Host anemone size as a determinant of social group size and structure in the orange clownfish (Amphiprion percula)

Study site and species

This study was undertaken in April 2016 on the inshore coral reefs near the Mahonia Na Dari Research and Conservation Centre, near the town of Kimbe, in Kimbe Bay, Papua New Guinea (05°26′S, 150°52′E) (Figs. 1A and 1B). It focused on an area of the fringing reef approximately five km long as well as 19 discrete emergent reefs ranging a distance of 0.2–1.4 km from the coastline (Fig. 1C). Kimbe Bay experiences a distinct wet season from December to January, and a windy season from June to July, but there is little variation in temperature year-round (Srinivasan, 2003).

This study focused on the orange clownfish, A. percula, which is associated primarily with Heteractis magnifica and Stichodactlya gigantea anemones in this region, usually at depths shallower than 12 m (Almany et al., 2007; Planes, Jones & Thorrold, 2009). Due to the relatively low numbers of S. gigantea on coastal reefs, this study focused primarily on A. percula groups associated with H. magnifica. A. percula lives in small social groups with a strong dominance hierarchy, whereby individuals of sequential rank follow a fixed size ratio (Buston, 2003a, 2003b, 2003c). Reproduction in A. percula is also restricted to the two highest rank individuals: the highest rank generally being the female and the second highest rank generally being the male, if they have achieved the size required for sexual maturation (Buston & Elith, 2011). Recruitment to the group is restricted until the lowest ranked juvenile reaches a critical size (Elliott, Elliott & Mariscal, 1995; Buston, 2003b). Because this study did not incorporate gonad analyses, and the length of certain largest individuals were potentially below the length for reproductive maturity, individuals measured are referred to by their size ranking: rank 1 (the largest individual), rank 2 (the second largest individual), and rank 3 (the third largest individual).

The research in this study was conducted in compliance with the Queensland Animal Care and Protection Act 2001, and with the consent of the Mahonia Na Dari Research and Conservation Center, and the Local Marine Management Committee and traditional owners in Kilu.

Field methods

The majority of the A. percula groups were located on the fringing reef and 19 adjacent reefs by systematic searches of all the reef area shallower than 20 m (Figs. 1B and 1C). The depth and GPS coordinates were recorded for all located anemones and A. percula groups. For each anemone, a photograph of the entire anemone, with calipers for scale, was taken to estimate the effective surface area. The number of A. percula in the anemone was recorded. All individuals in each social group were caught using hand nets, and the total length (TL) of each individual was measured using plastic calipers before releasing it back onto the host anemone.

Environmental variables

All anemones were mapped using QGIS software, and distance from shore was calculated as the linear distance between each anemone and the closest point on the shore. Using Digital Earth Watch software, tentacle crown surface area was calculated from the anemone photographs (Dixon, McVay & Chadwick, 2017). This process involved virtual drawing along the contour of the anemone, including specific tentacles, to capture the full extent of the surface area. Anemones that appeared to have contracted tentacles and oral discs when photographed were not included in statistical analyses of anemone size, as measurements of surface area would have been underestimated.

Group size and structure

Four variables were measured to quantify the social group structure of A. percula: (1) social group size, or the number of fish per anemone, (2) TL of rank 1 (the largest individual), (3) TL of rank 2 (the second largest individual), and (4) TL of rank 3 (the third largest individual). Only the lengths of the three largest individuals were assessed, because few A. percula groups contained more than three individuals per anemone in the area surveyed.

Effects of environmental variables on group size and structure

Separate linear regressions between each environmental variable (anemone size, depth, and distance from shore) and each social structure variable (group size, TL of rank 1, TL of rank 2, and TL of rank 3) were fitted. The environmental variables were log transformed to ensure a linear relationship with social structure variables and to ensure the data conformed to the assumptions of linear regression. The potential relationships between anemone size and the other environmental variables (depth and distance offshore) were also assessed.

As there were significant correlations between environmental variables, correlation coefficients were calculated and compared against thresholds of 0.7–0.8, levels recommended for conducting regressions without multicollinearity posing an issue (Green, 1979; Menard, 2002). Subsequently, the variance inflation factor (VIF) was calculated using the equation: VIFj = 1/(1−R_j²). The VIF was then compared against thresholds of 4–10, below which multicollinearity would not be a concern (Harrell, 2001; Guisan, Edwards & Hastie, 2002; Hair et al., 2006).

