Environmental biology of fishes, tập 93, số 1, 2012

2011 Abstract Competition has broad effects on fish and specifically the effects of competition on the prey capture kinematics and behavior are important for the assessment of future pre

Feeding ecology and prey preferences of a piscivorous fish

in the Lagoa do Peixe National Park, a Biosphere Reserve

in Southern Brazil

Fabiano Corrêa&Marlucy Coelho Claudino&

Rodrigo Ferreira Bastos&Sônia Huckembeck&

Alexandre Miranda Garcia

Received: 17 June 2010 / Accepted: 27 June 2011 / Published online: 13 July 2011

# Springer Science+Business Media B.V 2011

Abstract We investigated the diet, feeding strategy,

size-related dietary shifts and prey preferences of

South American Hoplias aff malabaricus in an

internationally recognized but poorly investigated

Biosphere Reserve in southern Brazil Fish were

caught between April 2008 and March 2009 using a

variety of fishing gear The analysis of 113 individuals

revealed a diet essentially composed of fish (16

species), particularly characid species (9) The diet

became more diverse and contained larger fish prey

with increasing predator size Feeding strategy

anal-ysis revealed a clear specialization towards the

consumption of fish However, individuals did not

prey upon particular prey species, instead

opportunis-tically consuming many different fish species, which

could be a strategy to avoid intraspecific competition

Characid species were the most important prey,

followed by poecillids A multi-gear sampling of the

ichthyofauna revealed that these prey species were the

most abundant (Characidae: 61.3%, Poeciliidae

18.8%) of the 14 fish families occurring at the study

site, suggesting that the predator exploits the most

abundant fish resources available rather than the rarer

fish prey These findings suggest that potential down controls exerted by H aff malabaricus in thissystem follow specific food web pathways that seem

top-to be mediated by the abundance of prey resources

Keywords Diet Feeding strategy Size-related dietaryshift Trahira Food niche Characins

Introduction

Piscivorous fish play an important role in the array ofecosystem services provided by ichthyofauna due to theirregulation of prey populations and, therefore, their impact

on trophic cascades (Helfman et al 2009) Suchinfluences can have crucial effects on the functioning

of aquatic ecosystems and the levels of biodiversity theysustain (Polis and Winemiller 1996; Mazzeo et al

2010) Temporal variations in predator abundance have

a profound impact on not only the abundance of aquaticprey but also on adjacent terrestrial ecosystems viaindirect interactions with terrestrial consumers (Hebert et

al.2008) Studying the feeding ecology of piscivorousfish is, therefore, a key step towards understanding theprocesses driving biodiversity in aquatic ecosystems.Such knowledge is critically relevant in conservationareas to guide managers and decision makers to copewith anthropogenic impacts, such as when exoticspecies threaten native ones (Zaret and Paine1973).Lagoa do Peixe National Park (LPNP) is a poorlyinvestigated conservation area in the coastal plain of

DOI 10.1007/s10641-011-9881-4

Instituto de Oceanografia, Laboratório de Ictiologia,

Universidade Federal de Rio Grande,

Campus Carreiros, Caixa Postal 474, CEP 96.201-900 Rio

Grande, RS, Brazil

e-mail: correafecologia@yahoo.com.br

e-mail: amgarcia@mikrus.com.br

southern Brazil which harbors many diverse and

productive aquatic fauna, including endangered fish

species such as the annual fish Austrolebias minuano

(Corrêa et al 2009) It was recognized as a Ramsar

Site and a UNESCO Biosphere Reserve in 1993,

largely due to its important ecological role as a

stopover site for migratory shorebirds in South

America (Tagliani 1995) Despite the importance of

the LPNP importance for conservation, no prior

information on fish trophic ecology or aquatic food

web structure is available for this unique ecosystem

The most abundant piscivorous fish in the

fresh-water wetlands of the LPNP is the South American

Hoplias aff malabaricus (Bloch, 1794) (Loebmann

and Vieira 2005) This perch-like piscivore of the

Erythrinidae family is distributed from Costa Rica

(Central America) to Southern South America

(Argentina) and is well known for having a strong

top-down structuring role on trophic food webs of

lentic aquatic systems (Winemiller 1989; Mazzeo et

al 2010) Although it is well adapted to lentic

environments, it is also found in small and large

rivers, especially in pools with abundant marginal

vegetation It is cryptically colored and captures its

prey by ambush and is active mainly at night

(Loureiro and Hahn 1996) It is considered a top

predator with trophic levels ranging from 2.4 to 4

(Garcia et al.2006; Rodríguez-Graña et al.2008)

Mesocosm experiments revealed that H aff

malabaricus can exert a strong control on

planktivo-rous fish with cascading effects on chlorophyll a,

water turbidity and zooplankton abundance (Mazzeo

et al 2010) These authors recommended, however,

that additional studies be carried out in whole-lake

systems because the simplified conditions and small

scale of mesocosm experiments hinder direct

extrap-olation Also, the high complexity of the food web

and spatial heterogeneity of natural aquatic systems in

subtropical and tropical regions represent additional

constraints and should be investigated However, prior

studies on the diet and feeding behavior of H aff

malabaricus have been restricted to streams and river

flood plains, with no known studies on subtropical

freshwater wetlands For example, field studies

con-ducted on streams and river flood plains have shown

that small fish usually comprise the main prey of H

aff malabaricus, although invertebrates, especially

insects, vegetation remains and organic detritus can

also be found in the stomach contents of H aff

malabaricus (Loureiro and Hahn 1996; Carvalho et

al 2002; Peretti and Andrian 2008; Corrêa andPiedras2009) Size-related diet shifts also occur, withsmall juveniles feeding primarily on invertebrates,while subadults and adults are piscivorous (Winemiller

1989) However, these prior studies did not uously show the fish prey preferences of the H aff.malabaricus under natural conditions because theauthors did not carry out a concomitant evaluation offish prey availability in the study site

unambig-Considering the crucial role that H aff icus can have on aquatic food webs and the lack ofinformation about its feeding ecology on subtropicalwetlands, the present study investigates its diet,feeding strategy and size-related dietary shifts in awetland located within an important conservation area

malabar-in southern Brazil Also, based on a concomitantmulti-gear sampling of the ichthyofauna in the studysite, our work reveals the prey preferences of this toppredator under natural field conditions

Material and methods

Study area

The LPNP was created in 1986 to protect crucialfeeding and resting sites for migratory birds along thecoast of Rio Grande do Sul state in southern Brazil(Fig 1) The LPNP has a humid subtropical climate,with mean temperatures ranging from 14.6°C inwinter to 22.2°C in summer, and a mean annualtemperature of 17.5°C The annual precipitationranges from 1150 to 1450 mm yr−1, with an annualmean of 1250 mm yr−1 (Tagliani 1995) The mainlagoon (locally known as ‘Lagoa do Peixe’) consistsprimarily of shallow areas (<50 cm), except in itschannel and at its communication with the AtlanticOcean in its central portion where it can reach a depth

of 2 m (Loebmann and Vieira2005) The connection

of the lagoon with the sea often remains closed forseveral months each year until excess freshwater,usually from rainfall during winter months, leads tothe opening of the mouth of the estuary and the re-establishment of its connection with the sea In someyears, the mouth of the estuary is artificially openedwith the use of machinery to promote the entrance oflarvae and juvenile shrimps and fish into the lagoon(Loebmann and Vieira2005)

The study site is located on a coastal plain with

low topography (< 20 m.a.s.l.) (Rambo 2000) under

strong influence of rainfall anomalies associated with

the El Niño Southern Oscillation phenomena (Grimm

et al 1998), which cause higher and lower than

average rainfall in this region during the El Niño and

La Niña episodes, respectively These rainfall

anoma-lies trigger hydrological changes and can have

profound effects on estuarine (Garcia et al 2001)

and freshwater fish assemblages in southern Brazil

(Garcia et al.2003)

Field collections, laboratory procedures and data

analysis

Specimens were obtained during a fish community

survey conducted monthly from April 2008 to May

2009 in a freshwater wetland located in the

northern reaches (31°65′10″S, 50°51′271″ W) of

the LPNP (Fig 1) In general, the physical features

were similar to those found on other permanentwetlands occurring in the northern and southernreaches of the park (Loebmann and Vieira2005) and

in nearby areas (Maltchik et al 2010) Individualswere caught using multiple types of fishing gear,including gillnets, beach seines and beam trawls.Two gillnet sets (4 × 2 m) with mesh sizes of 15, 20,

30 and 35 mm were employed from late afternoonuntil dawn in deeper waters (1.5–2.0 m) in order tocatch larger individuals (>200 mm) During theafternoon (from 1400 to 1700), one to two beachseine (9 m long, 2.4 m high, mesh size 13 mm wingsand 5 mm center) hauls were conducted in the openmargins of the wetland, and three beam trawl (mouth

of 1 × 1 m and size mesh of 5 mm) hauls wereemployed in the vegetated margins of the wetland ineach field trip The hauls conducted in the shallowerwaters (< 1.5 m) targeted mainly smaller individuals(< 200 mm) Some cryptic species were caught bydip net, which was used for approximately 15 min in

Fig 1 Southern Brazil (a)

and the location of the

Lagoa do Peixe National

Park (LPNP) (dashed line)

in the coastal plain (b) with

the position of the

freshwa-ter wetland in the northern

reaches of the park where

specimens of the Hoplias

aff malabaricus were

cap-tured (closed circle)

each field collection Aside from H aff malabaricus

specimens, all fish species collected in the

freshwa-ter wetland using the previously mentioned types of

fishing gear had their total size and numerical

abundance recorded in order to assess the general

relative abundance patterns of the ichthyofauna at

the study site Such concomitant surveys of the

abundance of the piscivorous fish and its fish prey

populations were used to infer prey availability and

patterns of resource exploitation by the H aff

malabaricus population

Immediately after their collection, individuals

were euthanized and placed on ice Later, they

were transported to the laboratory and stored in a

freezer After thawing, each individual was weighed

(Wt, g) and measured (total length, TL, mm) and,

after being surgically dissected, had their stomach

contents analyzed Food items were weighed on an

analytical balance with an accuracy of 0.0001 g

Food items found in the stomachs were identified

to the lowest possible taxonomic level using a

stereoscopic microscope When a food item was

intact, its weight (g) and total length (TL, mm)

were measured

Food resources found in non-empty stomachs were

quantified using the following parameters (Hyslop

1980): a) frequency of occurrence (%F), which

represented the percentage of the total number of

stomachs in which a particular food item was found,

b) numerical abundance (%N), which represented the

abundance in the percentage of a food item in relation

to the total abundance of all stomachs c) weight (%W)

of the item, which represented the weight (g) in

percentage of a food item in relation to the total

weight of all stomachs These parameters (%F,%N,%

W) were combined into the Index of Relative

Importance of Pinkas et al (1971) with the following

formula: IRI¼ %F»ð%Nþ %WÞ, which was

com-puted for each food item

In order to assess the influence of hydrological

changes in the study area, we analyzed diet and

feeding ecology during periods of lower and higher

precipitation In order to accomplish this, monthly

mean values of rainfall (square root transformed)

were analyzed with the Bray-Curtis similarity index in

order to distinguish temporal rainfall patterns during

the studied period Data were obtained in two

meteorological stations located in cities (Rio Grande

and Mostardas) near the study area (Fig 1)

Differ-ences in diet composition between rainfall periodsrevealed by the cluster analysis were evaluated byMDS (Multi-Dimensional Scaling) and ANOSIManalyses using the Primer software package (Clarkeand Gorley2006)

The feeding strategy of H aff malabaricus andthe characteristics of their food niche were evaluated

by the graphical method proposed by Amundsen et

al (1996) In this approach, prey-specific abundance

is defined as the percentage a prey taxon comprises

of all prey items in only those predators in which theactual prey occurs according to the formula Pi= (∑Si/

∑Sti) x 100, where Piis the prey-specific abundance

of prey i, Sithe stomach content (weight) comprised

of prey i, and Sti the total stomach content in onlythose predators with prey i in their stomach(Amundsen et al 1996) In the resulting diagram,%

F values are arranged in the x-axis and Pi values onthe y-axis The feeding strategy (generalist-specialistdichotomy) and the contribution of each individual

to the food niche structure is obtained by analyzingthe distribution of points (each one representing afood item) along a graph The population as a wholecan be considered a specialist when the prey pointsare positioned at the top right of the graph, whereastheir placement in the bottom half of the graphsuggests a population with a generalist feedingbehavior When points are located in the upper leftcorner of the diagram (i.e., with low%F and high Pi)they indicate individual opportunistic or specialistfeeding behavior of some individuals within thepredator population

The relationship between the weight (g) and totallength (TL, mm) of fish prey consumed and the totallength (TL, mm) of the predator was evaluated with anonlinear regression In the model calculation (PreyTL=

a * exp b * PredatorTL), the algorithm of Newton was used based on an iterative process using aconvergence criterion of 1.0−6 and baseline values(‘seeds’) for the regression constant (a) and theregression coefficient (b) of 1 and 0.01, respectively(Haimovici and Velasco2000; Cantanhêde et al.2009).Size-related dietary shifts were also analyzed based onthree size classes: 0–100, 101–200 and 201–350 mm

Gauss-TL, with 16, 20 and 18 individuals in each one,respectively Differences in the average total weight (g)

of the food found in the stomach content of these threesize classes were evaluated by the Kruskal-Wallis test(Zar1994)

Seasonal changes in rainfall

Rainfall showed strong variation across months, with

values ranging from 12.4 mm in November 2008 to

129.15 mm in August 2008 In general, the average

rainfall was higher during the winter and summer

(70.9 and 80.7 mm, respectively) than in the spring

and autumn (19.9 and 38.3 mm, respectively) Cluster

analysis applied to monthly rainfall values revealed

(at 75% of similarity) two main groups (Fig.2a) The

former were comprised of months with low

precipi-tation (< 25 mm), whereas the later by months with

moderate to high (25–100 mm) values (Fig.2b)

