1326 ARTICLE

Physical tidepool characteristics affect age- and size-class distributions and site fidelity in tidepool ( maculosus) S.J.S. Wuitchik, L.D. Harder, C.A. Meschkat, and S.M. Rogers

Abstract: Rapidly changing environments pose unique challenges to the resident organisms. Tidepools in coastal environments vary in biophysical characteristics spatially and temporally, and how they vary determines their short- and long-term suitability as habitats and therefore influence on the distributions of tidepool organisms. Biophysical effects on distribution could differ between age classes, depending on their intrinsic ontogenetic requirements and dominance relations. In this study, we inves- tigate the influence of physical pool characteristics on the site fidelity and population distribution of (Oligocottus maculosus Girard, 1856). We assessed short-term recapture of marked individuals and size-class distribution among four pool sets. The proportion of adults varied between pools primarily in association with water temperature and pool volume. Smaller adult and larger juvenile fish occupied warmer, small-volume pools, whereas larger adults occupied larger, cooler pools. Between 24% and 56% of marked fish were recaptured, with a higher probability of recapture in pools with “smooth” basins than in those with more rugose basins. Few fish moved among study pools, but the proportion of adults declined with repeated sampling, suggesting greater pool fidelity of juveniles. These results illustrate that intrinsic habitat features influence age- and size-class distributions in a resident tidepool sculpin species, with corresponding consequences for site fidelity.

Key words: abiotic influences, age structure, population distribution, site fidelity, size structure, Oligocottus maculosus, tidepool sculpin. Résumé : Les milieux faisant l’objet de changements rapides posent des défis particuliers pour les organismes qui y résident. Les caractéristiques biophysiques des bâches en milieux côtiers varient dans l’espace et le temps, et la nature de ces variations détermine leur adéquation à court et long terme comme habitats, influençant donc la répartition des organismes qui vivent dans ces bâches. Les effets biophysiques sur cette répartition peuvent varier selon la classe d’âge, en fonction de leurs exigences ontogéniques intrinsèques et de leurs relations de dominance. Nous examinons l’influence des caractéristiques physiques des bâches sur la fidélité au site et la répartition démographique de chabots de bâche (Oligocottus maculosus Girard, 1856). Nous avons évalué la recapture à court terme de

For personal use only. spécimens marqués et la répartition des classes d’âge dans quatre ensembles de bâches. La proportion d’adultes varie entre les bâches, principalement en association avec la température de l’eau et le volume de la bâche. De petits poissons adultes et de grands poissons juvéniles occupent des bâches plus chaudes et moins volumineuses, alors que de grands adultes occupent des bâches plus grandes et fraîches. De 24%à56%despoissons marqués ont été recapturés, la probabilité de recapture étant plus grande dans les étangs aux bâches « lisses » que dans ceux aux bassins plus rugueux. Peu de poissons se déplaçaient d’une bâche étudiée à une autre, mais la proportion d’adultes a baissé au fil des échantillonnages répétés, ce qui indiquerait une plus grande fidélité des juvéniles. Ces résultats illustrent le fait que des éléments intrinsèques de l’habitat influencent la répartition des classes d’âge et de taille chez une espèce de chabots qui réside dans des bâches, ainsi que les conséquences en découlant sur la fidélité au site. [Traduit par la Rédaction]

Mots-clés : influences abiotiques, structure par âge, répartition démographique, fidélité au site, structure par taille, Oligocottus maculosus, chabot de bâche.

Introduction (Horn et al. 1999; Richards 2011). In contrast, spatial variation in Many organisms have remarkable ability to contend with exten- physical and biological conditions may allow intertidal organisms to choose especially favourable sites (Gerking 1959; Green 1971;

Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18 sive variation in environmental conditions, even if conditions change rapidly (Palumbi 2001). In the rocky intertidal environ- Castellanos-Galindo et al. 2005). ment, organisms tend to respond to environmental heterogeneity Tidepools in rocky intertidal environments are subject to by zonation or relocation (Menge 1976; Zander et al. 1999). Tem- unique physical characteristics (Huggett and Griffiths 1986), in- poral variation in physical and chemical conditions between high cluding extensive variation in exposure to high-energy waves, and low tides requires that intertidal organisms either tolerate desiccation, extreme temperatures, and periodic restriction of the variation or move out of the intertidal zone during low tide habitat availability (Mahon and Mahon 1994). The mean and vari-

Received 17 October 2017. Accepted 27 April 2018. S.J.S. Wuitchik and S.M. Rogers. Department of Biological Science, University of Calgary, 2500 University Drive Northwest, Calgary, AB T2N 1N4, Canada; Bamfield Marine Sciences Centre, 100 Pachena Road, Bamfield, BC V0R 1B0 Canada. L.D. Harder. Department of Biological Science, University of Calgary, 2500 University Drive Northwest, Calgary, AB T2N 1N4, Canada. C.A. Meschkat. Bamfield Marine Sciences Centre, 100 Pachena Road, Bamfield, BC V0R 1B0 Canada; Department of Biological Sciences, University of Victoria, 3800 Finnerty Road, Victoria, BC V8P 4C2, Canada. Corresponding author: Sara J.S. Wuitchik (neé Smith) (email: [email protected]). Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from RightsLink.

Can. J. Zool. 96: 1326–1335 (2018) dx.doi.org/10.1139/cjz-2017-0297 Published at www.nrcresearchpress.com/cjz on 5 August 2018. Wuitchik et al. 1327

