Intercontinental Comparison of Fish Life History Strategies Along a Gradient of Hydrologic Variability Julian D

Intercontinental Comparison of Fish Life History Strategies along a Gradient of Hydrologic Variability Julian D. Olden* School of Aquatic and Fishery Sciences, Box 355020, University of Washington Seattle, Washington 98195, USA Mark J. Kennard Australian Rivers Institute, Griffith University, Nathan, Queensland 4111, Australia

Abstract.—The flow regime is considered the primary driver of physical processes in riverine ecosystems; thus we expect that the trait composition of fish assemblages might respond similarly to hydrologic variability, even at broad spatial scales. Here, we test the hypothesis that freshwater fish life history strategies on two continents (southern United States and eastern Australia) converge along gradients of hydrologic variability and primary productivity at the drainage scale. Our results show that the fishes of the United States and Australia conform to the three-dimensional adaptive space arising from the trade-offs among three basic demographic parameters of survival, fecundity, and onset and duration of reproductive life. Species from both continents represent the endpoints in adaptive space defining the periodic (19% versus 33% for the United States and Australia, respectively), opportunistic (69% versus 52%), and equilibrium life history strategies (12% versus 15%). We found evidence that fish life history composition of drainage basins in the two continents have converged across similar gradients of hydrologic variability and productivity despite phylogenetic and historical differences. Moreover, these relationships were largely consistent with predictions from life history theory. Increasing hydrologic variability has promoted the greater prevalence of opportunistic strategists (a strategy that should maximize fitness in environmental settings dominated by unpredictable environmental change) while concurrently minimizing the persistence of periodic-type species (a strategy typically inhabits seasonal, periodically suitable environments). Our study provides a conceptual framework of management options for species in regulated rivers because life history strategies are the underlying determinants for population responses to environmental change and therefore can be used to classify typical population responses to flow alteration or mitigation via environmental flow prescriptions.

Introduction nized (Naiman et al. 2008). Flow has been suggested to be the “master variable” that deter- The importance of hydrologic variability for mines pattern and process in rivers, thus limit- shaping the biophysical attributes and func- ing the distribution and abundance of species tioning of riverine ecosystems is well recog- and regulating ecological integrity (Poff et al. * Corresponding author: [email protected] 1997). Natural spatial variation in the hydro- 83 84 olden and kennard logic regime is influenced by variations in cli- Poff et al. 2006; Verberk et al. 2008; Frimpong mate and basin geology, topography, and veg- and Angermeier 2010, this volume). Species etation, which interact at multiple spatial and traits may be intercorrelated through physio- temporal scales to shape the physical template logical constraints, trade-offs (i.e., investments upon which ecological and evolutionary pro- in one trait leaving fewer resources available cesses operate in river ecosystems (Poff and for investment in another), or spin-offs (i.e., in- Ward 1990; Bunn and Arthington 2002; Lake vestments in one trait reduce costs or increase 2008). the benefits of investment in another trait), The natural flow-regime paradigm postu- resulting in the creation of life history strate- lates that the structure and function of riverine gies or tactics represented as sets of coevolved ecosystems, and the adaptations of their con- traits that enable a species to cope with a range stituent riparian and aquatic species, are dictat- of ecological problems (Stearns 1992). Com- ed by patterns of intra- and interannual varia- parative studies from a diverse array of fishes tion in river flows (Poff et al. 1997). A rich body in marine and freshwater systems have inde- of literature has demonstrated that the long- pendently identified three primary life his- term physical characteristics of flow variability tory strategies that represent the endpoints of have strong consequences at local to regional a triangular continuum arising from essential scales and at time intervals ranging from days trade-offs among the basic demographic- pa (ecological effects) to millennia (evolution- rameters of survival, fecundity, and onset and ary effects). Flow variability, including flood duration of reproduction (Winemiller 1989, and drought events, interacts with the under- 1992; Winemiller and Rose 1992; Vila-Gispert lying geology to shape the river’s physical and and Moreno-Amich 2002; Vila-Gispert et al. chemical templates and constrain assemblage 2002; King and McFarlane 2003; see Figure structure for fish (e.g., Lamouroux et al. 2002; 1). Winemiller and Rose (1992) synthesized Hoeinghaus et al. 2007; Kennard et al. 2007), these life history trade-offs and proposed the stream invertebrates (e.g., Poff and Ward 1989; following characteristic biological and habitat Vieira et al. 2004; Dewson et al. 2007), ripari- environmental attributes associated with the an plants (e.g., Nilsson et al. 1993; Naiman and three life history strategies. Periodic strategists Décamps 1997; Pettit et al. 2001), and riparian are large-bodied fishes with late maturation, invertebrates (e.g., Wenninger and Fagan 2000; high fecundity per spawning event, and low ju- Lambeets et al. 2008). Adaptations to natural venile survivorship (i.e., no parental care) and flow regimes include behaviors that enable in- typically inhabit seasonal, periodically suitable sects to avoid desiccation by droughts, fish life environments with large-space spatial (patchi- history strategies that are synchronized to take ness) and temporal (seasonality) heterogene- advantage of floodplain inundation, and plant ity. Opportunistic strategists are small-bodied morphologies that protect roots by jettisoning fishes with early maturation, low fecundity per seasonal biomass during floods (reviewed in spawning event, and low juvenile survivorship Lytle and Poff 2004). and typically inhabit habitats subjected to fre- There has been considerable interest in the quent and intense disturbances. Equilibrium identification of major axes of ecological strat- strategists are small to medium-bodied fishes egy variation in freshwater ecosystems based with moderate maturation age, low fecundity on trait correlations across large numbers of per spawning event, and high juvenile survivor- species (Winemiller 2005; Olden et al. 2006; ship (i.e., provides parental care) and typically intercontinental comparison of fish life histories 85

