Analysis of spatial niche structure in coexisting tidepool fishes: null models based on multi-scale experiments
2010; Wiley; Volume: 80; Issue: 1 Linguagem: Inglês
10.1111/j.1365-2656.2010.01749.x
ISSN1365-2656
AutoresSeiji Arakaki, Mutsunori Tokeshi,
Tópico(s)Plant and animal studies
ResumoJournal of Animal EcologyVolume 80, Issue 1 p. 137-147 Free Access Analysis of spatial niche structure in coexisting tidepool fishes: null models based on multi-scale experiments Seiji Arakaki, Corresponding Author Seiji Arakaki Present address: Sesoko Station, Tropical Biosphere Research Center, University of the Ryukyus, 3422 Sesoko, Motobu, Okinawa, Japan 905-0227. Correspondence author. E-mail: arakaki@ambl-ku.jpSearch for more papers by this authorMutsunori Tokeshi, Mutsunori Tokeshi AMBL, Kyushu University, Amakusa 863-2507, JapanSearch for more papers by this author Seiji Arakaki, Corresponding Author Seiji Arakaki Present address: Sesoko Station, Tropical Biosphere Research Center, University of the Ryukyus, 3422 Sesoko, Motobu, Okinawa, Japan 905-0227. Correspondence author. E-mail: arakaki@ambl-ku.jpSearch for more papers by this authorMutsunori Tokeshi, Mutsunori Tokeshi AMBL, Kyushu University, Amakusa 863-2507, JapanSearch for more papers by this author First published: 26 August 2010 https://doi.org/10.1111/j.1365-2656.2010.01749.xCitations: 18 AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Summary 1. Fundamental and realized spatial niches were investigated through a combination of laboratory and mesocosm experiments, field observations and null model analysis in three intertidal gobiid species (Bathygobius fuscus, Chaenogobius annularis and C. gulosus). Null models based on the results of single-species experiments were used to assess interspecific spatial use and coexistence on two different scales: (i) microhabitats within a tidepool ('microhabitat' scale); and (ii) distribution among a set of tidepools ('habitat-wide' scale). 2. Patterns of microhabitat use varied from single to paired treatments, depending on paired species. Realized overlap of microhabitat use was smaller than would be expected from single-individual situations for intraspecific combinations, but not for interspecific ones. 3. Patterns of tidepool occupancy (a measure of spatial niche breadth) in the mesocosm were influenced by interspecific interactions. Two Chaenogobius species, but not B. fuscus, decreased tidepool occupancy in the hetero-specific treatments compared with the mono-specific ones. For all interspecific combinations, spatial overlap (habitat-wide scale) was significantly lower than the values expected from mono-specific situations. The results also indicated a possible trade-off between competitiveness and growth efficiency in these fishes. 4. Interspecific spatial overlap in the field was similar to that in the mesocosm experiment and the pattern of coexistence of gobiids can be explained by the results of our experiments. 5. This study demonstrates that niches of intertidal fishes may experience modifications under the influence of species interactions and that null models based on controlled experiments can greatly facilitate the deciphering of such changes in niche structure. Introduction The niche concept has been central to community ecology, on which major theoretical and empirical studies have been undertaken (e.g. Grinnell 1914; Elton 1927; Hutchinson 1957; MacArthur 1958; Pulliam 2000; Chase & Leibold 2003; Odling-Smee, Laland & Feldman 2003; Tilman 2004; McGill et al. 2006; Pearman et al. 2008; Levine & HilleRisLambers 2009). Species coexistence cannot be discussed fully without referring to the niche and the aspects of resource utilization (Hutchinson 1957; Schoener 1974), both of which have important implications for species assembly, species abundance and competitive/non-competitive relations (Tokeshi 1999). The shape and volume of niche may be altered through biotic interactions, most notably competitive relations on ecological and evolutionary time scales (Connell 1983; Schoener 1983; Morin 1999; Pearman et al. 2008). Conversely, the consequences of competitive interactions may be revealed through a comparison between the fundamental and the realized niche, which would lead to a better understanding of the mechanics of community organization and species coexistence. Space is an important dimension of niche which may be shared or partitioned by co-occurring species (Schoener 1974; Tokeshi 1999), reinforcing the view that resource partitioning in terms of space is central to species coexistence (e.g. Rosenzweig 1981). Fundamental spatial niche depends on the ecophysiological and other traits of a species, forming the basis of species-specific occurrence patterns which in turn are closely influenced by requirements for food, reproduction, avoidance of predators/competitors and some adverse environmental conditions. The