Effect of rank 1 on social structure

The possible effects of the size of the rank 1 individual on group size and structure were analyzed using separate univariate regressions, with length of rank 1 as the independent variable, and the social group size, length of rank 2, and length of rank 3 as the dependent variable in each regression. The length of rank 1 was used as the social control factor, as previous studies have shown that female size (always the rank 1 individual) is a significant predictor of social group size and the size of subordinate individuals (Buston, 2003a; Mitchell & Dill, 2005).

Path analysis

A path analysis was conducted using structural equation modelling to determine the likely pathways that link anemone size to social group size and structure. Path analysis imposes directional causation from one variable to the next and includes the magnitude of such effects. Relationships in nature are often not solely unidirectional and, as in the case of the relationship between anemone and anemonefish, may involve feedback loops where both members yield influence on each other. Nevertheless, path analysis allows exploration of the strength of interactions between many interacting factors in a potentially complex model (Streiner, 2005). Therefore, use of the path analysis in this study provides a useful platform for assessing the magnitude of both environmental and social influences on the group structure of anemonefish.

The path analysis in this study was conducted in R using the Lavaan package and devised using the results from the individual regressions as a starting point. The causation directions were informed both from previous literature on the hierarchy of A. percula group structure (Buston, 2003a) as well as a basic assumption that anemone size may be influenced by both depth and distance offshore. The fit of the model was evaluated by examining the R² of the variables and comparing fit indices to accepted thresholds (e.g., a Comparative Fit Index of greater than 0.95 and a standardized root mean square residual below 0.08) (Browne & Cudeck, 1993; Hu & Bentler, 1999). Direct effects were determined by examining the path coefficients. Indirect effects were assessed by first determining the compound path and subsequently multiplying the coefficients of the component paths. The use of path analysis in this study is an a posteriori approach, fitting a model to a known data set for explanatory analysis, as opposed to creating a model prior to examining the data available.

General patterns

A total of 98 H. magnifica with resident fish were mapped and sampled on the 19 inshore reefs. Six of these anemones were not included in statistical analyses of anemone size as they were contracted when photographed, and therefore measurements of surface area would have been underestimated. The distances from shore of the 92 anemones included in statistical analyses ranged from 185 to 1,400 m, and the depths occupied ranged from 3.3 to 19 m. The size of A. percula social groups ranged from two to five individuals per anemone.

Are group size and structure positively related to anemone size?

Social group size, body length of rank 1, body length of rank 2, and body length of rank 3 all increased significantly with an increase in the log of anemone surface area (Fig. 2). Anemone size explained 28.3% of the variation in group size, 72.4% of the variation in rank 1 length, 65.5% of the variation in rank 2 length, and 48.25% of the variation in rank 3 length. All linear regressions were statistically significant (F_1,90 = 35.7, p < 0.001; F_1,90 = 236.1, p < 0.001; F_1,88 = 167.7, p < 0.001; F_1,72 = 67.14, p < 0.001).

Does anemone depth or distance offshore explain variation in group size and structure, and can this be attributed to spatial variation in anemone size?

There were significant, weak positive relationships between depth of the anemone and the body lengths of rank 1 rank 2, and rank 3 (F_1,90 = 17.3, p < 0.001; F_1,88 = 6.591, p = 0.012; F_1,72 = 5.054, p = 0.028; Fig. 3). These relationships explained 16.1% of the variation in rank 1 body length, 7.0% of the variation in rank 2 body length, and 6.6% of the variation in rank 3 body length, respectively. There was no significant relationship between social group size and depth of the anemone (although there was a positive trend).

Distance from shore had no significant effect on group size, the body length of rank 2, or the body length of rank 3 (Fig. 4). However, there was a weak positive relationship between body length of rank 1 and the log of the distance from shore, which explained 8.2% of the variation in the length of rank 1 (F_1,90 = 7.991, p = 0.006; Fig. 4).

There were significant positive relationships between the log of the anemone surface area and the log of the distance from shore, and the log of the depth (F_1,90 = 7.515, p = 0.007; F_1,90 = 20.35, p < 0.001 values; Fig. 5). Depth and distance from shore were weakly correlated with each other (r = 0.22, t_2,90 = 2.155, p = 0.034). A multiple regression with depth and distance from shore as independent variables explained 21.9% of the variation in anemone surface area (F_2,89 =12.51, p < 0.001). The VIF factor was 1.28, indicating that the collinearity among the environmental factors did not limit the ability to distinguish among variables.

Are the group size, and the size of the second and third largest individuals related to the size of the largest individual?