Figure 1. Triangular life history model depicting environmental gradients selecting for endpoint strategies defined by optimization of demographic parameters generation time, fecundity, and juvenile survivorship (modified from Winemiller 2005). Example species from the study regions and representative hydrologic regimes expected to favor each life history strategy are illustrated. Australian fish illustrations by B. J. Pusey; U.S. fish illustrations freely available on the Internet. inhabit environments with low variation in odic fish species in rivers of the Cote d’Ivoire habitat quality and strong biotic interactions. (Africa) compared to equilibrium species. By The demographic parameters discussed comparing trends in native and nonnative spe- above are direct reflections of the ways in cies distributions among fish life history strate- which fish allocate energy to reproduction, gies in the lower Colorado River basin (USA), and the three life history strategies of the con- Olden et al. (2006) found that century-long tinuum can be interpreted as being adaptive modifications in flow variability have likely with respect to relative variability and predict- promoted the spread of nonnative equilibrium ability of temporal and spatial variation in abi- strategies (favored in constant environments) otic environmental conditions, food availabil- while leading to greater distributional declines ity, and predation pressure (Winemiller 2005). of native species located along the periodic- Hydrological variability plays a dominant role opportunistic continuum (strategies favored in shaping physical processes in riverine eco- in more unpredictable and variable environ- systems, and a number of recent studies have ments). Tedesco et al. (2008) found higher supported the association between hydrology proportions of periodic species in highly sea- and fish life history strategies. Tedesco and sonal drainage basins of western Africa (e.g., Hugueny (2006) found that regional climate rivers with short and predictably favorable (presumably related to hydrologic variability) seasons), whereas more hydrologically stable induced greater population synchrony of peri- basins with a wet season of several months 86 olden and kennard were dominated by equilibrium strategists. Col- tunistic-periodic-equilibrium trichotomy. The lectively, these studies (and others) suggest that freshwater fish faunas differ strikingly between the distributions of freshwater fishes are shaped, the two regions, with the U.S. fauna being ex- at least in part, by interactions between life his- tremely diverse and with low endemism in tory strategies and patterns of predictability and comparison to the comparatively depauperate variability in hydrologic regimes. and highly endemic Australian fauna. Our pri- Enhancing our mechanistic understand- mary objective was to test the hypothesis that ing of the functional linkages between envi- the relationship between fish life history pat- ronmental drivers of fish species distributions terns, hydrologic variability and net primary will provide a foundation for predicting exist- productivity (NPP) in the United States and ing and future impacts of altered flow regimes, Australia is concordant with predictions from invasive species, and land-use change. Gaining life history theory, as proposed by Winemiller this knowledge is predicated upon examining (1995, 2005) and McCann (1998) (Figure 1). how fish assemblage structure varies in re- Specifically, by virtue of their small size and sponse to flow variability and testing the gen- rapid turnover rates, we predict that oppor- erality of these relationships across large spatial tunistic species should maximize fitness (and scales. By using the “habitat templet” concept thus be more prevalent) in drainage basins proposed by Southwood (1977), and applied characterized by productive and frequently to riverine systems by Townsend and Hildrew disturbed habitats (i.e., high hydrologic vari- (1994), we predict that because life history ability and high NPP). By contrast, periodic strategies are synchronized with long-term hy- and equilibrium strategists are predicted to be drologic dynamics, even geographically distant more common in drainage basins exhibiting regions will presumably favor the persistence low-flow variability and either more produc- of species with similar traits if subjected to sim- tive habitats for periodic strategists (i.e., low ilar historical flow regimes. If the functional hydrologic variability and high NPP) or less characteristics of organisms and communities productive habitats for equilibrium strategists are predictable from features of their environ- (i.e., low hydrologic variability and low NPP). ment, then convergence in these characteris- For periodic strategists this may be seen as a tics across regions implies the existence of key, consequence of the storage effect (i.e., long life repeated evolutionary mechanisms respon- stage buffers times of low recruitment: Warner sible for these relationships (Schluter 1986). and Chesson 1985) and high annual fecundi- Community convergence driven by similar ties, and for equilibrium strategies as a result environmental selective forces would be sup- of low adult mortality rates that enable them to ported by observations of similar relationships survive on lower resource biomass (McCann between spatial variations in community life 1998). By taking a functional approach, our history composition along environmental gra- study underscores the theoretical expectation dients (Lytle and Poff 2004). that species traits promoting local persistence In the present study, we explore this ques- will change along environmental gradients, tion by providing an intercontinental examina- thus facilitating the comparison of regions tion of patterns in fish functional composition separated by large geographic distances and across a gradient of hydrologic variability and aiding the development of broadly applicable productivity in the southern United States generalizations regarding patterns of fish life and eastern Australia, focusing on the oppor- history strategies. If fish species show similar intercontinental comparison of fish life histories 87 ecological responses to the same environmen- basin differences (see below). We selected drain- tal challenges irrespective of their ancestry, we age basins in the United States (n = 56 basins) expect that hydrologic variability has selected and Australia (n = 56 basins) that are located for similar community patterns in fish life his- within the temperate climate types Cfa and Cfb, tory strategies despite originating from com- representing the most prevalent climate that is pletely different regional species pools at op- common to both continents (Figure 2A). These posite sides of the world. climate regions are characterized by a temperate climate, without a dry season, and experience Methods either a hot summer with the hottest month $228C (Cfa) or a warm summer with the hot- Study Region test month less than 228C and $ 4 months We used an updated global map of the Köppen- where temperatures exceed 108C (Cfb). All Geiger climate classification (Peel et al. 2007) to drainage basins in the United States discharge to identify study regions in the United States and the Gulf of Mexico and are located in the Ohio Australia, having experienced similar regional River, Tennessee River, lower Mississippi River, climate (average annual air temperature and to- and coastal regions of southeastern states, as tal precipitation) and runoff regimes. By select- well as including small areas of the upper Mis- ing study regions from the same climate type, we sissippi, Missouri, Arkansas, White, and Red were facilitating the direct comparison of fish as- rivers (Figure 2B). Drainage basins in Australia semblages from drainage basins that have likely include all major eastern coastal rivers between experienced a similar range of historical climatic the Fitzroy River in the north and the Tarwin and runoff conditions while still exhibiting inter- River in the south (Figure 2C).

Figure 2. (A) Global map of the Köppen-Geiger climate classification (from Peel et al. 2007) depicting the two study regions in the United States and Australia (squares) that are located within the temperate climate types Cfa and Cfb. (B) Fish species richness of drainage basins in the United States. (C) Fish species richness of drainage basins in Australia. 88 olden and kennard Species Distributions States and Australia. We collated data for 14 ecological and life history attributes that could We assembled present-day species lists of na- be justified on the basis of our current state tive freshwater fishes (i.e., excluding nonnative of knowledge and information available for translocated native species) for drainage basins the majority of species pools (Table 1). These located in our study region (hereafter termed traits were divided into five categories accord- the “United States” and “Australia” for brevity, ing to body morphology—maximum total although specifically referring to the south- body length, shape factor, and swim factor ern United States and eastern Australia). We (following Webb 1984); behavior—substrate defined a freshwater fish in a relatively broad preference and vertical position; life history— sense to include those species that can repro- longevity, age at maturation (female), length at duce in freshwater and diadromous species maturation (female), total fecundity, egg size, that spend the majority of their lives in fresh- spawning frequency, reproductive guild (fol- waters. We excluded numerous species with lowing Balon 1975), and parental care (follow- strong marine or estuarine affinities that may ing Winemiller 1989); and trophic—trophic enter freshwaters for only short periods of time. feeding guild according to adult feeding mode United States distribution data on 559 native based on published diet analyses. Parental care fish species (from 36 families and 104 genera) was quantified as the Sx for i = 1–3, where x in 56 drainage basins were obtained from the i 1 = 0 if no special placement of zygotes, x = 1 NatureServe Database (NatureServe 2004), 1 if special placement of zygotes, x = 2 if both which summarizes data compiled by state 1 zygotes and larvae maintained in nest, x = 0 if natural heritage programs, museum records, 2 no parental protection of zygotes or larvae, x published literature, and expert opinion. The 2 = 1 if brief period of protection by one sex (<1 database was originally compiled at the 8-digit month), x = 2 if long period of protection by hydrologic unit code (representing a hierarchi- 2 one sex (>1 month) or brief care by both sexes, cal system of drainages) but was converted to x = 4 or lengthy protection by both sexes (>1 the four-digit hydrologic unit code scale to en- 2 month), x = 0 if no nutritional contribution to sure the completeness of species records. Na- 3 larvae, x = 2 if brief period of nutritional con- tive freshwater fish species distribution data 3 tribution to larvae (<1 month), x = 4 if long for Australia was assembled using information 3 period of nutritional contribution to larvae from Pusey et al. (2004), as well as by reference (1–2 months), and x = 8 if extremely long pe- to data sets held by state government agencies, 3 riod of nutritional contribution to larvae (>2 regional experts, existing literature, and inter- months). rogation of museum records. Fish occurrence Trait assignments were based on a multi- data for Australia was obtained for 66 native tiered data collection procedure. First, trait fish species (from 25 families and 45 genera) data were collected from species accounts in in 56 drainage basins. the comprehensive texts of the regional fish faunas: United States (references available upon Species Traits request) and Australia (Pusey et al. 2004 and We used the scientific literature and elec- references therein). Second, we used species tronic databases to provide a comprehensive descriptions from the primary literature, state functional description of the native freshwa- agency reports, university reports, and gradu- ter fishes of the study regions in the United ate theses. Third, we obtained data from elec- intercontinental comparison of fish life histories 89 Table 1. Ecological and life-history traits quantified for freshwater fishes of the United States and Australia. Trait Description Abbreviation Body morphology Maximum body length Maximum total body length (cm) MBL Shape factor Ratio of total body length to maximum body depth ShapeF Swim factor Ratio of minimum depth of the caudal peduncle to the SwimF maximum body depth Behavior Substrate preference Coarse (rocks, cobble, coarse gravel) Coa Fine (sand, fine gravel) Fin Silt/mud Sil General Gen Vertical position Benthic Ben Nonbenthic Non-Ben Life history Longevity Maximum potential life span (years) Long Age at maturation Mean age at maturation (years) MatAge Length at maturation Mean total length at maturation (cm) MatLen Total Fecundity Total number of eggs or offspring per breeding season Fecund Egg size Mean diameter of mature (fully yolked) ovarian oocytes (mm) EggS Spawning frequency Single spawning per lifetime SinL Single spawning per season SinS Multiple (batch/repeat/protracted) spawning per season MulS Reproductive guild Nonguarders (open substratum spawners) NgOS Nonguarders (brood hiders) NgBH Guarders (substratum choosers) GSC Guarders (nest spawners) GNS Bearers (internal) BI Bearers (external) BE Parental care Metric representing the total energetic contribution of parents ParentC to their offspring sensu Winemiller (1989) Trophic Trophic guild Herbivore-detritivore (ca. >25% plant matter) HeDe Omnivore (ca. 5–25% plant matter) Omni Invertivore Inve Invertivore–piscivore InvePisc