ation of environmental conditions in tidepools depend on their ing sizes, which are patchily distributed (Fig. 1). Four sets of three elevation, size, and depth, as well as climate conditions and bio- connected tidepools with one main pool and two smaller acces- logical communities (Metaxas and Scheibling 1993; Rankin and sory pools (i.e., 12 pools total) were selected as putative sites for Jensen 1993). The surface temperature, pH, and partial pressure of O. maculosus (Fig. 1). Two pool sets were located on the southwest of oxygen of tidepools all increase, on average, during diurnal low the main islet (SW1 and SW2; Table 1) and two were on the north- tides, but decrease during nocturnal low tides (Rankin and Jensen east side (NE1 and NE2; Table 1). Each set of pools was sampled on 1993). The extent of diurnal variation can vary seasonally; for three consecutive days: the southwest sets from 5 to 7 July and the example, temperature varies most during summer (Morris and northeast sets from 10 to 12 July. Daily sampling began at morning Taylor 1983), which tends to drive differential vertical distribution low tide and ended just before the subsequent high tide. All sam- of intertidal fishes (Zander et al. 1999). pled pools were fully exposed during low tide and fully immersed To cope with such conditions, tidepool residents must be able to during high tide. maintain homeostasis during rapid changes in salinity (Evans The community composition of fishes in rocky tidepools can be et al. 1999; Gibbons et al. 2016), temperature (Fangue et al. 2011), relatively consistent between seasons, suggesting community sta- and oxygen availability (Yoshiyama and Cech 1994), as well as bility and resilience over short temporal scales (Grossman 1986; avoiding accidental stranding and exposure at low tide (Gibson Castellanos-Galindo et al. 2005). To justify our assumption of com- 1999). Despite evolutionary and plastic responses to these rapid munity stability over time, we calculated the effective number of environmental changes, overall organism diversity and species species from a Shannon entropy index in the sampling area dur- richness in pools correlates negatively with tidepool elevation ing the year of collection and 2 years later. We used species collec- (Huggett and Griffiths 1986; Castellanos-Galindo et al. 2005). Ecto- tion data from the Bamfield Marine Sciences Centre biodiversity therms respond to variation in their thermal environment (Fry database (http://biodiv.bamfieldmsc.com/) for ray-finned fishes in 1947; Baird and Krueger 2003) and this response can differ be- a 0.4 km radius of Wizard Islets from the years 2013 and 2015. The tween life stages (Ward et al. 2010). This ontogenetic shift in hab- radius of 0.4 km covered the entire intertidal zone of both islets itat choice may lead to their selection of specific tidepools without including data from the larger channels on either side of affecting both age and size structures of local aggregations. the islets. The collection season (2013) and the following collection Tidepools in the rocky intertidal environment vary in their season (2015) had a similar sampling effort and a long enough physical characteristics (Huggett and Griffiths 1986) and the vari- temporal scale to assess if the community was stable over multi- ation that organisms are exposed to can result in restricted habi- ple years, We used these data to calculate a Shannon entropy tat use (Mahon and Mahon 1994). Motile organisms may choose to index, which was then converted to the effective number of spe- or be forced to leave the intertidal zone during tidal exchanges or cies (following Jost 2007). seek shelter within the intertidal until the next tide (Gibson 1999). The pool sets differed in their physical and biological character- Regardless of whether intertidal organisms leave the intertidal istics. Pools were located from 0.95 to 3.1 m above lower low water zone during tidal exchanges or seek shelter until the next tide, large tide (LLWLT) in the exposed intertidal zone of the islet, with they are forced into a periodicity that depends on tidal activity. varying volumes and temperatures but fairly consistent salinities The frequency, timing, and extent of movement within the inter- (Table 1). Most pools hosted primarily O. maculosus, with only one tidal zone are closely tied to physiological needs of the organism mosshead sculpin (Clinocottus globiceps (Girard, 1858)) and one ju- and the tidal forces experienced by each pool. For organisms that venile striped seaperch (Embiotoca lateralis Agassiz, 1854) observed remain in the intertidal zone during tidal exchange, tidepools act during one sampling day in two different pools. Tidepools also For personal use only. as microcosms (Gibson 1999), making them ideal habitats to as- contained diverse algal and assemblages. Due to re- sess the ecological and evolutionary responses to environmental quirements outlined by a Huu-ay-aht First Nations heritage inves- heterogeneity. tigation permit, we were unable to sample lethally to place The tidepool sculpin (Oligocottus maculosus Girard, 1856) (Cotti- specimens from Wizard Islets in a museum collection. dae) is a common occupant of tidepools along rocky shores and nearshore environments of the eastern Pacific Ocean. Oligocottus Tidepool features maculosus is an ideal species to test the influences of physical char- We measured rugosity, elevation above LLWLT, and volume acteristics on habitat choice, because it lives in environmentally once for each pool in all pool sets, whereas temperature and sa- heterogeneous tidepools and exhibits site fidelity (Green 1971). linity were measured daily. All measurements were taken at the Homing varies in association with intrinsic factors (e.g., fish age) low tide, which varied day to day but was between 0530 and 0630. and extrinsic factors (e.g., wave exposure within pools) (Craik Elevation was measured with a stadia rod and laser level (Crain 1981; Yoshiyama et al. 1992). Homing generates site fidelity, or the Enterprises, Inc., Mound City, Illinois, USA; Johnson Level and tendency of fishes to remain in their home pools over consecutive Tool, Mequon, Wisconsin, USA). Temperature and salinity data tidal cycles, rather than moving to other equally suitable sites were measured with a salinity, conductivity, and temperature probe (YSI Inc., Yellow Springs, Ohio, USA). Statistical analyses Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18 (Gerking 1959). The occurrence of site fidelity suggests that indi- viduals choose optimal tidepools; however, the role of physical consider mean temperature and salinity for the seven sampling pool features in that choice remains poorly understood in O. maculosus. days. Relative rugosity (terrain complexity) was estimated as the Therefore, we studied the local distribution of O. maculosus to ratio of surface area to planar area, both measured with transect (i) characterize physical features for occupied tidepools, (ii) char- tape (following Raffaelli and Hawkins 1999). The rugosity factor acterize the effects of these features on variation in sculpin size was calculated as the ratio of the distance covered by a flexible and age compositions among tidepools, and (iii) test whether site tape that followed the bathymetry of the pool end to end to the fidelity allows sculpin to take advantage of specific tidepool fea- distance covered by a tape stretched between the end points tures. (Gibson 1999). This ratio allows relative comparison between pools. Materials and methods Fish capture and tagging Site description Each sampling day we caught O. maculosus from each pool by This study was conducted during July 2013 on Wizard Islets dipnet at low tide (between 0.4 and 0.6 m above LLWLT) until no (48°51=29.86== N, 125°09=35.04== W) near the Deer Group Islands off fishes remained in a pool. Pools were heavily sampled until mul- the coast of Bamfield, British Columbia, Canada (Fig. 1). The rug- tiple researchers were confident that replicate sweeps with no ged topography of the Wizard Islets traps many tidepools of vary- capture or sight indicated that all the fishes were captured in

Published by NRC Research Press 1328 Can. J. Zool. Vol. 96, 2018

Fig. 1. Map of Vancouver Island with Victoria (48°25=43== N, 123°21=56== W), British Columbia, Canada, for reference and inset aerial photograph of Wizard Islets (48°51=29.86== N, 125°09=35.04== W) with locations of northeast and southwest sets of tidepools indicated. Photo credit: K.R. Holmes (Hakai Institute). For personal use only. Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18