Diet composition and seasonal changes

A total of 113 individuals with sizes (TL) ranging

from 26 to 364 mm had their stomach contents

analyzed, and nearly half of them (52.1%) wereempty The analysis of the food content of the non-empty stomachs revealed 20 items (Table 1) Fishwere the most conspicuous items in the diet of H aff.malabaricus and included 15 species belonging to 6distinct fish families Fragments of insects and plantswere also found in the stomach contents, but they hadcomparatively lower contributions in terms of numer-ical abundance (%N) and biomass (%W) (Table1).Characins were the most frequent and abundantprey found in the diet, with Cheirodon interruptusbeing the fish prey with the highest numericalabundance The spotted livebearer Phalloceros cau-dimaculatus (Poeciliidae) was the second mostnumerous fish in the diet of H aff malabaricus.The numerical abundance patterns of these two mainfish prey (characins and poecilids) matched therelative abundance of the two most abundant fishfamilies at the study site closely The multi-gearsampling of the ichthyofauna revealed that Characi-Fig 2 Dendogram showing

the similarity (Bray-Curtis

index) among mean values

of rainfall (square root

transformed) in each month

from April 2008 to March

2009 Data obtained from

two meteorological stations

located in the cities of Rio

Grande and Mostardas (see

hori-zontal line denotes

similari-ty at 75% (a) Mean values

of rainfall during the low

and moderate-high periods

obtained in the cluster

analysis (b)

dae (61.3%) and Poeciliidae (18.8%) were the most

abundant of the 14 fish families occurring at the

study site (Fig 3) Other fish prey found in the diet

belonged to the Heptapteridae and Cichlidae

fami-lies In contrast, these families were more important

in terms of their %W than their %N contribution

(Table1) Accordingly, these prey were not abundant

in the study site but were characterized by their

greater sizes (especially cichlids), when compared to

characins and poecillids (Fig.3) The other eight fish

families that occurred in small proportions (< 5%) at

the study site were not found in the diet of H aff.malabaricus (Table 1)

No differences were observed in the diet position of individuals caught during low andmoderate-high rainfall periods (Fig 4) This patternwas corroborated by the ANOSIM, which did notreveal statistically significant differences betweenthe periods (Global R = 0.49, p =−0.006) Therefore,

com-we pooled our dataset for the subsequent dataanalysis on feeding strategy and size-related dietaryshifts

Actinopterygii Characiformes Characidae

Cyprinodontiformes Poeciliidae

Cyprinodontiformes Rivulidae

Perciformes Cichlidae

Others

Table 1 Frequency of

occurrence (%F), numerical

abundance (%N), weight

(%W) and index of relative

importance (%IRI) of the

food items found in the

stomach content of Hoplias

aff malabaricus in the

‘Lagoa do Peixe’ National

Park (LPNP), southern

Brazil

Feeding strategy and size related dietary shift

The feeding strategy analysis revealed that, as awhole, the studied population had a specialist feedingstrategy towards the consumption of fish (Fig 5a),resulting in a narrow niche width as depicted by thebox inserted in Fig 5a However, when the feedingstrategy was analyzed considering only fish prey,

Fig 4 MDS ordination of the composition of the diet of

Hoplias aff malabaricus during periods of low (L) and

moderate-high (M-H) rainfall periods Data values were

square-root transformed, and the Bray-Curtis similarity was

used

Fig 3 a Numerical abundance (%) and number of species in

each family (parentheses) and b average total length (TL, mm)

of fish species in the study site based on a multi-gear survey

conducted concomitantly with the collection of the species

studied here (Hoplias aff malabaricus)

Fig 5 Feeding strategy diagram for individuals of the Hoplias aff malabaricus in the Lagoa do Peixe National Park with fish prey pooled as a single food category (a) or separately by species (b) Prey-specific abundance (Pi) is plotted against frequency of occurrence (%F) of food items in the diet of Hoplias aff malabaricus The inserted box represents a conceptual diagram of a resource niche width characterized by

a high between-phenotype component (sensu Amundsen et al.

were considered: Heptapterus sympterygium (Hepsyp), opoecilus melanotaenia (Cynmel), Hoplias aff malabaricus (Hopmal), Rhamdia quelen (Ranque), organic matter (OrgMat), fish remains (FisRem), Phalloceros caudimaculatus (Phacau), Astyanax jacuhiensis (Astjac), Cheirodon interruptus (Cheint), Astyanax spp (Astspp), Pimelodella australis (Pimaus), Char- acidae (Charac), Plant fragments (Plant), Insects (Insect)

Cyn-there was a high variability in prey consumption

between individuals (phenotypes) Some individuals

preyed upon certain fish prey, such as the catfishes

Heptapterus sympterygium and Rhamdia quelen, the

annual fish Cynopoecilus melanotaenia and juveniles

of the H aff malabaricus, resulting in low frequency

of occurrence (%F) and higher prey-specific

abun-dance values (Pi) in the diagram (Fig 5b) This

pattern revealed a higher between-phenotype

compo-nent to the niche width of the studied species and,

consequently, greater partitioning of the food

resour-ces (mainly fish) between individuals of the predator

population, as depicted by the box inserted in Fig.5b

There were conspicuous size-related dietary shifts

in terms of diet composition and prey sizes consumed

by H aff malabaricus; with increasing predator size,

the diet became more diverse (Table 2, Fig 6) and

comprised larger fish prey (Fig 7) There was a

significant increase in prey diet richness with the

increase in size (TL, mm) (Kruskal-Wallis, H: 6.61,

df: 2, p<0.037) (Fig.6) that was related to an increase

in the consumption of characid species by larger

individuals (Table 2) There was also a significant

positive correlation between prey (mm TL) and

predator sizes (mm, TL) (p<0.00) (Fig 7), which

was mainly associated with the consumption of larger

prey, such as the freshwater catfishes Pimelodella

australis (average size: 41.5 mm TL) and Rhamdia

quelen (138.0 mm TL) and the cichlid Australoheros

facetus (73.5 mm TL), by larger individuals (201–

350 mm TL) (Table3)

The tendency to consume larger prey was also

observed in the consumption of characins Altogether

with the consumption of smaller characins like

Cheirodon interruptus (27.7 mm) and Astyanax sp

(28.5 mm), larger specimens (201–350 mm TL)

included in their diet larger characins such as

Astyanax eigenmanniorum (42.0 mm), A jacuhiensis

(48.5 mm), Hyphessobrycon luetkenii (39.5 mm), H

bifasciatus 39.0 mm) and Hyphessobrycon sp

(34.5 mm), which were absent in the diet of smaller

H aff malabaricus specimens (< 200 mm TL)

(Table 3)

Discussion

Our results revealed that H aff malabaricus, in this

subtropical wetland, has a carnivorous feeding mode

focused primarily on fish Among fish prey, characinswere the dominant group followed by poeciliids Thepredominance of characins in the H aff malabaricusdiet has been described by earlier studies with thespecies (Winemiller 1989; Bistoni Mlos et al.1995;Loureiro and Hahn 1996; Corrêa and Piedras 2009).For example, Loureiro and Hahn (1996) showed thatcharacins were the main food of this species in theSegredo reservoir (South Brazil), and a similarfinding was reported by Carvalho et al (2002) for apopulation inhabiting the Vermelho River in Pantanal(Central Brazil) These studies, however, did notmention whether diet composition was related toresource availability The concomitant sampling of theichthyofauna at the current study site showed that H.aff malabaricus seems to exploit the most abundantfish resources available The two most abundant fishfamilies in the study site (Characidae and Poeciliidae)were clearly the most frequent and abundant in thediet of H aff malabaricus Conversely, most of theleast abundant (<1%) fish families at the study site,such as Sternopygidae, Gymnotidae, Hypopomidae,Crenuchidae, Synbranchidae, Anablepidae, were notfound in the diet of H aff malabaricus This suggeststhat this species exploits the most abundant fishresources available rather than rare fish prey Habitatuse by both predator and prey also helps explain thehigher predation on Characidae and Poeciliidaespecies Hoplias aff malabaricus is strongly associ-ated with habitats highly structured by riparian forest

or aquatic plants (Luz-Agostinho et al 2008), wheresmall fish such as characins and poecillids usuallyseek refuge from predators (Petry et al.2003; Peliciceand Agostinho 2006) Such habitat seems to favorambush predators such as the species studied here(Almeida et al.1997, Luz-Agostinho et al.2008).Although moderately abundant (~5%), the Callich-thydae (Callichthys callichthys) and Curimatidae(Cyphocharax voga) families were also absent in the

H aff malabaricus diet In contrast with Characidaeand Poeciliidae specimens that are typically foundfeeding in the water column (Ceneviva-Bastos andCasatti 2007; Abilhoa et al 2008), specimens ofCallichthydae and Curimatidae found in the study sitewere bottom dwellers typically feeding on detritusnear the substrate (Corrêa and Piedras 2008) Based

on their absence in the stomach content of H aff.malabaricus, we speculate that H aff malabaricusshows preference (or higher capture success rate) for

fish prey foraging in the water column, like the

Characidae and Poecilidae species, rather than bottom

dweller species

The analysis of the feeding strategy of the H aff

malabaricus population revealed a clear specialization

towards the consumption of fish when considering

only major food categories However, when the feeding

strategy was analyzed focusing on only the foodcategory ‘fish’, the analysis showed that individualsdid not prey upon a particular fish species, but, rather,had a high degree of opportunism in the consumption

of different fish species This pattern could arise fromthe fact that the H aff malabaricus is a typical ambush(sit-and-wait) predator preying upon prey species

Table 2 Frequency of occurrence (%F), numerical abundance

(%N), weight (%W) and index of relative importance (%IRI) of

the food items found in the stomach content of three size

Park (LPNP), southern Brazil n=number of individuals

entering its striking zone (Winemiller 1989), with no

particular specialization for a particular fish species

This opportunistic feeding behavior has been

previ-ously observed in this species by other authors

(Cassemiro et al 2005; Novakowski et al.2007)

In a broader context, the generalist-specialist

dichotomy and the food-niche structure can be

considered in terms of between- and

within-phenotype (individual) contributions For instance, a

population with a generalist feeding strategy could

have a trophic niche with a high between-phenotype

component, when the individuals are preying upon

different resources, or a high within-phenotype

component, when most of the individuals are

con-suming the same prey species (Amundsen et al

1996) Bolnick et al (2003) have argued thatbetween-phenotype variation in diet and resourceuse patterns are more widespread among animalpopulations than previously supposed Future research

is needed to evaluate if the observed high degree ofindividual-level food niche variation in H aff malabar-icus is an outcome of its ambush feeding strategy andits potential role in avoiding conspecific competition.The analysis of the stomach contents of the studiedspecies revealed a clear size-related dietary shiftbetween diet composition and richness and the size

Fig 7 Relationship between total length (TL, mm) of the fish

prey and the predator, based on the fish species found in the

stomach contents of Hoplias aff malabaricus in the Lagoa do

Peixe National Park n=39

Table 3 Average values of weight (W, g), total length (mm TL) and total width (mm TW) of the food items found in the

Peixe’ National Park (LPNP), southern Brazil

Actinopterygii Characiformes Characidae

Cyprinodontiformes Poeciliidae

Cyprinodontiformes Rivulidae

Perciformes Cichlidae

Fig 6 Average number (+%95 C.I.) of different food items

(diet richness) found in the stomach content of three size

Hoplias aff malabaricus in the Lagoa do Peixe National Park

of the prey consumed Dietary shifts have been

observed in H aff malabaricus populations occurring

in the Llanos Venezuelanos, with juveniles preying

upon insects and shifting to fish as they develop into

adults (Winemiller 1989) A similar decrease in the

importance of insects with increasing predator size

was also observed in the present study However, the

present study also revealed a positive relationship

between sizes of both the fish prey and the predator

Larger individuals (> 200 mm TL) had a richer diet

comprised of larger fish prey such as freshwater

catfishes and cichlids when compared with smaller (<

200 mm TL) H aff malabaricus specimens that

preyed mainly on smaller characins and poeciliids

The consumption of these larger prey species could be

related to greater efficiency in capturing larger and

more mobile prey and could compensate for the

greater energy demands of adult individuals that need

to perform activities such as spawning, nest

construc-tion and egg guarding (Prado et al 2006)

Alterna-tively, this trend could be explained by changes in

mouth gape as the predator increases in size Usually,

oral jaw dimensions and, in some cases, pharyngeal

gape are the primary determinants of the prey size

ranged consumed by predatory fishes (Helfman et al

2009) For instance, Pusey et al (2000) have shown

that interspecific and ontogenetic variation in diet are

strongly related to differences in body size and mouth

gape In this case, we would expect that larger

individuals of H aff malabaricus would be able to

consume larger prey species regardless of their

nutritional status Future studies are necessary to

evaluate the relative role of these hypotheses on

explaining the size-related dietary shift trend observed

for this species

In conclusion, our findings show that the South

American subtropical wetland species, H aff

mala-baricus, preys mainly upon fish regardless of dry or

wet weather conditions and feeds opportunistically on

the most abundant prey species available The

ubiquitous and smaller-sized (< 70 mm) characins

and poecillids, usually found in the water column near

or within riparian vegetation, constituted its main

prey, rather than the less abundant, and in most cases,

benthic species Size-related dietary shifts occurred

and were characterized by the consumption of larger

prey such as catfish and cichlid species These

findings suggest that fish predation exerted by H

aff malabaricus in this system follows specific

food-web pathways that seem to be mediated by theabundance of prey resources Additionally, the rela-tionship between prey size and predator size adds alayer of complexity to this picture Future fieldmesocosm or whole-system manipulative experi-ments, based on the insights provided by the currentwork, should be carried out in order to assess thestrength of these food web pathways and theirpotential effects on secondary production of subtrop-ical wetlands harboring H aff malabaricus popula-tions Such information could provide a crucialbackground in order to enhance ongoing conservationplans for protecting this Biosphere Reserve fromincreasing human activities in its vicinities such asagriculture, fishery and the cultivation of exoticspecies

Laboratory of the Oceanographic Institute of the Federal sity of Rio Grande (FURG) for their assistance in the field, especially Vinicius Condini; the Conselho Nacional de Desenvol- vimento Científico e Tecnológico, CNPq (Grant No 482920/ 2007-6) and International Foundation of Science, IFS (Grant No A/4419-1) for providing financial support and the ICMBIO for providing permit (14523-2 and 14523-4) for sample collections.