tronic databases available on the World Wide length at maturation were recorded for female Web, including FishBase, and U.S. state natural specimens when possible. heritage programs. Fourth, expert knowledge was used to assign values to a small number Environmental Variables of trait states that could not be obtained from the previous methods (mainly inferred from Basin-level life history composition was mod- congenerics). Trait values were represented by eled using contemporary environmental fac- ordinal, nominal, or continuous data (Table tors selected a priori, based on previous em- 1). Ordinal and nominal traits were assigned pirical demonstrations of their importance as a single state based on a majority of evidence determinants of global patterns of fish species rule according to adult preferences, and medi- richness (e.g., Oberdorff et al. 1997; Guégan et an values for continuous traits were used when al. 1998). These variables included total - sur ranges were presented. Trait values for age and face area of the drainage basin (km2) and mean 90 olden and kennard annual runoff (million liters per square kilo- ensure a similar historical range of flow condi- meter: ML/km2) as measures of habitat avail- tions but potentially some level of interbasin ability and diversity, net terrestrial primary differences in hydrologic variability within productivity (NPP, metric tons of carbon per and between study regions. Also, note that km2) as a measure of measure of available ener- data availability precluded the quantification gy, and coefficient of variation in annual runoff of intra-annual (seasonal) variability in dis- as a measure of hydrologic variability. Net ter- charge. However, we note that at the continen- restrial primary productivity was used because tal scale, there is a strong correlation between net aquatic primary productivity was not avail- inter-annual variability (used in this study) and able. Notably, by only selecting drainage basins intra-annual variability in discharge: United that are located in the same climatic region, States (R = 0.65, P < 0.05, n = 420, data source: we eliminated the need to include variables Olden and Poff 2003) and Australia R( = 0.78, such as latitude, air temperature, and annual P < 0.05, n = 830, data source: Kennard et al., precipitation to represent broad-scale climate in press). Below, we discuss the methodology conditions. Annual NPP was estimated using used to quantify basin runoff and hydrologic the Carnegie Ames Stanford Approach carbon variability. model (the amount of carbon produced by For the United States, U.S. Geological Sur- ecosystems per 0.25 degree grid cell), where vey daily streamflow data for 1900–2002 were the average value of all grid cells in each basin used to estimate runoff (streamflow per unit was computed using a geographic information area) for the hydrologic cataloging units in the system. The NPP data set is distributed by the study region, following the approach of Krug Columbia University Center for International et al. (1987). This data set is available from Earth Science Information Network and was the U.S. Geological Survey (http://water.usgs. obtained from Imhoff et al. (2004). Briefly, the gov/waterwatch) and was created by first iden- model incorporates satellite and climate data tifying stream gauges with the following char- (1982–1998) to estimate the fixation and re- acteristics: (1) complete data for each water lease of carbon based on a spatially and tem- year, (2) drainage basins contained entirely porally resolved prediction of NPP in a steady within a cataloging unit, and (3) drainage areas state (described in detail by Imhoff and Boun- that did not overlap with the drainage basins oua 2006). We chose to apply a global model of other gauges in the same cataloging unit. to estimate NPP rather than continental mod- The average daily flow for the water year then els that are readily available for both the Unit- was divided by the drainage basin area for each ed States (MODIS: Terra Dynamic Simulation gauge to calculate runoff, and a drainage basin Group) and Australia (Australian Natural Re- area weighted average runoff value was com- sources Data Library: CSIRO) to control for puted for each cataloging unit. An adequate inherent differences in NPP predictions aris- number of gauges were identified using the ing from differences in model structure. above criteria only for some cataloging units On an interannual basis, variability in ba- during some years. For the remaining catalog- sin runoff will act as a qualifier of disturbance ing units, the runoff estimates were based on and habitat availability and suitability in river- (1) gauges with small drainage basins with an ine ecosystems. We selected drainage basins unknown degree of overlap, (2) runoff values from the United States and Australia located in for neighboring cataloging units, and (3) aver- same temperate climate type (Cfa and Cfb) to age runoff values for larger accounting units. intercontinental comparison of fish life histories 91 We refer the reader to Krug et al. (1987) for 0.45–2.10) than in the United States (mean = more details. 0.56, range = 0.21–1.81). For Australia, drainage basin runoff was estimated using discharge data modeled using Data Analyses the water balance module of the GROWEST program (Hutchinson et al. 2004). GROW- Accounting for phylogeny.—It is expected EST operates on a weekly time step, converting that species share similar life history attri- monthly input rainfall and evaporation data to butes through descent from common ances- weekly values via cubic Bessel interpolation. try (Fisher and Owens 2004). It is therefore The water balance module is conceptualized necessary to account for phylogenetic effects as a single “bucket” model. It adds rainfall to when exploring patterns in ecological data the previous soil storage and removes it by because nonindependence among species means of evapotranspiration. The soil water violates the assumption of standard statistics surplus or “runoff” is the rainfall exceeding that errors are uncorrelated and may result in “bucketful” after allowing for evapotranspira- biased parameter estimates and increased type tion. Monthly rainfall and pan evaporation es- I error rates if phylogeny is not taken into ac- timates were generated at a grid spacing of 0.01 count. Phylogenetically informed analyses degrees (approximately 1 km) using elevation require an estimate of the phylogenetic rela- values derived by resampling the 9 s DEM ver- tionships among the taxa concerned. Current sion 3 with bilinear interpolation and monthly approaches for controlling the effects of phy- climate surface coefficients for a 30-year pe- logeny typically involve the method of inde- riod from 1971 to 2000. We refer the reader to pendent contrasts (Felsenstein 1985); howev- Stein et al. (2008) for more details. Monthly er, this technique cannot accommodate mixed and annual catchment water balance values variables types that are present in our data set were calculated from the accumulated totals of and requires a complete phylogeny (i.e., esti- the upstream grid cell runoff estimates for each mates of the branch lengths associated with catchment. Although we recognize that differ- the phylogenetic tree) that is currently not ences exist in how runoff was modeled in the available. Therefore, we estimated the degree United States and Australia, the spatial reso- of phylogenetic relatedness following Tedesco lution of global runoff models are too coarse et al. (2008), where all species within the same to provide accurate estimates of runoff at the genus were assigned a value of 1.0, all species watershed scale. within the same family a value of 0.5, all spe- The range of variation in mean annual cies within the same order a value of 0.33, and runoff (USA: 19.5–698.1 ML/km2, AUS: all species from different orders a value of 0.25. 23.3–734.2 ML/km) and NPP measured The resulting phylogenetic similarity matrix (USA: 2.6–10.7 metric tons of carbon/km2, was then converted to a phylogenetic dissimi- AUS: 2.9–10.1 metric tons of carbon/km2) larity matrix (i.e., 1 similarity) and used to ac- was similar for both regions, although drainage count for phylogeny in subsequent analyses. basins in the United States tended to be larger Similarities in species’ trait characteristics.— (mean = 35,584 km2, range = 9,453–84,433 We summarized similarities in species’ trait km2) than in Australia (mean = 6,727 km2, characteristics by conducting a principal coor- range = 110–141,281 km2), and runoff in Aus- dinate analysis (PCoA) on the species-by-trait tralia was more variable (mean = 1.07, range = matrix. PCoA is a statistical methodology to 92 olden and kennard explore and to visualize similarities/dissimi- Species life history strategies.—We exam- larities in multivariate data by optimally rep- ined the life history strategies of 257 fish spe- resenting the variability of a multidimensional cies in the United States (limited to those spe- data matrix (here, the species-by-trait matrix) cies in which complete life history trait values in ordination space with reduced (i.e., lower) were available but representative of the entire dimensionality. A species similarity matrix pool of U.S. species) and 66 fish species in Aus- according to the 14 biological traits was cal- tralia. We employed two approaches to quan- culated using Gower’s similarity coefficient, a tify the community composition of basins metric able to accommodate mixed data types with respect to the opportunistic, periodic, or and missing values (Legendre and Legendre equilibrium strategies. This analysis did not ac- 1998). These properties make Gower’s coef- count for phylogeny, given its weak influence ficient an appropriate choice for our analysis on patterns of life history trait variation (see given that the data set contains nominal, ordi- result above). nal, and continuous traits and that it was not First, we evaluated the fish life history possible to assign values for all traits to all spe- continuum model of Winemiller and Rose cies given the lack of scientific knowledge (Ta- (1992) by plotting species’ positions in rela- ble 1). Prior to conducting a PCoA, we used tion to three life history axes: (1) ln matura- a modified version of the eigenvector method tion size (a surrogate of maturation age that is presented by Diniz-Filho et al. (1998) and ap- highly correlated with maturation size in our plied by Olden et al. (2006) to partition the study, Pearson’s R = 0.82 across all species); total variance in the biological trait distance (2) ln mean fecundity; and (3) investment matrix into its phylogenetic and specific com- per progeny, calculated as ln(egg diameter + ponents using a Mantel test. We regressed the 1) + ln(parental care + 1). Each species was phylogenetic distance matrix (see Accounting assigned to one of the three life history strate- for phylogeny) against the trait distance matrix gies (opportunistic, periodic, or equilibrium) (based on Gower’s similarity) using a Mantel by calculating the Euclidean distance in tri- test to derive a residual matrix that represents variate life history space between the spe- trait similarities among species after control- cies’ position and the strategy endpoints and ling for phylogenetic relatedness. The Mantel designating the species to the closest strategy. test showed a low correlation between the trait Strategy endpoints where defined as the fol- and phylogenetic distance matrices (Mantel’s lowing: opportunistic (minimum fecundity, standardized R = 0.279, P = 0.244), indicating minimum juvenile investment, and mini- only a marginal degree of phylogenetic con- mum maturation size), periodic (maximum straint. PCoA was then performed on the re- fecundity, minimum juvenile investment, and sidual similarity matrix to summarize the dom- maximum maturation size), and equilibrium inant patterns of variation among the biological (mean fecundity, maximum juvenile invest- traits and examine functional similarities and ment, and maximum maturation size). This differences among species. Only the first two calculation was based on normalized trait val- principal components were statistically signifi- ues (i.e., standardized range between 0 and 1 cant (P < 0.05, based on the broken-stick rule: for each trait) to ensure equal contributions Legendre and Legendre 1998), and they were of the three life history parameters. This ap- used to facilitate visual interpretation of the re- proach represents an advance over past stud- sulting plots. ies that have assigned species to life history intercontinental comparison of fish life histories 93 strategies according to arbitrary, but ecologi- Relationships of basin life history composi- cally relevant, thresholds (i.e., Tedesco et al. tion and environmental variables.—We used 2006; Hoeinghaus et al. 2007). Basin-level multiple linear regression analysis to model life history composition was summarized as basin-level life history composition (sum- the proportional species richness for each life marized as the distance-weighted propor- history strategy. tional strategy richness: see Species life his- Second, we recognize that many species tory strategies) as a function of drainage basin occupy immediate positions in life history area, mean annual runoff, mean NPP, and space, thus questioning the validity of a strict hydrologic variability. Analyses were con- classification of species into the opportunist- ducted separately for the United States and periodic-equilibrium trichotomy. We ad- Australia, and basin area, runoff, and NPP dressed this issue by computing a distance- were log-transformed prior to analysis. Fol- weighted measure of proportional species lowing Schluter (1986) and Lamouroux et richness for each life history strategy. This al. (2002), basin-scale convergence was ana- was accomplished by normalizing the Euclid- lyzed by comparing the effects of hydrologic ean distances in trivariate life history space variability and productivity on fish life history between species and each strategy endpoint traits, where a similar effect (based on direc- (i.e., standardized range between 0 and 1 tion and significance) of these factors within for each distance), computing the inverse of both continents indicates convergence. these values (so that larger values represent a greater “weight” or affinity of a species to- Results ward a life history strategy) and summing Trait Composition of U.S. and Australian the weights for each strategy in each basin Fishes based on the species present. Basin-level life history composition was summarized as the Freshwater fishes from the study regions in proportional composition of each life history United States and Australia varied in their strategy. In essence, this method produces morphological, ecological, and life history at- distance-weighted proportional species rich- tributes. The first two principal components ness for each life history strategy. of the PCoA, after controlling for phylogeny, Last, we summarized patterns of life his- explained 35.6% of the total variation repre- tory diversity by conducting a principal com- sented by the 14 traits; additional axes were ponent analyses (PCA) on seven life history nonsignificant and did not alter the inter- traits that included maximum body size, lon- pretation of results. The first principal axis gevity, length at maturation, maturation age, identified a trait gradient that generally con- fecundity, egg size, and parental care (Table trasted Australian fishes occurring mostly in 1). All traits are continuous and were log(x + the right-hand side of the ordination from the 1)-transformed prior to the analysis to account U.S. fishes that occupied the entire trait space for heteroscedasticity. The resulting ordination (Figure 3). This axis describes a morphologi- summarizes life history similarities among fish cal and life history gradient, with large-bod- species and enabled the identification of a re- ied, long-lived, late maturing, highly fecund duced set of axes that best separated species species with low parental care at high values along the opportunistic-periodic-equilibrium and smaller-bodied, short-lived, early matur- gradient. ing, low fecund species with high parental 94 olden and kennard