Published by NRC Research Press Wuitchik et al. 1329

Table 1. Summary of the characteristics of the main pool of each tidepool set, including physical attributes, visible implant elastomer (VIE) tag colour, and the proportion of the observed tidepool sculpin (Oligocottus maculosus) aggregation that was considered itinerant juvenile (standard length <30 mm). Tag colour Temperature Salinity Elevation Volume Juvenile Tidepool set Sampling date Latitude Longitude and position (°C; mean ± SE) (ppt; mean ± SE) (m) (m3) Rugosity (%) SW1 5–7 July 2013 48°51.478=N 125°09.625=W Blue, right 20.3±2.81 25.1±3.05 3.02 0.07 1.15 48.2 SW2 5–7 July 2013 48°51.488=N 125°09.621=W Blue, left 19.0±1.21 28.1±0.23 2.93 1.06 1.68 33.6 NE1 10–12 July 2013 48°51.539=N 125°09.509=W Pink, right 17.9±0.40 29.0±0.36 1.65 4.37 1.29 0.0 NE2 10–12 July 2013 48°51.540=N 125°09.504=W Pink, left 18.1±0.88 29.9±0.56 1.16 1.00 1.15 7.0

these small tidepools. Pools were then observed for movement in randomly selected fish from the 78 sampled from this set from case any fishes remained. When the researchers were satisfied adult to juvenile, resulting in a “jittered” age composition for pool with their estimate of complete capture, sampled fishes were NE1C of p = 0.987. place in shaded, pool-specific holding containers with seaweed Second, we considered the joint effects of pool temperature (T) cover and individuals were anesthetized with 20 mg/L of clove oil and volume (V) on adult composition for the 12 pools during first in seawater (following Javahery et al. 2012). Fishes were identified samples using nonlinear regression. Nonlinear analysis was nec- while anesthetized prior to tagging; the following features were essary because these environmental variables had largely asymp- identified for each fish to be included as O. maculosus: single forked totic effects. Specifically, we fit the following regression model: opercular spine, absence of cirrus on nasal spine, definitive band- ing across eye, presence of cirri along lateral line, presence of Ϫ Ϫ Ϫ (1) L ϭ L Ϫ (L Ϫ L )e a(T 16) ϩ e bV sparse cirri on top of head (not as numerous as fluffy sculpin ∞ ∞ 16 (Oligocottus snyderi Greeley, 1898) but more than Puget Sound

sculpin (Ruscarius meanyi Jordan and Starks, 1895)), and a clear where L is the logit of adult frequency, L∞ is the asymptotic mean separation between first and second dorsal fins with a U-shaped logit at high temperatures, L16 is the expected logit for 16 °C (be- profile of tissue between terminal ends of the first dorsal fin rays. low the minimum observed temperature), and a and b are coeffi- In total, 668 unique O. maculosus were sampled with 420 adult cients governing the effects of variation in pool temperature and fish anesthetized, tagged, and returned to the tidepools. Fish were volume. Given an estimated logit, Lˆ, the estimated proportion of removed from the clove oil bath when they did not respond to ˆ ˆ adults in a pool was eL/͑1 ϩ eL͒. The fit of eq. 1 was compared with mechanical stimulation or handling, but still ventilated their gills those of the GLM described about and with variants of eq. 1 that unaided. All captured fish were measured for standard length. Following Craik (1978), we considered fish <30 mm long to be excluded the effects of pool temperate or volume using Akaike’s ⌬ (itinerant) juveniles and larger fish to be (resident) adults. All information criterion (AIC) or the AIC difference ( AIC) from the adult fish were tagged using a visible implant elastomer (VIE) tag best-fitting model. (Northwest Marine Technology, Shaw Island, Washington, USA) Size distribution on the ventral surface. Each set of pools was assigned a distinct We illustrate the distributions of standard length with both

For personal use only. colour tag and body placement (Table 1). Fish were placed in con- histograms and empirical probability density distributions. The tainers until they regained normal swimming behaviour (approx- imately 20–30 min) and were released into the pool from which latter were estimated with kernel methods (Sheather 2004) using they were caught. All protocols were conducted in accordance normally distributed kernels and Sheather and Jones’ plug-in ap- with the Canadian Council on Care. proach to identify the appropriate bandwidth, as implemented in Repeated sampling and tagging generated different classes of the UNIVARIATE procedure of SAS version 14.1 (SAS Institute Inc. 2015). ® fish. On the first sampling day, all M1 captured fish were marked and released into their pool. On the second day, the captured fish We analyzed effects on (undisturbed) variation in standard length among pools using fish caught during first sampling days included R1 recaptures of the M1 fish and M2 unmarked fish, which were marked in the same manner (including the same pool- at pools. Two sets of analyses were conducted with GLM that accommodated lognormal distributions. The first analysis consid- specific tag colour) as the M1 fish. On the third day, we recaptured R2 marked fish, which could have included some of the R1 fish that ered the joint effects of fish age class and pool identity and in- had been captured on days 1 and 2; some of the M1–R1 fish that cluded age class, pool set, and pool nested with set as fixed factors. were captured on the first day, but not the second day; and some This analysis initially also considered the age × set and age ×

of the M2 fish that were first caught on the second day. pool(set) interactions, but the latter was excluded, as it did not Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18 explain significant variation. A second pair of analyses assessed Statistical methods the effects of pool physical characteristics (ln-transformed) on Age distribution mean fish size, with separate analyses for juveniles and adults. We analyzed the age composition (juveniles versus adults) of Backward elimination (P < 0.05) was used to exclude nonsignifi- tidepools in two ways. First, we assessed overall variation in the cant independent variables. Analyses for both age classes identi- proportion of adults among sampled fish among pools using gen- fied multiple significant independent variables. To illustrate these eralized linear models (GLM) with the binomial distribution and effects, we present pool means that have been adjusted using the logit link function. All of these analyses considered pool set and fitted model with the values of all variables other than the vari- pools nested within sets as fixed factors. One analysis assessed able of interest replaced by their means. natural age composition using only data from first samples at pools, which were unaffected by possible effects of prior sampling Mark–recapture estimates of pool fidelity disturbance. To examine disturbance effects, an additional analy- The fidelity of fish to specific pools and its relation to pool charac- sis included sampling day (1, 2, 3) as a fixed factor. Because only teristics can be assessed based on the numbers of unmarked and adults occupied pool NE1C, the adult proportion of p = 1 hampered marked fish caught during second and third census days. Specifi- estimation of the associated logit, ln[p/(1 – p)], and standard errors. cally, physical pool characteristics could affect the probability that a We circumvented this problem by changing the age of a single fish was recaptured in the same pool, r. The recapture fate of an