Univer-References

Abilhoa V, Duboc LF, Damil PAF (2008) A comunidade de peixes de um riacho de Floresta com Araucária, alto rio Iguaçu, sul do Brasil Revista Brasileira de Zoologia

Almeida VLL, Hahn NS, Vazzoler AEAM (1997) Feeding

Amundsen PA, Gabler HM, Staldavik FJ (1996) A new approach to graphical analysis of feeding strategy from

Bistoni Mlos A, Haro JG, Gutiérrez M (1995) Feeding of Hoplias malabaricus in the wetlands of Dulce River (Cordoba, Argentina) Hydrobiology 316:103–107 Bolnick DI, Svanbäck R, Fordyce JA, Yang LH, Davis JM, Hulsey CD, Forister ML (2003) The ecology of individ- uals: incidence and implications of individual specializa-

Cantanhêde GS, Fugi R, Hahn NS (2009) Variation in prey selection of a piscivorous fish after the impoundment of a neotropical reservoir: prey size and type J Fish Biol

Carvalho NL, Fernandes CHV, Moreira VES (2002) ção de Hoplias malabaricus (Bloch, 1794) (Osteichtyes, Erythrinidae) no rio Vermelho, Pantanal Sul mato-

Alimenta-Grossense Revista de Brasileira de Zoociências 4:227–

236

Cassemiro FAZ, Hahn NS, Delariva RL (2005) Estrutura trófica

da ictiofauna, ao longo do gradiente longitudinal do

reservatório de Salto Caxias (rio Iguaçu, Paraná, Brasil)

no terceiro ano após o represamento Acta Sci Biol Sci

Ceneviva-Bastos M, Casatti L (2007) Oportunismo alimentar

de Knodus moenkhausii (Teleostei, Characidae): uma

espécie abundante em riachos do noroeste do Estado de

Clarke KR, Gorley RN (2006) Primer v 6: user manual/tutorial.

PRIM ER-E Ltd, Plymouth

Corrêa F, Piedras SRN (2008) Alimentação de Cyphocharax

voga (Hensel, 1869) (Characiformes, Curimatidae) no

Arroio Corrientes, Pelotas, Rio Grande do Sul, Brasil.

Corrêa F, Piedras SRN (2009) Alimentação de Hoplias aff.

malabaricus (Bloch, 1794) e Oligosarcus robustus

Menezes, 1969 em uma lagoa sob influência estuarina,

Pelotas, RS Biotemas 22:121–128

Corrêa F, Garcia AM, Loebmann D, Claudino MC, Bastos RF,

Vieira JP (2009) Pisces, Cyprinodontiformes, Rivulidae,

Austrolebias minuano (Costa and Cheffe, 2001): new

species record at Lagoa do Peixe National Park, Rio

Garcia AM, Vieira JP, Winemiller KO (2001) Dynamics of the

shallow-water fish assemblage of the Patos Lagoon

estuary (Brazil) during cold and warm ENSO episodes J

Garcia AM, Raseira MB, Vieira JP, Winemiller KO, Grimm AM

(2003) Spatiotemporal variation in shallow-water

freshwa-ter fish distribution and abundance in a large subtropical

Garcia AM, Hoeinghaus DJ, Vieira JP, Winemiller KO, Marques

DMLM, Bemvenuti MA (2006) Preliminary examination of

food web structure of Nicola Lake (Taim Hydrological

System, south Brazil) using dual C and N stable isotope

analyses Neotropical Ichthyology 2(4):279–284

Grimm AM, Ferraz SET, Gomes J (1998) Precipitation

anomalies in southern Brazil associated with El Niño and

Haimovici M, Velasco G (2000) Relações comprimento/peso de

peixes teleósteos marinhos do sul do Brasil com uma

avaliação de diferentes métodos de ajuste Atlantica

Gorman OT, Locke B, Madenjian CP, Roseman EF (2008)

Restoring piscivorous fish populations in the Laurentian

Great Lakes causes seabird dietary change Ecology

Helfman G, Collette BB, Facey DH, Bowen BW (2009) The

diversity of fishes: biology, evolution, and ecology

Wiley-Blackwell, Oxford

Hyslop EJ (1980) Stomach contents analysis; a review of

methods and their application J fish Biol 17:411–429

Loebmann D, Vieira JP (2005) Distribuição espacial e

abun-dância das assembléias de peixes na Lagoa do Peixe, RS,

Brasil Revista Brasileira de Zoologia 22:667–675

Loureiro VE, Hahn NS (1996) Dieta e atividade alimentar da

traíra, Hoplias malabaricus (Bloch, 1794) (Osteichtyes,

Erythrinidae) nos primeiros anos de formação do vatório de Segredo-Paraná Acta Limnológ Brasiliensia 8:195–205

reser-Luz-Agostinho KDG, Agostinho AA, Gomes LC, Júlio HF (2008) Influence of food pulses on diet composition and trophic relationships among piscivorous fish in the Upper

Maltchik L, Lanés LEK, Stenert C, Medeiros ESF (2010) Species-area relationship and environmental predictors of fish communities in coastal freshwater wetlands of

Mazzeo N, Iglesias C, Mello FT, Borthagaray A, Fosalba C, Ballabio R, Larrea D, Vilches J, García S, Pacheco JP, Jeppesen

E (2010) Trophic cascade effects of Hoplias malabaricus (Characiformes, Erythrinidae) in subtropical lakes food webs:

Novakowski GC, Hahn NS, Fugi R (2007) Alimentação de peixes piscívoros antes e após a formação do reservatório de Salto Caxias, Paraná, Brasil Biota Neotropica 7:149–155 Pelicice FM, Agostinho AA (2006) Feeding ecology of fishes associated with Egeria spp patches in a tropical reservoir, Brazil Ecol Fresh Fish 15:10–19

Peretti D, Andrian IF (2008) Feeding and morphological analysis

of the digestive tract of four species of fish (Astyanax altiparanae, Parauchenipterus galeatus, Serrasalmus mar- ginatus and Hoplias aff malabaricus) from the upper

Petry P, Bayley PB, Markle DF (2003) Relationships between fish assemblages, macrophytes and environmental gra- dients in the Amazon River Floodplain J Fish Biol

Pinkas L, Oliphant MS, Iverson ILK (1971) Food habits of albacore, bluefin tuna, and bonito in California waters.

Polis GA, Winemiller KO (1996) Food webs: integration of patterns and dynamics Chapman and Hall, New York Prado CPA, Gomiero LM, Froehlich O (2006) Spawning and parental care in Hoplias malabaricus (Teleostei, Characi- formes, Erythrinidae) in the Southern Pantanal, Brazil Braz J Biol 66(2B):697–702

Pusey BJ, Arthington AH, Read MG (2000) The dry-season diet of freshwater fishes in monsoonal tropical rivers of Cape York Peninsula, Australia Ecol Freshwat Fish

Tagliani PRA (1995) Estratégia de Planificação Ambiental para

o Sistema Ecológico da Restinga da Lagoa dos Patos Planície Costeira do Rio Grande do Sul PhD Thesis, Universidade Federal de São Carlos

-Winemiller KO (1989) Ontogenetic diet shifts and resource partitioning among piscivorous fishes in the Venezuelan Llanos Environ Biol Fish 26:177–199

Zar JH (1994) Biostatistical analysis, 2ªth edn Prentice Hall, USA

Zaret TM, Paine RT (1973) Species introduction in a tropical

The effects of intraspecific competition on the prey capture

behavior and kinematics of the bluegill sunfish,

Lepomis macrochirus

Janne A Pfeiffenberger&Philip J Motta

Received: 5 August 2010 / Accepted: 30 June 2011 / Published online: 27 July 2011

# Springer Science+Business Media B.V 2011

Abstract Competition has broad effects on fish and

specifically the effects of competition on the prey

capture kinematics and behavior are important for the

assessment of future prey capture studies in bony

fishes Prey capture kinematics and behavior in bony

fishes have been shown to be affected by temperature

and satiation The densities at which bony fish are

kept have also been shown to affect their growth,

behavior, prey selection, feeding and physiology We

investigated how density induced intraspecific

com-petition for food affects the prey capture kinematics of

juvenile bluegill sunfish, Lepomis macrochirus High

speed video was utilized to film five bold individuals

feeding at three different densities representing

differ-ent levels of intraspecific competition We

hypothe-sized that: (1) the feeding kinematics will be faster at

higher levels of competition compared to lower levels

of competition, and (2) bluegill should shift from

more suction-based feeding towards more ram-based

feeding with increasing levels of competition in order

to outcompete conspecifics for a prey item We foundthat, with increased intraspecific competition, preycapture became faster, involving more rapid jawopening and therefore greater inertial suction, shortermouth closing times, and shorter gape cycles Fur-thermore, the attack velocity of the fish increased withincreasing competition, however a shift towardsprimarily ram based feeding was not confirmed Ourstudy demonstrates that prey capture kinematics areaffected by the presence of conspecifics and futurestudies need to consider the effects of competition onprey capture kinematics

Keywords Prey capture kinematics Competition Bluegill Density Ram-suction

DOI 10.1007/s10641-011-9885-0

Department of Integrative Biology,

University of South Florida,

Department of Biological Sciences,

Florida Institute of Technology,

150 W University Boulevard,

Melbourne, FL 32901, USA

conspecifics at lower predator densities (Stoner and

Ottmar 2004) However, the effects of competition

on prey capture kinematics of fishes have not been

investigated

Fishes can capture food by ram, suction, or biting,

or a combination of the three (Liem1980; Norton and

Brainerd 1993) During prey capture, modulation of

prey capture kinematics of bony fishes occurs in

response to position, size and elusiveness of prey

(Nyberg1971; Wainwright and Lauder1986; Nemeth

1997a) Ram feeding involves high attack velocities

to overtake the prey item whereas suction feeding

involves lower attack velocities and precise

position-ing to suck the prey item into the mouth and usually a

combination of these two feeding modes is utilized by

fish (Nemeth1997a,b; Higham2007) With

increas-ing intraspecific competition, we would expect

pred-ators, such as the bluegill sunfish Lepomis

macrochirus, to approach the prey at higher attack

velocities and to capture the prey without braking

before employing a more ram dominated prey

capture

Bluegill sunfish, Lepomis macrochirus Rafinesque

1819 are native to and widespread in North America

Bluegill sunfish are considered one of the highest

performing suction feeders (Carroll et al 2004) and

have been the focus of extensive research in prey

capture kinematics and performance (Gillis and

Lauder 1995; Ferry-Graham et al 2003; Higham et

al 2005a; Higham et al.2006; Holzman et al 2008)

and feeding ecology (Werner1977; Mittelbach1981;

Osenberg et al 1988; Savino et al 1992; Brogowski

et al 2005) Increased intraspecific competition for

food has been assumed to affect the growth of bluegill

living in high density populations (Wiener and

Hanneman 1982) which is further supported by the

general assumption that almost all conspecifics

compete for food and habitat since they occupy the

same niche (Ward et al.2006)

Previous studies have shown that temperature

(Wintzer and Motta 2004; Devries et al 2006),

satiation (Sass and Motta 2002), prey type (Norton

1991) and body size (Richard and Wainwright 1995)

can affect the prey capture kinematics of fishes, and

that higher stocking densities and competition reduce

the growth and survival rate of fishes (Houde 1977;

Anderson et al.2002)

The goal of this study is to examine the effects of

intraspecific competition on the prey capture

kine-matics of bluegill sunfish Specifically, it is esized that: (1) bluegill sunfish will feed faster athigher levels of competition compared to lower levels

hypoth-of competition, and (2) bluegill sunfish will shift fromprimarily suction feeding towards more ram feedingwith increasing levels of intraspecific competition.This study will address whether future studies shouldtake into account the effects of competition on preycapture kinematics

Materials and methods

Study organismThirty five juvenile bluegill sunfish (8.9–10.0 cmSL) were caught by cast netting from the Hills-borough River drainage in Hillsborough County,Florida, USA The animals were housed in 40-literaquaria individually at 22°C with a 12:12 lightperiod and were acclimated for 2–4 weeks Thetanks were screened on the sides and the back withwhite paper such that the animals could not see eachother The front of the tank remained unscreened foracclimation of the animals to feed under cameralighting and while filming Animals were fed dailywith a mix of live and freeze-dried Artemia sp.through a clear vertically oriented PVC-tube (2.57 cmdiameter) that was positioned in the center of theaquarium with its opening approximately 4 cm belowthe water surface

Experimental procedures

To vary the levels of intraspecific competition, threemicrocosms with different numbers of conspecificswere established In the aquarium lacking competitionthere was only one fish, low levels of competitionwere represented by the focal fish plus two otherconspecifics, and high competition was represented

by the focal fish plus four other conspecifics.Consequently, the most aggressive bluegill wasselected as the focal fish for the trials The focal fishwas transferred to the filming tank alone or with otherfish, depending on the induced competition level Allfish were housed individually in 40-liter aquariabefore being moved to the filming tank (40-liters,

51 cm L×31 cm H×26 cm D at 22°C), which wasscreened on the sides and back such that the fish

could only see the fish which were present in the

filming tank The fish were allowed to acclimate for

24 h after transfer and were not fed during this time,

after which they were filmed

Using two High-Speed cameras, a Redlake

Motion-scope PCI 2000S and a Redlake MotionMotion-scope PCI 500

(Redlake MASD, Inc., San Diego, CA, USA), five

prey-capture events per treatment per focal fish were filmed at

250 fields per second to obtain a lateral and a dorsal

view of the prey-capture event The dorsal image was

captured by one camera mounted vertically at 90° over

the water surface and directly over the feeding tube The

second camera position in front of the aquarium

recorded a lateral view of the focal fish during prey

capture The prey items, live Artemia sp (average

0.5 cm length) were introduced one item at a time

through the clear PVC tube suspended in the middle of

the tank Prey items were introduced approximately

5 min apart when the fish had moved away from the

feeding tube Only prey capture events in which the

focal fish was lateral to the camera were used for

analysis A maximum of ten Artemia sp were fed to

the subject fish during any trial in order to eliminate

possible effects of satiation on the prey capture

kinematics (Sass and Motta 2002), although only the

first five successful prey capture events were used for

the analysis Other fish present in the tank only

consumed two prey items throughout all the trials As

five successful prey capture events were not always

obtained on 1 day, filming resumed after 24 h, during

which time the fish were not fed After five successful

filming events, each focal fish was returned to its

individual holding tank alone Each fish was only used

for one competition level treatment and was then

released or euthanized according to University

proto-col, except for the focal fish which experienced all

competition levels Each focal animal was filmed five

times for three different competition treatments (no

competition, low competition and high competition)

