Behavioral differences within and among populations of an African cichlid found in divergent and extreme environments

Behavioral differences within and among populations of an African cichlid found in divergent and... Animals are increasingly faced with human-induced stressors that vary in space and time, thus we can expect population-level divergence in behaviors that help animals to cope with environmental change. However, empirical evidence of behavioral trait divergence across environmental extremes is lacking. We tested for variation in behavioral traits among 2 populations of an African cichlid fish (Pseudocrenilabrus multicolor victoriae Seegers, 1990) that experience extremes of dissolved oxygen (DO) and turbidity and are known to vary in a number of physiological and life history traits associated with these stressors. Using a common garden rearing experiment, F1 pro- geny from wild-caught parents originating from a swamp (low DO, clear) and a river (high DO, tur- bid) were reared in high DO, clear water. Predator simulation assays were conducted to test for (1) variation in boldness, general activity, and foraging activity between populations, (2) differences in correlations between behaviors within and across populations, and (3) repeatability of behaviors. There was strong evidence for divergence between populations, with swamp fish being more bold (i.e., leaving refuge sooner after a simulated predator attack) and active (i.e., spent more time out of refuge) than river fish. Across populations there were positive correlations between foraging ac- tivity and both boldness and general activity; however, within populations, there was only a strong positive relationship between foraging activity and boldness in the river population. Here, we have demonstrated that populations that originate from drastically different environments can produce progeny that exhibit measurable differences in behaviors and their correlated relationships even when reared under common conditions. Key words: behavioral correlation, behavioral syndrome, boldness, hypoxia, pace of life, turbidity. Environmental heterogeneity can have a strong impact on phenotyp- et al. 2004; Mittlebach et al. 2014; Mitchell and Biro 2017). And ic diversity, driving divergence among populations through plasticity like many traits, aspects of personality (e.g., boldness, aggressive- and local adaptation (Hendry 2015; Edelaar et al. 2017). As an en- ness) can be under strong selection when faced with environmental vironment shifts through time and space, behavioral traits are often extremes (Dubuc-Messier et al. 2017). For example, Brown et al. a species’ first line of defense in response to change. This is especially (2005) found that populations of a poeciliid fish Brachyrhaphis epis- true under increasingly rapid, human-induced environmental change copi were consistently bolder if they were from high-predation ver- (Sih et al. 2011). Furthermore, we now understand that behavior is sus low-predation sites across multiple rivers. Thus, we might often a repeatable and heritable trait and that individuals consistent- expect populations to diverge in personality when they experience ly differ in personality within populations of a single species (Sih opposite ends of environmental extremes. V C The Author(s) (2018). Published by Oxford University Press. 1 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 2 Current Zoology, 2018, Vol. 0, No. 0 Personality traits have been increasingly found to covary with excessive sediment run-off that is pulsed into the system during in- physiological traits and together may promote species persistence tense rain events. Swamps, on the other hand, tend to offer more sta- under environmental extremes (Dingemanse and Wolf 2010; Sih ble environmental conditions due the buffering effects of dense et al. 2015). For example, individuals that display high aggressive- aquatic vegetation. ness, high activity, and are bold (e.g., proactive types) have higher Populations of P. multicolor show strong phenotypic differences resting metabolic rates than less aggressive, less active, and shy indi- in morphological, physiological, and life-history traits in response to viduals (e.g., reactive types) (Careau et al. 2008; Huntingford et al. extreme differences in DO and turbidity. For example, in response 2010). Boldness, in particular, has also been shown to positively to low DO, P. multicolor from swamps tend to have on average covary with metabolic scope (e.g., the difference between resting 10% smaller brains and 56% larger gill surface area (Chapman and and active metabolic rates) (Killen et al. 2014; Binder et al. 2016). Chapman 2003; Chapman et al. 2008; Crispo and Chapman 2010; For example, bold bluegill sunfish Lepomis macrochirus have Weins et al. 2014). Swamp fish also have lower resting metabolic greater metabolic scope and exhibit greater aerobic capacity for rates and higher metabolic scope, the latter likely afforded by larger locomotor activity than shy sunfish (Binder et al. 2016). These asso- gill surface areas and thus a higher capacity for metabolic perform- ciations have provided empirical evidence for the pace-of-life syn- ance (Reardon and Chapman 2010). Fish from the swamp and those drome which predicts that behavior, physiological, and life-history reared under hypoxic conditions also have smaller broods and brood traits coevolve in response to correlational selection pressures (Re ´ ale for a shorter period of time compared with populations from high et al. 2007, 2010). Spatial and temporal heterogeneity in environ- DO, turbid rivers (Reardon and Chapman 2009). The pace-of-life mental conditions thus has the potential to create geographic vari- hypothesis would therefore predict that P. multicolor from low DO/ ation in a suite of behavioral, physiological, and life-history traits low turbidity swamp populations would fall closer to the slow end (Atwell et al. 2014; Montiglio et al. 2014). Understanding the role of the pace-of-life continuum, and river fish (or fish reared under of environmental extremes in shaping population divergence in a normoxic conditions) closer to the fast end of the continuum. suite of traits may be particularly important for understanding how However, by rearing fish from both populations in a common envir- species adapt to changing environments. onment free of low DO and turbidity stressors, we predicted that In aquatic ecosystems, hypoxia (i.e., low dissolved oxygen, DO) swamp fish would show behaviors more consistent with a relatively and turbidity (i.e., suspended particulates) are metabolically and faster pace of life due to their adaptations for life under energetically visually challenging conditions, respectively, that have been shown constraining low DO conditions. From a physiological standpoint, individually to generate divergence in behavioral, life-history, and P. multicolor from swamp populations show enhanced metabolic physiological traits in aquatic organisms (e.g., van der Sluijs et al. abilities when tested under high DO conditions, compared with river 2011; Chapman 2015). For example, in contrast to normoxic envi- fish (Reardon and Chapman 2010). Thus, swamp fish should be, on ronments, hypoxic environments tend to favor fish that are less ac- average, more bold and active than high DO/high turbidity river tive (Abrahams et al. 2005; Gotanda et al. 2011; McNeil et al. populations when reared under common garden conditions. This 2016), have relatively larger gills and smaller brains (Chapman et al. might be especially true if river fish also have higher perceived pre- 2008; Crispo and Chapman 2010), lower metabolic rates (Reardon dation risk in a clear environment compared with the turbid envi- and Chapman 2010; Crocker et al. 2013), and shorter brooding ronments they typically encounter and thus may reduce activity, times due to the challenge of extracting oxygen at very low concen- increase vigilance, and be shyer in clear waters than they would nor- trations. Environments with anthropogenically elevated turbidity mally be. Precedent use of common garden experimental designs to levels also impose a suite of potential stressors on fish (e.g., Bruton reduce the effects of environmental stressors when testing for per- 1985; Newcombe and MacDonald 1991) and notably pose visual sonality and pace of life syndromes have been suggested and applied challenges for aquatic organisms (e.g., Utne-Palm 2002; van der in previous studies. For example, Ursza ´ n et al. (2015) used agile frog Sluijs et al. 2011). For example, Dugas and Franssen (2011, 2012) tadpoles Rana dalmatina as a model organism to provide evidence found that red shiner Cyprinella lutrensis populations from turbid for pace of life and behavioral consistencies at different ontogenetic waters have larger eyes and more intense nuptial color patterns than stages. Additionally, a common garden experimental design has clear-water populations, suggesting that such adaptions allow for been used to study the pace of life of stonechat Saxicola torquata better vision and more visible signals, respectively, when the water populations (Wikelski et al. 2003). By raising individuals from di- is turbid. In contrast to other traits, however, variation in personal- vergent environments in a common garden experimental design, en- ity across extremes of these 2 environmental stressors has been over- vironmental stressors originally present in the home environment looked in recent literature. are released from these individuals, thus any observed differences In this study, we compared behavioral traits (e.g., general activ- may allow us to draw some conclusions about the adaptive nature ity, foraging behavior, and boldness under predation risk) of 2 popu- of trait divergence. lations of the haplochromine cichlid Pseudocrenilabrus multicolor Finally, boldness and activity have frequently been found to be victoriae Seegers, 1990, from divergent environmental conditions. repeatable (i.e., consistent individual differences in behavior; Bell We also examined whether individuals exhibited consistent individ- et al. 2009) and also correlated within populations among a number ual differences in these behavioral traits (i.e., personality) both with- of species (Wilson and McLaughlin 2007; Mazue ´ et al. 2015); how- in and across populations. This species is broadly distributed across ever, we know much less about whether there is geographic vari- the Nile River basin of East Africa where it is found living under a ation in personality and correlated behaviors within species (except variety of environmental conditions, the extremes of which include see Bell 2004, van Dongen et al. 2010). In 1 example, Fraser et al. the dense interior of hypoxic (low DO), clear swamps to normoxic (2001) found a positive correlation between boldness and movement (high DO) but turbid lake edges and rivers (e.g., Chapman et al. (with higher growth rate in bold individuals) in Trinidadian killifish 2002; Crispo and Chapman 2008; McNeil et al. 2016). Rivulus hartii from high predation areas, but a lack of correlation in Additionally, river sites experience fluctuating environmental condi- populations from predator-absent sites. Therefore, we also exam- tions due to degradation of the surrounding landscape leading to ined whether boldness under predation risk and activity are Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 3 correlated behaviors within P. multicolor populations and whether to the new aquarium for 2 weeks before trials began. All fish housed these correlations were different across populations. Our experimen- in experimental aquaria were held in the same controlled greenhouse tal design therefore allowed us to test for population-level variation unit, under equivalent ambient light conditions, and were fed in behavioral traits and assess whether there is geographic variation the same food once a day ad libitum. Prior to the start of the behav- in personality and correlated behaviors. Together, this research adds ioral assays, food was withheld for 24-h to standardize hunger. To to the body of work to date on P. multicolor that has focused on reduce stress from individually housing a social cichlid species, ex- population-level responses to individual stressors (e.g., DO or tur- perimental aquaria were positioned to allow for visual interactions bidity), while also examining within-population, individual vari- between individuals in neighboring aquaria. To eliminate the ability ation in behavioral traits that are important in responses to rapidly of fish to observe behavioral trials of other fish, an opaque barrier changing environments (Sih et al. 2011; Tuomainen and Candolin was used to visually isolate the selected experimental aquarium 2011). from surrounding aquaria housed on the same shelf. This barrier was placed surrounding the aquarium before the acclimation period began and was not removed until the conclusion of the behavioral Materials and Methods trial. Collection and rearing of study organisms Adult P. multicolor were collected from 2 sites within the Mpanga Trial procedure River basin (Uganda, Africa) that differ dramatically in levels of DO Before the start of a trial, a movable transparent, acrylic barrier was and turbidity. Bwera is a dense Cyperus papyrus swamp (hereafter placed into the aquarium, dividing it into 2 zones: an accessible area swamp) characterized by low DO (monthly mean6 SE ¼ 0.28 where the fish can swim freely to and from refuge (“refuge zone”; 1/ mgL O 6 0.1; Crispo and Chapman 2008) and clear, but tannin- 3 refuge þ 1/3 open space), and an inaccessible area which the fish stained water, typically <5.0 NTU (mean6 SE point-in-time meas- could not enter (“food zone”, 1/3 open space) (Figure 1, Appendix urements during time of fish collection: 1.09 NTU6 0.03). Bunoga Figure A1). Once the clear barrier was in place, fish were given is a river site characterized as having, on average, higher DO 15 min to acclimate. The trial began by adding food (0.275 g6 0.02 (monthly mean6 SE: 8.5 mgL O 6 0.1; McNeil et al. 2016) and SE food added per trial) to the isolated “food zone” in order to pro- higher turbidity (mean6 SE point-in-time measurements during mote fish to leave refuge. Once the fish was observed to leave refuge time of fish collection: 13.25 NTU6 0.1) compared with the and swim towards the partition and food, we simulated a predator swamp. Fish collected from swamp and river sites were transported attack using a model invertebrate predator [i.e., a large plastic insect from Uganda