Social group size, body length of rank 2, and body length of rank 3 each increased significantly with an increase in body length of rank 1 (F_1,90 = 36.74, p < 0.001; F_1,88 = 315.4, p < 0.001; F_1,72 = 152.2, p < 0.001; Fig. 6). Rank 1 body length explained 29.0% of the variation in group size, 78.2% of the variation in rank 2 length, and 67.9% of the variation in the length of rank 3.

Does anemone size directly determine group size or is this mediated by the size of the largest individual?

The model produced by path analysis explained between 58% and 78% of the variation in the continuous group structure variables. The model fit the data well, with a Comparative Fit Index of 0.992, a Tucker-Lewis Index of 0.986, a root mean square error of approximation of 0.052, and a standardized root mean square residual of 0.051 (Fig. 7). Depth and distance from shore maintained weak, but direct, effects on anemone surface area. Anemone surface area had a strong direct effect on rank 1 body length, and rank 1 body length had strong, direct effects on the rank 2 body length and rank 3 body length. Through its effects on subordinate lengths, rank 1 maintained an indirect, yet important effect on social group size. Within this model, the direct effect of the anemone surface area on rank 1 body length was the strongest, direct interaction. However, anemone surface area also had positive, indirect effects on social group size and structure, via its effect on rank 1 body length (Fig. 7).

Critical importance of anemone size

Our results confirm those of previous studies showing that anemone size has a major influence on the size and structure of social groups of anemonefishes (Fricke, 1980; Fautin, 1992; Mitchell & Dill, 2005; Dixon, McVay & Chadwick, 2017). Anemone size was a good predictor of group size, and the lengths of rank 1, rank 2, and rank 3 individuals present on the anemone. The strong relationship between anemone size and rank 1 length is consistent with previous work (Elliott & Mariscal, 2001; Mitchell & Dill, 2005). Buston (2003a) did not find this relationship for A. percula in Madang, Papua New Guinea, suggesting some spatial variation in this phenomenon. It is important to note that the strength of this interaction is likely to be reinforced by reciprocal benefits that anemonefish bestow on their hosts, including defense (Porat & Chadwick-Furman, 2004; Godwin & Fautin, 1992) and provision of nutrients (Porat & Chadwick-Furman, 2005; Cleveland, Verde & Lee, 2010; Roopin & Chadwick, 2009; Roopin et al., 2011).

Role of depth and distance offshore

Although the regression analysis suggested depth did not have a direct effect on social group size, deeper anemones certainly supported larger rank 1 and rank 2 individuals. Because the highest rank individuals tend to be the breeding pair, and fecundity increases with body size (Saenz-Agudelo et al., 2015), this relationship may have important consequences for reproductive potential. The depth effect is most likely related to the reduction in light with increasing depth. It is known that anemones have diel contraction patterns in response to the availability of light, with tentacles that have high densities of zooxanthellae responding to levels of light intensity (Sebens & DeRiemer, 1977). Anemones in shallow areas may retain retracted tentacles due to the presence of potentially high levels of photosynthetically active radiation, which can adversely affect the anemone and its Symbiodinium by inhibiting photosynthesis (Mass et al., 2010). Anemones at greater depths may conversely expand their oral disc to capture more light for photosynthesis, as light intensity is reduced at greater depths. However, anemones have also been shown to increase the concentration of their Symbiodinium at depth as a way to increase their photosynthetic capability (Dixon et al., 2014), and thus an alternative explanation is that anemones might actually be larger at greater depths, due to reduced wave action, increased nutrient availability, or other environmental parameters (Sebens, 1982). Either way, the outcome is a greater surface area available for A. percula on deeper anemones.

Distance from shore also had a weak positive effect on rank 1 length, again, most likely as a result of the effect of distance offshore on anemone surface area. Anemones that were further offshore tended to be slightly larger. The presence of smaller anemones closer to shore may be due to the likely higher levels of sedimentation near the fringing reefs. Increased sediment has previously been linked to a reduction in photosynthetic ability and an increase in tissue bleaching in anemones (Goreau, 1964; Rogers, 1990). Anemones produce mucus to remove sediment (Patton, 1979), which may be energetically costly and result in reduced growth in waters with higher sediment loads. Higher levels of sediment could also cause the anemone to contract to prevent accumulation of sediment. Conditions further offshore may also foster more favorable environments in terms of nutrient availability, current patterns, and, more broadly, a reduction in anthropogenic stressors (Smith et al., 2008). One or both of these mechanisms would lead to a decrease in anemone surface area with proximity to the shore and a smaller area for A. percula to inhabit.