This trial was repeated for a total of five focal fish

(n=5) The focal animals were identified from

con-specifics by individual differences in their color

patterns and markings Prey capture events in which

the mouth of the animal touched the feeding tube were

not considered successful prey capture events and were

not accounted for since this might have an effect on the

prey capture kinematics Due to possible chemical cues

left by the fish while filming, the water was completely

changed after each filming session

Data analysis

Feeding events were downloaded to a computer andanalyzed using MaxTRAQ Software version 1.87 (Inno-vision Systems, Inc., Columbiaville, Michigan, USA).The following kinematic variables were analyzed fromthe lateral sequences: peak gape of the mouth (PeakGape); time to open the mouth (TOpen); time to close themouth once reaching peak gape (TClose); and total time

of the gape cycle (TCycle) Peak Gape is defined as themaximum distance between the two most anterior points

on the upper and lower jaw during the prey-captureevent TOpen, as described by Sanford and Wainwright(2002) eliminates the variability of early mouth opening

in bony fishes before prey-capture and is measured asthe time from 20% to 95% of peak gape TCloseis thetime from reaching Peak Gape to closing the mouth and

is measured from the field the mouth starts closing afterreaching Peak Gape to the field at which the mouthreaches 20% of Peak Gape while closing The 20% wasselected in order to be consistent with the TOpenmeasurement and also to eliminate possible variation

in mouth closing The gape cycle is measured as thetime from the field in which the mouth reaches 20% ofPeak Gape while opening to the field in which themouth reaches 20% of Peak Gape while closing Inaddition to the kinematic data the lateral filmingsequences were used to determine x and y coordinatesfor the predator and the prey to determine the distancemoved by the predator and prey for each prey captureevent (Norton and Brainerd 1993) The distances thepredator and the prey (DPredatorand DPrey) moved in thetime from reaching 20% of the peak gape while openingthe mouth until the prey disappeared in the mouth of thepredator were determined by tracking a spot on theopercle of the fish and the head of the Artemia sp.during the prey capture sequence In addition, thevelocities of the predator and the prey (VPredator and

VPrey) were determined for the duration of the preycapture event (TPrey), which is determined as the timefrom when the mouth reaches 20% of the peak gapewhile opening the mouth until the prey disappears in themouth of the predator In addition, the running averagefor the change in velocity of the predator (AInstantaneous)was determined for each field of 10 sequential fieldsbefore the prey entered the mouth of the predator andused to determine if the predator maintained, increased

or decreased its velocity prior to prey capture for threesubjects under the different competition levels An

increase in velocity between sequential fields therefore

indicates the fish is accelerating, whereas a decrease

indicates deceleration In this manner the net

acceler-ation or deceleracceler-ation of the predator was calculated

across the 10 fields The small sample size of subjects

occurred because this variable was determined post

hoc after the video sequences had been trimmed in

length

To quantify the ram and suction components in the

prey capture events, the velocity and the distance

moved of the predator (VPredator and DPredator) was

used as an indicator for the ram component and the

velocity and the distance moved of the prey (VPrey

and DPrey) and the time to open the mouth (TOpen) as

an indicator for the suction component

Statistical analysis

All variables obtained in this study were subjected to

a Two-Way Repeated Measures ANOVA using

Sig-maplot Software version 11.0 (Systat Software, Inc.,

Chicago, Illinois, USA) to account for significant

differences (P<0.05) across treatments in a model

with repeated measurements for each focal animal

The research was approved and followed the

guide-lines set forth by the University of South Florida,

Institutional Animal Care and Use Committee

(IACUC protocol #W3402)

Results

Behavior

The bold focal fish exhibited aggressive behavior

towards conspecifics once the feeding tube was

placed over the tank During low and high

competi-tion scenarios, the focal animals constantly bit the

conspecifics and chased them away as they

approached the feeding tube When the prey was

introduced into the filming tank, the focal animal

would rapidly swim towards the prey, capture it and

resume chasing away conspecifics which approached

the feeding tube

Data analysis

Two-Way Repeated Measures ANOVA revealed that

increased intraspecific competition resulted in

signif-icant differences (P<0.05) in six out of nine variablesmeasured in the bold individuals (Table 1) in thisstudy The field specific velocity (AInstantaneous) failednormality even with transformation and underwent anonparametric Friedman’s Repeated MeasuresANOVA on Ranks Time to open the mouth (TOpen)decreased with increasing competition, when compar-ing no competition to low competition (P=0.011) and

no competition to high competition (P=0.003); Time

to close mouth (TClose) and the gape cycle (TCycle)decreased as well when comparing no to highcompetition (P=0.023; Fig 1) Peak Gape showed

no significant difference with increasing competition.The distance of the prey (DPrey; Fig 2) decreasedwith increasing intraspecific competition, being differ-ent when comparing no competition to low competi-tion (P=0.040) and no competition to high competition(P=0.003), whereas the distance of the predator

across treatments The velocity of the prey (VPrey;Table 1) showed no significant differences withincreasing competition, however the velocity of thepredator (VPredator; Fig 2) increased with increasingintraspecific competition being different when compar-ing no competition to low competition (P=0.033) and

no competition to high competition (P=0.005) Theduration of the prey capture event (TPrey; Fig 2)decreased with increasing intraspecific competition,

Table 1 Average values for kinematic variables during prey capture for five Lepomis macrochirus Values are for five feeding events each Superscript letters denote significant differences across treatments.*indicate variables with significant differences (P<0.05), (ns) indicates no significant difference

Fig 1 Average values

and ±1 SE for a) time to open

the mouth (TOpen), b) time to

close the mouth (TClose) and

during prey capture for five

Lepomis macrochirus, at

three levels of competition.

The lines denote no

significant difference

between treatments

-20 -30

70

No Competition Low Competition High Competition

Fig 2 Average values and ±1 SE for a) the distance the prey

the prey capture event (TPrey) for five Lepomis macrochirus d)

Average velocity for ten fields prior to prey capture ±1 SE for three Lepomis macrochirus The lines in figures A, B, and C denote no significant differences between treatments

0.70

No DPredator vs No DPrey

Fig 3 The distances moved

by predator (DPredator) and

prey (DPrey) during prey

capture plotted against each

other at three levels of

competition for five

Lepomis macrochirus

being different when comparing no competition to low

competition (P=0.035) and no competition to high

competition (P=0.005) The average change in velocity

of the predator for ten fields prior to prey capture

signifi-cantly with increasing levels of competition, indicating

acceleration at high levels of competition, no

signifi-cant change in velocity (+2.6 cm s−1) at low

competition and a negative change in velocity

(−10.4 cm s−1), at no competition indicating

deceler-ation of the focal fish (P=0.028)

Discussion

This study, the first to investigate the effects of

intraspecific competition on the prey capture

kinematics in fish, demonstrates that bluegill

sunfish exhibit faster mouth opening and closing

times, shorter gape cycles and an increase in

predator approach velocity with increasing levels

of intraspecific competition In contrast to

previ-ous studies, prey capture kinematics have been

shown to become slower (time to reach maximum

gape and lower jaw depression, time to close the

mouth) and last longer (duration of bite) at lower

water temperatures (Wintzer and Motta 2004;

Devries et al 2006) and prey capture kinematics

are slower (lower jaw depression, max gape

distance, hyoid depression and recovery) with

increasing satiation (Sass and Motta 2002)

How-ever, this is the first study to reveal that intraspecific

competition for food modulates prey capture

kine-matics in fishes

Competition in fishes has been widely studied,

including competition for food (Booth and Beretta

2004; Schleuter and Eckmann 2006; Ward et al

2006), habitat (Almany 2004; Hasegawa and

Maekawa 2006; Kahl and Radke 2006) and

repro-duction (Taborsky1998; Stoltz and Neff 2006; Plath

et al.2008) Bluegill sunfish competing for the prey

item in this study exhibited two forms of food

competition: scramble and contest competition In

scramble competition the focal fish reaches the prey

item before other fish, whereas in contest competition

the focal fish aggressively displaces its competitors

while pursuing the prey item (Ward et al 2006)

Under both levels of competition the focal bluegill

sunfish reached the Artemia sp prey first, and chased

away conspecifics which were approaching thefeeding tube

During suction feeding the rapid expansion of thebuccal cavity generates a flow of water into the mouth

of the fish (Liem 1980; Norton and Brainerd 1993;Carroll et al.2004; Day et al.2005) and faster mouthopening times, as observed in this study (Fig.1), can

be interpreted as an increase in buccal expansion ratewhich leads to lower sub-ambient pressures within thebuccal cavity and higher flow speeds of the water infront of the mouth (Sanford and Wainwright 2002),therefore indicating an increase in inertial suctionforce with increasing levels of competition Fish mayperform compensatory suction to counter the bowwave generated by the increased velocities of thepredator (Van Damme and Aerts1997; Ferry-Graham

et al.2003; Higham et al.2005a), which could movethe prey away from the approaching predator Arecent study on bluegill sunfish demonstrated theformation of this bow wave when approaching preyand how suction feeding reverses the flow and drawsthe prey into the mouth (Holzman and Wainwright

2009) Therefore, more rapid mouth opening withincreased competition may be related to the increasedvelocity of the predator and the need for greatercompensatory suction to counter the effects of a bowwave generated by faster velocities during preycapture Faster mouth opening may also be due inpart to the higher predator approach velocity and theforce exerted on the opening mouth by the water.The gape cycle was found to become shorter induration with increasing levels of intraspecific com-petition, which indicate faster prey capture events.Therefore, the hypothesis that bluegill sunfish feedfaster at higher levels of competition compared tolower levels of competition was confirmed

Bluegill sunfish are characterized as ram andsuction feeders (Carroll et al.2004; Day et al.2005;Higham et al 2005b; Higham 2007) and utilized acombination of ram and suction during prey capture

in this study The bluegill sunfish exhibited higherapproach velocities during the prey capture event

intraspecific competition which are characteristic ofram feeders and are usually utilized when capturingelusive prey (Webb and Skadsen 1980; Norton1991;Porter and Motta 2000) The increased velocity isindicative of an increase of the ram component duringprey capture However, the apparent increase in

suction force with increasing competition is indicative

of an increase in the suction component during prey

capture An increase in inertial suction force usually

manifests itself by drawing the prey towards the

mouth from a greater distance and at a higher velocity

However, the distance the prey moved decreased

because the bluegill continued to accelerate as it

approached the prey to engulf it The velocity of the

prey did not increase despite more rapid mouth

opening because the fish was perhaps employing

greater compensatory suction to overcome the bow

wave it was generating in front of the mouth

Therefore it is concluded that both the ram and

suction components increased with increasing levels

of competition, however it was not possible to

confirm the second hypothesis that bluegill shift from

using primarily suction feeding to ram feeding with

increasing levels of competition

Recent studies in fish behavior investigating the

bold-shy continuum have found that fish exhibiting

bold behavior are more willing to take risks (Wilson

et al.1993; Webster et al.2009) Individuals

display-ing higher risk behavior increased their reproductive

success as well as increasing predation risk which

could lead to early mortality whereas shy individuals

will exhibit behavior to avoid predation (Wilson et al

1993; Webster et al.2009) The focal bluegill sunfish

in this study were selected for bold behavior and it is

likely that the level of boldness could affect the prey

capture kinematics, since the latency to attack prey

has been shown to decrease in bold individuals

(Webster et al.2009) This suggests that bold animals

may exhibit faster prey capture times compared to shy

animals

In conclusion, this study demonstrated that prey

capture kinematics of bold bluegill sunfish are

affected by the presence of conspecifics while

competing for food Increasing intraspecific

com-petition resulted in faster mouth opening and

closing, shorter gape cycle time and increased

predator velocity during prey capture However a

shift from suction feeding towards ram feeding with

increasing intraspecific competition was not

con-firmed in this study

Porter Family Foundation and the University of South Florida

Office of Undergraduate Research for funding this research as

well as Timothy Higham for offering comments on this

manuscript We also would like to thank Ralph Turingan for his help with the experimental setup, Richard Tankersley for his help with the statistical analysis as well as Kyle Mara, Maria Laura Habegger, Lisa Whitenack, Samantha Mulvany and Tanya Brunner with their help throughout this study.

Booth DJ, Beretta GA (2004) Influence of recruit condition on food competition and predation risk in a coral reef fish Oecologia 140:289–294

Brogowski Z, Siewert H, Keplinger D (2005) Feeding and growth responses of bluegill fish (Lepomis macrochirus) at various pH levels Pol J Environ Stud 14:517–519 Carroll AM, Wainwright PCH, Huskey SH, Collar DC, Turingan RG (2004) Morphology predicts suction feeding

3881 Day SW, Higham TE, Cheer AY, Wainwright PC (2005) Spatial and temporal flow patterns during suction feeding of bluegill sunfish (Lepomis macrochirus) by particle image

Devries MS, Wainwright PC, Beaupre SJ (2006) The effects of acute temperature change on prey capture kinematics in largemouth bass, Micropterus salmoides Copeia 2006:

Ehlinger TJ (1990) Habitat choice and phenotype-limited feeding efficiency in bluegill: individual differences and trophic polymorphism Ecology 7:886–896

Ferry-Graham LA, Wainwright PC, Lauder GV (2003) tification of flow during suction feeding in bluegill sunfishes Zoology 106:159–168

Quan-Gillis GB, Lauder GV (1995) Kinematics of feeding in bluegill sunfish: is there a general distinction between aquatic capture and transport behaviors J Exp Biol 198:

Hasegawa K, Maekawa K (2006) Effect of habitat components

on competitive interaction between native white-spotted

480 Higham TE (2007) Feeding, fins and braking maneuvers: locomotion during prey capture in centrarchid fishes J

Higham TE, Day SW, Wainwright PC (2005a) Sucking while swimming: evaluating the effects of ram speed on suction generation in bluegill sunfish Lepomis macrochirus using digital particle image velocimetry J Exp Biol 208:2653– 2660

Higham TE, Malas B, Jayne BC, Lauder GV (2005b) Constraints on starting and stopping: behavior compen- sates for reduced pectoral fin area during braking of the bluegill sunfish Lepomis macrochirus J Exp Biol 208:

Higham TE, Day SW, Wainwright PC (2006) The pressures of

suction feeding: the relation between buccal pressure and

induced fluid speed in centrarchid fishes J Exp Biol

209:3281–3287

Holzman RA, Wainwright PC (2009) How to surprise a

copepod: strike kinematics reduce hydrodynamic

distur-bance and increase stealth of suction-feeding fish Limnol

Holzman RA, Collar DC, Day RD, Bishop KL, Wainwright

PC (2008) Scaling of suction-induced flows in bluegill:

morphological and kinematic predictors for the

2668

Houde ED (1977) Food concentration and stocking density

effects on survival and growth of laboratory-reared larvae

of bay anchovy Anchoa mitchilli and lined sole Achirus

Kahl U, Radke RJ (2006) Habitat and food resource use of

perch and roach in a deep mesotrophic reservoir: enough

space to avoid competition? Ecol Freshw Fish 15:48–56

Liem KF (1980) Acquisition of energy by teleosts: Adaptive

mechanisms and evolutionary patterns In: Ali MA (ed)

Environmental physiology of fishes Plenum, New York,

Mittelbach GG (1981) Foraging efficiency and body size: a

study of optimal diet and habitat use by bluegills Ecology

Nemeth DH (1997a) Modulation of attack behavior and its

effect on feeding performance in a trophic generalist fish J

Nemeth DH (1997b) Modulation of buccal pressure during prey

capture in Hexagrammos decagrammus (Teleostei:

Norton SF (1991) Capture success and diet of cottid fishes: the

role of predator morphology and attack kinematics.