Figure 3. Two-dimensional ordination plot resulting from the principal coordinate analysis on the 14 biological traits for the fish species pools of the United States and Australia. (A) Fish species biplot where solid symbols represent the United States and open symbols represent Australia. (B) Eigenvector plot of the traits with the highest combined loadings (>0.50) on the first two principal components. Full descriptions of trait abbreviations are in Table 1. care at low values. Axis I also separated spe- States and Australia are located in positive cies with generalist feeding behaviors (om- axis I space, thus exhibiting trait similarities nivore, insectivore-piscivore) and low swim despite lacking a common ancestry (Figure factors (i.e., stronger swimming ability) in 3). Other important discriminators of trait positive ordination space from specialist in- variation among species included vertical po- vertivores with high swim factors (i.e., weaker sition (benthic versus nonbenthic), spawning swimming ability) in negative ordination frequency (single versus multiple spawning space (Figure 3). While notable differences events per season), and substrate preference in continental species pools do exist, a large (coarse versus fine grain). number of species from both the United Univariate trait comparisons support intercontinental comparison of fish life histories 95

Figure 4. Univariate comparisons of the fish species pools in the United States (black bars) and Australia (gray bars) according to (A) maximum body length, (B) swim factor, (C) total fecundity, (D) relative maturation (age of maturity/longevity), (E) reproductive guild, (F) parental care, and (G) trophic guild. continental patterns revealed in the multi- tively older age (calculated as age of maturity variate analyses. The United States contained divided by longevity) compared to the Aus- a greater proportion of fish species that were tralian fish species pool (Figure 4D). Aus- smaller-bodied (Figure 4A), exhibit lower tralian fishes showed a slight trend toward fecundity (Figure 4C), and mature at a rela- exhibiting lower swim factors compared to 96 olden and kennard fishes of the United States (Figure 4B). The United States supporting a relatively greater majority of United States fishes (>70%) - ei number of invertivores (Figure 4G). ther hide or guard their eggs during reproduction, whereas more than 65% of Australian Life History Strategies of U.S. and species spawn on open substrate and exhibit Australian fishes no egg guarding (Figure 4E), a pattern also reflected in the relatively greater amount of Australian and United States freshwater fishes parental care afforded by fish species of the conformed to the life history continuum mod- United States compared to Australia (Figure el defined by the demographic parameters of 4F). Only small differences were observed survival, fecundity, and onset and duration of for trophic guilds, with Australia containing reproduction (Figure 5A). We found strong relatively more piscivorous species and the evidence for the triangular adaptive surface

