Mar Biol (2017) 164:24 DOI 10.1007/s00227-016-3053-1
ORIGINAL PAPER
Anemonefish personalities influence the strength of mutualistic interactions with host sea anemones Philip F. P. Schmiege1 · Cassidy C. D’Aloia2 · Peter M. Buston1
Received: 11 September 2016 / Accepted: 4 December 2016 / Published online: 30 December 2016 © Springer-Verlag Berlin Heidelberg 2016
Abstract The anemone–anemonefish mutualism is one of the most iconic marine mutualisms. For decades, anemonefishes have been known to protect anemones from predators, while anemones provide safe havens for anemonefishes. More recently, it has been suggested that the number of anemonefish influences the survival, growth, and asexual reproduction of anemones. Here, we build on those findings, investigating the effect of four variables (fish number, fish biomass, fish shyness, and anemone colony area), on anemone growth and asexual reproduction. The interaction between Amphiprion percula and Entacmaea quadricolor was used as a tractable system in a controlled aquarium setting. Fish and anemones were monitored in 60 tanks for 18 months, and we recorded all variables at 6-month intervals. We performed single-measure analyses and found that fish shyness, defined as the time spent in the vicinity of the anemone, significantly predicts anemone growth over the entire experiment. Further, we performed repeated-measure analyses and found that both fish shyness and initial anemone colony area significantly predict anemone growth per time period. These data suggest that behavioral variation among individual fish may be an important driver of anemone growth. More generally, this study highlights the importance of behavioral traits in mediating the strength of Responsible Editor: D. Goulet. Reviewed by N. E. Chadwick and undisclosed experts. * Philip F. P. Schmiege
[email protected] 1
Department of Biology, Boston University, 5 Cummington Mall, Boston, MA 02215, USA
2
Biology Department, Woods Hole Oceanographic Institution, 266 Woods Hole Road, Falmouth, MA 02543, USA
interspecific interactions such as mutualisms and suggests that such effects should be accounted for when investigating the dynamics of interacting populations.
Introduction Mutualisms are defined as interspecific interactions that benefit both interactors, potentially fostering biodiversity (Schmitt and Holbrook 2003; Bascompte and Jordano 2007). Some classic examples of mutualisms include flowers and their pollinators (Darwin 1859), leaf-cutter ants and their symbiotic fungi (Weber 1966), and plants and mycorrhizae (Medve 1978). While early investigations tended to focus on specialist mutualisms (involving only two species), there is growing evidence that the majority of mutualisms in the world are generalist mutualisms involving more than two species (Bascompte 2009; Stier et al. 2012). Thus, pairwise mutualistic interactions are embedded within a larger network of mutualistic interactions (Bascompte and Jordano 2013). The commonness and potential fragility of these mutualistic networks have affected how we think about conservation biology (Bascompte and Jordano 2007; Rezende et al. 2007; Bascompte 2009). The protection of mutualistic species must also include the protection of their mutualistic partners (Bronstein 2009). While many mutualism studies have focused on terrestrial systems, one of the most iconic generalist mutualisms is in the marine environment between sea anemones, anemonefishes, and zooxanthellae (Allen 1972; Fautin 1986, 1991; Fautin and Allen 1997; Ollerton et al. 2007). On a gross scale, the presence of all three partners has been found to be necessary for each other’s survival (Mariscal 1970b; Porat and Chadwick-Furman 2004). Zooxanthellae, dinoflagellates that reside in the anemone’s tissue, use
13
24 Page 2 of 9
carbon dioxide, water, and energy from the sun to produce carbohydrates for the anemone (Fautin 1991). Anemonefishes defend anemones from predators that might eat their tentacles, by chasing and biting the predators (Fautin 1991; Godwin and Fautin 1992; Fautin and Allen 1997). Anemones, in turn, defend anemonefishes from their own predators, via nematocysts in the tentacles. These stinging cells harm predators, while anemonefishes are immune to the effects of these stings due to their mucous coating (Mariscal 1970a; Brooks and Mariscal 1984; Mebs 1994). On a finer scale, the number of anemonefish is positively associated with the anemone’s growth and asexual reproduction (Godwin and Fautin 1992; Porat and Chadwick-Furman 2004; Holbrook and Schmitt 2005; Porat and Chadwick-Furman 2005; Frisch et al. 2016). There are three plausible proximate mechanisms underlying these effects: (1) anemonefish produce carbon, nitrogen, and phosphorus as waste products, which can be taken up by the anemones and zooxanthellae, leading to an increase in