Norton SF, Brainerd EL (1993) Convergence in the feeding

mechanics of ecomorphologically similar species in the

Centrarchidae and Cichlidae J Exp Biol 176:11–29

Nyberg DD (1971) Prey capture in the largemouth bass Am

Midl Nat 86:128–144

Osenberg CW, Werner EE, Mittelbach GG, Hall DJ (1988)

Growth patterns in bluegill (Lepomis macrochirus) and

pumpkinseed (L gibbosus) sunfish: environmental

varia-tion and the importance of ontogenetic niche shifts Can J

Plath M, Richter S, Tiedemann R, Schlupp I (2008) Male fish

deceive competitors about mating preferences Curr Biol

Porter HT, Motta PJ (2000) A comparison of prey capture

behavior and kinematics in three ram feeding fishes Am

Richard BA, Wainwright PC (1995) Scaling the feeding mechanism of largemouth bass (Micropterus salmoides): kinematics of prey capture J Exp Biol 198:419–433 Sanford CP, Wainwright PC (2002) Use of sonomicrometry demonstrates the link between prey capture kinematics and

3457 Sass GG, Motta PJ (2002) The effects of satiation on prey capture kinematics in the largemouth bass, Micropterus

Savino JF, Marschall EA, Stein RA (1992) Bluegill growth as modified by plant density: an exploration of underlying

Stoner AW, Ottmar ML (2004) Fish density and size alter Pacific halibut feeding: implications for stock assessment.

J Fish Biol 64:1712–1724

Van Damme J, Aerts P (1997) Kinematics and functional morphology of aquatic feeding in Australian side-necked

Wainwright PC, Lauder GV (1986) Feeding biology of fishes: patterns of variation in the feeding mechanism.

Ward AJW, Webster MM, Hart PJB (2006) Intraspecific food

Webb PW, Skadsen JM (1980) Strike tactics of Esox Can J

Webster MM, Ward AJW, Hart PJB (2009) Individual boldness affects interspecific interactions in sticklebacks Behav Ecol Sociobiol 63:511–520

Werner EE (1977) Competition and habitat shift in two sunfishes (Centrarchidae) Ecology 58:869–876

Werner EE, Hall DJ (1979) Foraging efficiency and habitat switching in competing sunfishes Ecology 60:256–264 Wiener JG, Hanneman WR (1982) Growth and condition of bluegills in Wisconsin lakes: effects of population density

Wilson DS, Coleman K, Clark AB, Biederman L (1993) bold continuum in pumpkinseed sunfish (Lepomis gibbo- sus): an ecological study of a psychological trait J Comp

Wintzer AP, Motta PJ (2004) The effects of temperature on prey capture kinematics of the bluegill sunfish, Lepomis macro- chirus: implications for feeding studies Can J Zool

Acoustic diversity in Lake Malawi ’s rock-dwelling cichlids

Patrick D Danley&Martin Husemann&

Justin Chetta

Received: 20 July 2010 / Accepted: 4 July 2011 / Published online: 17 August 2011

# Springer Science+Business Media B.V 2011

Abstract The cichlids of Lake Malawi are one of the

world’s most species rich and phenotypically diverse

groups of extant vertebrates The extraordinary

variability of this group’s color patterns, reproductive

behaviors, and trophic morphologies are well

docu-mented More recently, an additional axis of

pheno-typic diversity has been identified Lake Malawi

cichlids have been shown to use species-specific

acoustic communication in both aggressive and

reproductive encounters However, documentation of

acoustic signals used by this group is limited to a

small number of taxa observed within the confines of

the laboratory This study examines the acoustic

signals produced by six species spanning four genera

of rock-dwelling cichlids recorded in their natural

habitat, the shallow waters surrounding Thumbi West

Island, Lake Malawi Four acoustic parameters were

quantified and compared between species: trill

dura-tion, number of pulses per trill, pulse duradura-tion, andpulse period Using these characteristics, sympatricspecies within the genus Maylandia were easilydistinguished Furthermore, a comparison of this data

to previously published acoustic data reveals possiblegeographic dialects within species

Keywords Mate choice Sound production Cichlidae Reproductive isolation Speciation Metriaclima

Introduction

The cichlids of Lake Malawi have undergone one ofthe most extensive and rapid radiations identified todate Since the formation of the lake basin 2 MYA,well over 800 species of cichlid fish have divergedfrom a single common ancestor Most fish in thissystem can be grouped into one of two major clades:the rock-dwelling cichlids and the sand dwellingcichlids (Albertson et al 1999) These clades areroughly equal in diversity and intense sexual selection

is believed to have played a significant role ingenerating the extraordinary species richness in bothgroups (Danley and Kocher 2001; Streelman andDanley 2003) As a result, these fish have become amodel system for examining recent and rapid speci-ation events (Kocher2004; Genner and Turner2005).Within the rock-dwelling cichlids, one of the mostlikely and conspicuous targets of this intense selective

pressure is male nuptial color pattern Male color

pattern is highly variable in Lake Malawi cichlids,

and, as a result, has been the focus of many

theoretical and empirical studies on the evolution

and diversification of species (van Oppen et al.1998;

Carleton et al.2006; Carleton2009; Kidd et al.2006)

However, additional cues such as olfaction and

acoustics have been suggested to play an important

role in mate choice (Robinson et al.1998; Knight and

Turner 1999; Amorim et al 2003; Amorim et al

2004; Plenderleith et al.2005; Cole and Stacey2006;

Smith and van Staaden 2009) Furthermore, mate

choice experiments revealed that visual cues alone are

not sufficient to maintain species boundaries (Blais et

al.2009) Hence, non-visual communication could be

of higher importance for the maintenance of species

boundaries within this system than currently thought

Fishes can produce sounds in various ways

Specialized skeletal muscles, filaments, pharyngeal

jaws and teeth (see Rice and Lobel 2002, 2004;

Amorim 2006 for review) can all be used to make

sounds These sounds are then often amplified

through the use of the swimbladder as a resonance

body Still, it is not exactly known how the majority

of cichlid fishes produce sounds Ripley and Lobel

(2004) and Rice and Lobel (2004) suggested that the

pharyngeal jaw, its attached muscles and the

swim-bladder play an important role for sound production

in the Lake Malawi cichlid Tramitichromis

interme-dius Lanzing (1974) proposed a similar mechanism

for the related species Oreochromis mossambicus

Longrie et al (2009) showed that O niloticus

produces sound during a backward movement of the

pelvic and pectoral girdles and a forward movement

of the second pterygiophore of the anal fin Still it is

not clear which sound production mechanisms apply

for the acoustic signalling in the rock-dwelling

cichlids of Lake Malawi

Sounds produced by cichlids appear to act in a

variety of social interaction Variations in acoustic

signals may be used for identification of conspecific

mates and the identification of male quality (Simões

et al 2008a) Amorim et al (2004), Longrie et al

(2008) and Simões et al (2008a) demonstrated that

cichlids produce sounds during antagonistic

encoun-ters and territorial defense as well in courtship

Likewise, females produce sounds as warning or

aggressive signals towards each other (Simões et al

2008a) Still, courtship appears to be the most

important situation in which acoustic signals areemitted (Simões et al 2008b) These observationsare consistent with the use of acoustic signals in awide variety of fish systems (see Lobel 1992, andAmorim 2006 for reviews) They also illustrate thepotential for these signals to contribute to reproduc-tive isolation

Recently, interest in cichlid acoustic tion has increased and those few studies performed todate have recorded sounds in captivity (but see Lobel

communica-1998) These studies have shown that species can bedifferentiated based on acoustic characteristics such astrill duration, number of pulses per trill, pulse period,pulse duration, and interpulse interval (Nelissen1975,

1977, 1978; Lobel 1998,2001; Amorim et al 2004,

2008; Simões et al 2006, 2008a) This study is thefirst to directly examine the variation of acoustic traitsacross several species of rock-dwelling cichlids in awild population Here, we analyze four parameters ofacoustic signals produced by six closely related,sympatric rock-dwelling cichlids of Lake Malawi

We compare our data with previously publishedstudies on Lake Malawi cichlids, some of whichexamined populations of the same species fromdifferent locations

Materials and methods

We examined male acoustic diversity of six speciesspanning four genera of Lake Malawi cichlids:Cynotilapia afra, Labeotropheus fuelleborni, Maylan-dia aurora, Maylandia callainos, Maylandia zebra,and Petrotilapia nigra (Debate has surrounded theappropriate genus name for those species belonging towhat we refer to here as Maylandia Other authorsmay refer to this genus as either Pseudotropheus orMetriaclima) All recordings were made in theshallow water (<5 m) around Thumbi West Island,Malawi (14° 01′27.58″ S 34°49′ 25.55″ E)

Males were observed for 20 min prior to recording

to identify breeding caves Breeding caves wereidentified through observing the focal male attempt-ing to lead a receptive female to a specific area withinthe rocky substrate Previous studies have demon-strated that breeding caves are species-specific andare occupied by a single territorial male (Hert 1989;Danley 2001; Jordan et al 2010) The hydrophonewas suspended in the center of male breeding caves

by wedging the hydrophone wire in the surrounding

rocks Breeding caves were approximately 30 cm×

15 cm×25 cm [length × width × depth (Danley

2001)] Given the placement of the hydrophone, we

assume that all the recorded acoustic data were

produced by the focal territorial male during courtship

Visual data necessary to validate this assumption are not

available All acoustic signals generated in this cave

were recorded for the subsequent 3.5 h

Recordings were made using a HTI- 96 MIN

hydrophone (sensitivity:−163.9 dB 1v/uPA;

frequen-cy response: 2 Hz −30 kHz) and a Shure FP11

amplifier Sounds were recorded using TDK IEC1

Type 1 cassette tapes with a Sony TCM-DV200DV

Recordings were digitized using Audacity, converted

to ‘.wav’-format and input into RavenPro 1.3

soft-ware (Charif2003) Only those trills with waveforms

distinct from background noise were analyzed

Oscil-lograms, spectrograms (frequency v time), and power

spectra (frequency vs power) were generated and

cross-referenced to measure the following four

parameters for all species: trill duration, pulse

duration, pulse period, and number of pulses per trill

(Amorim et al 2004) (Fig 1) Because the distance

between the hydrophone and the freely behaving focal

males were not controlled, center frequency

(geomet-ric mean between a lower and an upper frequency

threshold) and amplitude of the pulses were not

analyzed

All statistical analyses were run using R 2.9.0 (The

R Foundation for Statistical Computing) All data

were log transformed prior to analysis One-wayanalysis of variance (ANOVA) was used to test forsignificance in differences in each of the acousticcomponents Turkey’s HSD test was used for pairwise comparisons

Results

We were able to record sounds produced by onefocal male of each target species Given theplacement of the hydrophone, it is assumed thatthe recorded sounds were produced during court-ship, however visual confirmation of this assump-tion is not available The number of analyzabletrills recorded during the 3.5 h observation periodvaried between species (Table 1): C afra N = 5, L.fuelleborni N = 4, M aurora N = 14, M callainos N=

17, M zebra N = 12 and P nigra N= 6 Given the lownumbers of recorded trills of C afra, L fuelleborniand P nigra, data collected from these species arepresented but were not subjected to statisticalanalysis

Sounds produced by the Maylandia speciesshowed statistically significant differences in threeparameters: trill duration (Fig 2a; F2,34= 23.55, p<0.0001), pulse duration (Fig 2b, F2,37= 100.16, p<0.0001), and pulse period (Fig.2c, F2, 39= 92.76 p<0.0001) Pulses per trill (Fig 2d; F2, 40 = 2.83; p=0.07) were not significantly different across the threeMaylandia species

Fig 1 Typical M zebra trill; trill duration, pulse period, and pulse duration are indicated: a Oscillogram (amplitude in kU versus time

in seconds), b Spectrogram (frequency in kHz versus time in seconds), c Detail of a single pulse

Trill duration of M zebra [763.26 +/− 407.11 ms

(mean +/− SD)] was significantly longer than that of

M aurora (375.77 +/− 125.35 ms, p=0.01) or M

callainos (204.69 +/− 107.67 ms, p<0.001) (Fig.2b)

Maylandia aurora has a significantly longer trill

duration than M callainos (p<0.001) The remaining

species had trill durations of intermediate length: C

afra (446.2 +/− 189.87 ms), L fuelleborni (363.75 +/−

148.29 ms) and P nigra (422.33 +/− 217.36 ms)