Figure 5. Life history diversity of freshwater fishes from southern United States (solid triangles) and eastern Australia (open triangles). (A) Species are located on a two-dimensional triangular surface, embedded in the three-dimensional space established by fecundity (ln scale), length at maturity (ln scale) and juvenile survivorship (equal to ln (egg diameter + 1) + ln (parental care + 1)). The vertices of this triangular surface define three endpoint strategies: opportunistic, periodic, and equilibrium; and intermediate strategies are recognized within the gradient of life histories. (B) and (C) illustrate a three-dimensional linear plane fitted to all species (indicated by circles). intercontinental comparison of fish life histories 97 anchored by the opportunistic, periodic, and (dewfish Tandanus tandanus, Family: Plotosi- equilibrium life history strategies (Figure 5B, dae; Arius graeffei, Family: Ariidae). Despite 5C), where a two-dimensional linear plane these examples, the large majority of species showed a good fit to fish species of Australia exhibited immediate positions in life history (Pearson’s R = 0.65, P < 0.001, n = 66), the space. United States (R = 0.76, P < 0.001, n = 257), Multivariate ordination of fish species ac- and both countries pooled (R = 0.71, P < 0.001, cording to their life history attributes identified n = 323). For the United States, a conspicuous two major gradients of trait variation repre- cluster of species were located in close proxim- sented by the first two PCA axes that explained ity to the opportunistic endpoint, although all 81.5% of the variance in the original seven life endpoint and intermediate strategies were well history characteristics (Figure 6). The first represented. By contrast, the majority of Aus- principal component revealed a dichotomy tralian fishes occupied positions in life history between periodic strategists (i.e., relatively space along a linear axis connecting the oppor- larger species with a long life span, late [and tunistic and periodic endpoints (i.e., fewer spe- old] maturation, and high fecundity) occur- cies with equilibrium characteristics). ring on the positive scale, from opportunistic Species from both the United States and strategists (i.e., relatively smaller species with a Australia occupied the extreme opportunistic, short life span, early [and young] maturation, periodic, and equilibrium endpoints, although and low fecundity) occurring on the negative the majority of species take up intermediate scale. The second principal component identi- positions in life history space. From the United fied a dominant gradient of life history traits States, examples of more opportunistic strate- that contrasts species exhibiting high paren- gies include select species of shiners (Cyprinel- tal care, large egg size, and low fecundity with la spp. and Notropis spp., Family: Cyprinidae) positive scores from species showing minimal and topminnows (Fundulus spp., Family: Fun- parent care, small egg size, and relatively higher dulidae); periodic strategists include suckers fecundity with negative scores. This represents (Ictiobus spp. and Catostomus spp., Family: Ca- a dichotomy between species considered equi- tostomidae) and pickerel (Esox spp., Family: librium strategists versus periodic and oppor- Esocidae); and equilibrium strategists include tunistic strategists (Figure 6). black basses (Micropterus spp., Family: Cen- trarchidae) and catfishes Ameiurus ( spp. and Basin Life History Composition along a Noturus spp., Family: Ictaluridae). From Aus- Hydrologic and Productivity Gradient tralia, examples of opportunistic strategists include select species of gudgeons (Hypseleotris Drainage basins of the United States and Aus- spp., Family: Eleotridae) and rainbowfishes tralia showed similarity in their composition (Melanotaenia spp., Family: Melanotaeniidae); of opportunistic life history strategies, on av- periodic strategists include basses (Macquaria erage representing 50–60% of a basin’s fish as- spp., Family: Percichthyidae) and grunters semblage (Figure 7). By contrast, equilibrium (leathery grunter Scortum hillii and barred strategists were relatively more prevalent in the grunter Amniataba percoides, Family: Tera- United States compared to Australia, whereas pontidae); and equilibrium strategists include periodic strategists were more common in lungfish (Australian lungfish Neoceratodus for- Australia compared to the United States. Both steri, Family: Ceratodontidae) and catfishes approaches to quantifying life history compo- 98 olden and kennard

Figure 6. Two-dimensional ordination plot resulting from the principal component analysis on the seven life history traits for the fish species pools of the United States and Australia. Species are identified by solid symbols (United States) and open symbols (Australia). Traits with the highest combined loadings (>0.50) on the first two principal components are illustrated. Full descriptions of trait abbreviations are in Table 1.

sition of the drainage basins produced similar periodic strategists. We found no statistical results. The relationship between proportional difference in the slope of these associations strategy richness and distance-weighted pro- between the two continents. In contrast to portional strategy richness was significantly life history theory, increasing hydrologic vari- positive for periodic (Pearson’s R = 0.97, P < ability in Australia (but not the United States) 0.001), opportunistic (R = 0.76, P < 0.001), was related to greater proportions of equilib- and equilibrium strategies (R = 0.93, P < rium strategists. Net primary productivity was 0.001). Given the correlation between these positively concordant with the proportion of estimates of life history composition (and sim- periodic strategists in Australian basins and ilar results from multiple regression analyses), with the proportion of opportunistic strate- we report on the results based on the distance- gists in the United States and was negatively weighted proportional strategy richness. concordant with the equilibrium strategy for Results from the multiple regression analy- both countries, supporting predictions from sis provided strong support for the influence of life history theory. One exception, however, hydrologic variability on fish life history trait was the significant negative relationship be- composition in the United States and Australia tween NPP and the proportion of species con- (Table 2). In agreement with predictions from sidered opportunistic in Australia. In contrast life history theory, hydrologic variability for to flow variability, the slopes of this relation- both countries was positively (and significant- ship differed for opportunistic and periodic ly) associated with the prevalence of opportu- strategies but were not significant different for nistic strategists and negatively associated with equilibrium strategies (Table 2). These results intercontinental comparison of fish life histories 99

Figure 7. Ternary plot illustrating the relative proportion of periodic, opportunistic and equilibrium strategists in drainage basins of the United States (n = 56, solid symbol) and Australia (n = 56, open symbol). show that the association between hydrologic species pools. This is particularly relevant for variability and life history of fishes was similar our study of the fish faunas of southern United in the United States and Australia (two out of States and eastern Australia, which only have three strategies based on directionality and sig- a single species/genus/family in common, nificance) but weaker concordance with pro- flathead mullet Mugil cephalus. Despite widely ductivity (one out of three strategies). different biogeographic histories, our study found that patterns of life history composition Discussion of river basins responded similarly along a gradient of hydrologic variability. Riverine biodiversity is generated and main- In support of previous research for fresh- tained by geographic variation in stream pro- water fishes in a diversity of environments cesses and fluvial disturbance regimes, which (Winemiller 1989; Winemiller and Rose 1992; largely reflects regional differences in climate, Vila-Gispert and Moreno-Amich 2002; Vila- geology, and topography. When investigating Gispert et al. 2002; Olden et al. 2006, Tedesco fish community–environment relationships et al. 2008), the fishes of the United States and at large spatial scales, a functional perspective Australia examined in our study conform to using biological traits allows the comparison the three-dimensional adaptive space arising of species compositions that naturally differ from the interrelationships among three basic due to biogeographic constraints on regional demographic parameters of survival, fecundi- 100 olden and kennard

* * * * * * * * Comparison of estimates P 0.013 0.476 0.001 0.001 0.002 0.043 0.048 0.042 0.006 <0.001 <0.001 <0.001 = 5.57, P < 0.001 = 5.57, = 7.10, P < 0.001 = 7.10, = 11.66, P < 0.001 = 11.66, 4,51 4,51 4,51 4,51 Australia 4,51 4,51

0.115 0.391 0.720 0.585 1.034 0.498 0.642 –0.418 –1.134 –0.361 –0.605 –0.443 Estimate = 0.358, F = 0.358, = 0.303, F = 0.303, = 0.478, F = 0.478, 2 2 2 R R R P < 0.05, thus providing support for community convergence in P 0.181 0.010 0.174 0.011 0.105 0.524 0.005 0.449 0.695 0.444 <0.001 <0.001 P < 0.001 = 6.74, P = 0.010 = 3.70, P < 0.001 = 6.34, 4,51 4,51 4,51 4,51 United States