distribution of species and, consequently, species composition and coexistence in a community reflect some modifications of the fundamental spatial niches of different species. Possible effects of competition on habitat use (i.e. changes in spatial niche), in particular, have often been inferred through experimental manipulations of competitor abundance in various assemblages (e.g. Crowell & Pimm 1976; Black 1979; Hairston 1980; Pacala & Roughgarden 1982; Elmberg et al. 1997; Lee & Silliman 2006) including fishes (e.g. Werner & Hall 1976, 1977; Munday, Jones & Caley 2001; Schofield 2003). However, assessment of changes in spatial niche is not straightforward with experiments, as biological realism may be sacrificed and density and species identity cannot always be independently controlled in mixed-species situations. In this respect, incorporation of null model approaches may be of some value, but there has been no attempt to explore the possibility of combining manipulative experiments with null model analyses to elucidate patterns of niche shifts. Despite numerous ecological studies on intertidal fishes, mechanisms and relative importance of different types of biotic interactions in community organization still remain obscure (see Gibson & Yoshiyama 1999). As the number of tidepools available at low tide time is limited, space is probably the most important niche dimension for intertidal fishes. Indeed, previous studies noted differences in spatial use (e.g. Gibson 1972; Davis 2000) and the possible roles of intra-/interspecific competition for space in structuring intertidal fish communities (e.g. Stephens et al. 1970; Nakamura 1976; Mayr & Berger 1992; Pfister 1995, 2006; Faria & Almada 2001). There is, however, a paucity of information on variable patterns of intra- and interspecific space use on different scales of intertidal habitat. The objective of this work was to investigate the patterns of fundamental and realized spatial niche in intertidal gobiid species through a combination of laboratory and mesocosm experiments and null model analysis. Here, fundamental spatial niche was interpreted as spatial patterns shown by individuals of a species, either singly or as group of individuals. Null models were constructed from data derived from single-species/individual experiments in which there was no influence of inter-/intraspecific interactions. Spatial niche was considered on two different scales (microhabitat and habitat-wide) to examine the varied influences of biotic interactions on the patterns of space utilization and species coexistence in this assemblage. In assessing the influences of biotic interactions, growth was also taken into account alongside changes in space use. Materials and methods Study site and fish species Field observation and fish collection were carried out on a moderately wave-exposed rocky shore of Shikizaki (32°31′N, 130°01′E) on the Amakusa-Shimoshima Island, south-western Japan. The study site had a tidal amplitude of about 3·5 m and was exposed for a distance of about 80 m seaward at the low tide. Tidepools of various sizes formed on this relatively flat rocky shore. The observation area and the collection site were separated by >100 m and an extra large tidepool (c. 3 × 104 m2) in between. Thus, there was no or negligible effect of fish collection on the observation area. Gobiidae and Blenniidae were particularly abundant in the study site, with the former being dominant at all tidal levels, especially in the upper intertidal, constituting about 90% of tidepool fish community (Arakaki & Tokeshi 2006). Three common gobiid species chosen for the present study, Bathygobius fuscus (Rüppell), Chaenogobius annularis Gill and Chaenogobius gulosus (Sauvage), comprised over 80% of tidepool fish fauna throughout the year. All three species are adapted for benthic life with similar maximum body sizes (c. 10 cm total length), compressed body shape and cryptic coloration, and are known to consume small benthic animals (Dôtu 1955; Sasaki & Hattori 1969). Field observations on tidepool use were conducted in the same season as the mesocosm experiment (June–August 2003, see below). A total of 34 tidepools of various sizes (surface area 500–9500 cm2) were chosen from the mid to upper intertidal zones [170–255 cm above mean lower low water (MLLW)]. Species and their abundances were recorded by direct visual observation at day-time new moon spring low tide. We confirmed that this method had a similar accuracy of sampling as fish collection using anaesthetics (Arakaki & Tokeshi 2006). Laboratory experiment on microhabitat use A laboratory experiment was carried out to examine the patterns of microhabitat utilization under the influence of biotic interactions. The present experiment with paired individuals complemented our previous experiment (Arakaki & Tokeshi 2005) that dealt solely with single-individual