to The Ohio State University in July 2014 and held by (9.5 cm long) resembling a natural predator, Belostomatidae spp.] population under ambient light: dark cycles in aquaria located with- to illicit a fear response and scare the fish back into refuge. To initi- in Kottman Hall greenhouse (2021 Coffey Rd., Columbus Ohio), ate the predator attack, the model predator was introduced into the approved under The Institutional Animal Care and Use Committee “open water” zone of the aquarium where it was moved back and (IACUC) Protocol # 2014A00000055. forth to mimic searching behavior of the predator and thus scare the To create F1 progeny from wild-caught P. multicolor, 1 male fish. Once the fish retreated into refuge, the predator remained in and 3 female fish from the same population were placed in an iso- the aquarium for an additional 5–10 s before being removed. Upon lated aquarium to allow for natural reproduction. The maternal predator removal the clear barrier was also removed to allow fish mouth brooding behavior of P. multicolor allowed us to visually de- access to introduced food. All trials were video-recorded with a termine when a female started holding a brood (e.g., widely dis- V Canon Vixia HF R600 HD camcorder for post-trial analysis. Each tended jaw, refusal of food, and anti-social behavior). Once a female trial lasted 300 s from the time the simulated predator attack ended was noted to be mouth brooding she was removed and allowed to and barrier restricting food access was removed. further develop her brood in an isolated aquarium. The male was also replaced to maintain independent broods (i.e., because only one male was available per reproduction aquarium we knew the identity Behavior assays of each parent for each brood). This process allowed us to create To quantify and compare the behavior of P. multicolor individuals eight independent F1 broods (four broods with different parents per from each population, boldness, general activity, and foraging activ- population) of P. multicolor using N ¼ 16 fish (4 males: 4 females ity were quantified following a simulated predator attack as in pre- per population). Offspring were housed in a common garden envir- vious studies (Fraser and Gilliam 1987; Fraser et al. 2001; Wright onment for 3–6 months under normoxic (mean DO6 SE: 8.70 et al. 2006; Harcourt et al. 2009; Toms et al. 2010). We measured mgL O 6 0.76) and low turbidity (mean turbidity6SE: 0.49 boldness as the fish’s latency to leave refuge following the simulated NTU6 0.03 SE) conditions. Juvenile fish were fed Hikari First predator attack. We calculated boldness as 1 minus the proportion Bites daily ad libitum for 1 month, then weaned onto adult food of time before the fish emerged from the refuge following a simu- (crushed Tetra TetraMin Tropical Crisps) for the remainder of the lated predator attack. study. The fish was considered outside of refuge when it moved 1.5 Individuals from single broods were housed together in a single body lengths (measured in cm) away from the refuge. To measure aquarium (N ¼ 8 brood aquaria) until fish reached 3 months of age, general activity, we measured the proportion of time that an individ- after which 5 fish per brood per population (5 fish  4 broods  2 ual spent outside of refuge once it emerged from the refuge follow- populations; N ¼ 40 fish total) were randomly selected to be housed ing the simulated predator attack. individually in 19-L glass aquaria (Figure 1). Each aquarium con- Finally, we quantified foraging activity as the number of food tained a sponge filter and refuge (small plastic plant) placed at one pecks counted following the simulated predator attack and initial end of the aquarium such that one-third of the aquarium contained emergence from refuge. A single food peck was defined as a fast, for- refuge habitat and the remaining two-thirds was open water. Once ward strike at a food particle with an open mouth (masticating food placed in an experimental aquarium, fish were allowed to acclimate which was already in the mouth was not considered a food peck). Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 4 Current Zoology, 2018, Vol. 0, No. 0 Figure 1. Visual representation of the experimental aquaria. Aquaria are set up so 1/3 of the aquarium is an area of refuge (refuge zone), 1/3 of the aquarium is open water where the model predator is introduced to scare fish back into refuge, and the remaining 1/3 of the aquarium is the area where food is introduced (food zone). The food zone is initially blocked by a transparent removable barrier, then after the predator has been removed, the barrier is removed to allow fish to have access to food. To quantify the repeatability (i.e., consistent individual differen- identification) were determined to be nonsignificant, they were subse- ces in behavior) of these 3 behaviors (boldness, general activity, and quently removed from the model as well. foraging activity), we tested each individual fish 3 times, with a min- To test if there were correlations between the 3 measured behav- imum of 1 day between trials. At the conclusion of the final trial, we iors across populations (while accounting for differences in body measured the mass (g) and standard length (SL; cm) of each fish so size), we used separate multiple linear regressions (including log SL that we could account for differences in age and size between popu- as an additional predictor variable) to test each pair of behaviors. lations. We also calculated Fulton’s condition factor, K ¼ 100 To evaluate if the correlations between these behaviors differed be- [where W ¼ mass (g) and L ¼ standard length (cm); Fulton 1904; tween the swamp and river populations, we calculated Pearson cor- Froese 2006], to test if broods reared in the greenhouse at different relation coefficients separately for each population and then used times of the year displayed similar weight–length ratios. Fisher’s z-transformation to compare the 2 correlation coefficients from each population (Zar 1999). To test whether there were consistent individual differences in be- Statistical analyses havior, we calculated the repeatability, r, for each of the 3 behaviors Fish used for experimental trials varied significantly in age due to logis- via univariate ANOVA with individual as the fixed factor tical constraints; on average swamp fish were younger than river fish and behaviors as the dependent variables. We used the resulting (swamp: 116 days6 3.22 SE; river: 161 days6 4.28 SE;analysis of mean square values (MS and MS ) in the equation: within among variance (ANOVA): F ¼ 69.274, P< 0.001). This meant that, on 1,38 2 2 2 2 r ¼ s =ðs þ s Þ, where s is the variance among individuals, average, fish from each population were also of unequal size when A A A ðMS  MS Þ=n ,and s is the variance within individuals, MS A W 0 W tested: swamp fish were significantly smaller (mean mass: (Bell et al. 2009). It is important to include the term n for the number 0.91 g6 0.06 SE, mean standard length (SL): 2.95 cm6 0.06 SE)than of observations per individuals because repeatability changes system- river fish (mean mass: 1.27 g6 0.04 SE, mean SL: 3.38 cm6 0.04 SE; atically as the number of measurements per group increases, thus not ANOVA: mass: F ¼ 22.522, P<0.001; SL: F ¼ 34.502, 1,38 1,38 incorporating the number of observations per individual calculates re- P<0.001). Although the age and size of the 2 populations were differ- peatability incorrectly (Lessells and Boag 1987). The standard error ent, there was no significant difference in condition: Fulton’s condition, of repeatability was calculated using the following equation: K ¼ 3.456 0.07 SE for swamp fish and for river fish K ¼ 3.296 0.08 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SE (ANOVA: F ¼ 2.127, P ¼ 0.153). Although this demonstrates 1,38 21ðÞ  r ðÞ 1 þðÞ k  1 r that condition factor was similar across populations, we decided to in- SE r ¼ ½kkðÞ  1ðÞ N  1 clude standard length as a covariate in the following analyses to ac- count for the size differences apparent between populations and any where r is repeatability, k is number of measurements per individual ontogenetic behavioral shifts (see Bell and Stamps 2004). (or n ), and N is the number of individuals tested (Becker 1984). To test for population-level differences in behavioral responses to Analyses were performed in SPSS (version 24) and R statistical soft- the simulated predator attack, individual fish behaviors (i.e., boldness, ware (using Rstudio; version 1.1.419). general activity, and foraging activity) were averaged over 3 trials conducted per fish. For each behavioral response, we used an analysis of covariance (ANCOVA) with log-transformed SL as a covariate to Results test for differences in mean behaviors between swamp and river fish. Nonsignificant interaction terms (population  log SL) were removed Behavioral differences between populations from the models. Because some individuals were genetically related Variances for boldness and general activity were not homogeneous be- (i.e., siblings randomly selected from each brood), we also included tween populations, with behavioral variances being larger in the river brood identification (N ¼ 4 broods per population) as a random effect relative to the swamp population (Levene’s test for homogeneity of in our ANCOVA analysis. If the random effects (i.e., brood variances: boldness: F ¼ 6.113, P ¼ 0.018; activity: F ¼ 5.451, 1,38 1,38 Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 5 P ¼ 0.025), thus values for both were arcsine transformed, improving populations for general activity, with swamp fish being more active variance homogeneity for ANCOVA analysis. Foraging activity was than river fish (ANCOVA: F 5.390, P ¼ 0.026) (Figure 2B). 1, 36 ¼ not transformed since variances were homogeneous between the Finally, foraging activity did not differ among populations swamp and river population (Levene’s test for homogeneity of varian- (ANCOVA: F ¼ 1.888, P ¼ 0.178) (Figure 2C). However, the re- 1,37 ces: F ¼ 0.035, P ¼ 0.852). We included log SL as a covariate in the sult becomes significant (ANOVA: F ¼ 4.50, P ¼ 0.04) if we re- 1,38 1,38 ANCOVA models to account for any differences in size between popu- move the nonsignificant log SL covariate (F ¼ 0.045, P ¼ 0.833), 1,37 lations (see first paragraph of Statistical analysis for description of size suggesting that swamp fish had the tendency to peck at food particles differences). The interaction between population and log SL was not more often when out of refuge than did river fish (Table 1). significant for boldness and foraging activity and was therefore removed from those models; however, population  SL was a signifi- Correlated behaviors within and across populations cant interaction for general activity and was retained in the model (see Results of the multiple linear regression indicated that there was a Table 1). The random effect of brood number was not significant for significant positive relationship between boldness and foraging ac- boldness, general activity, or foraging activity and was therefore tivity across populations (Pearson’s r ¼ 0.399, t ¼ 2.172, removed from all models (Table 1). P ¼ 0.036), but no effect between log SL and foraging activity We found behavioral differences between the 2 populations des- (Pearson’s r¼0.248, t ¼0.597, P ¼ 0.554). Within each popu- pite being reared under common garden conditions and largely con- lation, we only found a positive correlation between foraging activ- sistent with pace of life predictions based on known divergent ity and boldness for river fish (Pearson’s r ¼ 0.633, t ¼ 3.467, characters. First, we found that swamp fish left refuge significantly P ¼ 0.003) (Figure 3A), but not for the swamp fish (Pearson’s earlier than did river fish (ANCOVA: F ¼ 12.36, P ¼ 0.001) 1,37 r¼0.028, t ¼0.118, P ¼ 0.907) (Figure 3A). The relationship (Table 1, Figure 2A) suggesting that on average swamp fish are more between boldness and foraging activity was significantly different bold than river fish. Second, there was a significant difference between between the 2 populations (Z¼2.26, P ¼ 0.024). Across populations, there was a significant positive relationship between general activity and foraging activity (Pearson’s r ¼ 0.376, Table 1. Results of ANCOVAs testing population-level differences t ¼ 2.40, P ¼ 0.021), but no relationship between log SL and forag- in boldness, general activity, and foraging activity across swamp ing activity (Pearson’s r¼0.248, t ¼1.47, P ¼ 0.151). Within and river populations. 37 each population, there was no significant correlation between forag- Behavior Model df F P* ing activity and general activity for either population (swamp: Pearson’s r ¼ 0.408, t ¼ 1.894, P ¼ 0.074; river: Pearson’s r ¼ 0.293, Boldness Population 1, 37 12.357 0.001 t ¼ 1.301, P ¼ 0.210) (Figure 3B). The relationship between general log SL 1, 37 0.007 0.934 activity and foraging activity was not significantly different between Brood number 6, 31 1.139 0.363 Population  log SL 1, 36 2.751 0.106 the swamp and river populations (Z ¼ 0.38, P ¼ 0.701) General Activity Population 1, 36 5.390 0.026 Finally, we found no relationship between boldness and general log SL 1, 36 3.060 0.089 activity across populations (Pearson’s r ¼ 0.222, t ¼ 1.28, Brood number 6, 30 1.576 0.219 P ¼ 0.207), but there was a significant negative relationship between Population  log SL 1, 36 4.782 0.035 log SL and general activity (Pearson’s r¼0.417, t ¼2.73, Foraging Activity Population 1, 37 1.888 0.178 P ¼ 0.01). Within each population there was not a significant rela- log SL 1, 37 0.045 0.833 tionship between boldness and general activity for either population Brood number 6, 31 1.149 0.358 (swamp: Pearson’s r ¼ 0.304, t ¼ 1.24, P ¼ 0.232; river: Pearson’s Population  log SL 1, 36 0.080 0.779 r¼0.033, t ¼0.789, P ¼ 0.44) (Figure 3C) and the correlation *Values in bold are significant at a ¼ 0.05. coefficients did not differ (Z ¼ 1.01, P ¼ 0.311). Figure 2. Behavioral responses in 2 different F1 P. multicolor populations, swamp (open bars) and river (solid bars). Estimated marginal means (6 SE)of (A) bold- ness (arcsine transformed); (B) General activity (arcsine transformed); and, (C) foraging activity. *Values are significantly different at a ¼ 0.05. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 6 Current Zoology, 2018, Vol. 0, No. 0 population differences in behavior might be associated with differ- ences in physiological traits and life-history strategies that may also diverge in response to human-induced rapid environmental change, HIREC (Sih et al. 2011). Here, we found divergent variation in be- havioral traits between 2 populations that experience extremes of hypoxia and turbidity and also found evidence that some, but not all, behaviors were repeatable and correlated within populations. Under common garden conditions (i.e., released from major stres- sors) fish originating from the hypoxic, clear swamp habitat that have been shown to have higher metabolic scope were, on average, more bold and active than fish from the normoxic, turbid river that have lower metabolic scope. Across populations, we found positive correlations between boldness and foraging activity, as well as be- tween foraging and general activity. Thus, individuals that were more active foragers were also bolder and generally more active. However, within populations, boldness and foraging activity were only positively correlated within the river population, but not corre- lated within the swamp population. Boldness and foraging activity were also repeatable behaviors within the river population, but only foraging activity was a repeatable behavior in the swamp popula- tion. Our results suggest that P. multicolor populations that experi- ence different extremes of environmental stressors behave differently. We discuss these results in the context of population- and individual-level behavior trait differences in fish known to be di- vergent in a number of other traits associated with 2 globally im- portant environmental stressors, hypoxia and turbidity. The observed differences in boldness and general activity be- tween populations that are physiologically adapted to deal with extremes in oxygen availability and water clarity is consistent with the pace-of-life concept, which predicts that individuals with a higher metabolic scope will be more bold and active than individuals with a lower metabolic scope. However, this finding requires consid- eration of the testing conditions in our experiment relative to the en- Figure 3. The relationships between (A) foraging activity and boldness, vironmental conditions in the home habitat of each population. In (B) foraging activity and general activity, and (C) boldness and general activ- their home environment, swamp fish face energetic challenges asso- ity for the swamp (open circles, long dashed lines) and river (closed circles, ciated with low DO (Chapman 2015) and have adapted (through solid line) populations, and with regression lines for both populations pooled both plastic and genetic mechanisms) to these environments by hav- (small dashed line). ing lower routine metabolic rates, larger gills, and smaller brains than river fish from high DO habitats (Reardon and Chapman Repeatability of boldness and activity 2009; Crispo and Chapman 2010). Furthermore, swamp fish have The swamp and river populations both exhibited significant repeat- also been shown to be less active and display fewer reproductive ability in at least 1 behavior; however, there were differences be- behaviors than river fish when in their low DO home environment tween the populations in repeatability values. The swamp (e.g., Gotanda et al. 2011; McNeil et al. 2016). Therefore, it might population showed significant repeatability in foraging behavior make sense to expect swamp fish to demonstrate a slower pace of (r ¼ 0.3476 0.147 SE; ANOVA: F ¼ 2.596, P ¼ 0.006), but 19,60 life relative to river fish, but only under home conditions. In con- did not show significant repeatability in bold behavior trast, under the novel rearing conditions (i.e., high DO/clear) of this (r ¼ 0.1386 0.146 SE; ANOVA: F ¼ 1.495, P ¼ 0.140) or gen- 19,60 study, we predicted that when released from the constraint of a low eral activity (r ¼ 0.1286 0.145 SE; ANOVA: F ¼ 1.479, 19,60 DO environment, swamp fish would be bolder and more active P ¼ 0.147). The river population also showed significant repeatabil- given their adaptations to an energetically challenging environment ity in foraging behavior (r ¼ 0.3746 0.145 SE; ANOVA: (i.e., due to adaptations to low DO that would increase performance F ¼ 2.794, P ¼ 0.003) as well as boldness (r ¼ 0.3496 0.146 SE; 19,60 in high DO conditions, e.g., Reardon and Chapman 2010). Ursza ´n ANOVA: F ¼ 2.632, P ¼ 0.005), but did not show significantly 19,60 et al. (2015) also suggest that a fixed behavioral strategy, such as we repeatability for general activity (r ¼ 0.0496 0.138 SE; ANOVA: see in the swamp population, would be less energetically expensive F ¼ 1.149, P ¼ 0.345). 19,60 compared with greater intra-individual variation. Based on this rea- soning, river fish adapted to a more turbid environment might there- fore be relatively less active compared with swamp fish in high DO Discussion and additionally, may be less bold if clear water induces a fear re- Animals are increasingly faced with human-induced stressors that sponse (e.g., see Lima and Dill 1990). Indeed, this is what we found: vary in space and time, thus we can expect population-level diver- fish from the swamp left refuge sooner after a simulated predator at- gence in behaviors that help animals to cope with environmental tack, were more active throughout the trial, and displayed marginal- change that varies across the landscape. We also expect that ly higher foraging activity levels than fish originating from the river Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 7 (lower metabolic scope, slower life history) (Figure 2). These popu- It is also possible that other ecological factors, both abiotic (e.g., lation differences suggest a relationship between behavior and life temperature, flow rate) and biotic (e.g., density, competition effects, history, and that low DO and high turbidity are strong selective predator abundance), may contribute to behavioral trait divergence agents in this system. Evidence of other environmental extremes across P. multicolor populations. While our knowledge of the abiot- driving intraspecific, population-level variation in behavioral traits, ic factors that vary between swamp and river habitats (such as DO, and relationships between behaviors and/or physiological and life- turbidity, temperature) is extensive and comes from monthly sam- history traits is accumulating (e.g., Bell 2004; Moran et al. 2016; pling over 15 years (L.J. Chapman, personal communication), our Dubuc-Messier et al. 2017). knowledge of the biological characteristics of each habitat is less While our results suggest a relationship between behavior trait complete. For example, while we know that fish from both types of differences and population of origin, and a possible link to known habitat experience similar suites of predators (e.g., wading birds, divergent physiological traits, they do not allow us to differentiate snakes, and Clarias catfish), we do not know whether there are dif- between the mechanisms driving this association. For example, ferences in predation pressure exerted by these predators across the observed differences in behavior could be due to physiological adap- 2 habitats (e.g., due to differences in density, habitat complexity, tations to these stressors. In common carp Cyprinus carpio, for in- etc.). Variation in predation pressure is well known to drive stance, fish with higher metabolic rates and lower cortisol receptor population-level differences in behavior (e.g., Endler 1980) and in expression in the brain were found to be bolder and more aggressive, correlations between suites of behavior (e.g., Fraser et al. 2001). suggesting that variation in physiology is tightly linked to behavioral Future work distinguishing among these and other mechanisms will expression (Huntingford et al. 2010). In a direct test of the link be- greatly enhance our understanding of the variables that shape diver- tween physiological and behavioral traits, Binder et al. (2016) found gent behavior (Bell 2007). that individual bluegill sunfish L. macrochirus with higher aerobic Our results also suggest that correlations between boldness and capacity were bolder than individuals with lower aerobic capacity. foraging activity is not a general pattern across P. multicolor popula- Additionally, there is evidence that oxidative stress resulting from tions and may be associated with the differences in the selective pres- prolonged exposure to elevated stress hormones (i.e., glucocorti- sures experienced in swamp and river habitats. Unfortunately, coids) is linked to behavioral coping strategies (Costantini et al. comparing only 2 populations from different environments cannot 2011). However, in each of these examples, we still do not know if definitively tell us what factors drive these differences. However, physiological differences drive behavioral expression or vice versa. our results in combination with past work demonstrating other trait We know that oxygen concentration and turbidity level are likely differences across multiple populations of P. multicolor (e.g., strong selective agents driving divergence between swamp and river Chapman et al. 2000; Reardon and Chapman 2010; Crispo and populations given previous work and the results presented here, and Chapman 2010) suggest that differences in DO and water clarity we also know that environmental conditions in the swamp tend to might influence whether behaviors are correlated or not. More be relatively stable whereas they fluctuate more in rivers. In swamp broadly, this may suggest that correlations between behaviors may populations, the relative stability of the environment in concert with depend on context. For example, a positive correlation between consistently low DO would favor traits that help minimize energy boldness and foraging activity might be particularly important in expenditure, for example, lower metabolic rates, lower activity, and river environments that are much more turbid, on average, than bold behavior since attaining resources needs to be efficient. swamps, but experience fluctuating turbidity. For instance, during However, this combination of an energetically challenging but stable periods of high turbidity it is possible that the perceived risk of pre- environment is unlikely to favor tight linkages between traits if those dation is low and so there is an advantage for bold individuals that linkages might constrain energetic coping mechanisms. On the other are more active foragers. Alternatively, periods of lower turbidity, hand, in river populations, we might expect that the visual impair- when predation risk might be higher because fish are more visible to ment resulting from high turbidity favors decreased predator vigi- predators, individuals with a strong anti-predator response that lance or assessment of predation risk (Lima and Dill 1990), but spend more time hiding than feeding might be favored. It is then during periods of high water clarity (as imposed in our rearing envir- possible that the relative stability of the clear, low DO swamp sites onment) this might create a higher perceived risk and thus favor shy do not favor such a trade-off which is why we do not see a correl- behavior. ation between boldness and foraging activity in that population. In a Alternatively, differences in both behavior and physiology could similar examination of boldness and activity coupled with aggres- be due to plasticity. Binder et al. (2016) suggest that plasticity in aer- sion in threespine stickleback Gasterosteus aculeatus, Bell (2004) obic metabolism could account for the positive correlation between found differences in mean behaviors across very different environ- metabolic rate and boldness. While the expression of behavior trait ments (e.g., high flow and temperature variation in an unregulated differences here under common garden conditions suggests some stream vs. low variation in a damned and regulated stream). She heritability, we also know that P. multicolor can display behavioral also discovered that in sticklebacks from the more variable environ- flexibility. For example, in a reciprocal rearing study, P. multicolor ment, behaviors were often correlated across contexts, whereas males displayed higher rates of aggression under turbid conditions there were no significant correlations between behaviors in fish orig- regardless of whether they originated from swamp or river popula- inating from the more stable population. Our results suggest that tions or if they were reared under clear or turbid conditions (Gray there are population-level differences in the way that behavioral et al. 2012). Similarly, Reardon and Chapman (2010) demonstrate traits have (co)evolved, similar to the differences found across developmental plasticity in routine metabolic rate when swamp fish stickleback populations by Bell (2004). were reared in high or low DO. Further explorations of how mul- Consistent behavioral variation, or repeatability, is known to tiple stressors that vary among populations might shape behavior occur within and between individuals across time or ecological con- and that directly test for associations between behavioral and texts (Dall et al., 2004; Sih et al., 2004; Sih and Bell, 2008). In our physiological traits could be informative under the pace-of-life study, river-origin fish displayed repeatability across 2 contexts concept. (boldness and foraging activity) whereas the swamp-origin fish only Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 8 Current Zoology, 2018, Vol. 0, No. 0 displayed repeatability in foraging activity. Divergent habitats, as References well as habitats with large environmental variability, may favor dif- Abrahams MV, Robb TL, Hare JF, 2005. Effect of hypoxia on opercular dis- ferences in behavioral traits and the repeatability of behavior, plays: evidence for an honest signal? Anim Behav 70:427–432. assuming limited plasticity. In this system, swamp sites exhibit lower Atwell JW, Cardoso GC, Whittaker DJ, Price TD, Ketterson ED, 2014. variation in environmental factors (i.e., are more stable), whereas Hormonal, behavioral, and life-history traits exhibit correlated shifts in rela- environmental factors tend to fluctuate in the river habitats (e.g., tion to population establishment in a novel environment. Am Nat 184: Crispo and Chapman 2010). Behaviors which show relatively low 147–160. Becker WA, 1984. A Manual of Quantitative Genetics. Pullman: Academic within-individual variance compared with between-individual vari- Enterprises. ance are more repeatable (Bell et al. 2009). If, however, an environ- Bell AM, 2004. Behavioural differences between individuals and ment has relatively low environmental