Role of the largest individual

The size of the rank 1 individual (in most cases—the female) exhibited strong relationships with rank 2 body size, rank 3 body size, and the group size. This is likely the outcome of the strong social hierarchy in anemonefish groups, and our results are consistent with other work on anemonefishes (Fricke & Fricke, 1977; Fricke, 1979, 1983; Godwin, 1994; Buston, 2003a; Iwata et al., 2008). Mitchell & Dill (2005) concluded that social group size in the sister species, A. ocellaris, is predominantly controlled by the dominant female (rank 1), based on the effect of female size on the size structure of the group. Buston (2003a) has previously established the importance of social control on individual body lengths in A. percula, with a fixed size difference between fishes that are sequentially ranked in the social hierarchy. With such fixed size ratios, the larger the rank 1 individual, the more individuals can be accommodated in the group. The mechanism controlling group size in anemonefish is likely forcible eviction or competitive exclusion (Buston, 2003b), leading to excluded fish finding refuge in potentially less preferable hosts (Huebner et al., 2012). Lower social rank has been associated with lower survivorship in A. percula, likely due to competitive eviction by higher ranked individuals (Buston, 2003c). Subordinate individuals may in turn be reducing their food intake to avoid exceeding size thresholds that could lead to eviction, as observed in other space-limited fishes (Wong et al., 2008). Hence, the size of the group and its structure may be determined by complex top-down and bottom-up interactions between competing individuals in the social hierarchy, including the species of host anemone.

Distinguishing direct and indirect effects

It is very difficult to tease apart the relative importance of environmental factors (e.g., anemone size) and social factors (e.g., size of individuals by rank) in determining social structure variables, such as group size or the size of the subordinate fishes, since all the variables are correlated. Anemone size may directly influence group size through strict space limitation and associated density-dependent processes, or indirectly via the effect of anemone size on rank 1 size (Mitchell & Dill, 2005). Buston (2003a) concluded that the dominant female’s body length (rank 1 length) controlled the body lengths of the subordinates, because anemone diameter did not have a significant effect. Mitchell & Dill (2005) reached similar conclusions on the basis of stronger relationships with female size, compared with anemone size. Our path analysis strongly suggested that both direct and indirect effects are involved. Anemone surface area had a strong direct effect on rank 1 size and a slightly weaker effect on rank 2 size. The size of rank 1, in turn had a strong direct effect on the sizes of rank 2 and rank 3, while maintaining a substantial indirect effect on social group size through its effects on subordinates. Hence, anemone area had considerable indirect effects on all of the variables governing group size and structure. Further, the total effect (direct + indirect) of anemone size on rank 2 size was comparable to the direct effect of rank 1. The path analysis also demonstrated that the environmental variables (distance from shore, depth) have indirect effects on all four group structure variables through their influence on anemone surface area, although the effect of depth is arguably greater than that of distance from the coastline.

Future directions

While the path analysis helps resolve the likely cause-effect links and direct vs indirect relationships, ultimately this complexity will need to be resolved by experimentation. Experiments have previously demonstrated that anemonefish provide important benefits to their hosts through a multitude of mechanisms. While Godwin & Fautin (1992) demonstrated the importance of A. percula for protection of anemones, other potential benefits, such as nutrient transfer and aeration, have yet to be tested for this species and need be resolved. Observation of the behavior of different individuals in such experimental designs may also aid in revealing the exact mechanisms responsible for control of group size and structure.

Further investigation is needed to understand the reasons underlying the effects of depth and distance offshore on social group structure. Variations in depth are coupled with variations in light intensity, temperature, and pressure (Levinton, 1982). Similarly, distance from the coastline often varies with other factors such as nutrient concentration, sediment runoff, and anthropogenic impacts (Fabricius, 2005; Wenger et al., 2015). It would be tractable to experimentally manipulate other environmental variables, such as light intensity and temperature, to fully understand how each variable influences the host anemones, and consequently, the structure of A. percula social groups.

There is increasing evidence that shallow water anemones experience severe bleaching after warm water events (Saenz-Agudelo et al., 2011; Hobbs et al., 2013). Bleaching not only reduced size and abundance of anemones (Hobbs et al., 2013), but further resulted in lower egg production and recruitment of anemone fish (Saenz-Agudelo et al., 2011). Higher temperatures have also been linked to breeding disruption, egg degradation, and possible connectivity among anemonefish populations (Munday et al., 2008, 2009). These effects are likely to have flow-on effects on the structure of clownfish social groups. Further attention should be directed to evaluating ongoing effects of climate change on the ecology and behavior of fishes that are so closely linked to the quality of their habitat.