protein synthesis and zooxanthellae density (Fitt and Cook 2001; Roopin et al. 2008; Roopin and Chadwick 2009; Cleveland et al. 2011; Roopin et al. 2011); (2) anemonefish move within their host anemone’s tentacles, decreasing boundary layer thickness and increasing gas (oxygen) and nutrient exchange (Liberman et al. 1995; Szczebak et al. 2013); and (3) anemonefish defend their host anemone’s tentacles from predators, allowing the anemones to expand their tentacles for longer periods of time, thus enhancing nutrient acquisition from the water column and photosynthesis (Godwin and Fautin 1992; Porat and ChadwickFurman 2004). From the fish’s perspective, living in a large anemone is associated with higher growth rates, which in turn are associated with higher reproductive success, though whether this is a causal relationship remains unclear (Buston 2002; Buston and Elith 2011). In addition to these well-documented mechanisms focused on nutrient exchanges, individual behavioral traits, i.e., personality traits, may also play a key role in mediating the anemone–anemonefish mutualism. Personality traits are defined as behaviors that are consistent within an individual across time and context, though they can vary between individuals (Reale et al. 2007). Researchers have begun to note that individual behavioral variation can have profound impacts on various ecological phenomena, including the roles that those individuals play within their populations and communities (Bolnick et al. 2003; Sih et al. 2004; Reale et al. 2007; Bolnick et al. 2011). Classically, the study of interspecific interactions has considered individuals within each population to be identical, ignoring variation in individual behavior (Lomnicki 1988). Recently, terrestrial studies have shown that factors such as foraging behavior and diet breadth influence the strength of these interactions (Riechert 1991; Johnson et al. 2008; Pruitt and
13
Mar Biol (2017) 164:24
Krauel 2010). Moreover, when behavioral temperament is taken into account, these interactions can radically change, e.g., amensalism to a commensalism or mutualism (Pruitt and Ferrari 2011). Thus, the anemone–anemonefish mutualism offers an interesting system for exploring the potential effects of individual-level behavioral variation on a marine mutualistic interaction. While recent work has emphasized the correlation between fish number and host anemone growth, asexual reproduction, and survival (Porat and Chadwick-Furman 2004; Holbrook and Schmitt 2005; Mitchell and Dill 2005; Porat and Chadwick-Furman 2005; Frisch et al. 2016), additional alternative hypotheses remain to be tested. First, larger fish are known to produce more waste and nutrients (Roopin et al. 2008; Bermudes et al. 2010). Anemones may therefore grow and divide more if they host greater anemonefish biomass. Second, nutrients are transferred between the fish and the anemone via the water surrounding the anemone (Cleveland et al. 2011; Verde et al. 2015). Therefore, it is plausible that anemones will grow and divide more if they host fish that spend more time in the vicinity of the anemone. Fish shyness, defined here as the time spent in the vicinity of the anemone, is a consistent personality trait in anemonefishes (Wong et al. 2013) and many other fishes (Conrad et al. 2011). Finally, anemone size has been shown to be positively correlated with division in the sea anemone Anthopleura elegantissima in both the laboratory and the field (Sebens 1980, 1982). Therefore, it is plausible that such correlations will also exist in other anemone populations and should be controlled for in studies of the anemone– anemonefish interactions. In this study, we used a controlled laboratory setup in which we paired anemonefish (Amphiprion percula) with sea anemones (Entacmaea quadricolor). While this is not a naturally occurring interaction, previous studies have shown that E. quadricolor is a generalist anemone, interacting with many different species of anemonefish (Fautin 1986; Ollerton et al. 2007). Additionally, in the absence of a naturally occurring host, anemonefishes will associate with a surrogate host (Arvedlund and Takemura 2005). Therefore, this tractable study system helps to provide insight into generalist interactions between anemonefishes and host anemones. To test the effect of behavior on a mutualistic interaction, we studied variation in host anemone growth and asexual reproduction as a result of (1) fish number, (2) fish biomass, (3) fish shyness, and (4) anemone colony size. With this study, we build on previous findings by considering the effect of behavior on this iconic interaction. We find that variation in fish shyness is positively correlated with anemone growth. We aim to highlight the importance of including variation in behavioral traits in the study of ecological interactions.
