermal variable (minimum, maximum, mean, and standard deviation) differed during the hottest part of the day (1300 h) on the underside of the rock vs. the substrate using separate one-way anovas. To understand the temperature overlap between these two microhabitats, we graphed the mean range of temperature variation (minimum and maximum) and overlap at each time period throughout the day. We tested for nest-site selection based on thermal cues by comparing the egg sac temperature to the thermal variables of the surface on which the egg sac was laid during the hottest part of the day (minimum, maximum, mean and standard deviation). Finally, we tested whether the relationship between mean and maximum temperatures differed during the hottest part of the day between the underside and substrate using ancova with maximum temperature as the covariate. All statistical analyses were performed using systat version 11.0 (SYSTAT Software Inc., Chicago, Illinois, USA).


Of 971 individual rocks sampled, 304 (39·4%) were used by Hemicloea, of which 62 (20·4%) were used as nest sites. Hemicloea spiders began nesting in late spring (October–November), and the eggs continued to incubate through mid-summer (January). Most spider egg sacs were glued to the substrate beneath the rock (98·5%), with only one egg sac located on the underside of the rock (χ2 = 63·06, d.f. = 1, < 0·0001).

Spiders used rocks non-randomly with respect to size and thickness, canopy cover and transmitted solar radiation (manova, Wilks’ Lambda = 0·92, F8,1512 = 8·44, < 0·0001). Spiders used rocks and nested beneath rocks that were larger and thinner than unused rocks, but were similar to each other (rock area: F2,759 = 4·47, = 0·01; rock thickness: F2,759 = 12·53, < 0·0001; all Fisher’s PLSD indicated that spider rocks and nest rocks were similar and significantly larger and thinner than unused rocks; Fig. 2). Canopy cover (F2,759 = 12·05, < 0·0001) and transmitted solar radiation differed significantly among rock types (F2,759 = 9·07, < 0·0001). Rocks used by spiders had more open canopies and received more transmitted solar radiation than did either spider nest rocks or unused rocks (mean canopy openness ± SE = 62·01 ± 1·11%, 58·99 ± 2·06% and 55·64 ± 0·70%, respectively; mean transmitted solar radiation = 25·33 ± 0·34, 24·93 ± 0·66 and 23·58 ± 0·23 mols m−2 per day respectively; Fisher’s PLSD indicated that spider rocks had significantly more open canopies and received more solar radiation than either nest rocks or unused rocks, which did not differ significantly).

Figure 2.

 Attributes of rocks used by Hemicloea spiders, as nest sites by Hemicloea spiders, or unused rocks, showing differences in (a) rock size (area) and (b) thickness. Shown are box plots; the lower bound of each box represents the first quartile, the middle is the median, the upper bound is the third quartile, and the error bars represent minimum and maximum values. In both cases, unused rocks were different from the other categories (Fisher’s PLSD, all < 0.05), while spider and nest rocks were similar to one another (all > 0.56).

The three rock types differed significantly in terms of the ways in which thermal regimes beneath rocks changed over time (rock type: F2,129 = 22·32, < 0·0001; time: F15,1935 = 670·0, < 0·0001; rock type × time interaction: F30,1935 = 26·05, < 0·0001; Fig. 3). Nest-site rocks were warmer than unused rocks, but were cooler than spider rocks during the hottest part of the day (Fig. 3). Daily maximum and mean temperatures beneath rocks differed significantly over time (day) by rock type (maximum, rock type: F2,129 = 32·68, < 0·0001; time: F51,6579 = 1989·0, < 0·0001; rock type × time interaction: F51,6579 = 15·42, < 0·0001; Mean, rock type: F2,129 = 22·96, < 0·0001; time: F51,6579 = 5100·0, < 0·0001; rock type × time interaction: F51,6579 = 12·79, < 0·0001; Fig. 4). The broad pattern was that rocks used by spiders, and nest rocks were significantly warmer than unused rocks, whereas thermal differences between spider and nest rocks were minor (<0·5 °C; see Fig. 4).