variability, it is possible that two populations of stickleback Gasterosteus aculeatus. J Evol Biol 18: this would select for a specific behavioral type that is better suited 464–473. for the environment, thus reducing between-individual variance. Bell AM, 2007. Future directions in behavioural syndromes research. Proc R Based on our tests for homogeneity of variance between popula- Soc B Biol Sci 274:755–761. tions, we found that in the swamp population 2 of the 3 behaviors Bell AM, Hankison SJ, Laskowski KL, 2009. The repeatability of behaviour: a (boldness and general activity) had lower between-individual vari- meta-analysis. Anim Behav 77:771–783. ance compared with the river population. In the swamp population, Bell AM, Stamps JA, 2004. Development of behavioural differences between low environmental variation in addition to low DO, may especially individuals and populations of sticklebacks Gasterosteus aculeatus. Anim Behav 68:1339–1348. favor a more fixed behavioral type, given energetic constraints Binder TR, Wilson AD, Wilson SM, Suski CD, Godin J-GJ et al., 2016. Is there (Ursza ´ n et al. 2015). Alternatively, environments with relatively a pace-of-life syndrome linking boldness and metabolic capacity for locomo- large environmental variation may select for a suite of different tion in bluegill sunfish? Anim Behav 121:175–183. behaviors and possibly increase between-individual variance. Brown C, Jones F, Braithwaite V, 2005. In situ examination of Intraspecific divergence in behavior across populations experi- boldness-shyness traits in the tropical poeciliid Brachyraphis episcopi. Anim encing environmental extremes has been found in a number of taxa Behav 70:1003–1009. (e.g., Brown et al. 2005; Moran et al. 2016). As an example, van Bruton MN, 1985. The effects of suspensoids on fish. Hydrobiologia 125: Oers et al. (2004) found that great tits Parus major show heritable 221–241. variation in risk-taking behavior when high- and low risk-taking Careau V, Thomas D, Humphries MM, Re ´ ale D, 2008. Energy metabolism lines were selected in the laboratory over 2 generations. Similarly, and animal personality. Oikos 117:641–653. Chapman LJ, 2015. Low-oxygen lifestyles in extremophilic fishes. In: Reisch R, Brown et al. (2005) found heritable variation in boldness among Plath M, Tobler M, editors. Extremophile Fishes - Ecology and Evolution of populations experiencing different levels of predation in a tropical Teleosts in Extreme Environments. Heidelberg: Springer. 9–33. poecilliid fish, Brachyraphis episcopi, when F1s from 2 populations Chapman LJ, Albert J, Galis F, 2008. Developmental plasticity, genetic differ- were reared in the laboratory and exposed to different predation entiation, and hypoxia-induced tradeoffs in an African cichlid fish. Open regimes. By investigating the underlying behavioral responses from Evol J 2:75–88. an individual perspective within divergent populations we gain a Chapman LJ, Chapman CA, 2003. Fishes of the African rain forests: emerging mechanistic understanding of how behavioral responses, linked to and potential threats to a little-known fauna. In: Chrisman TL, Chapman physiological adaptation to environmental change, might contribute LJ, Chapman CA, Kaufman LS, editors. Conservation, Ecology, and to population persistence under human-induced rapid environmen- Management of African Fresh Waters. Gainesville: University Press of tal change, HIREC (Koolhaas et al. 1999; Dingemanse and Wolf Florida. 176–209. Chapman LJ, Chapman CA, Nordlie FG, Rosenberger AE, 2002. 2010; Frost et al. 2013; Sih et al. 2015, Wong and Candolin 2015). Physiological refugia: swamps, hypoxia tolerance and maintenance of fish Here, we have demonstrated behavioral trait divergence between popu- diversity in the Lake Victoria region. Comp Biochem Phys A Mol Integr lations and that trade-offs between behavioral traits vary among these Physiol 133:421–437. populations. Remaining to be investigated is how these behavioral trait Chapman LJ, Galis F, Shinn J, 2000. Phenotypic plasticity and the possible differences translate into advantages in the respective home environ- role of genetic assimilation: hypoxia-induced trade-offs in the morphologic- ments of swamp and river fish. al traits of an African cichlid. Ecol Lett 3:387–393. Costantini D, Marasco V, Møller AP, 2011. A meta-analysis of glucocorti- coids as modulators of oxidative stress in vertebrates. J Comp Physiol B Acknowledgments 181:447–456. Crispo E, Chapman LJ, 2008. Population genetic structure across dissolved We thank the Commissioner of Fisheries Resources Management and oxygen regimes in an African cichlid fish. Mol Ecol 17:2134–2148. Development, Uganda, for permission to export live specimens, the Uganda Crispo E, Chapman LJ, 2010. Geographic variation in phenotypic plasticity in National Council for Science and Technology (UNCST) for research permis- response to dissolved oxygen in an African cichlid fish. J Evol Biol 23: sion, and Makerere University Biological Field Station for logistical support. 2091–2103. We additionally thank Dr. Lauren Chapman for use of research facilities in Crocker CD, Chapman LJ, Martinez ML, 2013. Hypoxia-induced plasticity in Uganda (Lake Nabugabo Research Station) and mentorship throughout this the metabolic response of a widespread cichlid. Comp Biochem Phys B 166: project; Ugandan field assistants for invaluable help whilst in Uganda; and, J. 141–147. Odegard for assistance with fish care. Dall SRX, Houston AI, McNamara JM, 2004. The behavioural ecology of per- sonality: consistent individual differences from an adaptive perspective. Ecol Lett 7:734–739. Funding Dingemanse NJ, Wolf M, 2010. Recent models for adaptive personality differ- ences: a review. Philos T Roy Soc B 365:3947–3958. Funding was provided by the American Cichlid Association Guy D. Jordan Dubuc-Messier G, Re ´ ale D, Perret P, Charmantier A, 2017. Environmental Research Fund and The Ohio State University Undergraduate Student heterogeneity and population differences in blue tits personality traits. Government Academic Enrichment Grant to R.C.O., The Ohio State Behav Ecol 28:448–459. University Office of International Affairs Faculty Seed Grant to S.M.G. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 9 Dugas MB, Franssen NR, 2011. Nuptial coloration of red shiners Cyprinella Moran NP, Mossop KD, Thompson RM, Wong BB, 2016. Boldness in ex- lutrensis is more intense in turbid habitats. Naturwissenschaften 98: treme environments: temperament divergence in a desert-dwelling fish. 247–251. Anim Behav 122:125–133. Dugas MB, Franssen NR, 2012. Red shiners Cyprinella lutrensis have larger Newcombe CP, MacDonald DD, 1991. Effects of suspended sediments on eyes in turbid habitats. Can J Zool 90:1431–1436. aquatic ecosystems. N Am J Fish Manage 11:72–82. Edelaar P, Jovani R, Gomez-Mestre I, 2017. Should I change or should I go? Re ´ ale D, Reader SM, Sol D, McDougall PT, Dingemanse NJ, 2007. Phenotypic plasticity and matching habitat choice in the adaptation to envir- Integrating animal temperament within ecology and evolution. Biol Rev 82: onmental heterogeneity. Am Nat 190:506–520. 291–318. Endler JA, 1980. Natural selection on color patterns in Poecilia reticulata. Re ´ ale D, Garant D, Humphries MM, Bergeron P, Careau V et al., 2010. Evolution 34:76–91. Personality and the emergence of the pace-of-life syndrome concept at the Fraser DF, Gilliam JF, 1987. Feeding under predation hazard: response of the population level. Phil Trans R Soc B 365:4051–4063. guppy and hart rivulus from sites with contrasting preadtion hazard. Behav Reardon EE, Chapman LJ, 2009. Hypoxia and life-history traits in a eurytopic Ecol Sociobiol 21:203–209. African cichlid. J Fish Biol 75:1795–1815. Fraser DF, Gilliam JF, Daley MJ, Le AN, Skalski GT et al., 2001. Explaining Reardon EE, Chapman LJ, 2010. Energetics of hypoxia in a mouth-brooding leptokurtic movement distributions: intrapopulation variation in boldness cichlid: evidence for interdemic and developmental effects. Physiol Biochem and exploration. Am Nat 158:124–135. Zool 83:414–423. Froese R, 2006. Cube law, condition factor and weight-length relationships: Sih A, Bell AM, 2008. Insights for behavioral ecology from behavioral syn- history, meta-analysis and recommendations. J Appl Ichthyol 22:241–253. dromes. In: Brockmann HJ, Roper TJ, Naguib M, Wynne-Edwards KE, Frost AJ, Thomson JS, Smith C, Burton HC, Davis B et al., 2013. Barnard C et al., editors. Advances in the Study of Behavior. San Diego: Environmental change alters personality in the rainbow trout Elsevier Academic Press Inc. 227–281. Oncorhynchus mykiss. Anim Behav 85:1199–1207. Sih A, Bell AM, Johnson JC, Ziemba RE, 2004. Behavioral syndromes: an Fulton TW, 1904. The Rate of Growth of Fishes. 22nd Annual Report, intergrative overiew. Q Rev Biol 79:241–277. Fisheries Board of Scotland, Edinburgh. Sih A, Ferrari MCO, Harris DJ, 2011. Evolution and behavioural responses to Gotanda KM, Reardon EE, Chapman LJ, 2011. Hypoxia and male behaviour human-induced rapid environmental change. Evol Appl 4:367–387. in an African cichlid Pseudocrenilabrus multicolor victoriae. J Fish Biol 78: Sih A, Mathot KJ, Moiro ´ n M, Montiglio P-O, Wolf M et al., 2015. Animal 2085–2092. personality and state - behaviour feedbacks: a review and guide for empiri- Gray SM, McDonnell LH, Cinquemani FG, Chapman LJ, 2012. As clear as cists. Trends Ecol Evol 30:50–60. mud: turbidity induces behavioral changes in the African cichlid Toms CN, Echevarria DJ, Jouandot DJ, 2010. A methodological review of Pseudocrenilabrus multicolor. Curr Zool 58:146–157. personality-related studies in fish: focus on the shy-bold axis of behavior. Int Harcourt JL, Sweetman G, Johnstone RA, Manica A, 2009. Personality J Comp Psychol 23: 1–25. counts: the effect of boldness on shoal choice in three-spined sticklebacks. Tuomainen U, Candolin U, 2011. Behavioural responses to human-induced Anim Behav 77:1501–1505. environmental change. Biol Rev 86:640–657. Hendry AP, 2015. Key questions on the role of phenotypic plasticity in Ursza ´ n TJ, To¨ro ¨ k J, Hettyey A, Garamszegi LZ, Herczeg G, 2015. eco-evolutionary dynamics. J Hered 107:25–41. Behavioural consistency and life history of Rana dalmatina tadpoles. Oecol Huntingford FA, Andrew G, Mackenzie S, Morera D, Coyle SM et al., 2010. 178:129–140. Coping strategies in a strongly schooling fish, the common carp Cyprinus Utne-Palm AC, 2002. Visual feeding of fish in a turbidenvironment: physical carpio. J Fish Biol 76:1576–1591. and behavioural aspects. Mar Freshw Behav Phy 35:111–128. Killen SS, Mitchell MD, Rummer JL, Chivers DP, Ferrari MC et al., 2014. van der Sluijs I, Gray SM, Amorim MCP, Barber I, Candolin U et al., 2011. Aerobic scope predicts dominance during early life in a tropical damselfish. Communication in troubled waters: responsesof fish communication sys- Funct Ecol 28:1367–1376. tems to changing environments. Evol Ecol 25:623–640. Koolhaas JM, Korte SM, De Boer SF, Van Der Vegt BJ, Van Reenen CG et al., van Dongen WFD, Maldonado K, Sabat P, Va ´ squez RA, 2010. Geographic 1999. Coping styles in animals: current status in behavior and stress-physi- variation in the repeatability of a personality trait. Behav Ecol 21: ology. Neurosci Biobehav R 23:925–935. 1243–1250. Lessells CM, Boag PT, 1987. Unrepeatable repeatabilities: a common mistake. van Oers K, de Jong G, van Noordwijk AJ, Kempenaers B, Drent PJ, 2004. Auk 104:116–121. Contribution of genetics to the study of animal personalities: a review of Lima SL, Dill LM, 1990. Behavioral decisions made under the risk of preda- case studies. Behaviour 142:1185–1206. tion: a review and prospectus. Can J Zool 68:619–640. Weins KE, Crispo E, Chapman LJ, 2014. Phenotypic plasticity is maintained Mazue ´ GPF, Dechaume-Moncharmont F-X, Godin J-GJ, 2015. Boldness - ex- despite geographical isolation in an African cichlid fish Pseudocrenilabrus ploration behavioral syndrome: interfamily variability and repeatability of multicolor. Integr Zool 9:85–96. personality traits in the young of the convict cichlid Amatitlania siquia. Wikelski M, Spinney L, Schelsky W, Scheuerlein A, Gwinner E, 2003. Slow Behav Ecol 26:900–908. pace of life in tropical sedentary birds: a common-garden experiment on McNeil GV, Friesen CN, Gray SM, Aldredge A, Chapman LJ, 2016. Male col- four stonechat populations from different latitudes. Proc R Soc Lond B 270: our variation in a eurytopic African cichlid: the role of diet and hypoxia. 2383–2388. Biol J Linn S 118:551–568. Wilson ADM, McLaughlin RL, 2007. Behavioural syndromes in brook charr, Mitchell DJ, Biro PA, 2017. Is behavioural plasticity consistent across Salvelinus fontinalis: prey-search in the field corresponds with space use in different environmental gradients and through time? Proc R Soc B 284: novel laboratory situations. Anim Behav 74:689–696. 20170893. Wong BBM, Candolin U, 2015. Behavioral responses to changing environ- Mittlebach GG, Ballew NG, Kjelvik MK, 2014. Fish behavioral types and ments. Behav Ecol 26:665–673. their ecological consequences. Can J Fish Aquat Sci 71:927–944. Wright D, Nakamichi R, Krause J, Butlin RK, 2006. QTL analysis of behav- Montiglio PO, Garant D, Bergeron P, Messier GD, Re ´ ale D, 2014. Pulsed ioral and morphological differentiation between wild and laboratory zebra- resources and the coupling between life-history strategies and exploration fish Danio rerio. Behav Genet 36:271–284. patterns in eastern chipmunks Tamias striatus. J Anim Ecol 83:720–728. Zar JH, 1999. Biostatistical Analysis. Upper Saddle River: Prentice-Hall Inc. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 10 Current Zoology, 2018, Vol. 0, No. 0 APPENDIX Figure A1. Experimental progression of an actual trial: (A) “food addition” phase, which lured the tested fish (circled in red) out of refuge in attempt to consume the introduced food on the adjacent side of the clear barrier, (B) “predator introduction” phase introduces the model predator which scares test fish back into ref- uge, (C) “behavioral observation” phase removes the predator and clear barrier to allows fish to have access to food. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Current Zoology Oxford University Press

Behavioral differences within and among populations of an African cichlid found in divergent and extreme environments

Free
10 pages

Loading next page...
 
/lp/ou_press/behavioral-differences-within-and-among-populations-of-an-african-R0hvq7kPIi
Publisher
Oxford University Press
Copyright
© The Author(s) (2018). Published by Oxford University Press.