Page 3 of 9 24
Mar Biol (2017) 164:24
Materials and methods Source populations The investigation was carried out at Boston University, Boston, MA, between December 2012 and July 2014. All clownfish, Amphiprion percula, used in this investigation were wild-caught in Papua New Guinea and supplied by Quality Marine. All fish were non-breeders when they arrived to the laboratory, and they remained non-breeders throughout the experiment. Removal or harvesting of non-breeders is considered to be a sustainable practice, because they do not contribute to population growth (Buston 2004; Planes et al. 2009). All anemones, E. quadricolor, used in this investigation were supplied by the Boston University Marine Program teaching laboratories: These are healthy anemone populations that have been in the BUMP teaching laboratories for many years (lamentably, their origin is unknown). While E. quadricolor is not a natural host of A. percula in the field, the individuals used in this experiment readily associated with each other in the laboratory. Housing Fish and anemones were housed in sixty 113.5-L (30 gallon) tanks (91 cm × 45 cm × 30 cm) divided evenly over four environmentally independent racks, resulting in 15 tanks per rack (rack IDs: A, B, C, and D). Each rack had its own continuous flow of recirculating saltwater (approximately 4400 gallons per hour; Instant Ocean Salt) and had its own pump (Reeflo Hammerhead), protein skimmer (My Reef Creations Pro 2), and UV water treatment system (30,000 uWs). The abiotic conditions in each rack were kept similar to those that would be found on coral reefs in Papua New Guinea: salinity—32.5 ± 1.58 ppt; temperature—27.3 ± 0.16 °C; pH—8.30 ± 0.34. The water quality, including salinity, pH, and temperature, was monitored 24/7 by a Profilux computer controller, and manual testing for dissolved phosphate and ammonia was done once every 2 weeks (Salifert and Red Sea test kits). Each tank had a standardized setup. The lighting consisted of two T5 24-W bulbs whose spectra color mimics the natural environment (Sunlight Supply T5 TekLights). These lights provided the anemones with an average photosynthetically active radiation (PAR) of between 150 and 170 umol quanta m−2 s−1 at the bottom of the tank and were on a timer giving the fish a 12-h day. This is similar to other laboratory studies (Roopin and Chadwick 2009), and also similar to field values for E. quadricolor which ranges between 50 and 700 umol quanta m−2 s−1 (Dixon et al. 2014). Each tank had one half inch of sand on the bottom,
and a 15-cm2 ceramic tile toward one end, with a reef rock placed on the tile. Because anemones tend to attach to reef rock, this setup allowed the location of the anemones in all the tanks to be controlled, minimizing the variation of abiotic factors, e.g., PAR or water currents, experienced by the anemones that might confound our results. Experimental setup Initially, 96 fish and 60 anemones were distributed into the 60 thirty-gallon tanks: 36 tanks had two fish; 24 tanks had one fish; and all tanks had one anemone. At the start of the experiment, the average fish size was 36.29 ± 2.70 mm in standard length (range 29.4–41.6 mm, n = 96) and 1.7 ± 0.36 g body mass (range 1.0–2.6 g, n = 96). The average anemone size was 18.58 ± 11.47 cm2 at the start of the experiment (range 6.17–72.82 cm2, n = 60). The fish and the anemones were paired haphazardly with no bias toward fish or anemone initial size. Fish were fed once per day, six days per week, with New Life Spectrum marine formula 1-mm pellets. The fish were observed during each feeding, and any uneaten pellets were siphoned out of the tank within the first hour after feeding so the anemones did not use those nutrients (sensu Roopin and Chadwick 2009). In an effort to further control anemone nutrient intake, anemones were not fed during this 18-month project. By not feeding them, we are able to more clearly test for the effect of fish behavior and fishderived nutrients on the anemones. Anemone metrics Anemone measurements were taken four times during the project: t0 in December 2012–January 2013, t1 in June 2013–July 2013, t2 in December 2013–January 2014, and t3 in June 2014–July 2014. The number of anemone clones in each tank was counted at each time. Changes in clone number (defined here as asexual reproduction) were calculated by subtracting the number of anemones at time n from the number of anemones at time n + 1 (sensu Holbrook and Schmitt 2005). Anemone colony size was obtained by measuring the tentacular crown length and width of each clone with calipers, estimating surface area as an ellipse, and then combining these calculations for total colony area (Buston 2003a; Porat and Chadwick-Furman 2004; Holbrook and Schmitt 2005; Cleveland et al. 2011). At each time point, the tentacular crown was measured on three days, with a day in-between each measurement. The mean was used for statistical analyses to account for day-to-day variation in the anemone’s behavior, i.e., how contracted or elongated its tentacles were. Anemone colony growth was calculated by subtracting total colony area at time n from
13
24 Page 4 of 9