0.579 1.137 0.091 0.637 0.589 –0.077 –0.257 –1.140 –1.200 –0.300 –0.631 –0.051 Estimate = 0.346, F = 0.346, = 0.225, F = 0.225, = 0.332, F = 0.332, 2 2 2 R R R

– – – + + + Prediction ) ) ) ) ) ) 2 2 2 2 2 2 Flow variability NPP (metric tons of carbon) Model Model Runoff (ML/km Flow variability NPP (metric tons of carbon) Flow variability NPP (metric tons of carbon) Model: Equilibrium Strategy Basin area (km Model: Periodic Strategy Model: Periodic Basin area (km Table 2. Results from the multiple regression models relating proportional life-history strategy richness to basin area (log-transformed), annual runoff annual (log-transformed), area basin to richness strategy life-history proportional relating models regression multiple the from Results 2. Table predictions theory Life-history log-transformed). productivity primary (NPP, net annual and variation) of (coefficient variability flow (log-transformed), difference significant no indicates * 1. Figure and 2005) (1995, Winemiller follow and presented are strategies endpoint with associated factors the of of regression estimates (slopes) between the United States and Australia based on Model: Opportunistic Strategy life-history composition. Runoff (ML/km Runoff (ML/km Model Source of variation Basin area (km intercontinental comparison of fish life histories 101 ty, and onset and duration of reproductive life. headwater streams are very similar in North Our analyses revealed that species from both America and Europe, despite their indepen- countries represent the endpoints defining the dent origins. Through a series of exploratory periodic, opportunistic, and equilibrium life analyses, the authors proposed that fish faunas history strategies, as well as occupying inter- in Europe and western North America shared mediate positions in life history space. A large greater trait similarity compared to eastern and majority of American fishes were characterized western fishes of North America. Research by by the suite of opportunistic attributes, includ- Winemiller and colleagues have showed that ing small body size, short-lived, and low batch the freshwater fishes of South America and fecundity with multiple reproductive batches North America both encompass the trilateral per breeding season. By contrast, the majority of life history space defined by the opportunistic- Australian fishes were located along a life history periodic-equilibrium trichotomy (Winemiller axis connecting the opportunistic endpoint to 1989; Winemiller and Rose 1992). This asso- the midpoint of the edge attaching the periodic ciation was further explored by Vila-Gispert and equilibrium endpoints. This axis defines a et al. (2002) who found consistency in basic gradient of evolutionary “bet-hedging” (sensu life history patterns among 301 fish species Cole 1954) considered adaptive in relatively from different regions of Europe, North Amer- unpredictable environments where conditions ica, and South America. Species from South are occasionally so bad that recruitment fails America showed a greater tendency toward the entirely (Stearns 1992). The bet-hedging suite opportunistic suite of traits, whereas North of traits is also associated with relatively greater American and European fishes occupied im- mobility to promote rapid recolonization after mediate positions along an axis linking the op- disturbance, a pattern supported by our -find portunistic and periodic strategies. ing that Australian fishes are characterized by Drawing definitive conclusions from smaller swim factors compared to fishes of the this study is complicated by the fact that fish United States, indicating a stronger swimming species data were collected from entire con- ability (Webb 1984). Last, we found that ap- tinents, thus ignoring spatial variation in en- proximately three-quarters of fishes in the Unit- vironmental conditions that shape species ed States exhibited relatively more parental care traits and assemblage functional composi- by hiding or guarding their eggs during repro- tion. The present study tackles this problem duction, compared to Australia where approxi- by comparing fish faunas from regions expe- mately two-thirds of the species spawn on open riencing similar long-term climatic regimes. substrate and exhibit no egg guarding. Based on a hard classification of species, we Our study, together with other large- found that the southern United States and scale scale comparisons of freshwater fish eastern Australia contained similar fractions traits, provides strong evidence that essential of equilibrium species (12% versus 15%, re- trade-offs produce intercontinental similari- spectively), whereas the United States con- ties in life history strategies despite divergent tained a greater proportion of opportunistic phylogenies. For example, in the edited book species (69% versus 52%) but less periodic Community and Evolutionary Ecology of North species (19% versus 33%). However, it is im- American Stream Fishes, Moyle and Herbold portant to note that the majority of species in (1987) reported that the body size and life his- both continents exhibit immediate strategies tory characteristics of fish communities in cold between these extremes. 102 olden and kennard We found that hydrologic variability and and nutrients for future bouts of reproduc- regional climate were related to patterns of fish tion (Winemiller and Rose 1992). Because life history variation in the United States and freshwater environments exhibit either spatial Australia. These relationships were generally or temporal variability that is to some degree the same between countries and consistent predictable, the periodic production schedule with predictions from life history theory for allows a fitness payoff when reproduction co- hydrologic variability but were much more incides with favorable environmental condi- variable for NPP. Hydrologic variability (both tions. Given that drought and flood conditions countries) and NPP (United States only) were represent significant agents of physical change positively related to basins with higher preva- in freshwater ecosystems (Lake 2008), small, lence of opportunistic strategists, a life history but frequent, extreme flow events that induce that should maximize fitness in environmental mortality of early life stages of periodic-type settings dominated by unpredictable environ- species may significantly influence population mental change that cause frequent density- abundance and species distributions. independent mortality, followed by rapid In contrast to life history theory, we found colonization (Winemiller 1995, 2005). Many no support for the predicted negative relation- opportunistic species in southern United States ship between hydrologic variability and the and eastern Australia are associated with shal- distribution of equilibrium strategists in Aus- low riffle and marginal habitats, where changes tralia (a negative but nonsignificant relation- in hydrologic regimes, including floods and ship was observed for the United States). The droughts, induce major alterations in water equilibrium strategy optimizes juvenile survi- depth, substrate characteristics, and habitat vorship by appointing a greater amount of ma- availability (Hoeinghaus et al. 2007; Kennard terial into each individual egg and/or the pro- et al. 2007). In the absence of intense preda- vision of parental care (Winemiller and Rose tion and resource limitation (likely to exist in 1992). However, in support of theory—which warmer, productive regions) that promote low predicts that the equilibrium configuration of juvenile density dependence, opportunistic- life history traits should be favored in high ju- type species should be favored due to their venile density-dependent and resource-limited high turnover (short juvenile phase of the life environments (McCann 1998)—we observed cycle) and high annual fecundity (multiple a significant negative relationship between spawning events) (McCann 1998). net primary productivity and the prevalence Our findings also provide strong support of equilibrium species in basins of both the for the negative influence of hydrologic vari- United States and Australia. This indicates that ability on the prevalence of periodic strategists freshwater ecosystems located in less produc- in basins of both the United States and Aus- tive watersheds are characterized by relatively tralia. The periodic strategy maximizes fecun- more equilibrium-type species. It is important dity by delaying maturity (to achieve sufficient to note that in contrast to hydrologic variabil- size and to acquire resources) and producing ity, there are strong predictions of how equilib- many smaller eggs as a tactic to increase ju- rium-periodic life history strategies respond to venile survivorship. According to life history patterns of hydrologic seasonality (i.e., predict- theory, large body size of periodic strategists ability). For example, in support of theory, Te- enhances adult survivorship during subopti- desco et