situations; results from both were used in the present study. Individuals of three species used in the experiment were collected using a hand net from the collection site. Fishes were carefully brought back to the laboratory in a container within 10 min of collection. All individuals were identified to species but sexes were not separated, as Chaenogobius spp. could not be sexed on the basis of external morphology. Fishes were selected for similar body sizes (mean TL ± 1SD, B. fuscus = 38·5 ± 4·2 mm, C. annularis = 36·1 ± 2·2 mm and C. gulosus = 39·4 ± 2·6 mm). Each species was kept separately in aquaria (62 × 38 × 15 cm) containing a bed of small stones with a flow-through seawater system for 7 days prior to the experiment. The fish were fed daily with pieces of shrimps, crabs and molluscs collected from the field. The acclimatization period of 7 days was considered appropriate for these gobiid fishes, though shorter than in other studies (Mayr & Berger 1992; Griffiths 2002; Schofield 2003). The experiments were conducted in rectangular opaque aquaria (37 × 25 × 14 cm, corresponding in size to small natural tidepools), each of which was subdivided into four equal quarters by thin retention walls (2 cm high, 0·8 mm thick) and a square stone plate (10 × 10 × 1 cm) was placed in the centre to serve as a rooftop such that a fish could use the shaded space below as shelter (Fig. 1a). Each quarter of an aquarium was allotted with one of four types of substrate, (i) sand (diameter, 0·25–1·0 mm); (ii) coarse sand (1·0–5·6 mm); (iii) gravel (5·6–19 mm); and (iv) bare rock (stone tile); the positioning of the four substrates in an aquarium was determined randomly for each setup. Substrates were collected from the field and separated using sieves of different mesh sizes in the laboratory. Stone tiles used as bare rock were preconditioned in the field for a year prior to the experiment. All sessile organisms attached were scraped off and all substrates were dried under the sun for 2 days before the experiment. The aquaria with substrates were conditioned with running seawater before each experiment trial. Figure 1Open in figure viewerPowerPoint Experimental setup of (a) the microhabitat experiment and (b) the mesocosm experiment. (a) shows an aquarium with a stone plate in the centre as shelter for fishes and different substrates in four areas. (b) shows details of an artificial tidepool with tilted sides and the arrangement of 24 of these in the mesocosm arena. Six combinations (three intraspecific and three interspecific) were investigated in the laboratory experiment. Prior to experiment, fishes were chosen arbitrarily from the maintenance aquaria and introduced into the centre of an experimental aquarium 6 h before the actual observation. This imitated the situation where a fish moved into a new tidepool before low tide under natural tidal cycles. Eight pairs of each interspecific combination were introduced into separate aquaria (i.e. two fish per aquarium) to serve as replicates, while there were four pairs for each intraspecific combination. Water depth was kept at 5 cm to imitate natural tidepool conditions at low tide; a fish could move freely between substrates. Aquaria were covered with a transparent plate to prevent gobies from escaping. A 1-h observation was carried out for each pair of fish during which positions in the aquarium (a total of 10 categories: four substrates × outside/inside the shelter, on top of the shelter, on the side wall) were recorded at 2-min intervals during day-time (between 13:00 and 15:00 hours) or night-time (21:30–23:30 hours). Thus for each individual, a total of 30 position data were obtained. A hand torch covered by a red film was used for night-time observations. It was turned off during the intervals of checking fish positions to minimize disturbance; fish apparently behaved normally (i.e. no sign of escape/stress responses such as quick changes in movement) throughout an observation session. Fishes were recognized individually by species, relative body size and the peculiarities of their colour patterns. When a fish changed its location at the time of observation, only the first position was recorded. If a fish was on a boundary of zones, its location was determined according to where the majority of body mass existed and/or the position of its head. After each observation individuals were returned to maintenance aquaria; all fishes were released to the field on completion of the experiments. Mesocosm experiment on habitat-wide spatial use A medium-scale artificial tidepool (mesocosm) experiment was conducted to investigate habitat-wide space utilization. The mesocosm was established on a semi-natural intertidal shore 1·5 km from the field site that was protected from wave exposure by concrete wall but had natural tidal fluctuations. An area of 6·3 × 5·4 m was enclosed by a net (5 mm mesh) of 1·7 