the total colony area at time n + 1 (sensu Holbrook and Schmitt 2005). As long as anemone clones remained on the rocky island, they were kept in the tank so we could observe total growth over the 1.5 years. If anemones left the rock during a given time period, they were excluded from that time period, placed back on the rock for reattachment, and used for subsequent time periods if they remained attached to the rock. A total of 17 out of 60 tanks were removed from the single-measure analysis, while 27 out of 180 tanks were removed from the repeated-measure analysis. Fish metrics Fish measurements were taken at each time point (t0, t1, t2, t3), once the anemone measurements were completed. The number of fish in each tank was counted at each time point. Fish mass was measured by weighing the fish individually to 0.1 g using an electronic microbalance. The time that each fish spent within one body length of the anemone, i.e., fish shyness, was measured by videotaping each fish for 12 min in the morning (0900–1200) and in the afternoon (1300–1700). The cameras used were Kodak Playsport 1080p and were set up on tripods facing the tanks. Because a blind was not set up around filming, the first 2 min of each video was disregarded to allow for acclimation (sensu Wong et al. 2013), and no one entered the laboratory during filming. A period of 10 min has been shown to be sufficiently long enough to measure this behavioral trait and demonstrate its consistency in Amphiprion ocellaris (Wong et al. 2013) and A. percula (Medina and Buston unpublished data). The videos were then watched once for rank 1 individuals and once for rank 2 individuals. The rank 1 and 2 fish were easily identifiable because of unique stripe markings on each fish that were recorded when the fish entered the laboratory (Nelson et al. 1994; see Electronic Supplementary Materials of Buston 2003b). In addition, rank 1 fish are larger and would eventually become the dominant breeding female, while rank 2 fish are smaller and would become the dominant breeding males in the clownfish hierarchy (Buston 2003c). We averaged the morning and afternoon measures of shyness for each individual, i.e., rank 1 shyness = (rank 1 morning shyness + rank 1 afternoon shyness)/2. The metric ‘shyness’ used in this study was the sum of rank 1 shyness and rank 2 shyness, i.e., the combined amount of time that the two fish spent within one body length of the anemone. Average time spent in close proximity to the anemone was the average of t1, t2, and t3 (t0 could not be used due to an incomplete dataset for this measure at this time). Over the course of this experiment, only 2 fish died, and these tanks were excluded for the duration of the investigation.
13
Mar Biol (2017) 164:24
Statistical analysis Of the initial 60 tanks, 51 were usable for time period 1 (t0– t1), 54 were usable for time period 2 (t1–t2), 48 were usable for time period 3 (t2–t3), and 43 were usable across the entire year and a half. This difference in numbers was either due to fish dying (n = 2) or anemones moving off rocks (all others) and thus elimination from that specific time period. For the single-measure analysis, we used the latter 43 groups to predict anemone growth and asexual reproduction across the year and a half; for the repeated-measure analysis, we used data from t0, t1, and t2 to predict anemone growth and asexual reproduction across each 6-month time period. All statistical analysis was conducted in R v.3.2.4 (R Core Team 2016). First, we investigated the effect of the anemonefish on the growth of their host anemones using linear regression models. Specifically, we investigated the relationship between the independent variables fish number, fish biomass (g), fish shyness (s), initial anemone colony size (cm2), rack ID (A, B, C, or D), and the dependent variable anemone growth (cm2). For the single-measure analysis, we investigated the effect of these variables measured at t0 on anemone growth over the entire 18 months; for the repeated-measure analysis, we investigated the effect of these variables measured at t0, t1, and t2 on anemone growth over each 6-month time period. Tank ID was added as a random effect for the repeated-measure analysis to control for the lack of independence among the three measures from each tank. Second, we investigated the effects of the anemonefish on the asexual reproduction of their host anemones using logistic regression models. Specifically, we investigated the relationship between the independent variables fish number, fish biomass, fish shyness, initial anemone colony area, rack ID, and the dependent variable anemone asexual reproduction. Anemone reproduction was treated as a binary variable with the anemones either dividing (1) or not (0). For the single-measure analysis, we investigated the effect of these variables measured at t0 on reproduction over the entire 18 months; for the repeated-measure analysis, we investigated the effect of these variables measured at t0, t1, and t2 on reproduction over each 6-month time period. Tank ID was added as a random effect for the repeated-measure analysis. For all analyses, we used backwards stepwise regression with an information theoretic model selection framework. Models were ranked based on their corrected Akaike information criterion (AICC) scores. We initially considered all models within 2 AICC units of the top-supported model. Next, considering this subset of top models, we adopted a parsimony approach and excluded models with predictors that explained a negligible amount of variation in the dependent variable, i.e., uninformative predictors (sensu Arnold 2010).