Published by NRC Research Press 1330 Can. J. Zool. Vol. 96, 2018

Fig. 2. Relative frequency distribution (histograms) and probability Fig. 3. Relations of the least-squares mean (±SE) proportion of adult density distribution (curves) for standard length of tidepool sculpin tidepool sculpin (Oligocottus maculosus) in individual tidepools to (Oligocottus maculosus) in four sets of sampled tidepools. Fish with (A) water temperature, (B) pool volume, and (C) sampling day. The standard length <30 mm were considered itinerant juveniles and results in panels A and B are based on first sampling days only, with were not tagged. Fish larger than this threshold were expected to grey and white symbols distinguishing southwest and northeast exhibit site fidelity and were tagged. pools, respectively. Mean and SE values are back-transformed from logit results, hence the asymmetrical SE values. The fitted curves were estimated by nonlinear regression and are plotted for (A) mean water temperature or (B) mean pool volume. For personal use only. Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18

Published by NRC Research Press Wuitchik et al. 1331

Fig. 4. Relations of mean (±SE) standard length of juvenile (squares) and adult (circles) tidepool sculpin (Oligocottus maculosus) caught during first sampling days to (A) tidepool location, (B) pool elevation, (C) pool volume, (D) pool temperature (adults only), (E) pool salinity (juveniles only), and (F) pool rugosity (juveniles only). Grey and white symbols distinguish southwest and northeast tidepools, respectively. Curves and mean values in panels B–F are based on the fits of separate generalized linear models for juveniles and adults, adjusted for the effects of other physical characteristics included in the respective models (see Table 2). Juveniles were absent from the largest pool, NE1C. Mean and SE values are back-transformed from ln results, hence the asymmetrical SE values. For personal use only. Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18

individual fish on a census day is dichotomous (recaptured or not) Alternatively, the recapture probability could vary stochasti- and the number of previously marked fish, M, sets the maximum cally among sets of pools, which would generate greater variation number of recaptured fish in each sample. If all fish in a pool have (overdispersion) in the proportion of recaptured fish than ex- the same recapture probability and captures (and recaptures) occur pected from a binomial process. We considered this possibility by independently, then the number of (recaptured) marked fish, R,ina allowing r to vary (between 0 and 1) around a mean of ¯r according sample should follow a binomial distribution to a beta distribution, resulting in a beta-binomial distribution:

M Ϫ ⌫(M ϩ 1)⌫(R ϩ a)⌫(M Ϫ R ϩ b) (2) B(R|M, r) ϭ ͩ ͪrR(1 Ϫ r)M R (3) BB(R|M, ¯r, ␾) ϭ R ⌫(R ϩ 1)⌫(M Ϫ R ϩ 1)⌫(a)⌫(b)⌫(M ϩ a ϩ b)

Published by NRC Research Press 1332 Can. J. Zool. Vol. 96, 2018

where ⌫(z) is the incomplete gamma function of z, ␾ Ͼ 0 depicts Table 2. Statistics from generalized linear models assessing the ef- the variation in r about ¯r, a ϭ ¯r/␾ and b ϭ ͑1 Ϫ ¯r͒/␾ (Richards 2008). fects of tidepool physical characteristics on the standard length of For either distribution, the effects of pool characteristics on juvenile and adult tidepool sculpin (Oligocottus maculosus) captured recapture probabilities can be incorporated by representing r (or ¯r) during first sampling days. as a specific function of those characteristics, as defined by a set of Partial regression parameters, ␪. As only four sets of pools were sampled, we maxi- Physical characteristic coefficient ± 95% CI t df mally examined the effect of one pool characteristic (tempera- Juvenile ture, salinity, elevation, volume, or rugosity) at a time. Therefore, Elevation 0.253±0.074 3.42*** 87 we considered the following alternatives for ␪: Volume –0.059±0.018 3.21** 87 (i) a common proportion of recaptured fish in samples 2 and 3, Salinity 16.78±3.88 4.33*** 87 ␪ ␪ ␪ Rugosity 0.622±0.167 3.73*** 87 regardless of pool characteristics, = 2 = 3 ={r}; (ii) different proportions of recaptured fish in samples 2 and 3 Adult Elevation 0.141±0.047 2.98** 355 that were unaffected by pool characteristics, ␪ ={r }, ␪ ={r }; 2 2 3 3 Volume 0.040±0.007 5.76*** 355 (iii) the proportion of recaptured fish varied with a specific Temperature –0.670±0.181 3.70*** 355 pool characteristic, X, according to a logistic function, r ϭ ␤ ϩ␤ ␤ ϩ␤ Note: Both analyses considered log-normal distributions and ln-transformed 0 1X ͑ ϩ 0 1X͒ e / 1 e , with common regression coefficients for sam- physical characteristics. 95% CI refers to 95% confidence interval. **, P < 0.01; ***, ␪ ␪ ␪ ples 2 and 3, = 2 = 3 ={b0, b1}; P < 0.001. (iv) the proportion of recaptured fish varied with a specific pool characteristic with different regression coefficients for samples 2 similar fish abundance, with a mean of 56.0 unique fish in the SW ␪ ␪ and 3, 2 ={b2,0, b2,1}, 3 ={b3,0, b3,1}. sets (mean SW1 = 66.7, mean SW2 = 45.3) and a mean of 55.3 unique fishes in the NE sets (mean NE1 = 39.7, mean NE2 = 71.3). Given a specific parameter set, the likelihood of a regression The proportion of adults (i.e., fish >30 mm) in a home pool model given the data are ranged from 52.5% in the SW sets to 100% in the NE sets (Table 1, Fig. 2). Based on fish captured during the first sampling day for ϭ ␪ ϩ ␪ L B(R2 |M1, 2)·B(R3 |M1 M2, 3) individual pools, the (jittered) proportion of adults varied signifi-

cantly among pool sets (F[3,439] = 7.33, P < 0.001) and marginally for the binomial distribution and among pools within sets (F[8,439] = 2.00, P < 0.05; Figs. 3A–3C). Proportionally more adults occupied NE than SW pools (F[1,439] = ϭ ␪ ␾ ϩ ␪ ␾ 16.86, P < 0.001; Figs. 3A, 3B), but the adult proportion did not L BB(R2 |M1, 2, )·BB(R3 |M1 M2, 3, ) differ between sets within these groups (P > 0.4 in both cases). The significant variation among pools within sets solely involved set ␾ for the beta-binomial distribution (note the common for both SW2 (F[2,439] = 4.74, P < 0.01). samples). The two terms in each of these expressions represent Overall variation in the adult proportion in first-day samples the likelihood contributions of recapture in the second sample of can be attributed largely to asymptotic effects of pool tempera- fish marked in the first sample and recapture in the third sample ture and volume (AIC 19.9 units smaller than for the preceding of any fish marked in the first or second sample, respectively. analysis). Adults were almost the sole occupants of cool (<17 °C)