m high and the bottom was cleared of large stones and covered by thin wood panels to create a uniform substrate with a natural slope. In the enclosure, artificial tidepools (46 × 30 × 7 cm) were arranged in two different tidal zones: 12 tidepools in the upper zone in two rows (205–219 cm above MLLW) and 12 in the lower in one row (192–196 cm MLLW) separated by 15 cm gaps (Fig. 1b). Each artificial tidepool had a dustpan-like shape with sloping sidewalls such that, when placed on a sloping mesocosm floor, the upper, open side was flush with the mesocosm floor so that fish could enter the pool without hindrance as the tide receded. One stone was placed at the upper entrance to serve as a landmark for fishes and stone tiles and stones were placed inside as microhabitats (Fig. 1b). Three gobiid species with similar size ranges as in the microhabitat experiment (mean ± 1SD: B. fuscus = 38·7 ± 3·0 mm, C. annularis = 37·1 ± 1·5 mm, C. gulosus = 39·1 ± 3·2 mm) were used in this mesocosm experiment. Body sizes of fishes were measured, after anaesthetizing, to the nearest 0·01 mm (TL) using digital callipers before and after introduction into the experimental arena. Fish individuals were marked by injecting visible implant fluorescent elastomer tags (Northwest Marine Technology, Shaw Island, Washington, USA) 1 week before the experiment. The tag was considered suitable for small fishes as it was known to have high retention (Griffiths 2002). Six different treatments [three mono-specific and three hetero-specific (two-species) combinations] were run as separate trials with a total of 48 individuals each (either 48 individuals of the same species or 24 individuals each of two species; two individuals per tidepool). In the case of interspecific combinations, different species were paired in each artificial tidepool at the time of introduction. Fishes could swim about freely and feed on small organisms in the mesocosm at high tide and retreat into artificial tidepools at low tide. Identity and the number of individuals in each artificial tidepool were recorded at consecutive low tides for a week after introduction. Individual that failed to use an artificial tidepool were collected and kept in the maintenance aquarium during low tide and released again into the mesocosm at the next rising tide. Few individuals that went missing during an experiment were not replaced. All fishes were released to the collection site after the experiments. Data analysis Null model analysis Null models were employed to evaluate the effects of biotic interactions. A single most crucial element of null modelling concerns the creation of 'null expectation', a pattern in which a focal biological phenomenon is absent. In the case of biotic interactions, the situation without such interactions cannot in principle be re-created from observed data as there is generally no guarantee that the aspect of relevance extracted from the data concerned has not been influenced by those interactions, past and present. In the present study our rationale was to construct null models based on experimental manipulations in which a single species/individual was present, thereby guaranteeing that the expected patterns were devoid of biotic interactions. For the microhabitat (laboratory) experiment, the null model involved randomization based on two single-individual experimental results to create null expectation of (pairwise) space use. For the habitat-wide (mesocosm) experiment, the null model was generated by randomly combining two single-species (not individual) experimental results. The choice of individual-based or species-based procedures was related to the fact that individual-based situations become less realistic as spatial scale increases; while a single-individual situation is relevant at microhabitat or 'within-pool' scale, it is not at habitat-wide or 'among-pool' scale. Note that these two analyses for different spatial scales involved different details of null model construction, as explained below. Microhabitat overlap (laboratory experiment) Overlap (Mβ) in microhabitat use was calculated as, where Px(i) and Py(i) are the proportional use of the ith (micro-) habitat by species/individual x and y, respectively. Proportional use was based on the 30 position data for each individual from the laboratory experiment. The index ranged from 0 (no overlap) to 1 (complete overlap) (Tokeshi 1986, 1999). Realized (observed) overlap values were compared with expected values derived from the data of single-individual experiment (conducted under the same setup, Arakaki & Tokeshi 2005), assuming that paired fishes behave in the same manner as in a single-individual situation (i.e. not influenced by other individuals). To generate the expected pattern, a computer simulation was performed in which fishes were re-distributed among ten microhabitats according to the probabilities equalling