ISSN
1674-5507
eISSN
2396-9814
D.O.I.
10.1093/cz/zoy027
Publisher site
See Article on Publisher Site

Abstract

Animals are increasingly faced with human-induced stressors that vary in space and time, thus we can expect population-level divergence in behaviors that help animals to cope with environmental change. However, empirical evidence of behavioral trait divergence across environmental extremes is lacking. We tested for variation in behavioral traits among 2 populations of an African cichlid fish (Pseudocrenilabrus multicolor victoriae Seegers, 1990) that experience extremes of dissolved oxygen (DO) and turbidity and are known to vary in a number of physiological and life history traits associated with these stressors. Using a common garden rearing experiment, F1 pro- geny from wild-caught parents originating from a swamp (low DO, clear) and a river (high DO, tur- bid) were reared in high DO, clear water. Predator simulation assays were conducted to test for (1) variation in boldness, general activity, and foraging activity between populations, (2) differences in correlations between behaviors within and across populations, and (3) repeatability of behaviors. There was strong evidence for divergence between populations, with swamp fish being more bold (i.e., leaving refuge sooner after a simulated predator attack) and active (i.e., spent more time out of refuge) than river fish. Across populations there were positive correlations between foraging ac- tivity and both boldness and general activity; however, within populations, there was only a strong positive relationship between foraging activity and boldness in the river population. Here, we have demonstrated that populations that originate from drastically different environments can produce progeny that exhibit measurable differences in behaviors and their correlated relationships even when reared under common conditions. Key words: behavioral correlation, behavioral syndrome, boldness, hypoxia, pace of life, turbidity. Environmental heterogeneity can have a strong impact on phenotyp- et al. 2004; Mittlebach et al. 2014; Mitchell and Biro 2017). And ic diversity, driving divergence among populations through plasticity like many traits, aspects of personality (e.g., boldness, aggressive- and local adaptation (Hendry 2015; Edelaar et al. 2017). As an en- ness) can be under strong selection when faced with environmental vironment shifts through time and space, behavioral traits are often extremes (Dubuc-Messier et al. 2017). For example, Brown et al. a species’ first line of defense in response to change. This is especially (2005) found that populations of a poeciliid fish Brachyrhaphis epis- true under increasingly rapid, human-induced environmental change copi were consistently bolder if they were from high-predation ver- (Sih et al. 2011). Furthermore, we now understand that behavior is sus low-predation sites across multiple rivers. Thus, we might often a repeatable and heritable trait and that individuals consistent- expect populations to diverge in personality when they experience ly differ in personality within populations of a single species (Sih opposite ends of environmental extremes. V C The Author(s) (2018). Published by Oxford University Press. 1 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 2 Current Zoology, 2018, Vol. 0, No. 0 Personality traits have been increasingly found to covary with excessive sediment run-off that is pulsed into the system during in- physiological traits and together may promote species persistence tense rain events. Swamps, on the other hand, tend to offer more sta- under environmental extremes (Dingemanse and Wolf 2010; Sih ble environmental conditions due the buffering effects of dense et al. 2015). For example, individuals that display high aggressive- aquatic vegetation. ness, high activity, and are bold (e.g., proactive types) have higher Populations of P. multicolor show strong phenotypic differences resting metabolic rates than less aggressive, less active, and shy indi- in morphological, physiological, and life-history traits in response to viduals (e.g., reactive types) (Careau et al. 2008; Huntingford et al. extreme differences in DO and turbidity. For example, in response 2010). Boldness, in particular, has also been shown to positively to low DO, P. multicolor from swamps tend to have on average covary with metabolic scope (e.g., the difference between resting 10% smaller brains and 56% larger gill surface area (Chapman and and active metabolic rates) (Killen et al. 2014; Binder et al. 2016). Chapman 2003; Chapman et al. 2008; Crispo and Chapman 2010; For example, bold bluegill sunfish Lepomis macrochirus have Weins et al. 2014). Swamp fish also have lower resting metabolic greater metabolic scope and exhibit greater aerobic capacity for rates and higher metabolic scope, the latter likely afforded by larger locomotor activity than shy sunfish (Binder et al. 2016). These asso- gill surface areas and thus a higher capacity for metabolic perform- ciations have provided empirical evidence for the pace-of-life syn- ance (Reardon and Chapman 2010). Fish from the swamp and those drome which predicts that behavior, physiological, and life-history reared under hypoxic conditions also have smaller broods and brood traits coevolve in response to correlational selection pressures (Re ´ ale for a shorter period of time compared with populations from high et al. 2007, 2010). Spatial and temporal heterogeneity in environ- DO, turbid rivers (Reardon and Chapman 2009). The pace-of-life mental conditions thus has the potential to create geographic vari- hypothesis would therefore predict that P. multicolor from low DO/ ation in a suite of behavioral, physiological, and life-history traits low turbidity swamp populations would fall closer to the slow end (Atwell et al. 2014; Montiglio et al. 2014). Understanding the role of the pace-of-life continuum, and river fish (or fish reared under of environmental extremes in shaping population divergence in a normoxic conditions) closer to the fast end of the continuum. suite of traits may be particularly important for understanding how However, by rearing fish from both populations in a common envir- species adapt to changing environments. onment free of low DO and turbidity stressors, we predicted that In aquatic ecosystems, hypoxia (i.e., low dissolved oxygen, DO) swamp fish would show behaviors more consistent with a relatively and turbidity (i.e., suspended particulates) are metabolically and faster pace of life due to their adaptations for life under energetically visually challenging conditions, respectively, that have been shown constraining low DO conditions. From a physiological standpoint, individually to generate divergence in behavioral, life-history, and P. multicolor from swamp populations show enhanced metabolic physiological traits in aquatic organisms (e.g., van der Sluijs et al. abilities when tested under high DO conditions, compared with river 2011; Chapman 2015). For example, in contrast to normoxic envi- fish (Reardon and Chapman 2010). Thus, swamp fish should be, on ronments, hypoxic environments tend to favor fish that are less ac- average, more bold and active than high DO/high turbidity river tive (Abrahams et al. 2005; Gotanda et al. 2011; McNeil et al. populations when reared under common garden conditions. This 2016), have relatively larger gills and smaller brains (Chapman et al. might be especially true if river fish also have higher perceived pre- 2008; Crispo and Chapman 2010), lower metabolic rates (Reardon dation risk in a clear environment compared with the turbid envi- and Chapman 2010; Crocker et al. 2013), and shorter brooding ronments they typically encounter and thus may reduce activity, times due to the challenge of extracting oxygen at very low concen- increase vigilance, and be shyer in clear waters than they would nor- trations. Environments with anthropogenically elevated turbidity mally be. Precedent use of common garden experimental designs to levels also impose a suite of potential stressors on fish (e.g., Bruton reduce the effects of environmental stressors when testing for per- 1985; Newcombe and MacDonald 1991) and notably pose visual sonality and pace of life syndromes have been suggested and applied challenges for aquatic organisms (e.g., Utne-Palm 2002; van der in previous studies. For example, Ursza ´ n et al. (2015) used agile frog Sluijs et al. 2011). For example, Dugas and Franssen (2011, 2012) tadpoles Rana dalmatina as a model organism to provide evidence found that red shiner Cyprinella lutrensis populations from turbid for pace of life and behavioral consistencies at different ontogenetic waters have larger eyes and more intense nuptial color patterns than stages. Additionally, a common garden experimental design has clear-water populations, suggesting that such adaptions allow for been used to study the pace of life of stonechat Saxicola torquata better vision and more visible signals, respectively, when the water populations (Wikelski et al. 2003). By raising individuals from di- is turbid. In contrast to other traits, however, variation in personal- vergent environments in a common garden experimental design, en- ity across extremes of these 2 environmental stressors has been over- vironmental stressors originally present in the home environment looked in recent literature. are released from these individuals, thus any observed differences In this study, we compared behavioral traits (e.g., general activ- may allow us to draw some conclusions about the adaptive nature ity, foraging behavior, and boldness under predation risk) of 2 popu- of trait divergence. lations of the haplochromine cichlid Pseudocrenilabrus multicolor Finally, boldness and activity have frequently been found to be victoriae Seegers, 1990, from divergent environmental conditions. repeatable (i.e., consistent individual differences in behavior; Bell We also examined whether individuals exhibited consistent individ- et al. 2009) and also correlated within populations among a number ual differences in these behavioral traits (i.e., personality) both with- of species (Wilson and McLaughlin 2007; Mazue ´ et al. 2015); how- in and across populations. This species is broadly distributed across ever, we know much less about whether there is geographic vari- the Nile River basin of East Africa where it is found living under a ation in personality and correlated behaviors within species (except variety of environmental conditions, the extremes of which include see Bell 2004, van Dongen et al. 2010). In 1 example, Fraser et al. the dense interior of hypoxic (low DO), clear swamps to normoxic (2001) found a positive correlation between boldness and movement (high DO) but turbid lake edges and rivers (e.g., Chapman et al. (with higher growth rate in bold individuals) in Trinidadian killifish 2002; Crispo and Chapman 2008; McNeil et al. 2016). Rivulus hartii from high predation areas, but a lack of correlation in Additionally, river sites experience fluctuating environmental condi- populations from predator-absent sites. Therefore, we also exam- tions due to degradation of the surrounding landscape leading to ined whether boldness under predation risk and activity are Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 3 correlated behaviors within P. multicolor populations and whether to the new aquarium for 2 weeks before trials began. All fish housed these correlations were different across populations. Our experimen- in experimental aquaria were held in the same controlled greenhouse tal design therefore allowed us to test for population-level variation unit, under equivalent ambient light conditions, and were fed in behavioral traits and assess whether there is geographic variation the same food once a day ad libitum. Prior to the start of the behav- in personality and correlated behaviors. Together, this research adds ioral assays, food was withheld for 24-h to standardize hunger. To to the body of work to date on P. multicolor that has focused on reduce stress from individually housing a social cichlid species, ex- population-level responses to individual stressors (e.g., DO or tur- perimental aquaria were positioned to allow for visual interactions bidity), while also examining within-population, individual vari- between individuals in neighboring aquaria. To eliminate the ability ation in behavioral traits that are important in responses to rapidly of fish to observe behavioral trials of other fish, an opaque barrier changing environments (Sih et al. 2011; Tuomainen and Candolin was used to visually isolate the selected experimental aquarium 2011). from surrounding aquaria housed on the same shelf. This barrier was placed surrounding the aquarium before the acclimation period began and was not removed until the conclusion of the behavioral Materials and Methods trial. Collection and rearing of study organisms Adult P. multicolor were collected from 2 sites within the Mpanga Trial procedure River basin (Uganda, Africa) that differ dramatically in levels of DO Before the start of a trial, a movable transparent, acrylic barrier was and turbidity. Bwera is a dense Cyperus papyrus swamp (hereafter placed into the aquarium, dividing it into 2 zones: an accessible area swamp) characterized by low DO (monthly mean6 SE ¼ 0.28 where the fish can swim freely to and from refuge (“refuge zone”; 1/ mgL O 6 0.1; Crispo and Chapman 2008) and clear, but tannin- 3 refuge þ 1/3 open space), and an inaccessible area which the fish stained water, typically <5.0 NTU (mean6 SE point-in-time meas- could not enter (“food zone”, 1/3 open space) (Figure 1, Appendix urements during time of fish collection: 1.09 NTU6 0.03). Bunoga Figure A1). Once the clear barrier was in place, fish were given is a river site characterized as having, on average, higher DO 15 min to acclimate. The trial began by adding food (0.275 g6 0.02 (monthly mean6 SE: 8.5 mgL O 6 0.1; McNeil et al. 2016) and SE food added per trial) to the isolated “food zone” in order to pro- higher turbidity (mean6 SE point-in-time measurements during mote fish to leave refuge. Once the fish was observed to leave refuge time of fish collection: 13.25 NTU6 0.1) compared with the and swim towards the partition and food, we simulated a predator swamp. Fish collected from swamp and river sites were transported attack using a model invertebrate predator [i.e., a large plastic insect from Uganda to The Ohio State University in July 2014 