For personal use only. These components are assumed to be independent, hence the pools, whereas in pools >18 °C, they represented approximately multiplication. two-thirds of the captured fish, regardless of temperature (Fig. 3A; Analysis of the preceding alternatives involved 24 regression temperature ⌬AIC = 20.3). The proportion of adults tended to in- models, which we fit using the NLMIXED procedure of SAS crease with pool volume and only adults were captured in the excep- version 14.1 (SAS Institute Inc. 2015). We identified the best-® tionally large NE1C pool (Fig. 3B; volume ⌬AIC = 12.1). fitting model(s) based on two criteria (Richards 2008). First, candidate models had ⌬ = AIC – AIC ≤ 6, where AIC is Size-specific fish distribution i i min i The length of tagged fish ranged from 30 to 80 mm (Fig. 2). For Akaike’s information criterion for model i and AICmin is the fit of the best-fitting model. Second, from these candidate models, fish captured on the first sampling day, standard length varied significantly among pool sets in a manner that differed between we excluded any model that was a more complicated version of a juveniles and adults (age × set interaction: F = 8.28, P < 0.001; better-fitting model (i.e., the latter is nested within the former). [3,435] Fig. 4A). For both age classes, fish of equivalent mean length oc- Results cupied the SW1, SW2, and NE2 pool sets (P > 0.05, Dunn–Šidák multiple comparisons), whereas the NE1 set was occupied by sig- Site and tidepool features nificantly larger adults (except compared with SW2), but signifi- Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18 The Shannon entropy index was 0.35 and 0.68 in 2013 and 2015, cantly smaller juveniles (P < 0.05, Dunn–Šidák multiple comparisons). respectively. By converting the Shannon entropy index to the In addition to these effects, overall fish size differed among pools

effective number of species, Wizard Islets effectively support 1.42 within sets (F[8,435] = 4.38, P < 0.001), because of significant varia- and 1.97 species in 2013 and 2015, respectively. tion within the SW sets (P < 0.001 in both cases), but not within the Physical characteristics differed among the main pools of each NE sets (P > 0.7 in both cases). Within-set variation occurred equiv- pool set (Table 1). In general, higher pools (2.93–3.02 m above alently for juveniles and adults (age × pool(set) interaction:

LLWLT) were less saline (25.1 ± 3.05 ppt, mean ± SE) and warmer F[7,428] = 1.53, P > 0.15). (20.3 ± 2.81 °C, mean ± SE) than lower pools (1.16–1.65 m about Separate analyses that considered the effects of pool physical LLWLT). Elevation and mean temperature correlated positively, characteristics on mean length for juveniles and adults found but not significantly among pools (Pearson product-moment cor- somewhat contrasting influences (Table 2, Figs. 4B–4F). Variation relation coefficient, r = 0.853, P = 0.147). Pool volume and rugosity in pool elevation had parallel effects for both age classes, with did not vary consistently with other characteristics. larger fish occupying higher pools (Fig. 4B). In contrast, pool vol- ume had contrasting effects, with large pools being occupied by Age-specific fish distribution smaller juveniles and larger adults than small pools (Fig. 4C). The Overall, 668 unique fish were caught and 420 unique adults mean size of adults, but not of juveniles, varied significantly with were tagged in the 12 pools of the 4 pool sets. Pool sets supported temperature, with larger adults occupying cooler pools (Fig. 4D). Two

Published by NRC Research Press Wuitchik et al. 1333

Table 3. The histories of tidepool sculpin (Oligocottus maculosus) capture and recapture during three consecutive sampling days for four tidepool sets. Day 2 Day 3 Day 1: Total New Total Tidepool set tagged captured Recaptured tagged % Recapture captured Recaptured % Recapture SW1 68 45 30 15 44.1 44 37 44.6 SW2 88 33 21 12 23.9 49 38 38 NE1 77 29 24 5 31.2 53 42 51.2 NE2 126 99 70 29 55.6 70 55 35.5

other characteristics significantly affected the mean size of juve- Table 4. Fits (⌬) of nonlinear regressions of variation in the propor- nile fish, but not of adults: on average, smaller juveniles occupied tions of recaptured tidepool sculpin (Oligocottus maculosus) on the char- pools with lower salinity and less rugosity (Figs. 4E–4F). acteristics of four tidepools. Binomial Beta-binomial Pool fidelity Capture and marking showed limited movement among study Independent variable Common Separate Common Separate pools, but they are correlated with an altered age structure of None 13.53 15.50 2.46 4.23 recaptured fishes. Only seven fish moved between pool sets; two Temperature 15.43 18.88 4.46 8.22 from SW1 to SW2, two from NE1 to NE2, and three from NE2 to Salinity 15.51 16.96 4.46 8.01 NE1 (i.e., 1.04% of unique fishes during study). During the 3-day Elevation 12.43 8.91 3.68 5.94 sampling periods, the (jittered) proportion of adult fish declined Volume 15.52 12.77 4.46 6.44 from 0.824 to 0.701 (F[2,1044] = 8.82, P < 0.001), after accounting for Rugosity 6.05 0 1.23 0.98 significant variation among sets (F[3,1044] = 13.69, P < 0.001) and ⌬ Note: = AICi – AICmin is provided for models based on binomial and beta- among pools within sets (F[8,1044] = 7.85, P < 0.001; Fig. 3C). The binomial distributions and common or separate regression coefficients for re-

significant component of this decline occurred after the initial captures on the second and third sampling days. AICi is Akaike’s information sampling day (F[1,1044] = 17.16, P < 0.001), as the adult proportion did criterion for model i and AICmin is the fit of the best-fitting model. not differ significantly between second and third sampling days (F[1,1044] = 0.21, P > 0.6). This effect of sampling day did not vary The size and age distributions of O. maculosus, as well as their site among pool sets (interaction: F[6,1022] = 0.23, P > 0.95) or among fidelity after initial sampling, varied significantly among tide- pools within sets (interaction: F[16,1022] = 1.05, P > 0.35). For adults, pools in association with their physical and chemical conditions. mean length did not vary significantly among sampling days Adults predominated in the two NE pool sets, which were dis- (F[2,767] = 1.44, P > 0.2). tinctly lower and cooler than the SW sets that were occupied by up Recapture success varied among pool sets from 23.9% to 55.6% to 40% juveniles (Fig. 3A). Larger juveniles tended to occupy more (Table 3) and differed significantly with pool characteristics and saline and more rugose pools (Figs. 4E, 4F), whereas larger adults sampling days. Of the four models that did not account for possi- tended to occupy cooler pools (Fig. 4D). Pool volume had contrast-