Page 5 of 9 24
Mar Biol (2017) 164:24
Results
Over this 18-month investigation, the average increase in total anemone area in each tank was 14.80 ± 14.29 cm2 and 88.37% of anemones grew, while 11.63% of anemones shrank. The linear regression analysis revealed that of the four independent variables (fish number, fish biomass, fish shyness, and anemone colony area) fish shyness was the sole significant predictor of anemone growth over the three time periods (P = 0.012, R2 = 0.12; Fig. 1). On average, anemones grew an additional 0.02 cm2 over the full 18 months for every additional second in the 10-min trial period that the fish spent in the anemone. The repeated-measure analysis revealed that fish shyness, initial anemone colony area, and time period were all significant predictors of anemone growth (R2 = 0.31; Fig. 2; Table 1). Anemones grew more when paired with fish that spent more time in the anemones (Fig. 2a). On average, colony surface area grew an additional 0.01 cm2 in 6 months for every additional second the fish spent in the anemone (Table 1). In contrast, anemones grew less when they were larger (Fig. 2b). For every additional cm2 in initial size, anemones grew 0.15 cm2 less (Table 1). Finally, we observed an effect of time period as anemones grew more in time period 1 (t0–t1) than in the subsequent two time periods (Fig. 2c; Table 1). Anemone asexual reproduction Over this 18-month investigation, 55.81% of anemones reproduced asexually. Over the course of 1.5 years, we lost 3 anemones, 28 anemones did not divide, 25 anemones reproduced once, 4 reproduced twice, and 1 anemone reproduced three times, revealing an average asexual reproduction rate of 0.63 ± 0.69 polyps/anemone per 18 months (the study population increased from 60 polyps at t0 to 93 polyps at t3). For anemone asexual reproduction, the logistic model analysis revealed that of the four independent variables (fish number, fish biomass, fish shyness, and anemone colony area) only initial anemone colony area was marginally significant in predicting anemone reproduction across the full 18 months (P = 0.074, R2 = 0.25; Fig. 3). As initial anemone colony area increased, there was an increased probability of anemone asexual reproduction over the timeframe of the entire experiment (Fig. 3). In contrast, none of the four independent variables predicted anemone asexual reproduction during each individual time period in the repeated-measure analysis (see the global model in Table 2).
Predicted Anemone Growth (cm 2)
Anemone Growth
60
40
20
0
0
300
600
900
Fish Shyness (sec)
Fig. 1 Predicted anemone growth as a function of fish shyness over the course of the entire experiment (18 months). Maximum possible value for fish shyness is 1200 s: It is the sum of the time spent in the vicinity of the anemone in 600 s by two fish. Fitted line, y = 0.021x + 4.221, is based on the best-fit linear model. Fish shyness was a significant predictor of anemone growth in the singlemeasure analysis (coefficient = 0.021; SE = 0.008; t value = 2.645; P value = 0.012)
Discussion There is a growing appreciation among population ecologists that intraspecific behavioral variation among individuals within populations can influence interspecific interactions between two or more species (Pruitt and Ferrari 2011). To understand how individual variation in behavior and other factors might combine to influence the strength of a marine mutualism, we investigated the interaction between an anemonefish (Amphiprion percula) and a surrogate host sea anemone (E. quadricolor). We investigated the effect of four variables (fish number, fish biomass, fish shyness, and anemone colony size) on anemone growth and asexual reproduction. We found that initial colony size was a significant predictor of anemone colony growth and a marginally significant predictor of anemone asexual reproduction. The association between anemone size and anemone growth was negative, indicating that small anemones grew more than large anemones (Fig. 2b). The association between anemone size and the probability of asexual reproduction was positive, indicating that large anemones had a greater propensity to divide than smaller anemones (Fig. 3). The latter result is consistent with past research, which demonstrated that large anemones tended to divide more readily than small anemones in low nutrient environments (Sebens 1980, 1982). This study indicates that future studies
13
24 Page 6 of 9
Mar Biol (2017) 164:24 Table 1 Repeated-measure analysis of anemone growth over each time period
(a) 17.5 15 12.5 10
250
750
500
1000
1250
Fish Shyness (sec)
(b) Predicted Anemone Growth (cm2)
Estimate
SE
t value
P value
Intercept Shyness Area Time period 2
12.463 0.010 −0.150 −12.582
3.477 0.004 0.064 3.061
3.584 2.759 −2.352 −4.111