Pulse duration showed significant differences

be-tween all Maylandia species (Fig.2b, F2,37= 100.16,

p<0.0001) Again, M zebra had the longest pulse

duration (16.11 +/− 4.60 ms) The pulse duration of

M callainos (5.32 +/− 0.99) was shorter than of M

zebra (p < 0.001), while M aurora (12.36 +/−

2.33 ms) had a pulse duration intermediate yet distinct

from either M zebra (p=0.03) or M callainos (p<

0.001) Pulse duration of C afra (9.32 +/− 1.19 ms),

L fuelleborni (9.23 +/− 0.76 s) and P nigra (10.0 +/−

3.0 ms) had an intermediate length and were most

similar to M aurora

Likewise, all Maylandia species could be

distin-guished based on pulse period (Fig 2c, F2, 39 =

92.758 p <0.001) A pattern similar as for pulse

duration was observed; M zebra produced sounds

with the longest pulse period (PP = 55.45 +/−

13.48 ms) PP of M aurora was of intermediate

length (PP = 34.54 +/− 8.57 ms), whereas M

callainos showed the shortest PP (17.28 +/−

4.25 ms) Turkey’s HSD test revealed that all pair

wise comparisons of Maylandia species were

signif-icantly different in PP with p<0.001 C afra (PP =

34.46 +/− 6.61 ms), L fuelleborni (PP = 34.73 +/−

5.68 ms) and P nigra (PP = 27.92 +/− 4.61 ms)

exhibited intermediate pulse periods

Species could not be distinguished based on the

number of pulses per trill (Fig.2d; F2, 40= 2.83; p=

0.07) Admittedly, M zebra (17.23 +/− 11.59)

produced more pulses per trill than M callainos(10.88 +/− 2.88, p=0.07), but little difference wasobserved between M aurora (11.77 +/− 3.52) and M.zebra and M callainos (p=0.21 and p=0.86, respec-tively) P nigra (18.33 +/−8.51) showed a similarnumber of trills compared to M zebra; numbers ofpulses per trill for C afra (13.4 +/− 5.86) and L.fuelleborni (11.5 +/− 3.87) were more similar to M.aurora

Discussion

Recently, the importance of non-visual modalities inthe cichlid mate choice has gained recognition.Acoustics and olfaction have been identified as themost important non-visual communication modes incichlids (Amorim et al 2004; Almeida et al 2005;Amorim et al.2008; Blais et al.2009) Chemical cuesare used in a wide array of social interactions,including the indication of social status (Barata et al

2007; Bender et al 2008), reproductive status offemales (Clement et al 2004; Almeida et al 2005;Miranda et al 2005) and agonistic interactions(Giaquinto and Volpato1997) Furthermore, olfactorycues appear to influence female cichlid mate choice(Plenderleith et al.2005; but see Venesky et al.2005,and Jordan et al 2003 for alternative views).Likewise, acoustic communication has been docu-mented as a part of a variety of social interactions

Table 1 Means and standard deviations of the acoustic parameters measured in this study

„

including antagonistic and courtship behavior

Previ-ous studies have shown that species can be

differen-tiated based on acoustic characteristics such as trill

duration, number of pulses per trill, pulse period,

pulse duration, and interpulse interval (Nelissen1975,

1977,1978; Lobel 1998,2001; Amorim et al.2004;

Simões et al.2006; Amorim et al.2008; Simões et al

2008a)

Our data are consistent with the hypothesis that

sympatric species vary in their acoustic signals and

that these signals may play a role in mate choice

among the cichlid fishes of Lake Malawi While

temporal discrimination of these fish has not been

specifically analyzed, studies of other hearing

gen-eralists suggest that cichlids are capable of detecting

the species specific differences in acoustic behavior

documented in these species (reviewed in Amorim et

al 2004) The data presented here indicate that

Maylandia, if capable of detecting the

species-specific temporal differences described here, may

use acoustic signals to distinguish con- from

hetero-specific males Our data suggest that the three

investigated Maylandia species can be distinguished

based on one or more acoustic parameters Of the

three Maylandia species examined, M zebra shows

the most distinct sound pattern Maylandia zebra has

the longest trill duration, longest pulse duration, and

longest pulse period relative to any of the examined

species

A comparison of our dataset to previously

pub-lished data suggests that acoustic variation exists not

only between different species, but also between

populations within species (Table 2) The sounds

produced by two of the six species examined in this

study have been examined in the lab Four datasets

from two locations are available for M zebra [3

Nkhata Bay (Amorim et al.2004,2008; Simões et al

2008a), 1 Thumbi West (present study), and four data

sets from two locations for M callainos 3 Nkhata Bay

(Amorim et al 2004, 2008; Smith 2007) 1 Thumbi

West (present study)]

While a statistical comparison of the data was not

possible due to the different ways of describing

variation given in the publications, we can identify

substantial differences in analyzed parameters within

and between species Within M callainos, trill

duration, pulse period, and pulse duration are

dis-tinctly shorter in the Thumbi West population

com-pared to Nkhata Bay The number of pulses per trill T

differs less among different populations, but appears

to be slightly higher for the Thumbi West population

(Table 2) Maylandia zebra from Thumbi West

produce a greater number of pulses per trill, longer

pulse duration and a shorter pulse period than

observed for specimens from Nkhata Bay (Table2)

The comparison of published data in the laboratory

and wild population data points toward two interesting

observations First, sympatric species differ in at least

one of the observed parameters Second, populations of

the same species at different locations appear to differ in

their acoustic patterns Hence, our data corroborates the

conclusions of previously published studies which

suggested that (1) acoustic signals may play a role in

mate choice and recognition and (2) acoustic signals are

diverging between geographically isolated populations

The presence of geographically differentiated dialects

may contribute to the reproductive isolation of

pop-ulations and hence contribute to the high rate of

speciation observed in this system More work is

needed to understand population level variation in

acoustic signals of Lake Malawi cichlids

Staaden for further information on the populations they

observed Emily Rapstine provided assistance in digitizing the

recordings Arthur Popper generously provided the hydrophone

and amplifier Thomas Kocher and Karen Carleton provided

PDD with the opportunity to record at Thumbi West Island We

are grateful to the University of Malawi and officials at Lake

Malawi National Park for providing the facilities and permits

necessary to conduct this work We also want to thank Aimee

Howe, other members of the Danley Lab, and two anonymous

reviewers for their comments on this manuscript.

References

Albertson RC, Markert JA, Danley PD, Kocher TD (1999)

Phylogeny of a rapidly evolving clade: The cichlid fishes

of lake malawi, east africa Proceedings of the National

Academy of Sciences of the United States of America

96:5107–5110

Almeida OG, Miranda A, Frade P, Hubbard PC, Barata EN,

Canário AVM (2005) Urine as a social signal in the

Mozambique Tilapia (Oreochromis mossambicus) Chem

Amorim MCP (2006) Diversity of sound production in fish In:

Ladich F, Collin SP, Moller P, Kapoor BG (eds) Communication

Amorim MCP, Fonseca PJ, Almeida VC (2003) Sound

production during courtship and spawning of Oreochromis

mossambicus male-female and male-male interactions J

Amorim MCP, Knight ME, Stratoudakis Y, Turner GF (2004) Differences in sounds made by courting males of three closely related Lake Malawi cichlid species J Fish Biol 65:1358–1371

Amorim MCP, Simões JM, Fonseca PJ, Turner GF (2008) Species differences in courtship acoustic signals among five Lake Malawi cichlid species (Pseudotropheus spp.) J Fish Biol 72:1355–1368

Barata EN, Hubbard PC, Almeida OG, Miranda A, Canário AVM (2007) Male urine signals rank in the Mozambique tilapia (Oreochromis mossambicus) BMC Biol 5:1–11 Bender N, Heg-Bachar Z, Oliveira RF, Canário AVM, Taborsky M (2008) Hormonal control of brood care and social status in a

Blais J, Plenderleith M, Rico C, Taylor MI, Seehausen O, van Oosterhout C, Turner G (2009) Assortative mating among Lake Malawi cichlid fish populations is not simply predict-

Carleton K (2009) Cichlid fish visual systems: mechanisms of

Carleton KL, Spady TC, Kocher TD (2006) Visual cation in East African cichlid fishes: diversity in a phylogenetic context In: Ladich F, Collin SP, Moller P, Kapoor BG (eds) Communication in fishes, vol 1 Science

Clarif RA, Waack AM, Strickman LM (2008) Raven v1.3.

cornell.edu/raven Clement TS, Grens KE, Fernald RD (2004) Female affiliative preference depends on reproductive state in the African

Cole TB, Stacey NE (2006) Olfactory responses to steroids in

an African mouth-brooding cichlid, Haplochromis burtoni

Danley PD (2001) The ecological behavioral, and genetic

rock-dwelling cichlids Ph.D Dissertation, University of New Hampshire

Danley PD, Kocher TD (2001) Speciation in rapidly diverging

Genner MJ, Turner GF (2005) The mbuna cichlids of Lake Malawi: a model for rapid speciation and adaptive

Giaquinto PC, Volpato GL (1997) Chemical communication, aggression, and conspecific recognition in the fish Nile

Hert E (1989) The function of egg-spots in an African

Jordan R, Kellogg KA, Juanes F, Stauffer, JR Jr (2003) Evaluation of female mate choice cues in a group of Lake Malawi mbuna (Cichlidae) Copeia 181–186

Jordan R, Mellor D, Wilt L, Gershenson D, Howe D (2010) Male interactions in a group of Malawi cichlids Ethol

Kocher TD (2004) Adaptive evolution and explosive

specia-tion: the cichlid fish model Nature Rev Genet 5:288–298

Lanzing WJR (1974) Sound production in the cichlid Tilapia

mossambica Peters J Fish Biol 6:341–347

Lobel PS (1992) Sounds produced by spawning fishes Environ

Lobel PS (1998) Possible species specific courtship sounds by

two sympatric cichlid fishes in Lake Malawi, Africa.

Lobel PS (2001) Acoustic behavior of cichlid fishes J

Longrie N, Fine ML, Parmentier E (2008) Innate sound

production in the cichlid Oreochromis niloticus J Zool

Longrie N, Van Wassenbergh S, Vandewalle P, Mauguit Q,

Parmentier E (2009) Potential mechanism of sound

production in Oreochromis niloticus (Cichlidae) J Exp

Miranda A, Almeida OG, Hubbard PC, Barata EN, Canário

AVM (2005) Olfactory discrimination of female

reproduc-tive status by male tilapia (Oreochromis mossambicus) J

Exp Biol 208:2037–2043

Nelissen MHJ (1975) Sound production by Simochromis

diagramma (Günther) (Pisces, Cichlidae) Acta Zool

Nelissen MHJ (1977) Sound production by Haplochromis

burtoni (Günther) and Tropheus moorii Boulenger (Pisces,

Nelissen MHJ (1978) Sound production by some

Tanga-nyikan cichlid fishes and a hypothesis for the

evolu-tion of their communicaevolu-tion mechanisms Behaviour

Plenderleith M, van Oosterhout C, Robinson RL, Turner GF

(2005) Female preference for conspecific males based on

olfactory cues in a Lake Malawi cichlid fish Biol Lett

Rice AN, Lobel PS (2002) Enzyme activities of pharyngeal jaw

musculature in the cichlid Tramitichromis intermedius:

implications for sound production in cichlid fishes J Exp Biol 205:3519–3523

Rice AA, Lobel PS (2004) The pharyngeal jaw apparatus of the Cichlidae and Pomacentridae: function in feeding and

Ripley JL, Lobel PS (2004) Correlation of acoustic and visual signals in the cichlid fish, Tramitichromis intermedius.

Robinson RR, Fernald RD, Stacey NE (1998) The olfactory system of a cichlid fish responds to steroidal compounds J

Simões JM, Duarte I, Fonseca PJ, Turner GF, Amorim MCP (2006) Acoustic behaviour in Malawian cichlids (Pseudo- tropheus, Cichlidae): potential cues for species recognition and intraspecific communication Razprave IV, razreda

Simões JM, Duarte IG, Fonseca PJ, Turner GF, Amorim MCP (2008a) Courtship and agonistic sounds by the cichlid fish Pseudotropheus zebra J Acoust Soc Am 124:1332– 1338

Simões JM, Fonseca PJ, Turner GF, Amorim MCP (2008b) African cichlid Pseudotropheus ssp males moan to female

Smith AR (2007) Is acoustic communication a candidate signal for sexual selection in Malawian Cichlids? M.A Thesis, Bowling Green State University

Smith AR, van Staaden MJ (2009) The association of visual and acoustic courtship behaviors in African cichlid fishes.

Streelman JT, Danley PD (2003) The stages of vertebrate

Van Oppen MJ, Turner GF, Rico C, Robinson RL, Deutsch JC, Genner MJ, Hewitt GM (1998) Assortative mating among rock-dwelling cichlid fishes supports high estimates of

Venesky MD, Andraso GM, Ropski SJ (2005) Behavior of male kenyi cichlids, pseudotropheus lombardoi, in response to visual and olfactory cues from females BIOS 76:77–83

Hierarchical and interactive habitat selection in response

to abiotic and biotic factors: The effect of hypoxia

on habitat selection of juvenile estuarine fishes

John T Froeschke&Gregory W Stunz

Received: 24 August 2010 / Accepted: 4 July 2011 / Published online: 27 July 2011

# Springer Science+Business Media B.V 2011

Abstract Habitat selection is a shared process among

animals where individuals choose areas that differ in

biotic and abiotic characteristics to maximize

individ-ual fitness We used manipulative laboratory

meso-cosm choice experiments to examine hierarchical

and interactive relationships influencing this habitat

selection process of estuarine fishes We assessed

selection among substrate, dissolved oxygen (DO)

concentration, food availability, and predation risk

using two common juvenile estuarine fish species,

pinfish (Lagodon rhomboides) and Atlantic croaker

(Micropogonias undulatus) For both species oxygen

concentration greatly influenced selection patterns;

fishes strongly avoided low DO conditions, while in

higher levels of DO factors such as substrate or food

influenced selection patterns However, both species

strongly avoided predators even when alternative

habitat was severely oxygen limited These results

suggest that predation risk may be the greatestdeterminant of habitat selection of the factorsconsidered Expansion of low DO areas in theworld’s oceans is a major anthropogenic disturbanceand is rapidly increasing Assessing impacts ofhypoxia on habitat usage of mobile organisms iscritical as changes in environmental metrics includ-ing predator distribution and DO levels may alterhabitat selection patterns disrupting critical ecosys-tem processes and trophic interactions Our resultsindicate that juvenile fishes may forgo emigrationfrom hypoxia due to predation risk If similarpatterns occur for juvenile fishes in estuaries theymay potentially suffer from reduced growth, repro-ductive output, and survivorship