al. (2008) found a higher proportion mal conditions and permits storage of energy of periodic strategists in highly seasonal basins intercontinental comparison of fish life histories 103 of West Africa that exhibit short and predict- ronment, yet the actual traits possessed by spe- able hydrologic season. The quantification of cies may differ because they originate from dif- flow regime predictability in our study regions ferent systematic groups (Verberk et al. 2008). is presently not feasible, but will be the subject Consequently, other aspects of functional and of future investigation. life history variation should vary across domi- We found moderate evidence that life his- nant environmental gradients in freshwater tory traits of freshwater fish communities in ecosystems. Previous studies have shown that southern United States and eastern Austra- the stability of flow regimes is directly related lia have converged across similar gradients of to taxonomic and trophic diversity (Horwitz hydrologic variability (but not productivity) 1978), whereby assemblages characterized by despite phylogenetic and historical differences generalists in hydrologically variable sites and between continents. Increasing hydrologic more specialist species in stable sites (Poff and variability has promoted the greater prevalence Allan 1995). Hoeinghaus et al. (2007) also re- of opportunistic strategists (and their associ- ported the importance of hydrologic stability ated biological traits) while concurrently mini- in shaping the functional attributes of fish fau- mizing the persistence of periodic-type spe- nas in rivers of Texas (United States). cies. Notably, no such relationship was found Fish life history strategies represent dif- for equilibrium strategists. Our findings at the ferent solutions to particular ecological prob- scale of the drainage basin correspond to the lems, thus providing a connection between results of Lamouroux et al. (2002) at the local species traits and environmental gradients in reach scale. Lamouroux et al. (2002) found freshwater ecosystems. Despite the powerful that, within continents, local hydraulic and geo- role that species traits have to play in ecologi- morphic variables were highly correlated with cal studies, it is important to recognize a num- fish trait proportions in streams of France and ber of limitations that are common to previous the United States (Virginia). Taken together, studies, including our own. First, there exists convergence in the life history characteristics a substantial temporal mismatch between the of fish assemblages across biogeographically time period over which environmental vari- distinct regions of the United States and Aus- ables were quantified (i.e., past century) and tralia may imply the existence of key, repeated the evolutionary time period over which fish evolutionary mechanisms responsible for the species have evolved and responded to envi- observed flow:trait relationships. ronmental change (i.e., millennia). For this rea- Despite the role of hydrologic variability son, it is perhaps not surprising that previous and productivity in shaping life history diversi- studies have had difficulty linking fish traits to ty in Australia and the United States, divergent descriptors of the physical environment. Sec- life history strategies are still expected in loca- ond, we acknowledge that intraspecific variations experiencing the same long-term environ- tions in life history traits do exist at large-spa- mental regimes. This occurs because species tial scales and that assigning a single trait value perceive the same environment very differently to a species may not appropriately represent from another and thus use their environment this variation. However, Blanck and Lamour- at a range of different spatial and temporal oux (2007) found that species fecundity and scales (Winemiller 2005). Moreover, species life history traits related to body length varied “assigned” to the same life history strategy may more among species than between populations have similar functional relations to their envi- of the same species for European freshwater 104 olden and kennard fishes, suggesting that the relative position of communities away from opportunistic strate- species in opportunistic-periodic-equilibrium gies and toward more periodic-equilibrium life history space is unlikely to be significantly forms (Olden et al. 2006). affected by intraspecific trait variation. Third, data availability required us to quantify inter- Acknowledgments annual hydrologic variability (i.e., variation in mean annual discharge) rather than intra- We gratefully acknowledge Meryl Mims and annual seasonal variability, which we might ex- two anonymous reviewers for helpful com- pect to play a larger role in shaping at life histo- ments, Zachary Shattuck for helping populate ry composition of fish communities (Tedesco the fish trait database, Janet Stein for provision et al. 2008). Fourth, our grain of analysis was of environmental data, and Tarmo Raadik and the river basin, which clearly encompasses a Tom Raynor for providing unpublished fish great deal of environmental heterogeneity. Fu- distributional data. Funding support was pro- ture investigations that link river hydrology to vided by the USGS Lower Colorado River Ba- reach-scale fish assemblages are likely to yield sin Aquatic GAP Program (JDO) and a post- greater insight into inter-continental commu- doctoral fellowship from the Australian Rivers nity convergence in life history composition. Institute, Griffith University (MJK). JDO and In conclusion, balancing human and eco- MJK conceived and developed the idea for the system needs for freshwater will require that manuscript and assembled the data sets, JDO our view of the ecology of rivers and the im- conducted the data analysis, and JDO and pacts associated with water use and infrastruc- MJK wrote the manuscript. tural development be placed in the context of the natural flow regime and deviations from the References natural condition. Our study provides a con- Balon, E. K. 1975. Reproductive guilds of fishes: a ceptual framework of management options for proposal and definition. Journal of the Fisheries species in regulated rivers because life history Research Board of Canada 32:821–864. Blanck, A., and N. Lamouroux. 2007. Large-scale in- strategies are the underlying determinants for traspecific variation in life-history traits of Eu- population responses to environmental change ropean freshwater fish. Journal of Biogeography (King and McFarlane 2003; Winemiller 2005) 34:862–875. and therefore can be used to classify typical Bunn, S. E., and A. H. Arthington. 2002. Basic prin- population responses to flow alteration or mit- ciples and ecological consequences of altered flow regimes for aquatic biodiversity. Environ- igation via environmental flow prescriptions. mental Management 30:492–507. This can provide vital management advice for Cole, L. 1954. The population consequences of life defining flow targets and predicting fish spe- history phenomena. Quarterly Review in Biol- cies responses to flow restoration. Given that ogy 29:103–137. hydrologic variability (this study) and predict- Dewson, Z. S., A. B. W. James, and R. G. Death. 2007. A review of the consequences of decreased flow ability (Tedesco et al. 2008) act as landscape for instream habitat and macroinvertebrates. filters for fish life history traits, we expect that Journal of the North American Benthological dam management practices will have persis- Society 26:401–415. tent effects on fish faunas at local and regional Diniz-Filho, J. A. F., C. E. R. De Sant’ana, and L. M. Bini. 1998. An eigenvector method for estimating scales. In the long term, dam-induced dampen- phylogenetic inertia. Evolution 52:1247–1262. ing of flow variability resulting from operations Felsenstein, J. 1985. Phylogenies and the compara- for flood control and water supply may shift tive method. American Naturalist 125:1–15. intercontinental comparison of fish life histories 105 Fisher, D. O., and I. P. F. Owens. 2004. The compara- Krug, W. R., W. A. Gebert, and D. J. Graczyk. 1987. tive method in conservation biology. Trends in Preparation of average annual runoff map of the Ecology and Evolution 19:391–398. United States, 1951–80. U.S. Geological Survey Frimpong, E. A., and P. L. Angermeier. 2010. Trait- Open-File Report 87–535, 414. based approaches in the analysis of stream fish Lake, P. S. 2008. Flow-generated disturbances and communities. Pages 109–136 in K. B. Gido and ecological responses: floods and droughts. Pages D. A. Jackson, editors. Community ecology of 75–92 in P. J. Wood, D. M. Hannah, and J. P. Sa- stream fishes: concepts, approaches, and tech- dler, editors. Hydroecology and ecohydrology: niques. American Fisheries Society, Symposium past, present and future. Wiley Press, New York. 73, Bethesda, Maryland. Lambeets, K., M. L. Vandegehuchte, J. Maelfait, and Guégan, J.-F., S. Lek, T. Oberdorff. 1998. Energy D. Bonte. 