For personal use only. ble effects of a particular pool characteristic, the beta-binomial ing effects on the proportions and sizes of juveniles and adults in distribution fit better than the binomial distribution (Table 4), pools, with small pools being occupied by relatively more, larger indicating heterogeneous recapture probabilities among pools. juveniles and smaller adults than large pools (Figs. 3B, 4C). The Only variation among pools in rugosity provided a better fit to the ⌬ proportion of adults captured decreased with each resampling data. In particular, the best-fitting model (Table 4, =0)ac- event (Fig. 3C). During the day after initial censusing, but not counted for different effects of pool rugosity between second and subsequently, the proportion of recaptured fish in censuses varied third samples and involved the binomial distribution. That the negatively with basin rugosity (Figs. 5A, 5B). These results reveal simpler binomial distribution was adequate in this case indicates related features of habitat use and site fidelity by O. maculosus. that the modelled effects of rugosity largely accounted for the extra-binomial variation among the four pools. The hetero- Pool occupancy geneous effects of rugosity involved less chance of recapture in The nonrandom associations of sculpin age and size with the rugose than in “smooth” pools during the second sample (Fig. 5A; physical characteristics of the tidepools that they occupied, sug- partial regression coefficient ± 95% confidence interval: b[20] = gest active habitat selection. Oligocottus maculosus is well suited to 2.632 ± 2.005, b[21] = –2.332 ± 1.550), but no significant effect during life in tidepools, as it tolerates rapid shifts in salinity, tempera- the third sample (Fig. 5B; b[30] = –0.261 ± 1.679, b[31] = –0.081 ± 1.271). ture, and low-oxygen conditions (Todgham et al. 2005; Sloman Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18 This difference between sampling days arose largely because the et al. 2008; Mandic et al. 2009). Furthermore, this species exhibits recapture probability in rugose pools increased to equal that in cross-tolerance, with heat-shock tolerance enhancing tolerance to smooth pools on the final sampling day. successive osmotic stress and (or) hypoxia (McBryan et al. 2013). Despite such tolerance, individuals likely preferentially occupy Discussion favourable conditions, as allowed by physical, chemical, and bio- Homing by resident individuals promotes community consis- logical circumstances (Szabo 2002; Petty and Grossman 2004; tency and O. maculosus exhibits fidelity to the pool of initial cap- Sloman et al. 2008). The higher frequency of adults, especially ture, even when the surrounding habitat offers expansive larger individuals, in cooler pools (Figs. 3A, 4D) suggests avoid- undisturbed (i.e., not sampled) pools (Yoshiyama et al. 1992). The ance of heat stress. Similarly, the predominance of adults in large Shannon entropy index and estimate of effective number of spe- pools and the positive association of adult size and pool volume cies both suggest that given the robustness of site fidelity and (Figs. 3B, 4C) could be associated with greater food availability. homing in this species, the sets of study pools (including the In contrast, juveniles may be less able to implement their envi- adjacent accessory pools) reliably include the home ranges of in- ronmental preferences in the presence of competing (Szabo 2002), dividual fish (White and Brown 2013). These metrics are likely or even predatory (Koczaja et al. 2005), adults. On Wizard Islets, heavily influenced by the dominance of O. maculosus in the com- there are no recorded land-based or avian predators to account for munity over the years. persistent or semiregular pressure, thus the likelihood

Published by NRC Research Press 1334 Can. J. Zool. Vol. 96, 2018

Fig. 5. Relations of the proportion of recaptured tidepool sculpin (Oligocottus maculosus) to pool rugosity for the (A) second and (B) third samples.

of intraspecific competition across age classes is a feasible expla- The probability of recapture during the day after the initial nation of this distribution. Such avoidance is suggested by the census was particularly low in pools with rugose basins (Figs. 5A, relatively high frequency of juveniles in small pools (Fig. 3B), the 5B), whereas during the third sampling day, recapture probability contrasting relations of fish size to pool volume for juveniles and in rugose pools increased to match that in smoother pools. This is adults (Fig. 4C), and the tendency of larger juveniles to occupy potentially due to the fish being more successful at hiding in more pools with more rugose basins (Fig. 4F). The predominance of rugose pools and stopped attempting to hide by day 3; further large adults in large, cool tidepools, such as those in the NE sets, study over a longer temporal scale would be beneficial to eluci- may thus promote juvenile occupancy of higher, warmer pools, date the long-term patterns and mechanisms underlying these such as those in the SW sets. Alternatively, juveniles may prefer- observations. Since fishes exhibit significantly lower escape laten- entially occupy warmer pools as a means of speeding their growth rate (Hofmann and Fischer 2003), which may be beneficial even cies in familiar territory (Brown 2001), the residency effect de- when considering the trade-off of higher predation risk in shal- scribed by Maynard Smith and Parker (1976) is likely a very lower waters (Ward et al. 2010). Regardless of the mechanism, important driver in intertidal fishes returning to specific tide- these results demonstrate that the occupation of specific tide- pools (White and Brown 2013). Habitat structure may also be an pools reflect size and age structures of local fish assemblages important indicator of the presence of fishes to use cover and within pools. Such nonrandom habitat use, particularly by juve- access food resources in both juvenile and adult age classes (Levin niles, is likely influenced by life-history characteristics that could and Levin 1993; Szabo 2002). affect survival (Fangue et al. 2006, 2011). Conclusions Pool fidelity Our study attempted to demonstrate the influence of physical As illustrated by the recapture frequencies (Figs. 5A, 5B) and the environments of tidepools on the habitat use patterns in the For personal use only. limited observed movement between pool sets, although O. maculosus, with results showing short-term site fidelity and O. maculosus move out of tidepools on rising tides (Green 1971), body-size-associated distribution patterns that were associated individuals often return to the same tidepool at low tide. The with physical aspects of the environment to some extent. Al- observed recapture frequencies are similar to those recorded be- though our results illustrate a role for intrinsic habitat features tween consecutive years (Gerbacher and Denison 1930; Yoshiyama that influence age- and size-class distributions in a resident tide- et al. 1992). Such site fidelity implies preference for particular pools and familiarity with the topography of the home range pool sculpin species, with corresponding consequences for site (Gibson 1999), though sampling during times of particularly fidelity, there were several caveats that should be addressed in strong tidal exchange may have impacted this. In addition to site future efforts to elucidate these observed trends. First, the time fidelity, the smallest O. maculosus individuals typically travel far- span of the study could be extended to include more extensive thest between low tides (Green 1971). This movement pattern may temporal sampling with additional unique marking of individuals contribute to smaller juveniles occupying lower pools, as they are to facilitate more accurate characterization of the habitat selec- submerged longest between low tides. Whether younger fish gain tion and home ranges of the species, as well as factors that influ- spatial awareness of their habitat by travelling farther before ence them. For example, data from a longer temporal period that choosing a home range (Craik 1981) remains unknown. encompasses enough time for an individual to show ontogenetic Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18 Although many tagged fish were recaptured at the same pool habitat shifts are desirable to elucidate the size-related habitat use during subsequent censuses, most were not recovered (Figs. 5A, patterns. Ideally, all sites should include similar range of variation 5B) and presumably moved elsewhere. Consequently, fish abun- in observed physical factors. Additional site-specific factors (e.g., dance in the sampled pools declined 40% after the first census wave exposure, current, recruitment variation in past years) could (Table 3). These results indicate that the initial census disturbed also be incorporated to explain the pattern of habitat use and fish, causing them to perceive the pool in which they were cap- occurrence. Translocation experiments would also be useful to tured as unsuitable, at least temporarily. Adults responded most determine the likelihood of return to pool of capture and to track strongly to disturbance, as they represented about 70% of cap- tured fish in the second and third samples, compared with 82% in the movement patterns. Although our data have spatial and tem- the initial sample. In addition to these responses, the 30% and 20% poral limitations precluding inference of general trends, our re- of the fish captured during the second and third censuses, respec- sults highlight a role for intrinsic habitat features on age- and tively, had not been captured previously (Table 3). Thus, the tag- size-class distributions in a resident tidepool sculpin species, with ging results demonstrate that O. maculosus adults do not exhibit the corresponding consequences for site fidelity suggesting that immutable site fidelity, but instead move between pools in close continued efforts to understand the nature of rocky intertidal proximity if conditions decline locally or improve elsewhere environments will contribute to a better understanding of en- (Green 1971). vironmental heterogeneity on the ecology of tidepool fishes.