Keywords Habitat selection Hypoxia Abiotic Biotic Predation Dissolved oxygen

Introduction

Habitat selection is a nearly universal process amonganimals where individuals must choose among hab-itats that differ in biotic and abiotic characteristics(Johnson1980; Huey1991) The hierarchy of habitatselection for an individual should reflect factorspotentially limiting an individual’s fitness For exam-ple, ideal habitats would provide high net energeticreturn rate (to maximize growth and reproduction)and low mortality (Gilliam and Fraser 1987) These

DOI 10.1007/s10641-011-9887-y

Harte Research Institute for Gulf of Mexico,

Studies and Department of Life Sciences,

Texas A&M University-Corpus Christi,

Gulf of Mexico Fishery Management Council,

2203 N Lois Ave Suite 1100,

Tampa, FL 33607, USA

conditions rarely exist in nature, and individuals must

choose between sub-optimal environments to balance

physiological performance with predation risk

Habitat selection patterns of fishes have been

well-described and offer a good model to test the

relative roles of abiotic versus biotic habitat

characteristics For example, estuaries are

recog-nized as high quality habitat types as they are

food-rich, structurally complex, and provide refuge from

predation (Beck et al 2001) However, human

activities have dramatically altered both abiotic and

biotic properties of coastal ecosystems at an alarming

rate (Altieri2008; Halpern et al.2008) For example, in

many coastal estuaries, predator densities have been

dramatically reduced (Lotze et al 2006), important

habitats such as seagrass meadows and oyster reefs

have declined (Levin and Stunz 2005), and dissolved

oxygen (DO) concentrations are falling (i.e., hypoxia)

worldwide (Rabalais et al.2007; Diaz and Rosenburg

2008; Montagna and Froeschke 2009) These habitat

alterations may influence habitat selection of

ecologi-cally important species and could precipitate

large-scale community changes in marine ecosystems

(Utne-Palm et al 2010)

A critical abiotic attribute often compromised in

marine ecosystems is dissolved oxygen (DO) In

aquatic systems, low oxygen levels are a widespread

problem (Diaz and Rosenburg 2008), not limited to

particular areas, and interact with biotic habitat

characteristics in very complex ways (Breitburg

2002; Rabalais et al 2002; Altieri 2008), ultimately

altering the distribution of individuals (Lenihan et al

2001; Bell and Eggleston 2005; Utne-Palm et al

2010) Successful avoidance of hypoxia (DO

concen-tration <2 mg O2l−1) is dependent upon movement

responses and physiological tolerances which are

typically species specific (Pihl et al.1991) In aquatic

ecosystems, low levels of oxygen have been

associ-ated with reduced abundance, biomass, diversity,

growth, and have also been attributed to population

declines of some estuarine fishes (Eby et al 2005;

Long and Seitz 2008; Vaquer-Sunyer and Duarte

2008; Montagna and Froeschke 2009) Mitigation of

human impacts to ecosystems requires an improved

understanding of attributes animals use for habitat

selection, particularly as it relates to importance of

abiotic and biotic factors on habitat choice

Hypoxia is becoming a more common

phenome-non in estuaries and is thought to be increasing in

occurrence worldwide including the Gulf of Mexico(Diaz and Rosenburg 2008) The Gulf of Mexicosupports a variety of economically and ecologicallyimportant species that may be impacted by declining

DO levels The seasonal presence of a large dead zone

in the Gulf of Mexico at the mouth of the Mississippiriver is well documented and also occurs withinestuarine waters along the Texas coast (Montagna andRitter 2006) The presence of a hypoxic zone hasbeen documented annually in Corpus Christi Bay,Texas since 1988 and there is a long-term trend ofdecreasing dissolved oxygen levels in these waters(Applebaum et al.2005) This hypoxic zone also hadreduced diversity of both benthic and mobile organ-isms (including fishes) and lower abundance andbiomass of epifauna (Montagna and Froeschke2009).Pinfish (Lagodon rhomboides) and Atlantic croaker(Micropogonias undulatus) are two of the mostabundant estuarine fishes in the Gulf of Mexico andshallow coastal estuaries comprise important nurseryhabitat for these species (Reese et al.2008) Both can

be found in a variety of habitats including seagrassmeadows and non-vegetated substrate and are sensitive

to hypoxia (Wannamaker and Rice2000) Thus, thesespecies provide excellent models for testing hypothesesabout factors influencing habitat selection patterns ofestuarine fishes

The goal of this study was to assess the relativeimportance of biotic and abiotic factors on habitatselection Specifically, we compared the relativeimportance of predator density, substrate, and foodavailability with varying levels of DO concentration

to assess ecological impacts of declining oxygenconcentrations on habitat selection patterns We used

a series of replicated laboratory mesocosm choiceexperiments with two ecologically important estuarinefishes as models

Methods

Collection of study organismsJuvenile Atlantic croaker (“croaker”) and pinfish werecollected from shallow estuarine habitats in Aransas andCorpus Christi Bays, Texas USA using bag seines.Mean standard length (SL) of croaker and pinfish was:(mean ± standard error) 38±5 mm SL and 35±3 mm SLrespectively Animals were held at wet laboratory

facilities in aerated and filtered 38-l aquaria Fish were

maintained in holding tanks at 23–25°C and 30–35 psu;

DO concentrations were maintained between 6.0 and

6.5 mg O2l−1 Fish were acclimated to aquaria for at

least 3 d prior to experimental procedures and fed

frozen mysid shrimp (mysid sp.) daily to satiation and

kept on a 12-h light/dark photoperiod

Experimental design

We used a sequential series of replicated experimental

mesocosm trials to test for habitat selection patterns for

both species at varying levels of predator density,

substrate type, DO concentration, and food Replicated

two-way choice trials were completed in a

225-cm×60-cm×75-cm mesocosm filled to 25 cm with filtered

seawater (Fig.1) Washed sand was used to simulate

non-vegetated habitat (Stunz et al 2001), and shoal

grass (Halodule wrightii) was simulated using an

artificial seagrass unit (ASU) Artificial seagrass units

are replicate seagrass habitat and are created from

polyethylene ribbon attached to a mesh base placed

under a sand substrate ASU’s were used as a proxy forseagrass because preliminary trials on selection pat-terns of juvenile pinfish and croaker showed nosignificant selection patterns between shoal grassand the ASU’s (one sample t-test, pinfish t=1.88, df=7,p=0.10, croaker t=0.95, df=9, p=0.36)

A DO gradient was established using a Plexiglassdivider, and releasing nitrogen and oxygen gas intoeach chamber of the mesocosm During experimentaltrials, the divider was raised 60 mm to permit fishmovement (Wannamaker and Rice 2000) Dissolvedoxygen concentrations were measured at the centerand at each end of the mesocosm prior to introduction

of experimental fish and at the conclusion of eachexperiment using a Thermo Scientific Orion 3-Star

DO meter (Thermo Electron Corporation Beverly,

MA USA) Trials were not run if DO levels deviatedmore than ± 0.40 mg O2 −1 from target levels.Wannamaker and Rice (2000) demonstrated thatpinfish and croaker could detect and avoid thehypoxia in a laboratory mesocosm and our prelimi-nary experiments showed similar patterns

Fig 1 a Diagram of experimental mesocosm set up prior to

fish introduction where the seagrass treatment is placed in the

low DO treatment, and b after the fish are introduced c

Experimental set-up prior to fish introduction where the sand treatment is placed in the low DO treatment, and d set-up after the fish are introduced

Experimental procedure

To determine the relative importance of substrate and

DO concentrations on habitat selection patterns, we

established a range of substrate-DO combinations

First, we conducted a substrate control experiment to

determine selection patterns based on the habitat type

(seagrass vs non-vegetated) without a DO treatment

(both chambers 6.0 mg O2 −l) We then conducted a

series of experiments, each testing fish habitat

selection patterns between two substrate-DO

combi-nations (Fig 1) Experiments were conducted at DO

levels from 1 to 6 mg O2 −l to identify potential

graded, threshold or interactive responses Trials were

first set up with the low DO treatment within the

seagrass substrate and high DO treatment within the

non-vegetated substrate Reciprocal experiments were

completed by establishing low oxygen treatment in

non-vegetated substrate and high oxygen

concentra-tion in seagrass

For all experiments, three fish (approximating

natural densities; Stunz et al 2002) were randomly

selected from holding aquaria for each trial Fish were

released in the center of the mesocosm as close as

possible to the divider near the bottom and acclimated

for 10 min prior to each experiment The acclimation

period occurred after the divider was raised to permit

movement and exploration of all treatments For

every trial the locations of all three fish were averaged

into a single response as individual fish within a trial

were not independent The low and high oxygen

chambers of the mesocosm were re-assigned for each

replicate to eliminate bias toward any particular area

Observations were made by a single observer >2 m

away from the tank with the observer peering over an

opaque barrier to prevent observer interference to fish

behavior During preliminary trials, this method of

observation did not elicit a startle or other behavioral

responses from the fish Since these experiments were

relatively short-term, we were able to make live direct

observations during the entirety of the experiment

without disturbing the fish

To assess the relative importance of food availability

relative to hypoxia for both species, the experimental

mesocosm was modified by placing a food enriched

treatment (mysid shrimp) in one chamber, while food

was absent in the other Fish were fasted for 24 h and

three fish were placed in the mesocosm with DO

concentration at 4.0 mg O −l One chamber was

enriched with 6.07±0.07 g (mean ± standard error) offrozen mysid shrimp per trial and habitat choicewas monitored every 20 s for 10 min afteracclimation (n = 6) Excess food was removed afterevery trial A second experiment (n=6) was repeated,but mysid shrimp were placed in a low oxygentreatment (1.0 mg O2 l−l), while the other chamberremained at 4.0 mg O2 l−l but without the mysidshrimp food treatment

The relative influence of predator-presence and DOconcentration on habitat selection patterns was alsoexamined by introducing predatory fish Three sub-adultred drum (Sciaenops ocellatus) (mean ± standard error)190±8 mm SL were used for the predation trials.Predators were acclimated for 48 h The Plexiglasspartition separating sides of the mesocosm wasmodified by cutting eight 4-cm round holes near thebottom of the divider to permit movement of juvenilefishes throughout the entire mesocosm but restrictedthe movement of the red drum predators to onechamber We performed preliminary trials that showedjuvenile fish would readily pass through holes in thepartition For the initial experiment (n=6) both sections

of the mesocosm were set to 4.0 mg O2l−1 A secondexperiment (n = 6) was conducted in which thepredatory red drum were placed in the 4.0 mg O2

l−1 DO treatment and the side without predators set

to 1.0 mg O2l−1

Statistical analyses

The response variable for all habitat selection trialswas the mean proportion of time three fish spent ineach chamber in the mesocosm Residuals were notnormally distributed, therefore, data were arc-sinsquare root transformed and tested against the nullhypothesis of 50% of the time spent in each side ofthe mesocosm using a two-tailed, one-sample, stu-dent’s t-test (α=0.05) All data management andanalyses were conducted using SAS 9.1.3 software(SAS Institute, Inc., 2000)

Results

Habitat selection experiments

In normoxic conditions (both chambers 6 mg O2l−1),both pinfish and croaker displayed significant

habitat preferences Pinfish selected seagrass treatment

(t= 3.5, df = 5, p = 0.016) (Fig 2a), while croaker

selected non-vegetated bottom (t = 4.5, df = 5, p =

0.007) (Fig.2c) Based on significant habitat selection

preference patterns in previous trials, we designed

experiments to test the importance of habitat type (e.g.,

seagrass and non-vegetated bottom) and oxygen

concentration on selection patterns For pinfish, both

habitat type and DO levels influenced selection

patterns Dissolved oxygen exerted greater influence

on selection patterns during hypoxic conditions (i.e.,

DO≤2.0 mg O2 l−1), while habitat type was more

important during moderate hypoxia or normoxic

conditions (i.e., DO≥2.0 mg O2l−1; Fig.2a)

Despite the preference for vegetated habitat in the

control experiment, pinfish avoided the low

oxygen-seagrass treatment when DO levels were decreased to

1.0 mg O2l−1and selected the alternative 4.0 mg O2

l−1 sand bottom treatment (t=3.8, df=5, p=0.011)

(Table 1) A similar pattern was observed during

1.0 mg O2l−1seagrass vs 2.0 mg O2l−1sand bottom

experiment as pinfish displayed significant selectionfor increased oxygen treatment despite a relativelysmall difference in DO concentrations betweentreatments However, this pattern was not observedwhen oxygen levels were increased, simulatingmoderate levels of hypoxia During the 2.0 mg O2

l−1 seagrass vs 4.0 mg O2 l−1 non-vegetatedsubstrate experiment, pinfish selection patterns weremore variable, although fish spent more time in thelower oxygen seagrass chamber Overall, no signif-icant selection patterns were detected for thisexperiment (t = 1.1, df = 5, p = 0.33) In the 4.0 mg

O2 l−1 seagrass vs 6.0 mg O2 l−1 non-vegetatedsubstrate experiment, a similar pattern was observed asfish selected the lower oxygen seagrass chamber,but the response was variable and a significant selectionpattern was not detected (t=1.9, df=5, p=0.121).Habitat selection patterns of pinfish were alsodetermined when DO concentrations were reduced

in the sand bottom treatment relative to theirpreferred seagrass habitat In this situation, pinfish

a

d c

b

Fig 2 Mean ± SE percentage occurrence of pinfish and

croaker in each habitat*DO treatment combination Each

solid-open bar pair represents six replicate 30 min mesocosm

trials with the following treatments: a high DO-non-vegetated,

low DO-seagrass; b low DO- non-vegetated, high DO-seagrass;

c high DO- non-vegetated, low DO-seagrass; d low DO-sand, high DO-seagrass Significant results from one-sample Student’s t-tests are indicated by *=p<0.05 and ***=p <0.001

displayed significant selection patterns for the

increased DO-seagrass treatment, as this treatment

contained both the favored abiotic and biotic

conditions within one chamber of the mesocosm

(Fig 2b)

Habitat selection patterns were also determined for

croaker In contrast to pinfish, croaker is a habitat

generalist (Petrik et al.1999) However, the influence

of habitat type and DO concentration was similar

between both species Dissolved oxygen exerted

greater influence on selection patterns during hypoxic

conditions (e.g DO≤2.0 mg O l−1), while habitat

type was more important during moderate hypoxia ornormoxic conditions Habitat selection experimentswith croaker were investigated by placing the favoredsubstrate (non-vegetated bottom as determined frompreliminary trials) and high oxygen treatment inseparate chambers Similar to pinfish, croaker avoidedtheir preferred habitat type when DO levels were1.0 mg O2l−1 This was observed for both the 4.0 mg