the observed proportions of microhabitats occupied in the single-individual experiment (Table 1). This re-distribution process was repeated 30 times for each fish individual to generate a set of space-use (30 positions) data, which was then used to calculate spatial overlap for a particular pair of individuals. The whole procedure was replicated 1000 times with random selection of individuals to obtain the mean with 95% range for each intra-/interspecific combination. Table 1. Frequencies of microhabitat use (%, inside/outside shelter) by three gobiid species, under single-individual conditions Microhabitat category Sand Coarse sand Gravel Bare rock Above plate Side wall B. fuscus Day 6·7/30 1·7/17·9 12·1/23·8 0·8/2·1 1·7 3·3 Night 2·5/23·3 2·5/14·2 3·3/42·9 2·1/7·9 1·3 0 C. annularis Day 24·6/12·1 14·6/6·3 15/8·8 9·2/3·8 0·4 5·4 Night 5·8/12·1 5·8/7·9 15·4/21·7 2·9/5 2·9 20·4 C. gulosus Day 0/0 45·4/0·4 50/0 4·2/0 0 0 Night 42·5/5·4 0/2·5 29·6/5·4 12·9/0·4 0·4 0·8 Tidepool occupancy and overlap (mesocosm experiment) Tidepool occupancy, expressed as the proportion of tidepools occupied out of total, represents a measure of spatial niche breadth on the mesocosm scale. Realized tidepool occupancy and overlap in tidepool use under hetero-specific mixtures were compared with the expected occupancy/overlap values based on the mono-specific treatment in the mesocosm experiment. To derive expected values, the following procedure was taken: (1) data on 24 individuals of each species were randomly extracted from the mono-specific treatment (a total of 39 (B. fuscus), 43 (C. annularis) and 45 (C. gulosus) individuals, excluding those with incomplete spatial data), (2) for each pair of species (a total of 48 individuals), tidepool occupancy was calculated for (i) each species separately and (ii) the two species together ('combined' occupancy; note that this value is equal to or smaller than the smaller of 1·0 and the addition of two separate occupancy values, depending on the degree of overlap in spatial distribution between the two species), (3) for each pair of species, spatial overlap (Mβ) was calculated, using each tidepool as unit of habitat. In addition, 'observed' values of intraspecific overlap in tidepool use were obtained by randomly dividing the observed mono-specific set of 48 individuals into two groups (each with 24 individuals) and calculating the spatial overlap between the two groups. These procedures were repeated 1000 times to obtain the mean and 95% range of occupancy and overlap values. Although data were collected on 10–14 consecutive occasions (designated as t1 to t10/t14), only those of the last three occasions (t8–t10 or t12–t14) were subjected to this analysis, as these represented more reliable 'settled' patterns. Interspecific spatial overlap was also calculated for field-derived data, separately for each species combinations on monthly occasions. Change in tidepool use (mesocosm experiment) Effect sizes were calculated for three aspects of habitat-wide space utilization in the mesocosm experiment: (i) number of 'displaced' individuals (i.e. those located outside the 24 aquaria in the mesocosm), (ii) upper/lower positioning (the ratio of individuals located in the upper and lower tidepools) and (iii) level of aggregation, expressed as Morisita's index (Iδ). In meta-analyses, the 'effect size' (D) is calculated by standardizing the difference between 'control' and 'experimental' groups (Gurevitch et al. 1992; Gurevitch & Hedges 2001). In our analysis, the mean values in the hetero-specific treatment (Xe) and that in the mono-specific treatment (Xc) correspond to the 'experimental' and the 'control' group, respectively. Thus, positive/negative values of D imply departures of the hetero-specific situations from the mono-specific ones. We calculated the effect size D using the pooled standard deviation (S) of the mono-specific and the hetero-specific treatments: where Se, Sc,Ne and Nc are the standard deviation (S) and the number of cases (N) of the hetero-specific treatment (e) and the mono-specific treatment (c), respectively. The term J corrects for bias due to small sample size: Growth in the mesocosm experiment Growth rates, expressed as (TL2−TL1)/TL1 (where TL1 and TL2 are length at the start and the end of experiment, respectively), were compared between the mono-specific and the hetero-specific treatment using the Kruskal–Wallis test with Scheffe's post hoc test. For B. fuscus, the Mann–Whitney U-test was used to compare the mono-specific treatment (sample size was reduced to 37 due to missing values) and the mixture with C. annularis; the data on B. fuscus + C. gulosus combination were incomplete due to typhoon interruption. There were no significant differences in the initial mean size of introduced fishes among all treatment groups (anova, P > 0·05). Standardized mean differences of growth rates between