and held by (9.5 cm long) resembling a natural predator, Belostomatidae spp.] population under ambient light: dark cycles in aquaria located with- to illicit a fear response and scare the fish back into refuge. To initi- in Kottman Hall greenhouse (2021 Coffey Rd., Columbus Ohio), ate the predator attack, the model predator was introduced into the approved under The Institutional Animal Care and Use Committee “open water” zone of the aquarium where it was moved back and (IACUC) Protocol # 2014A00000055. forth to mimic searching behavior of the predator and thus scare the To create F1 progeny from wild-caught P. multicolor, 1 male fish. Once the fish retreated into refuge, the predator remained in and 3 female fish from the same population were placed in an iso- the aquarium for an additional 5–10 s before being removed. Upon lated aquarium to allow for natural reproduction. The maternal predator removal the clear barrier was also removed to allow fish mouth brooding behavior of P. multicolor allowed us to visually de- access to introduced food. All trials were video-recorded with a termine when a female started holding a brood (e.g., widely dis- V Canon Vixia HF R600 HD camcorder for post-trial analysis. Each tended jaw, refusal of food, and anti-social behavior). Once a female trial lasted 300 s from the time the simulated predator attack ended was noted to be mouth brooding she was removed and allowed to and barrier restricting food access was removed. further develop her brood in an isolated aquarium. The male was also replaced to maintain independent broods (i.e., because only one male was available per reproduction aquarium we knew the identity Behavior assays of each parent for each brood). This process allowed us to create To quantify and compare the behavior of P. multicolor individuals eight independent F1 broods (four broods with different parents per from each population, boldness, general activity, and foraging activ- population) of P. multicolor using N ¼ 16 fish (4 males: 4 females ity were quantified following a simulated predator attack as in pre- per population). Offspring were housed in a common garden envir- vious studies (Fraser and Gilliam 1987; Fraser et al. 2001; Wright onment for 3–6 months under normoxic (mean DO6 SE: 8.70 et al. 2006; Harcourt et al. 2009; Toms et al. 2010). We measured mgL O 6 0.76) and low turbidity (mean turbidity6SE: 0.49 boldness as the fish’s latency to leave refuge following the simulated NTU6 0.03 SE) conditions. Juvenile fish were fed Hikari First predator attack. We calculated boldness as 1 minus the proportion Bites daily ad libitum for 1 month, then weaned onto adult food of time before the fish emerged from the refuge following a simu- (crushed Tetra TetraMin Tropical Crisps) for the remainder of the lated predator attack. study. The fish was considered outside of refuge when it moved 1.5 Individuals from single broods were housed together in a single body lengths (measured in cm) away from the refuge. To measure aquarium (N ¼ 8 brood aquaria) until fish reached 3 months of age, general activity, we measured the proportion of time that an individ- after which 5 fish per brood per population (5 fish  4 broods  2 ual spent outside of refuge once it emerged from the refuge follow- populations; N ¼ 40 fish total) were randomly selected to be housed ing the simulated predator attack. individually in 19-L glass aquaria (Figure 1). Each aquarium con- Finally, we quantified foraging activity as the number of food tained a sponge filter and refuge (small plastic plant) placed at one pecks counted following the simulated predator attack and initial end of the aquarium such that one-third of the aquarium contained emergence from refuge. A single food peck was defined as a fast, for- refuge habitat and the remaining two-thirds was open water. Once ward strike at a food particle with an open mouth (masticating food placed in an experimental aquarium, fish were allowed to acclimate which was already in the mouth was not considered a food peck). Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 4 Current Zoology, 2018, Vol. 0, No. 0 Figure 1. Visual representation of the experimental aquaria. Aquaria are set up so 1/3 of the aquarium is an area of refuge (refuge zone), 1/3 of the aquarium is open water where the model predator is introduced to scare fish back into refuge, and the remaining 1/3 of the aquarium is the area where food is introduced (food zone). The food zone is initially blocked by a transparent removable barrier, then after the predator has been removed, the barrier is removed to allow fish to have access to food. To quantify the repeatability (i.e., consistent individual differen- identification) were determined to be nonsignificant, they were subse- ces in behavior) of these 3 behaviors (boldness, general activity, and quently removed from the model as well. foraging activity), we tested each individual fish 3 times, with a min- To test if there were correlations between the 3 measured behav- imum of 1 day between trials. At the conclusion of the final trial, we iors across populations (while accounting for differences in body measured the mass (g) and standard length (SL; cm) of each fish so size), we used separate multiple linear regressions (including log SL that we could account for differences in age and size between popu- as an additional predictor variable) to test each pair of behaviors. lations. We also calculated Fulton’s condition factor, K ¼ 100 To evaluate if the correlations between these behaviors differed be- [where W ¼ mass (g) and L ¼ standard length (cm); Fulton 1904; tween the swamp and river populations, we calculated Pearson cor- Froese 2006], to test if broods reared in the greenhouse at different relation coefficients separately for each population and then used times of the year displayed similar weight–length ratios. Fisher’s z-transformation to compare the 2 correlation coefficients from each population (Zar 1999). To test whether there were consistent individual differences in be- Statistical analyses havior, we calculated the repeatability, r, for each of the 3 behaviors Fish used for experimental trials varied significantly in age due to logis- via univariate ANOVA with individual as the fixed factor tical constraints; on average swamp fish were younger than river fish and behaviors as the dependent variables. We used the resulting (swamp: 116 days6 3.22 SE; river: 161 days6 4.28 SE;analysis of mean square values (MS and MS ) in the equation: within among variance (ANOVA): F ¼ 69.274, P< 0.001). This meant that, on 1,38 2 2 2 2 r ¼ s =ðs þ s Þ, where s is the variance among individuals, average, fish from each population were also of unequal size when A A A ðMS  MS Þ=n ,and s is the variance within individuals, MS A W 0 W tested: swamp fish were significantly smaller (mean mass: (Bell et al. 2009). It is important to include the term n for the number 0.91 g6 0.06 SE, mean standard length (SL): 2.95 cm6 0.06 SE)than of observations per individuals because repeatability changes system- river fish (mean mass: 1.27 g6 0.04 SE, mean SL: 3.38 cm6 0.04 SE; atically as the number of measurements per group increases, thus not ANOVA: mass: F ¼ 22.522, P<0.001; SL: F ¼ 34.502, 1,38 1,38 incorporating the number of observations per individual calculates re- P<0.001). Although the age and size of the 2 populations were differ- peatability incorrectly (Lessells and Boag 1987). The standard error ent, there was no significant difference in condition: Fulton’s condition, of repeatability was calculated using the following equation: K ¼ 3.456 0.07 SE for swamp fish and for river fish K ¼ 3.296 0.08 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SE (ANOVA: F ¼ 2.127, P ¼ 0.153). Although this demonstrates 1,38 21ðÞ  r ðÞ 1 þðÞ k  1 r that condition factor was similar across populations, we decided to in- SE r ¼ ½kkðÞ  1ðÞ N  1 clude standard length as a covariate in the following analyses to ac- count for the size differences apparent between populations and any where r is repeatability, k is number of measurements per individual ontogenetic behavioral shifts (see Bell and Stamps 2004). (or n ), and N is the number of individuals tested (Becker 1984). To test for population-level differences in behavioral responses to Analyses were performed in SPSS (version 24) and R statistical soft- the simulated predator attack, individual fish behaviors (i.e., boldness, ware (using Rstudio; version 1.1.419). general activity, and foraging activity) were averaged over 3 trials conducted per fish. For each behavioral response, we used an analysis of covariance (ANCOVA) with log-transformed SL as a covariate to Results test for differences in mean behaviors between swamp and river fish. Nonsignificant interaction terms (population  log SL) were removed Behavioral differences between populations from the models. Because some individuals were genetically related Variances for boldness and general activity were not homogeneous be- (i.e., siblings randomly selected from each brood), we also included tween populations, with behavioral variances being larger in the river brood identification (N ¼ 4 broods per population) as a random effect relative to the swamp population (Levene’s test for homogeneity of in our ANCOVA analysis. If the random effects (i.e., brood variances: boldness: F ¼ 6.113, P ¼ 0.018; activity: F ¼ 5.451, 1,38 1,38 Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 5 P ¼ 0.025), thus values for both were arcsine transformed, improving populations for general activity, with swamp fish being more active variance homogeneity for ANCOVA analysis. Foraging activity was than river fish (ANCOVA: F 5.390, P ¼ 0.026) (Figure 2B). 1, 36 ¼ not transformed since variances were homogeneous between the Finally, foraging activity did not differ among populations swamp and river population (Levene’s test for homogeneity of varian- (ANCOVA: F ¼ 1.888, P ¼ 0.178) (Figure 2C). However, the re- 1,37 ces: F ¼ 0.035, P ¼ 0.852). We included log SL as a covariate in the sult becomes significant (ANOVA: F ¼ 4.50, P ¼ 0.04) if we re- 1,38 1,38 ANCOVA models to account for any differences in size between popu- move the nonsignificant log SL covariate (F ¼ 0.045, P ¼ 0.833), 1,37 lations (see first paragraph of Statistical analysis for description of size suggesting that swamp fish had the tendency to peck at food particles differences). The interaction between population and log SL was not more often when out of refuge than did river fish (Table 1). significant for boldness and foraging activity and was therefore removed from those models; however, population  SL was a signifi- Correlated behaviors within and across populations cant interaction for general activity and was retained in the model (see Results of the multiple linear regression indicated that there was a Table 1). The random effect of brood number was not significant for significant positive relationship between boldness and foraging ac- boldness, general activity, or foraging activity and was therefore tivity across populations (Pearson’s r ¼ 0.399, t ¼ 2.172, removed from all models (Table 1). P ¼ 0.036), but no effect between log SL and foraging activity We found behavioral differences between the 2 populations des- (Pearson’s r¼0.248, t ¼0.597, P ¼ 0.554). Within each popu- pite being reared under common garden conditions and largely con- lation, we only found a positive correlation between foraging activ- sistent with pace of life predictions based on known divergent ity and boldness for river fish (Pearson’s r ¼ 0.633, t ¼ 3.467, characters. First, we found that swamp fish left refuge significantly P ¼ 0.003) (Figure 3A), but not for the swamp fish (Pearson’s earlier than did river fish (ANCOVA: F ¼ 12.36, P ¼ 0.001) 1,37 r¼0.028, t ¼0.118, P ¼ 0.907) (Figure 3A). The relationship (Table 1, Figure 2A) suggesting that on average swamp fish are more between boldness and foraging activity was significantly different bold than river fish. Second, there was a significant difference between between the 2 populations (Z¼2.26, P ¼ 0.024). Across populations, there was a significant positive relationship between general activity and foraging activity (Pearson’s r ¼ 0.376, Table 1. Results of ANCOVAs testing population-level differences t ¼ 2.40, P ¼ 0.021), but no relationship between log SL and forag- in boldness, general activity, and foraging activity across swamp ing activity (Pearson’s r¼0.248, t ¼1.47, P ¼ 0.151). Within and river populations. 37 each population, there was no significant correlation between forag- Behavior Model df F P* ing activity and general activity for either population (swamp: Pearson’s r ¼ 0.408, t ¼ 1.894, P ¼ 0.074; river: Pearson’s r ¼ 0.293, Boldness Population 1, 37 12.357 0.001 t ¼ 1.301, P ¼ 0.210) (Figure 3B). The relationship between general log SL 1, 37 0.007 0.934 activity and foraging activity was not significantly different between Brood number 6, 31 1.139 0.363 Population  log SL 1, 36 2.751 0.106 the swamp and river populations (Z ¼ 0.38, P ¼ 0.701) General Activity Population 1, 36 5.390 0.026 Finally, we found no relationship between boldness and general log SL 1, 36 3.060 0.089 activity across populations (Pearson’s r ¼ 0.222, t ¼ 1.28, Brood number 6, 30 1.576 0.219 P ¼ 0.207), but there was a significant negative relationship between Population  log SL 1, 36 4.782 0.035 log SL and general activity (Pearson’s r¼0.417, t ¼2.73, Foraging Activity Population 1, 37 1.888 0.178 P ¼ 0.01). Within each population there was not a significant rela- log SL 1, 37 0.045 0.833 tionship between boldness and general activity for either population Brood number 6, 31 1.149 0.358 (swamp: Pearson’s r ¼ 0.304, t ¼ 1.24, P ¼ 0.232; river: Pearson’s Population  log SL 1, 36 0.080 0.779 r¼0.033, t ¼0.789, P ¼ 0.44) (Figure 3C) and the correlation *Values in bold are significant at a ¼ 0.05. coefficients did not differ (Z ¼ 1.01, P ¼ 0.311). Figure 2. Behavioral responses in 2 different F1 P. multicolor populations, swamp (open bars) and river (solid bars). Estimated marginal means (6 SE)of (A) bold- ness (arcsine transformed); (B) General activity (arcsine transformed); and, (C) foraging activity. *Values are significantly different at a ¼ 0.05. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 6 Current Zoology, 2018, Vol. 0, No. 0 population differences in behavior might be associated with differ- ences in physiological traits and life-history strategies that may also diverge in response to human-induced rapid environmental change, HIREC (Sih et al. 2011). Here, we found divergent variation in be- havioral traits between 2 populations that experience extremes of hypoxia and turbidity and also found evidence that some, but not all, behaviors were repeatable and correlated within populations. Under common garden conditions (i.e., released from major stres- sors) fish originating from the hypoxic, clear swamp habitat that have been shown to have higher metabolic scope were, on average, more bold and active than fish from the normoxic, turbid river that have lower metabolic scope. Across populations, we found positive correlations between boldness and foraging activity, as well as be- tween foraging and general activity. Thus, individuals that were more active foragers were also bolder and generally more active. However, within populations, boldness and foraging activity were only positively correlated within the river population, but not corre- lated within the swamp population. Boldness and foraging activity were also repeatable behaviors within the river population, but only foraging activity was a repeatable behavior in the swamp popula- tion. Our results suggest that P. multicolor populations that experi- ence different extremes of environmental stressors behave differently. We discuss these results in the context of population- and individual-level behavior trait differences in fish known to be di- vergent in a number of other traits associated with 2 globally im- portant environmental stressors, hypoxia and turbidity. The observed differences in boldness and general activity be- tween populations that are physiologically adapted to deal with extremes in oxygen availability and water clarity is consistent with the pace-of-life concept, which predicts that individuals with a higher metabolic scope will be more bold and active than individuals with a lower metabolic scope. However, this finding requires consid- eration of the testing conditions in our experiment relative to the en- Figure 3. The relationships between (A) foraging activity and boldness, vironmental conditions in the home habitat of each population. In (B) foraging activity and general activity, and (C) boldness and general activ- their home environment, swamp fish face energetic challenges asso- ity for the swamp (open circles, long dashed lines) and river (closed circles, ciated with low DO (Chapman 2015) and have adapted (through solid line) populations, and with regression lines for both populations pooled both plastic and genetic mechanisms) to these environments by hav- (small dashed line). ing lower routine metabolic rates, larger gills, and smaller brains than river fish from high DO habitats (Reardon and Chapman Repeatability of boldness and activity 2009; Crispo and Chapman 2010). Furthermore, swamp fish have The swamp and river populations both exhibited significant repeat- also been shown to be less active and display fewer reproductive ability in at least 1 behavior; however, there were differences be- behaviors than river fish when in their low DO home environment tween the populations in repeatability values. The swamp (e.g., Gotanda et al. 2011; McNeil et al. 2016). Therefore, it might population showed significant repeatability in foraging behavior make sense to expect swamp fish to demonstrate a slower pace of (r ¼ 0.3476 0.147 SE; ANOVA: F ¼ 2.596, P ¼ 0.006), but 19,60 life relative to river fish, but only under home conditions. In con- did not show significant repeatability in bold behavior trast, under the novel rearing conditions (i.e., high DO/clear) of this (r ¼ 0.1386 0.146 SE; ANOVA: F ¼ 1.495, P ¼ 0.140) or gen- 19,60 study, we predicted that when released from the constraint of a low eral activity (r ¼ 0.1286 0.145 SE; ANOVA: F ¼ 1.479, 19,60 DO environment, swamp fish would be bolder and more active P ¼ 0.147). The river population also showed significant repeatabil- given their adaptations to an energetically challenging environment ity in foraging behavior (r ¼ 0.3746 0.145 SE; ANOVA: (i.e., due to adaptations to low DO that would increase performance F ¼ 2.794, P ¼ 0.003) as well as boldness (r ¼ 0.3496 0.146 SE; 19,60 in high DO conditions, e.g., Reardon and Chapman 2010). Ursza ´n ANOVA: F ¼ 2.632, P ¼ 0.005), but did not show significantly 19,60 et al. (2015) also suggest that a fixed behavioral strategy, such as we repeatability for general activity (r ¼ 0.0496 0.138 SE; ANOVA: see in the swamp population, would be less energetically expensive F ¼ 1.149, P ¼ 0.345). 19,60 compared with greater intra-individual variation. Based on this rea- soning, river fish adapted to a more turbid environment might there- fore be relatively less active compared with swamp fish in high DO Discussion and additionally, may be less bold if clear water induces a fear re- Animals are increasingly faced with human-induced stressors that sponse (e.g., see Lima and Dill 1990). Indeed, this is what we found: vary in space and time, thus we can expect population-level diver- fish from the swamp left refuge sooner after a simulated predator at- gence in behaviors that help animals to cope with environmental tack, were more active throughout the trial, and displayed marginal- change that varies across the landscape. We also expect that ly higher foraging activity levels than fish originating from the river Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 7 (lower metabolic scope, slower life history) (Figure 2). These popu- It is also possible that other ecological factors, both abiotic (e.g., lation differences suggest a relationship between behavior and life temperature, flow rate) and biotic (e.g., density, competition effects, history, and that low DO and high turbidity are strong selective predator abundance), may contribute to behavioral trait divergence agents in this system. Evidence of other environmental extremes across P. multicolor populations. While our knowledge of the abiot- driving intraspecific, population-level variation in behavioral traits, ic factors that vary between swamp and river habitats (such as DO, and relationships between behaviors and/or physiological and life- turbidity, temperature) is extensive and comes from monthly sam- history traits is accumulating (e.g., Bell 2004; Moran et al. 2016; pling over 15 years (L.J. Chapman, personal communication), our Dubuc-Messier et al. 2017). knowledge of the biological characteristics of each habitat is less While our results suggest a relationship between behavior trait complete. For example, while we know that fish from both types of differences and population of origin, and a possible link to known habitat experience similar suites of predators (e.g., wading birds, divergent physiological traits, they do not allow us to differentiate snakes, and Clarias catfish), we do not know whether there are dif- between the mechanisms driving this association. For example, ferences in predation pressure exerted by these predators across the observed differences in behavior could be due to physiological adap- 2 habitats (e.g., due to differences in density, habitat complexity, tations to these stressors. In common carp Cyprinus carpio, for in- etc.). Variation in predation pressure is well known to drive stance, fish with higher metabolic rates and lower cortisol receptor population-level differences in behavior (e.g., Endler 1980) and in expression in the brain were found to be bolder and more aggressive, correlations between suites of behavior (e.g., Fraser et al. 2001). suggesting that variation in physiology is tightly linked to behavioral Future work distinguishing among these and other mechanisms will expression (Huntingford et al. 2010). In a direct test of the link be- greatly enhance our understanding of the variables that shape diver- tween physiological and behavioral traits, Binder et al. (2016) found gent behavior (Bell 2007). that individual bluegill sunfish L. macrochirus with higher aerobic Our results also suggest that correlations between boldness and capacity were bolder than individuals with lower aerobic capacity. foraging activity is not a general pattern across P. multicolor popula- Additionally, there is evidence that oxidative stress resulting from tions and may be associated with the differences in the selective pres- prolonged exposure to elevated stress hormones (i.e., glucocorti- sures experienced in swamp and river habitats. Unfortunately, coids) is linked to behavioral coping strategies (Costantini et al. comparing only 2 populations from different environments cannot 2011). However, in each of these examples, we still do not know if definitively tell us what factors drive these differences. However, physiological differences drive behavioral expression or vice versa. our results in combination with past work demonstrating other trait We know that oxygen concentration and turbidity level are likely differences across multiple populations of P. multicolor (e.g., strong selective agents driving divergence between swamp and river Chapman et al. 2000; Reardon and Chapman 2010; Crispo and populations given previous work and the results presented here, and Chapman 2010) suggest that differences in DO and water clarity we also know that environmental conditions in the swamp tend to might influence whether behaviors are correlated or not. More be relatively stable whereas they fluctuate more in rivers. In swamp broadly, this may suggest that correlations between behaviors may populations, the relative stability of the environment in concert with depend on context. For example, a positive correlation between consistently low DO would favor traits that help minimize energy boldness and foraging activity might be particularly important in expenditure, for example, lower metabolic rates, lower activity, and river environments that are much more turbid, on average, than bold behavior since attaining resources needs to be efficient. swamps, but experience fluctuating turbidity. For instance, during However, this combination of an energetically challenging but stable periods of high turbidity it is possible that the perceived risk of pre- environment is unlikely to favor tight linkages between traits if those dation is low and so there is an advantage for bold individuals that linkages might constrain energetic coping mechanisms. On the other are more active foragers. Alternatively, periods of lower turbidity, hand, in river populations, we might expect that the visual impair- when predation risk might be higher because fish are more visible to ment resulting from high turbidity favors decreased predator vigi- predators, individuals with a strong anti-predator response that lance or assessment of predation risk (Lima and Dill 1990), but spend more time hiding than feeding might be favored. It is then during periods of high water clarity (as imposed in our rearing envir- possible that the relative stability of the clear, low DO swamp sites onment) this might create a higher perceived risk and thus favor shy do not favor such a trade-off which is why we do not see a correl- behavior. ation between boldness and foraging activity in that population. In a Alternatively, differences in both behavior and physiology could similar examination of boldness and activity coupled with aggres- be due to plasticity. Binder et al. (2016) suggest that plasticity in aer- sion in threespine stickleback Gasterosteus aculeatus, Bell (2004) obic metabolism could account for the positive correlation between found differences in mean behaviors across very different environ- metabolic rate and boldness. While the expression of behavior trait ments (e.g., high flow and temperature variation in an unregulated differences here under common garden conditions suggests some stream vs. low variation in a damned and regulated stream). She heritability, we also know that P. multicolor can display behavioral also discovered that in sticklebacks from the more variable environ- flexibility. For example, in a reciprocal rearing study, P. multicolor ment, behaviors were often correlated across contexts, whereas males displayed higher rates of aggression under turbid conditions there were no significant correlations between behaviors in fish orig- regardless of whether they originated from swamp or river popula- inating from the more stable population. Our results suggest that tions or if they were reared under clear or turbid conditions (Gray there are population-level differences in the way that behavioral et al. 2012). Similarly, Reardon and Chapman (2010) demonstrate traits have (co)evolved, similar to the differences found across developmental plasticity in routine metabolic rate when swamp fish stickleback populations by Bell (2004). were reared in high or low DO. Further explorations of how mul- Consistent behavioral variation, or repeatability, is known to tiple stressors that vary among populations might shape behavior occur within and between individuals across time or ecological con- and that directly test for associations between behavioral and texts (Dall et al., 2004; Sih et al., 2004; Sih and Bell, 2008). In our physiological traits could be informative under the pace-of-life study, river-origin fish displayed repeatability across 2 contexts concept. (boldness and foraging activity) whereas the swamp-origin fish only Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 8 Current Zoology, 2018, Vol. 0, No. 0 displayed repeatability in foraging activity. Divergent habitats, as References well as habitats with large environmental variability, may favor dif- Abrahams MV, Robb TL, Hare JF, 2005. Effect of hypoxia on opercular dis- ferences in behavioral traits and the repeatability of behavior, plays: evidence for an honest signal? Anim Behav 70:427–432. assuming limited plasticity. In this system, swamp sites exhibit lower Atwell JW, Cardoso GC, Whittaker DJ, Price TD, Ketterson ED, 2014. variation in environmental factors (i.e., are more stable), whereas Hormonal, behavioral, and life-history traits exhibit correlated shifts in rela- environmental factors tend to fluctuate in the river habitats (e.g., tion to population establishment in a novel environment. Am Nat 184: Crispo and Chapman 2010). Behaviors which show relatively low 147–160. Becker WA, 1984. A Manual of Quantitative Genetics. Pullman: Academic within-individual variance compared with between-individual vari- Enterprises. ance are more repeatable (Bell et al. 2009). If, however, an environ- Bell AM, 2004. Behavioural