O2l−1seagrass vs 1.0 mg O2l−1non-vegetated bottom(t=6.0, df=5, p=0.002) and 2.0 mg O2l−1seagrass vs.1.0 mg O2l−1non-vegetated bottom experiments (t=30.5, df= 5, p < 0.001) (Table 1, Fig 2c) When

Table 1 Summary of substrate-DO concentration mesocsom

experiments Preference indicated significant selection for one

chamber of the substrate-DO combination P-values are

indicated from one-sample t-tests against the null expectation

of 50% selection for each chamber Six replicates (n=6) were conducted for each substrate-DO combination for pinfish (Lagodon rhomboides) and croaker (Micropogonias undulatus)

minimum DO levels were increased in the sand bottom

chamber to 2.0 and 4.0 mg O2 l−1, respectively,

selection patterns were more variable Overall, croaker

selection patterns were similar to pinfish, as the mean

proportion of time spent in elevated DO-seagrass

chamber was higher; although, significant patterns

were not detected for either the 4.0 mg O2l−1seagrass

vs 2.0 mg O2l−1sand bottom (t=2.5, df=5, p=0.056)

or 6.0 mg O2 l−1 seagrass vs 4.0 mg O2 l−1 sand

bottom (t=1.7, df=5, p=0.151) experiments

Habitat selection patterns of croaker were also

determined during a reciprocal set of experiments

where oxygen levels were reduced in the seagrass

substrate relative to non-vegetated substrate Croaker

demonstrated significant avoidance of seagrass when

DO levels were reduced to 1.0 mg O2l−1within this

treatment This pattern was observed during both the

Food vs DO selection experiments

The addition of food resources influenced selectionpatterns of pinfish during moderate hypoxia (4.0 mg

O2 l−1) but did not influence selection patterns at1.0 mg O2l−1 Both chambers of the mesocosm wereinitially set to 4.0 mg O2l−1and a mysid shrimp foodsupplement was added to one side Pinfish showed a

Fig 3 Mean ± SE percentage of time (a) pinfish and (b)

croaker spent in each food enhanced*DO treatment

combina-tion Each solid-open bar pair represents six replicate 10 min.

treatment Mean percentage of time (c) pinfish and (d) croaker spent in each chamber of the mesocosm in the presence of a three red drum predators with two different DO*predator

t-tests are indicated by * = p <0.05

marginally significant preference for food (t=2.5, df=5,

p=0.052) (Fig.3a) During a second experiment food

treatment was placed in a reduced oxygen treatment

(1.0 mg O2l−1), while the other chamber remained at

4.0 mg O2 l−1 without a food supplement Pinfish

avoided the food enriched chamber and exhibited a

significant selection for the 4.0 mg O2 l−1 treatment

despite the lack of food (t=3.4, df=5, p=0.020) Both

experiments were repeated for croaker During

moder-ate hypoxia (4.0 mg O2 l−1) food enrichment did not

influence selection patterns (t=2.1, df=5, p=0.089),

and croaker avoided the food supplemented treatment

when placed in the low oxygen treatment (t=3.9, df=5,

p=0.011) (Fig.3b)

Predator presence vs DO selection experiments

The presence of predators exerted a strong influence on

selection patterns for both species at all levels of DO

concentration tested An initial experiment was

con-ducted to determine the effect of predator presence on

the habitat selection patterns in absence of a DO

difference (both chambers set to 4 mg O2 l−1) Both

species strongly avoided predators in the control

experiment (no DO concentration difference, pinfish,

t=6.4, df=5, p=0.001; croaker, t=6.6, df=5, p=0.001)

(Fig 3c-d) In a second experiment fish had a choicebetween predators with 4.0 mg O2l−1and no predatorswith 1.0 mg O2 l−1 Both species chose chamberswithout predators, despite the low oxygen concentra-tion (pinfish, t = 6.6, df = 5, p = 0.001; croaker, t = 4.5,

df = 5, p = 0.007)

The response of habitat selection patterns to DO,substrate, and food was strongly interactive (Fig 4).During hypoxic conditions, DO concentration was animportant determinant of habitat selection patterns.However, as oxygen levels increased, the relativeimportance on fish habitat selection decreased andsubstrate preference became more important As withsubstrate, food availability influenced selection pat-terns of pinfish at 4.0 mg O2 l−1, however, foodavailability was unimportant at 1.0 mg O2 l−1 Thepresence of predators exerted the greatest influence onhabitat selection Both species strongly avoidedpredators even when the alternative habitat wasseverely hypoxic (1.0 mg O2l−1)

Discussion

Habitat selection influences distribution, abundance, andpopulation dynamics of mobile organisms (Johnson

Fig 4 A general mechanistic hypothesis of habitat selection of

juvenile estuarine fishes based on our experimental

observa-tions Dissolved oxygen concentration was an important

determinant of habitat selection patterns when concentrations

(substrate) and food availability became increasingly important

with increasing oxygen concentration During low oxygen

conditions food availability or substrate type did not influence

habitat selection however the relative importance increased with DO concentration Patterns of habitat selection in response

to predation risk and the DO concentrations were similar between species Predation risk (dashed line), at least in the case of high predator density in this study, exerted the greatest influence on habitat selection patterns of juvenile estuarine fishes (of the factors examined in the current study) across all levels of DO concentration considered

1980; Levin et al 1997; Stunz et al 2001; Morris

2003) Preferential selection for high quality habitats

may increase growth rates or survivorship, ultimately

contributing disproportionately to adult populations

(Beck et al 2001) However, habitat quality may be

influenced by a myriad of abiotic and biotic factors

and improved management of marine resources

requires a detailed understanding of the mechanism

used by fishes to select the highest quality habitat

available (Morris 2003)

Abundance of estuarine organisms is typically higher

in structurally complex habitats such as seagrass

mead-ows (Jordan et al 1997; Levin et al 1997) Seagrass

meadows are particularly important habitats for newly

recruited juvenile fishes (Burfeind and Stunz2006) In

experiments without DO treatments, pinfish showed

significant selection preference for seagrass habitat

although croaker, a habitat ‘generalist’ preferred

non-vegetated habitat This is consistent with previous

laboratory experiments (Petrik et al 1999) and field

observations (Jordan et al 1997) for these species

Despite significant substrate preferences, both species

avoided their“preferred” substrate when placed in low

DO concentrations while at higher levels, adequate DO

levels allowed other factors such as “preferred”

substrate or prey availability to influence habitat usage

patterns In a Galveston TX estuary, both recruitment

and growth rates of pinfish were higher in seagrass as

compared to non-vegetated habitats (Levin et al.1997)

Juvenile red drum also grew significantly faster in

vegetated as compared to sand substrates in

experimen-tal field enclosures (Stunz et al 2002) Similar to the

substrate treatment, the addition of food only influenced

selection patterns of pinfish in absence of hypoxic

conditions Pinfish avoided the food enriched treatment

when placed in the low oxygen treatment, suggesting

that food availability is not a strong driver of habitat

selection in estuarine ecosystems where food is

typically abundant (Heck et al.2003)

The ability of estuarine organisms to detect and

avoid hypoxia in laboratory mesocosms was

previ-ously reported (Wannamaker and Rice2000; Stierhoff

et al 2009) As with the current study, avoidance

patterns were greatest when DO treatments were

below 2 mg O2 l−1 This coincides with the level at

which fish emigrate from hypoxic areas and is

associated with significant reductions in abundance

(Breitburg 2002), and diversity (Vaquer-Sunyer and

Duarte2008; Montagna and Froeschke2009) Habitat

selection patterns of flatfishes in the Gulf of Mexicowere altered by low DO levels with reduced habitatsuitability in regions with hypoxia and increasedsuitability in nearby refuges (Switzer et al 2009).This study suggests potential wide-scale alteration ofhabitat selection patterns due to hypoxia and suggeststhat this factor alone may induce emigration oravoidance of otherwise suitable habitats (Utne-Palm

et al.2010) In estuaries, hypoxia may reduce quality

of nursery habitat even if preferred habitat types andfood resources are abundant

Hypoxia impacts may be most severe for juveniles asthe dispersal potential may be limited due to their smallsize and increased risk of predation during movementaway from hypoxic zones In a study of intermittenthypoxia in Chesapeake Bay, juvenile fishes were lessable to escape than adults and mortality rates ofjuveniles was extremely high (Breitburg 1992) In-creased mortality rates of small fish due to hypoxiamay be associated with increased oxygen demands ofjuveniles, reduced swimming speeds (Breitburg1992),

or increased predation risk associated with emigration.Results from the current study provide further evidencefor predator-mediated habitat selection (Jordan et al

1997), and the critical role that predators play inecosystem regulation (Heck and Valentine2007) Longand Seitz (2008) reported increased susceptibility ofbenthic prey to predators from hypoxia in ChesapeakeBay However, Altieri (2008) suggests that responses tohypoxia may reduce predation and hypoxia tolerantspecies such as quahog clam (Mercenaria mercenaria)may benefit from non-lethal hypoxia events Howeverreductions in abundance and diversity of speciessensitive to hypoxia have been observed (Altieri

2008; Montagna and Froeschke 2009) If juvenilefishes forgo emigration from hypoxia due to predationrisk, they are subject to the physiological effects ofhypoxia and long-term impacts on fish populationsmay be observed Landry et al (2007) reported reducedreproductive output and Eby et al (2005) predictedlong-term population declines of croaker resultingfrom exposure to hypoxic conditions However,species specific responses to hypoxia is typical andhas been reported in both laboratory and field studies(Wannamaker and Rice2000; Montagna and Froeschke

2009; Switzer et al 2009) These results suggestthat environmental stressors such as hypoxia can beimportant determinants on community structure(Menge and Sutherland 1987; Lenihan et al 2001),

where some species may benefit but net declines in

diversity and resilience may be expected from

ecosys-tem stressors

Hypoxia may exert direct or indirect effects on

population dynamics of juvenile fishes Populations

may be affected directly from hypoxia either through

increased mortality or decreased recruitment due to

avoidance of hypoxic areas Indirect effects including

reduced growth rate, increased density dependent

competition in normoxic refuges and greater

preda-tion risk have been hypothesized previously While

indirect effects are more difficult to empirically

demonstrate (Heck and Valentine2007), they may exert

greater long-term effects on the population dynamics

and community structure of estuarine systems For

example, the rate of juvenile survival is often cited as

the best predictor of subsequent adult population size

(Caley et al.1996; Levin and Stunz2005) and hypoxia

induced exposure has been shown to reduce growth

rates of some fishes (Chabot and Dutil1999; Eby et

al 2005; Stierhoff et al 2006) and increase their

duration in critical life stages where predation risk is

high (Levin et al 1997; Levin and Stunz 2005)

Moreover, reduced growth rate may delay sexual

maturity and reduce total reproductive output leading

to long-term reductions in population size Growth

rates of croaker were reduced significantly inside a

hypoxic estuary and subsequent reductions of

demo-graphic rates were predicted to result in long-term

population declines in the estuary (Eby et al 2005)

Stierhoff et al (2006) reported reduced feeding and

growth rates of Paralichthys dentatus and

Pseudo-pleuronectes americanus due to moderate hypoxia

while similar results were also reported for Gadus

morhua (Chabot and Dutil 1999) Reduced growth

rates may ultimately lead to substantial reductions in

fisheries productivity, predator densities, and ultimately

a disruption of vital ecosystem links and trophic

interactions to the detriment of ecosystem based

management goals

Expansion of low oxygen areas is currently

consid-ered among the most damaging environmental

prob-lems (Diaz and Rosenburg 2008) This problem will

intensify as low oxygen zones increase both temporally

and spatially throughout coastal and estuarine regions

from enhanced nutrient deposition and warming

seas (Diaz and Rosenburg 2008; Vaquer-Sunyer and

Duarte2008) Assessing impacts of these changes on

habitat usage of mobile organisms is critical as

changes in environmental metrics including predatordistribution and DO levels may alter habitat selectionpatterns and reduce fitness levels of individuals andpotentially disrupting vital ecosystem links andtrophic interactions to the detriment of ecosystem-based management goals

the Texas Research Development Fund We thank the members

of the Fisheries Ecology Laboratory at Texas A&M Corpus Christi for their help with this project Special thanks to

University-M Reese and L Brown for their field and laboratory assistance.

References

Altieri AH (2008) Dead zones enhance key fisheries species by

Applebaum S, Montagna PA, Ritter C (2005) Status and trends

of dissolved oxygen in Corpus Christi Bay, Texas, U.S.A.

Beck MW, Heck KL, Able KW, Childers DL, Eggleston DB, Gillanders BM, Halpern B, Hays CG, Hoshino K, Minello

TJ, Orth RJ, Sheridan PF, Weinstein MR (2001) The identification, conservation, and management of estuarine and marine nurseries for fish and invertebrates Bioscience

Bell GW, Eggleston DB (2005) Species-specific avoidance responses by blue crabs and fish to chronic and episodic hypoxia Mar Biol 146:761–770

Breitburg D (2002) Effects of hypoxia, and the balance between hypoxia and enrichment, on coastal fishes and fisheries.

Breitburg DL (1992) Episodic hypoxia in Chesapeake Bay: Interacting effects of recruitment, behavior, and physical

Burfeind DD, Stunz GW (2006) The effects of boat propeller scarring intensity on nekton abundance in subtropical

Caley MJ, Carr MH, Hixon MA, Hughes TP, Jones GP, Menge

BA (1996) Recruitment and the local dynamics of open marine populations Annu Rev Ecol Syst 27:477 Chabot D, Dutil JD (1999) Reduced growth of Atlantic cod in

Diaz JR, Rosenburg R (2008) Spreading dead zones and consequences for marine ecosystems Science 321:926– 929

Eby LA, Crowder LB, McClellan CM, Peterson CH, Powers

MJ (2005) Habitat degradation from intermittent hypoxia:

261 Gilliam JF, Fraser DF (1987) Habitat selection under predation hazard: Test of a model with foraging minnows Ecology

Halpern BS, Walbridge S, Selkoe KA, Kappel CV, Micheli F,

Heinemann D, Lenihan HS, Madin EMP, Perry MT, Selig