the hetero-specific and the mono-specific treatment were also calculated in the same way as mentioned above (change in tidepool use). In this analysis, mean growth in the hetero-specific treatment (Xe) and that in the mono-specific treatment (Xc) correspond to the 'experimental' and the 'control' group, respectively. Results Microhabitat use and overlap (laboratory experiment) On the microhabitat scale, the patterns of substrate use under mixed situations departed significantly from those expected under single situations for all combinations and times of day (chi-squared test with Bonferroni correction, P < 0·05) (Table 2). In contrast, the patterns of shelter use showed no such difference (Mann–Whitney U-test with Bonferroni correction, corresponding to P > 0·05) except for the C. annularis + C. gulosus pair by daytime. Realized values of intraspecific overlap in microhabitat use were significantly lower than would be expected from single situations in two Chaenogobius species and marginally so in B. fuscus (Fig. 2). In contrast, realized values of interspecific overlap showed no significant departure from expected values, except the C. annularis + C. gulosus pair by daytime. Ca + Cg was the only combination showing a clear difference in overlap pattern between day and night. Table 2. Comparison of microhabitat use between single and mixed treatments of tidepool fishes. Changes in microhabitat use when mixed with other conspecific/heterospecific individual are tested for each focal species by the chi-squared test for substrate use and the Mann–Whitney U-test for shelter use. *P < 0·05 with Bonferroni correction Focal species Mixed species + B. fuscus + C. annularis + C. gulosus χ2 U χ2 U χ2 U B. fuscus Day 39·1* 22 26·1* 32 77·7* 29·5 Night 75·6* 21 26·3* 15 68·7* 18·5 C. annularis Day 30·1* 8·5 50·2* 14·5 117* 0* Night 81·7* 30·5 44·9* 28·5 103* 23·5 C. gulosus Day 54·8* 13·5 95·2* 31·5 52·1* 19 Night 122* 15·5 1499* 23 1958·5* 28·5 Figure 2Open in figure viewerPowerPoint Expected (bars, mean with 95% range) and realized (circles, mean ± 1SD) overlap in microhabitat use for different combinations of fish. Bf, Bathygobius fuscus; Ca, Chaenogobius annularis; Cg, C. gulosus. Tidepool occupancy and overlap (mesocosm experiment) Tidepool occupancy of B. fuscus under interspecific mixtures showed no significant departure from the mono-specific situation, while C. annularis and C. gulosus showed some significant departures (Fig. 3). Chaenogobius annularis had a significant reduction in tidepool occupancy when mixed with either B. fuscus or C. gulosus. Realized occupancy of C. gulosus varied depending on paired species: it was smaller when mixed with C. annularis but tended to be similar/slightly larger when mixed with B. fuscus than would be expected under mono-specific situations. Realized total (combined) occupancy of the B. fuscus + C. gulosus pair was larger than the expected value, while values for B. fuscus + C. annularis and C. annularis + C. gulosus pairs were smaller. Figure 3Open in figure viewerPowerPoint Expected (open circles, mean with 95% range) and realized (filled circles) values of tidepool occupancy under three different hetero-specific treatments (a), (b) and (c) of the mesocosm experiment. Left and right panels of each figure show tidepool occupancy of each species and the middle panel shows total (combined) occupancy of two species, under the designated hetero-specific treatment. Last three observations of each treatment (designated as t12–t14, t8–t10) are shown. In all hetero-specific treatments, realized overlap values in the mesocosm were significantly smaller than would be expected from the mono-specific treatments and similar to observed values in the field (Fig. 4). The observed values of overlap in the field were apparently small compared with expected values of the mesocosm except for the June data of B. fuscus + C. gulosus combination. Observed intraspecific spatial overlap had similar magnitudes in the three species, being slightly shifted towards larger values than the expected interspecific overlap of B. fuscus + C. annularis and B. fuscus + C. gulosus pairs. Figure 4Open in figure viewerPowerPoint Inter- and intraspecific overlap in tidepool use in the mesocosm experiment. Expected (open circles, mean with 95% range) and realized (filled circles) interspecific overlap of each hetero-specific treatment are shown, alongside calculated intraspecific overlap (half-tone symbols with dotted lines, mean with 95% range). Last three observations of each treatment (designated as t12–t14, t8–t10) are shown. Dotted horizontal lines indicate field-observed interspecific overlap in June, July and August. Change in tidepool use (mesocosm experiment) Three gobiid species showed contrasting patterns of tidepool use under the hetero-specific treatment compared with the mono-specific one (Fig. 5). Chaenogo
Referência(s)