2008. Understanding the impact of availability and habitat heterogeneity predict flooding on trait-displacements and shifts in global riverine fish diversity. Nature (London) assemblage structure of predatory arthropods 391:382–384. on river banks. Journal of Animal Ecology Hoeinghaus, D. J., K. O. Winemiller, and J. S. Birn- 10:1365–2656. baum. 2007. Local vs. regional influences on the Lamouroux, N., N. L. Poff, and P. L. Angermeier. structure of fish assemblages in Texas streams. 2002. Intercontinental convergence of stream Journal of Biogeography 34:324–338. fish community traits along geomorphic and Horwitz, R. J. 1978. Temporal variability patterns hydraulic gradients. Ecology 83:1792–1807. and the distributional patterns of stream fishes. Legendre, P., and L. Legendre. 1998. Numerical ecol- Ecological Monographs 48:307–321. ogy, 2nd edition. Elsevier Scientific, Amsterdam. Hutchinson, M. F., H. A. Nix, and C. McTaggart. Lytle, D. A., and N. L. Poff. 2004. Adaptation to natu- 2004. GROWEST version 2.0. Australian Na- ral flow regimes. Trends in Ecology and Evolu- tional University, Centre for Resource and En- tion 19:94–100. vironmental Studies, Canberra McCann, K. 1998. Density-dependent coexistence in Imhoff, M. L., L. Bounoua, T. Ricketts, C. Loucks, R. fish communities. Ecology 79:2957–2967. Harriss, and W. T. Lawrence. 2004. Global pat- Moyle, P. B., and B. Herbold. 1987. Life-history patterns in net primary productivity. Data distrib- terns and community structure in stream fishes uted by the Socioeconomic Data and Applica- of western North America: comparisons with tions Center (SEDAC). Available: http://sedac. eastern North America and Europe. Pages 25– ciesin.columbia.edu/es/hanpp.html (September 32 in W. J. Matthews, and D. C. Heins, editors. 2008). Community and evolutionary ecology of North Imhoff, M. L., and L. Bounoua. 2006. Exploring global American stream fishes. University of Oklaho- patterns of net primary production carbon sup- ma Press, Norman. ply and demand using satellite observations and Naiman, R. J., and H. Décamps. 1997. The ecology statistical data. Journal of Geophysical Research of interfaces: riparian zones. Annual Review of 111: D22S12, doi:10.1029/2006JD007377. Ecology and Systematics 28:621–658. Kennard, M. J., J. D. Olden, A. H. Arthington, B. J. Naiman, R. J., J. J. Latterell, N. E. Pettit, and J. D. Pusey, and N. L. Poff. 2007. Multi-scale effects Olden. 2008. Flow variability and the biophysi- of flow regime and habitat and their interaction cal vitality of river systems. Comptes Rendus on fish assemblage structure in eastern Austra- Geosciences 340:629–643. lia. Canadian Journal of Fisheries and Aquatic NatureServe. 2004. Downloadable animal datasets. Sciences 64:1346–1359. NatureServe Central Databases. Available: Kennard, M. J., B. J. Pusey, J. D. Olden, S. J. Mackay, J. www.natureserve.org/getData/dataSets/water- L. Stein, and N. Marsh. In press. Classification of shedHucs/index.jsp (August 2008). natural flow regimes in Australia to support en- Nilsson, C., E. Nilsson, M. E. Johansson, M. Dyne- vironmental flow management. Freshwater Bi- sius, G. Grelsson, S. Xiong, R. Jansson, and M. ology, doi: 10.1111/j.1365-2427.2009.02307.x. Danvind. 1993. Processes structuring ripar- King, J. R., and G. A. McFarlane. 2003. Marine fish ian vegetation. Current Topics in Botanical Re- life history strategies: applications to fishery search 1:419–431. management. Fisheries Management and Ecol- Oberdorff, T., B. Hugueny, and J. Guégan. 1997. Is ogy 10:249–264. there an influence of historical events on con- 106 olden and kennard temporary fish species richness in rivers? Com- ecological strategies? Journal of Animal Ecol- parisons between Western Europe and North ogy 46:337–365. America. Journal of Biogeography 24:461–467. Stearns, S. C. 1992. The evolution of life histories. Olden, J. D., and N. L. Poff. 2003. Redundancy and Oxford University Press, Oxford, UK. the choice of hydrologic indices for character- Stein J. L., M. F. Hutchinson, B. J. Pusey and M. J. Ken- izing streamflow regimes. River Research and nard. 2008. Ecohydrological classification based Applications 19:101–121. on landscape and climate data. Appendix 8 in B. Olden, J. D., N. L. Poff, and K. R. Bestgen. 2006. Life- J. Pusey, M. J. Kennard, J. L. Stein, J. D. Olden, history strategies predict fish invasions and ex- S. J. Mackay, M. F Hutchinson, and F. Sheldon, tirpations in the Colorado River basin. Ecologi- editors. Ecohydrological regionalisation of Aus- cal Monographs 76:25–40. tralia: a tool for management and science. In- Peel, M. C., B. L. Finlayson, and T. A. McMahon. novations Project GRU36, Final Report to Land 2007. Updated world map of the Köppen-Gei- and Water Australia. Available: http://lwa.gov. ger climate classification. Hydrology and Earth au/products/pn22591 (September 2008). System Sciences 11:1633–1644. Tedesco, P., and B. Hugueny. 2006. Life history strat- Pettit, N. E., R. H. Froend, and P. M. Davies. 2001. egies affect climate based spatial synchrony in Identifying the natural flow regime and the re- population dynamics of West African freshwa- lationship with riparian vegetation for two con- ter fishes. Oikos 115:117–127. trasting western Australian rivers. Regulated Tedesco, P., B. Hugueny, T. Oberdorff, H. H. Durr, Rivers Research and Management 17:201–215. S. Merigoux, and B. de Merona. 2008. River Poff, N. L., and J. V. Ward. 1989. Implications of hydrological seasonality influences life history streamflow variability and predictability for lot- strategies of tropical riverine fishes. Oecologia ic community structure: a regional analysis of 156:691–702. streamflow patterns. Canadian Journal of Fish- Townsend, C. R., and A. G. Hildrew. 1994. Species eries and Aquatic Sciences 46:1805–1818. traits in relation to a habitat templet for river Poff, N. L., and J. V. Ward. 1990. The physical habitat systems. Freshwater Biology 31:265–275. template of lotic systems: recovery in the context Verberk, W. C. E. P., H. Siepel, and H. Esselink. 2008. of historical pattern of spatio-temporal heteroge- Life-history strategies in freshwater macroin- neity. Environmental Management 14:629–646. vertebrates. Freshwater Biology 53:1722–1738. Poff, N. L., and J. D. Allan. 1995. Functional organi- Vieira, N. K. M., W. H. Clements, L. S. Guevara, and zation of stream fish assemblages in relation to B. F. Jacobs. 2004. Resistance and resilience of hydrological variability. Ecology 76:606–627. stream insect communities to repeated hydro- Poff, N. L., D. Allan, M. B. Bain, J. R. Karr, K. L. Pre- logic disturbances after a wildfire. Freshwater stegaard, B. D. Richter, R. E. Sparks, and J. C. Biology 49:1243–1259. Stromberg. 1997. The natural flow regime: a Vila-Gispert, A., and R. Moreno-Amich. 2002. Life- paradigm for river conservation and restora- history patterns of 25 species from European tion. Bioscience 47:769–784. freshwater fish communities. Environmental Bi- Poff, N. L., J. D. Olden, N. K. M. Vieira, D. S. Finn, ology of Fishes 65:387–400. M. P. Simmons, and B. C. Kondratieff. 2006. Vila-Gispert, A., R. Moreno-Amich, and E. Garcia- Functional trait niches of North American lotic Berthou. 2002. Gradients of life-history variation: insects: traits-based ecological applications in an intercontinental comparison of fishes. Reviews light of phylogenetic relationships. Journal of the in Fish Biology and Fisheries 12:417–427. North American Benthological Society 25:730– Warner, R., and P. L. Chesson. 1985. Coexistence 755. mediated by recruitment fluctuations: a field Pusey, B. J., M. J. Kennard, and A. H. Arthingon. guide to the storage effect. American Naturalist 2004. Freshwater fishes of north-eastern Aus- 125:769–787. tralia. CSIRO Publishing, Collingwood, Victo- Webb, P. W. 1984. Form and function in fish swim- ria, Australia. ming. Scientific American 251:58–68. Schluter, D. 1986. Tests for similarity and conver- Wenninger, E. J., and W. F. Fagan. 2000. Effect of gence of finch communities. Ecology 67:1073– river flow manipulation on wolf spider assem- 1085. blages at three desert riparian sites. Journal of Southwood, T. R. E. 1977. Habitat, the templet for Arachnology 28:115–122. intercontinental comparison of fish life histories 107 Winemiller, K. O. 1989. Patterns of variation in life Winemiller, K. O. 2005. Life history strategies, popu- history among South American fishes in sea- lation regulation, and implications for fisheries sonal environments. Oecologia 81:225–241. management. Canadian Journal of Fishery and Winemiller, K. O. 1992. Life-history strategies and Aquatic Sciences 62:872–885. the effectiveness of sexual selection. Oikos Winemiller, K. O., and K. A. Rose. 1992. Patterns of 62:318–327. life-history diversification in North American Winemiller, K. O. 1995. The structural and functional fishes: implications for population regulation. aspects of fish diversity. Bulletin Francais de la Canadian Journal of Fisheries and Aquatic Sci- Peche et de la Pisciculture 337/338/339:23–45. ences 49:2196–2218.