Published by NRC Research Press Wuitchik et al. 1335

Acknowledgements clove oil in fish (review). Fish Physiol. Biochem. 38(6): 1545–1552. doi:10.1007/ s10695-012-9682-5. PMID:22752268. We acknowledge that the Wizard Islets are a part of the Huu-ay-aht Jost, L. 2007. Partitioning diversity into independent alpha and beta compo- First Nations traditional territory and are grateful for the opportu- nents. Ecology, 88(10): 2427–2439. doi:10.1890/06-1736.1. PMID:18027744. nity to conduct our research in protected areas. We thank the Bam- Koczaja, C., McCall, L., Fitch, E., Glorioso, B., Hanna, C., Kyzar, J., Niemiller, M., field Marine Sciences Centre (BMSC) for the support of this research, Spiess, J., Tolley, A., Wyckoff, R., and Mullen, D. 2005. Size-specific habitat as well as K.R. Holmes (Hakai Institute) and I. McKechnie (University segregation and intraspecific interactions in banded sculpin (Cottus carolinae). Southeast. Nat. 4(2): 207–218. doi:10.1656/1528-7092(2005)004[0207:SHSAII]2. of Victoria and Hakai Institute) for the production of the high- 0.CO;2. resolution site map. We also thank T. Buser and D. Wuitchik for their Levin, P.S., and Levin, S. 1993. Habitat structure, conspecific presence and spatial reviews and edits that improved the manuscript. The research was variation in the recruitment of a temperate reef fish. Ecology, 94(2): 176–185. funded by a BMSC New Researcher Award (S.M.R.), Discovery Grants doi:10.1007/BF00341315. Mahon, R., and Mahon, S.D. 1994. Structure and resilience of a tidepool fish from the Natural Sciences and Engineering Research Council of assemblage at Barbados. Environ. Biol. Fishes, 41(1–4): 171–190. doi:10.1007/ Canada (NSERC; L.D.H. and S.M.R.), and a New Faculty Award from BF02197843. Alberta Innovates Technology Futures (S.M.R.). Mandic, M., Todgham, A.E., and Richards, J.G. 2009. Mechanisms and evolution of hypoxia tolerance in fish. Proc. R. Soc. B Biol. Sci. 276(1657): 735–744. References doi:10.1098/rspb.2008.1235. Maynard Smith, J., and Parker, G.A. 1976. The logic of asymmetric contests. Baird, O.E., and Krueger, C.C. 2003. Behavioral thermoregulation of brook and Anim. Behav. 24(1): 159–175. doi:10.1016/S0003-3472(76)80110-8. rainbow trout: comparison of summer habitat use in an Adirondack river, New York. Trans. Am. Fish. Soc. 132(6): 1194–1206. doi:10.1577/T02-127. McBryan, T.L., Anttila, K., Healy, T.M., and Schulte, P.M. 2013. Responses to Brown, C. 2001. Familiarity with the test environment improves escape re- temperature and hypoxia as interacting stressors in fish: implications for sponses in the crimson spotted rainbowfish, Melanotaenia duboulayi. Anim. adaptation to environmental change. Integr. Comp. Biol. 53(4): 648–659. Cogn. 4(2): 109–113. doi:10.1007/s100710100105. doi:10.1093/icb/ict066. PMID:23784697. Castellanos-Galindo, G.A., Giraldo, A., and Rubio, E.A. 2005. Community struc- Menge, B.A. 1976. Organization of the New England rocky intertidal community: ture of an assemblage of tidepool fishes on a tropical eastern Pacific rocky role of predation, competition, and environmental heterogeneity. Ecol. shore, Colombia. J. Fish Biol. 67(2): 392–408. doi:10.1111/j.0022-1112.2005.00735.x. Monogr. 46(4): 355–393. doi:10.2307/1942563. Craik, G.J.S. 1978. A further investigation of the homing behaviour of the inter- Metaxas, A., and Scheibling, R. 1993. Community structure and organization of tidal cottid, Oligocottus maculosus Girard. Ph.D. thesis, The University of British tidepools. Mar. Ecol. Prog. Ser. 98: 187–198. doi:10.3354/meps098187. Columbia, Vancouver. Morris, S., and Taylor, A.C. 1983. Diurnal and seasonal variation in physico- Craik, G.J.S. 1981. The effects of age and length on homing performance in the chemical conditions within intertidal rock pools. Estuar. Coast. Shelf Sci. intertidal cottid, Oligocottus maculosus Girard. Can. J. Zool. 59(4): 598–604. 17(3): 339–355. doi:10.1016/0272-7714(83)90026-4. doi:10.1139/z81-088. Palumbi, S.R. 2001. Humans as the world’s greatest evolutionary force. Science, Evans, D.H., Claiborne, J.B., and Kormanik, G.A. 1999. Osmoregulation, 293(5536): 1786–1790. doi:10.1126/science.293.5536.1786. acid–base regulation, and nitrogen excretion. In Intertidal fishes: life in two Petty, J.T., and Grossman, G.D. 2004. Restricted movement by mottled sculpin worlds. 1st ed. Edited by M.H. Horn, K.L.M. Martin, and M.A. Chotkowski. (Pisces: ) in a southern Appalachian stream. Freshw. Biol. 49(5): 631– Academic Press, San Diego, Calif. pp. 79–96. doi:10.1016/B978-012356040-7/ 645. doi:10.1111/j.1365-2427.2004.01216.x. 50006-8. Raffaelli, D., and Hawkins, S.J. 1999. Intertidal ecology. Springer, Dordrecht, the Fangue, N.A., Hofmeister, M., and Schulte, P.M. 2006. Intraspecific variation in Netherlands. doi:10.1007/978-94-009-1489-6. thermal tolerance and heat shock protein gene expression in common killi- Rankin, J.C., and Jensen, F.B. 1993. Fish ecophysiology. Springer, Dordrecht, the fish, Fundulus heteroclitus. J. Exp. Biol. 209(15): 2859–2872. doi:10.1242/jeb. Netherlands. doi:10.1007/978-94-011-2304-4. 02260. PMID:16857869. Richards, J.G. 2011. Physiological, behavioral and biochemical adaptations of Fangue, N.A., Osborne, E.J., Todgham, A.E., and Schulte, P.M. 2011. The onset intertidal fishes to hypoxia. J. Exp. Biol. 214(2): 191–199. doi:10.1242/jeb. temperature of the heat-shock response and whole-organism thermal tolerance 047951. PMID:21177940. are tightly correlated in both laboratory-acclimated and field-acclimatized tide- Richards, S.A. 2008. Dealing with overdispersed count data in applied ecology.