Figure 3.

 Daily thermal regimes beneath spider rocks, nest rocks and unused rocks. Shown are mean values (±SE) on 21 November 2009 (i.e. early summer). Dataloggers were in contact with both the substrate and underside of rocks and thus show averages of both surfaces.

Figure 4.

 Daily (a) maximum and (b) mean temperatures beneath spider rocks, nest rocks and unused rocks. Shown are daily values in October–November 2009 (i.e. early summer, n = 52 days). Dataloggers were in contact with both the substrate and underside of rocks and thus show averages of both surfaces. Error bars are not shown for simplicity.

The undersides of Hemicloea nest rocks were hotter than the substrate beneath the same rocks in terms of minimum (F1,32 = 25·8, < 0·0001), maximum (F1,32 = 12·49, = 0·001) and mean temperatures (F1,32 = 91·12, < 0·0001) during the warmest part of the day (Figs 1d,e and 5); however, the substrate had a higher standard deviation of temperature than did the underside of the covering rock (F1,32 = 12·02, = 0·002; Fig. 5). The undersides of rocks and the substrate beneath overlapped substantially in the temperatures available throughout the day, but each also provided a distinct range of temperatures unavailable in the other microhabitat (Fig. 6). Temperatures on the undersides of rocks were higher than the substrate for much of the day (Fig. 6). Despite the high overlap in temperatures between microhabitats, egg sacs were laid in areas where afternoon temperatures were always within the range of temperatures available only on the substrate (Fig. 6). Accordingly, egg sac temperatures were similar to mean substrate temperatures (location: F1,32 = 2·20, = 0·15; location × time interaction: F5,160 = 0·22, = 0·95; Fig. 6).

Figure 5.

 Comparison of temperatures on the underside or substrate beneath loose surface rocks used by Hemicloea major for nesting, shown during the warmest part of the day (1300 h on 9 December 2009) for the following thermal variables: minimum, maximum, mean and standard deviation. Shown are box plots; the lower bound of each box represents the first quartile, the middle is the median, the upper bound is the third quartile, and the error bars represent minimum and maximum values.

Figure 6.

 Available thermal regimes on the underside and substrate beneath rocks used by nesting spiders Hemicloea major. Shown are mean egg sac temperatures (±SE), along with temperatures available on the underside of rocks, temperature overlap between the underside of rocks and the substrate, and temperatures only available on the substrate on 9 December 2009. Regions of overlap were determined by graphing the mean minimum and maximum temperatures for the undersides of rocks and the substrate, and joining areas where these lines overlapped. Temperatures were measured using a thermal imaging camera.

The relationship between mean and maximum temperatures during the warmest part of the day differed significantly between the substrate and rock underside (ancova with maximum temperature as the covariate: F1,30 = 4·91, = 0·035; Fig. 6). Overall, both mean and maximum temperatures were substantially higher on the underside of the rock than on the substrate later (Fig. 6). Individual regressions on these data showed that the maximum temperature was dependent upon mean temperature for the rock underside (R2 = 0·71, F1,15 = 36·78, < 0·0001), but not for the substrate beneath (R2 = 0·17, F1,15 = 3·09, = 0·10; Fig. 6).


In species with maternal care, the decision of where to locate nests will influence the fitness of both mothers and offspring (Clutton-Brock 1991). In some spiders living in hot environments, females move egg cases within nest sites to avoid lethally high temperatures (e.g. Henschel, Ward & Lubin 1992). Because female flat-rock spiders do not move their egg sacs, they must position egg sacs carefully to prevent the eggs from overheating (cf. Li & Jackson 1996). Hence, we predicted that temperature would influence nest-site choice in this species. Intriguingly, females selected exposed hot rocks as nest sites (nest-site rocks were hotter than most available retreat-sites), but laid their egg sacs on microhabitats that were the mean available temperature (the substrate beneath the rock, rather than the underside of the surface rock; and within that selected habitat type, in relatively cool areas). Thermal imaging technology revealed that by 1300 h, the mean temperature of the underside of nest rocks was 41·4 °C, over 9 °C higher than the mean substrate temperature. Hence, by depositing egg cases on the substrate, females provided eggs with high temperatures that facilitate rapid embryonic development (Goldsbrough, Hochuli & Shine 2004) while at the same time avoided exposing embryos to high temperatures, which are potentially lethal to embryos (Li & Jackson 1996; Hanna & Cobb 2006). These results suggest that females use thermal cues not only to select suitable nest rocks, but also to identify suitable sites for egg deposition.