differences between individuals and ment has relatively low environmental variability, it is possible that two populations of stickleback Gasterosteus aculeatus. J Evol Biol 18: this would select for a specific behavioral type that is better suited 464–473. for the environment, thus reducing between-individual variance. Bell AM, 2007. Future directions in behavioural syndromes research. Proc R Based on our tests for homogeneity of variance between popula- Soc B Biol Sci 274:755–761. tions, we found that in the swamp population 2 of the 3 behaviors Bell AM, Hankison SJ, Laskowski KL, 2009. The repeatability of behaviour: a (boldness and general activity) had lower between-individual vari- meta-analysis. Anim Behav 77:771–783. ance compared with the river population. In the swamp population, Bell AM, Stamps JA, 2004. Development of behavioural differences between low environmental variation in addition to low DO, may especially individuals and populations of sticklebacks Gasterosteus aculeatus. Anim Behav 68:1339–1348. favor a more fixed behavioral type, given energetic constraints Binder TR, Wilson AD, Wilson SM, Suski CD, Godin J-GJ et al., 2016. Is there (Ursza ´ n et al. 2015). Alternatively, environments with relatively a pace-of-life syndrome linking boldness and metabolic capacity for locomo- large environmental variation may select for a suite of different tion in bluegill sunfish? Anim Behav 121:175–183. behaviors and possibly increase between-individual variance. Brown C, Jones F, Braithwaite V, 2005. In situ examination of Intraspecific divergence in behavior across populations experi- boldness-shyness traits in the tropical poeciliid Brachyraphis episcopi. Anim encing environmental extremes has been found in a number of taxa Behav 70:1003–1009. (e.g., Brown et al. 2005; Moran et al. 2016). As an example, van Bruton MN, 1985. The effects of suspensoids on fish. Hydrobiologia 125: Oers et al. (2004) found that great tits Parus major show heritable 221–241. variation in risk-taking behavior when high- and low risk-taking Careau V, Thomas D, Humphries MM, Re ´ ale D, 2008. Energy metabolism lines were selected in the laboratory over 2 generations. Similarly, and animal personality. Oikos 117:641–653. Chapman LJ, 2015. Low-oxygen lifestyles in extremophilic fishes. In: Reisch R, Brown et al. (2005) found heritable variation in boldness among Plath M, Tobler M, editors. Extremophile Fishes - Ecology and Evolution of populations experiencing different levels of predation in a tropical Teleosts in Extreme Environments. Heidelberg: Springer. 9–33. poecilliid fish, Brachyraphis episcopi, when F1s from 2 populations Chapman LJ, Albert J, Galis F, 2008. Developmental plasticity, genetic differ- were reared in the laboratory and exposed to different predation entiation, and hypoxia-induced tradeoffs in an African cichlid fish. Open regimes. By investigating the underlying behavioral responses from Evol J 2:75–88. an individual perspective within divergent populations we gain a Chapman LJ, Chapman CA, 2003. Fishes of the African rain forests: emerging mechanistic understanding of how behavioral responses, linked to and potential threats to a little-known fauna. In: Chrisman TL, Chapman physiological adaptation to environmental change, might contribute LJ, Chapman CA, Kaufman LS, editors. Conservation, Ecology, and to population persistence under human-induced rapid environmen- Management of African Fresh Waters. Gainesville: University Press of tal change, HIREC (Koolhaas et al. 1999; Dingemanse and Wolf Florida. 176–209. Chapman LJ, Chapman CA, Nordlie FG, Rosenberger AE, 2002. 2010; Frost et al. 2013; Sih et al. 2015, Wong and Candolin 2015). Physiological refugia: swamps, hypoxia tolerance and maintenance of fish Here, we have demonstrated behavioral trait divergence between popu- diversity in the Lake Victoria region. Comp Biochem Phys A Mol Integr lations and that trade-offs between behavioral traits vary among these Physiol 133:421–437. populations. Remaining to be investigated is how these behavioral trait Chapman LJ, Galis F, Shinn J, 2000. Phenotypic plasticity and the possible differences translate into advantages in the respective home environ- role of genetic assimilation: hypoxia-induced trade-offs in the morphologic- ments of swamp and river fish. al traits of an African cichlid. Ecol Lett 3:387–393. Costantini D, Marasco V, Møller AP, 2011. A meta-analysis of glucocorti- coids as modulators of oxidative stress in vertebrates. J Comp Physiol B Acknowledgments 181:447–456. Crispo E, Chapman LJ, 2008. Population genetic structure across dissolved We thank the Commissioner of Fisheries Resources Management and oxygen regimes in an African cichlid fish. Mol Ecol 17:2134–2148. Development, Uganda, for permission to export live specimens, the Uganda Crispo E, Chapman LJ, 2010. Geographic variation in phenotypic plasticity in National Council for Science and Technology (UNCST) for research permis- response to dissolved oxygen in an African cichlid fish. J Evol Biol 23: sion, and Makerere University Biological Field Station for logistical support. 2091–2103. We additionally thank Dr. Lauren Chapman for use of research facilities in Crocker CD, Chapman LJ, Martinez ML, 2013. Hypoxia-induced plasticity in Uganda (Lake Nabugabo Research Station) and mentorship throughout this the metabolic response of a widespread cichlid. Comp Biochem Phys B 166: project; Ugandan field assistants for invaluable help whilst in Uganda; and, J. 141–147. Odegard for assistance with fish care. Dall SRX, Houston AI, McNamara JM, 2004. The behavioural ecology of per- sonality: consistent individual differences from an adaptive perspective. Ecol Lett 7:734–739. Funding Dingemanse NJ, Wolf M, 2010. Recent models for adaptive personality differ- ences: a review. Philos T Roy Soc B 365:3947–3958. Funding was provided by the American Cichlid Association Guy D. Jordan Dubuc-Messier G, Re ´ ale D, Perret P, Charmantier A, 2017. Environmental Research Fund and The Ohio State University Undergraduate Student heterogeneity and population differences in blue tits personality traits. Government Academic Enrichment Grant to R.C.O., The Ohio State Behav Ecol 28:448–459. University Office of International Affairs Faculty Seed Grant to S.M.G. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 Oldham et al.  Cichlid behavior across extreme environments 9 Dugas MB, Franssen NR, 2011. Nuptial coloration of red shiners Cyprinella Moran NP, Mossop KD, Thompson RM, Wong BB, 2016. Boldness in ex- lutrensis is more intense in turbid habitats. Naturwissenschaften 98: treme environments: temperament divergence in a desert-dwelling fish. 247–251. Anim Behav 122:125–133. Dugas MB, Franssen NR, 2012. Red shiners Cyprinella lutrensis have larger Newcombe CP, MacDonald DD, 1991. Effects of suspended sediments on eyes in turbid habitats. Can J Zool 90:1431–1436. aquatic ecosystems. N Am J Fish Manage 11:72–82. Edelaar P, Jovani R, Gomez-Mestre I, 2017. Should I change or should I go? Re ´ ale D, Reader SM, Sol D, McDougall PT, Dingemanse NJ, 2007. Phenotypic plasticity and matching habitat choice in the adaptation to envir- Integrating animal temperament within ecology and evolution. Biol Rev 82: onmental heterogeneity. Am Nat 190:506–520. 291–318. Endler JA, 1980. Natural selection on color patterns in Poecilia reticulata. Re ´ ale D, Garant D, Humphries MM, Bergeron P, Careau V et al., 2010. Evolution 34:76–91. Personality and the emergence of the pace-of-life syndrome concept at the Fraser DF, Gilliam JF, 1987. Feeding under predation hazard: response of the population level. Phil Trans R Soc B 365:4051–4063. guppy and hart rivulus from sites with contrasting preadtion hazard. Behav Reardon EE, Chapman LJ, 2009. Hypoxia and life-history traits in a eurytopic Ecol Sociobiol 21:203–209. African cichlid. J Fish Biol 75:1795–1815. Fraser DF, Gilliam JF, Daley MJ, Le AN, Skalski GT et al., 2001. Explaining Reardon EE, Chapman LJ, 2010. Energetics of hypoxia in a mouth-brooding leptokurtic movement distributions: intrapopulation variation in boldness cichlid: evidence for interdemic and developmental effects. Physiol Biochem and exploration. Am Nat 158:124–135. Zool 83:414–423. Froese R, 2006. Cube law, condition factor and weight-length relationships: Sih A, Bell AM, 2008. Insights for behavioral ecology from behavioral syn- history, meta-analysis and recommendations. J Appl Ichthyol 22:241–253. dromes. In: Brockmann HJ, Roper TJ, Naguib M, Wynne-Edwards KE, Frost AJ, Thomson JS, Smith C, Burton HC, Davis B et al., 2013. Barnard C et al., editors. Advances in the Study of Behavior. San Diego: Environmental change alters personality in the rainbow trout Elsevier Academic Press Inc. 227–281. Oncorhynchus mykiss. Anim Behav 85:1199–1207. Sih A, Bell AM, Johnson JC, Ziemba RE, 2004. Behavioral syndromes: an Fulton TW, 1904. The Rate of Growth of Fishes. 22nd Annual Report, intergrative overiew. Q Rev Biol 79:241–277. Fisheries Board of Scotland, Edinburgh. Sih A, Ferrari MCO, Harris DJ, 2011. Evolution and behavioural responses to Gotanda KM, Reardon EE, Chapman LJ, 2011. Hypoxia and male behaviour human-induced rapid environmental change. Evol Appl 4:367–387. in an African cichlid Pseudocrenilabrus multicolor victoriae. J Fish Biol 78: Sih A, Mathot KJ, Moiro ´ n M, Montiglio P-O, Wolf M et al., 2015. Animal 2085–2092. personality and state - behaviour feedbacks: a review and guide for empiri- Gray SM, McDonnell LH, Cinquemani FG, Chapman LJ, 2012. As clear as cists. Trends Ecol Evol 30:50–60. mud: turbidity induces behavioral changes in the African cichlid Toms CN, Echevarria DJ, Jouandot DJ, 2010. A methodological review of Pseudocrenilabrus multicolor. Curr Zool 58:146–157. personality-related studies in fish: focus on the shy-bold axis of behavior. Int Harcourt JL, Sweetman G, Johnstone RA, Manica A, 2009. Personality J Comp Psychol 23: 1–25. counts: the effect of boldness on shoal choice in three-spined sticklebacks. Tuomainen U, Candolin U, 2011. Behavioural responses to human-induced Anim Behav 77:1501–1505. environmental change. Biol Rev 86:640–657. Hendry AP, 2015. Key questions on the role of phenotypic plasticity in Ursza ´ n TJ, To¨ro ¨ k J, Hettyey A, Garamszegi LZ, Herczeg G, 2015. eco-evolutionary dynamics. J Hered 107:25–41. Behavioural consistency and life history of Rana dalmatina tadpoles. Oecol Huntingford FA, Andrew G, Mackenzie S, Morera D, Coyle SM et al., 2010. 178:129–140. Coping strategies in a strongly schooling fish, the common carp Cyprinus Utne-Palm AC, 2002. Visual feeding of fish in a turbidenvironment: physical carpio. J Fish Biol 76:1576–1591. and behavioural aspects. Mar Freshw Behav Phy 35:111–128. Killen SS, Mitchell MD, Rummer JL, Chivers DP, Ferrari MC et al., 2014. van der Sluijs I, Gray SM, Amorim MCP, Barber I, Candolin U et al., 2011. Aerobic scope predicts dominance during early life in a tropical damselfish. Communication in troubled waters: responsesof fish communication sys- Funct Ecol 28:1367–1376. tems to changing environments. Evol Ecol 25:623–640. Koolhaas JM, Korte SM, De Boer SF, Van Der Vegt BJ, Van Reenen CG et al., van Dongen WFD, Maldonado K, Sabat P, Va ´ squez RA, 2010. Geographic 1999. Coping styles in animals: current status in behavior and stress-physi- variation in the repeatability of a personality trait. Behav Ecol 21: ology. Neurosci Biobehav R 23:925–935. 1243–1250. Lessells CM, Boag PT, 1987. Unrepeatable repeatabilities: a common mistake. van Oers K, de Jong G, van Noordwijk AJ, Kempenaers B, Drent PJ, 2004. Auk 104:116–121. Contribution of genetics to the study of animal personalities: a review of Lima SL, Dill LM, 1990. Behavioral decisions made under the risk of preda- case studies. Behaviour 142:1185–1206. tion: a review and prospectus. Can J Zool 68:619–640. Weins KE, Crispo E, Chapman LJ, 2014. Phenotypic plasticity is maintained Mazue ´ GPF, Dechaume-Moncharmont F-X, Godin J-GJ, 2015. Boldness - ex- despite geographical isolation in an African cichlid fish Pseudocrenilabrus ploration behavioral syndrome: interfamily variability and repeatability of multicolor. Integr Zool 9:85–96. personality traits in the young of the convict cichlid Amatitlania siquia. Wikelski M, Spinney L, Schelsky W, Scheuerlein A, Gwinner E, 2003. Slow Behav Ecol 26:900–908. pace of life in tropical sedentary birds: a common-garden experiment on McNeil GV, Friesen CN, Gray SM, Aldredge A, Chapman LJ, 2016. Male col- four stonechat populations from different latitudes. Proc R Soc Lond B 270: our variation in a eurytopic African cichlid: the role of diet and hypoxia. 2383–2388. Biol J Linn S 118:551–568. Wilson ADM, McLaughlin RL, 2007. Behavioural syndromes in brook charr, Mitchell DJ, Biro PA, 2017. Is behavioural plasticity consistent across Salvelinus fontinalis: prey-search in the field corresponds with space use in different environmental gradients and through time? Proc R Soc B 284: novel laboratory situations. Anim Behav 74:689–696. 20170893. Wong BBM, Candolin U, 2015. Behavioral responses to changing environ- Mittlebach GG, Ballew NG, Kjelvik MK, 2014. Fish behavioral types and ments. Behav Ecol 26:665–673. their ecological consequences. Can J Fish Aquat Sci 71:927–944. Wright D, Nakamichi R, Krause J, Butlin RK, 2006. QTL analysis of behav- Montiglio PO, Garant D, Bergeron P, Messier GD, Re ´ ale D, 2014. Pulsed ioral and morphological differentiation between wild and laboratory zebra- resources and the coupling between life-history strategies and exploration fish Danio rerio. Behav Genet 36:271–284. patterns in eastern chipmunks Tamias striatus. J Anim Ecol 83:720–728. Zar JH, 1999. Biostatistical Analysis. Upper Saddle River: Prentice-Hall Inc. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018 10 Current Zoology, 2018, Vol. 0, No. 0 APPENDIX Figure A1. Experimental progression of an actual trial: (A) “food addition” phase, which lured the tested fish (circled in red) out of refuge in attempt to consume the introduced food on the adjacent side of the clear barrier, (B) “predator introduction” phase introduces the model predator which scares test fish back into ref- uge, (C) “behavioral observation” phase removes the predator and clear barrier to allows fish to have access to food. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zoy027/4969520 by Ed 'DeepDyve' Gillespie user on 13 July 2018

Journal

Current ZoologyOxford University Press

Published: Apr 12, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off