For personal use only. pool (Oligocottus maculosus). Physiol. Biochem. Zool. 84(4): 341–352. doi: J. Appl. Ecol. 45(1): 218–227. doi:10.1111/j.1365-2664.2007.01377.x. 10.1086/660113. PMID:21743248. SAS Institute Inc. 2015. SAS. Version 14 [computer program]. SAS Institute Inc., Fry, F.E.J. 1947. Effects of the environment on animal activity. Publ. Ont. Fish. Cary, N.C. Res. Lab. 68. University of Toronto Press, Toronto, Ont. Sheather, S.J. 2004. Density estimation. Stat. Sci. 19(4): 588–597. doi:10.1214/ Gerbacher, W.M., and Denison, M. 1930. Experiments with in tide pools. 088342304000000297. Publ. Puget Sound Biol. Stn. 7: 209–215. Sloman, K.A., Mandic, M., Todgham, A.E., Fangue, N.A., Subrt, P., and Gerking, S.D. 1959. The restricted movement of fish populations. Camb. Philos. Richards, J.G. 2008. The response of the tidepool sculpin, Oligocottus maculosus, Soc. 34(2): 221–242. doi:10.1111/j.1469-185X.1959.tb01289.x. to hypoxia in laboratory, mesocosm and field environments. Comp. Biochem. Gibbons, T.C., Rudman, S.M., and Schulte, P.M. 2016. Responses to simulated Physiol. Part A Mol. Integr. Physiol. 149(3): 284–292. doi:10.1016/j.cbpa.2008.01.004. winter conditions differ between threespine stickleback ecotypes. Mol. Ecol. Szabo, A.R. 2002. Experimental tests of intercohort competition for food and 25(3): 764–775. doi:10.1111/mec.13507. PMID:26645643. cover in the tidepool sculpin (Oligocottus maculosus Girard). Can. J. Zool. 80(1): Gibson, R.N. 1999. Movement and homing in intertidal fishes. In Intertidal 137–144. doi:10.1139/z01-218. fishes: life in two worlds. 1st ed. Edited by M.H. Horn, K.L.M. Martin, and Todgham, A.E., Schulte, P.M., and Iwama, G.K. 2005. Cross-tolerance in the tide- M.A. Chotkowski. Academic Press, San Diego, Calif. pp. 97–125. doi:10.1016/ pool sculpin: the role of heat shock proteins. Physiol. Biochem. Zool. 78(2): B978-012356040-7/50007-X. 133–144. doi:10.1086/425205. PMID:15778933. Green, J.M. 1971. High tide movements and homing behaviour of the tidepool Ward, A.J.W., Hensor, E.M.A., Webster, M.M., and Hart, P.J.B. 2010. Behavioural sculpin Oligocottus maculosus. J. Fish. Res. Board Can. 28(3): 383–389. doi:10. thermoregulation in two freshwater fish species. J. Fish Biol. 76(10): 2287– 1139/f71-051. 2298. doi:10.1111/j.1095-8649.2010.02576.x. PMID:20557593.

Can. J. Zool. Downloaded from www.nrcresearchpress.com by UNIV CALGARY on 12/04/18 Grossman, G.D. 1986. Food resource partitioning in a rocky intertidal fish assem- White, G.E., and Brown, C. 2013. Site fidelity and homing behaviour in intertidal blage. J. Zool. (Lond.), 1(2): 317–355. doi:10.1111/j.1096-3642.1986.tb00642.x. fishes. Mar. Biol. 160(6): 1365–1372. doi:10.1007/s00227-013-2188-6. Hofmann, N., and Fischer, P. 2003. Impact of temperature on food intake and Yoshiyama, R.M., and Cech, J.J., Jr. 1994. Aerial respiration by rocky intertidal growth in juvenile burbot. J. Fish Biol. 63(5): 1295–1305. doi:10.1046/j.1095- fishes of California and Oregon. Copeia, 1994(1): 153–158. doi:10.2307/ 8649.2003.00252.x. 1446681. Horn, M.H., Martin, K.L.M., and Chotkowsi, M.A. 1999. Introduction. In Intertidal Yoshiyama, R.M., Gaylord, K.B., Philippart, M.T., Moore, T.R., Jordan, J.R., fishes: life in two worlds. 1st ed. Edited by M.H. Horn, K.L.M. Martin, and Coon, C.C., Schalk, L.L., Valpey, C.J., and Tosques, I. 1992. Homing behavior M.A. Chotkowski. Academic Press, San Diego, Calif. pp. 1–6. doi:10.1016/B978- and site fidelity in intertidal sculpins (Pisces: Cottidae). J. Exp. Mar. Biol. Ecol. 012356040-7/50002-0. 160(1): 115–130. doi:10.1016/0022-0981(92)90114-P. Huggett, J., and Griffiths, C. 1986. Some relationships between elevation, Zander, C.D., Nieder, J., and Martin, K. 1999. Vertical distribution patterns. In physico-chemical variables and biota of intertidal rock pools. Mar. Ecol. Prog. Intertidal fishes: life in two worlds. 1st ed. Edited by M.H. Horn, K.L.M. Martin, Ser. 29(2): 189–197. doi:10.3354/meps029189. and M.A. Chotkowski. Academic Press, San Diego, Calif. pp. 26–53. doi:10. Javahery, S., Nekoubin, H., and Moradlu, A.H. 2012. Effect of anaesthesia with 1016/B978-012356040-7/50004-4.

Published by NRC Research Press