In spiders, embryonic developmental rates increase with increasing temperature up until a threshold at which the embryos begin dying (Li & Jackson 1996). One problem that females face when selecting oviposition sites is the tight link between mean and maximum temperatures at potential egg-laying sites. Sites with higher mean temperatures (such as the underside of a rock) will increase developmental rates, but also will increase the chances of the embryos experiencing lethal temperatures for part of the day. In contrast, embryos deposited on the substrates are buffered against high temperatures (Fig. 6) and low temperatures at night (Feder 1997). This almost-inevitable link between mean and maximum temperatures may preclude nesting females of many species from selecting sites that provide optimal mean temperatures for their offspring (for an example with montane lizards, see Shine, Elphick & Barrott 2003).

Nest-site selection in egg-guarding species requires the female to balance the thermal costs and benefits to both herself and her offspring (Beuchat 1986; Beuchat & Ellner 1987). In flat-rock spiders, foraging success increases at higher temperatures (tested at 5 °C intervals from 20 to 40 °C in the laboratory; van den Berg 2009), which may explain why spiders choose hot rocks for retreat sites. The substantial thermal heterogeneity beneath individual rocks allows females to behaviourally thermoregulate without abandoning their nest site (Dial 1978; Huey et al. 1989; Kearney & Predavec 2000). On cold days, we observed female spiders pressing their bodies against the underside of the rock, presumably to heat up (as do some lizards: Dial 1978; Kearney & Predavec 2000). Females that behaviourally regulate body temperatures within their retreat site could maximize prey capture success whilst guarding their egg sacs against potential predators (e.g. Mora 1990; Machado & Oliveira 2002), thereby minimizing the risk of egg predation. Behavioural thermoregulation (and high thermal tolerances, van den Berg 2009) also may allow female spiders to remain under hot rocks year round, thereby avoiding the potential costs of territory acquisition, especially for newly maturing adults.

The nest-site selection tactic adopted by female flat-rock spiders (using moderate sites under the hottest rocks) is not the only way that they could provide their offspring with suitable incubation temperatures. For example, females potentially could select the hottest sites under cooler-than-average (shaded) rocks: these could result in egg incubation conditions very similar to those we have documented in the field. By doing so, however, a female spider would substantially reduce her own opportunities for behavioural thermoregulation, because the only other sites under such a rock would be cooler, not warmer, than those suitable for embryogenesis. It seems plausible that only by laying her eggs in the coolest sites under the hottest rocks can a female spider simultaneously satisfy the thermal optima both of herself and her developing offspring.

Future research could usefully document thermal tolerances of developing spider embryos in more detail, as well as exploring other systems in which egg-guarding parents need to balance their own thermal requirements (and perhaps, oxygen or hydric needs) against those of their offspring. The wide diversity of taxa that exhibit parental care (Clutton-Brock 1991) provides exciting opportunities to look for consistencies and divergences in maternal nest-site selection criteria in multiple, independently evolved systems.


We thank Betsy Roznik for help with fieldwork. Funding was provided by the Australian Reptile Park, Australian Research Council, Forests New South Wales, the Hawkesbury-Nepean Catchment Authority, New South Wales Department of Environment, Climate Change and Water, Royal Zoological Society of New South Wales (through an Ethel Mary Read Grant to D.A.P.), and Zoos Victoria. D.A.P. was sponsored by an Endeavour International Postgraduate Research Scholarship and an International Postgraduate Award funded by the Australian Department of Education, Science and Training and the University of Sydney.