Evolution, 58(6), 2004, pp. 1209±1224

THE EVOLUTION OF LARVAL MORPHOLOGY AND SWIMMING PERFORMANCE IN ASCIDIANS

MATTHEW J. MCHENRY1,2 AND SHEILA N. PATEK3,4 1The Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts 02138 2E-mail: [email protected] 3The Department of Integrative Biology, University of California, Berkeley, California 94720 4E-mail: [email protected]

Abstract. The complexity of organismal function challenges our ability to understand the evolution of lo- comotion. To meet this challenge, we used a combination of biomechanics, phylogenetic comparative analyses, and theoretical morphology to examine evolutionary changes in body shape and how those changes affected swimming performance in ascidian larvae. Results of phylogenetic comparative analyses suggest that coloniality evolved at least three times among ascidians and that colonial have a convergent larval morphology characterized by a large trunk volume and shorter tail length in proportion to the trunk. To explore the functional signi®cance of this evolutionary change, we ®rst veri®ed the accuracy of a mathematical model of swimming biomechanics in a solitary (C. intestinalis) and a colonial (D. occidentalis) species and then ran numerous simulations of the model that varied in tail length and trunk volume. The results of these simulations were used to construct landscapes of speed and cost of transport predictions within a trunk volume/tail length morphospace. Our results suggest that the reduction of proportionate tail length in colonial species resulted in improved energetic economy of swimming. The increase in the size of larvae with the origin of coloniality facilitated faster swimming with negligible energetic cost, but may have required a reduction in adult fecundity. Therefore, the evolution of ascidians appears to be in¯uenced by a trade-off between the fecundity of the adult stage and the swimming performance of larvae.

Key words. Ciona intestinalis, Distaplia occidentalis, kinematics, larvae, morphology, morphospace, urochordata.

Received September 11, 2003. Accepted March 15, 2004.

The biomechanical complexity of animal motion presents Bennett 1986; Jayne and Bennett 1989) may be confounded challenges for understanding broad patterns of locomotor by variation in performance caused by traits other than those evolution. Measures of locomotor performance typically have examined (Koehl 1996). Many mechanistic investigations a nonlinear dependency on numerous aspects of the mor- have attempted to resolve the causal relationships between phology and motion of an animal's body (McMahon 1984; traits and performance by developing mathematical models Alexander 2003; Biewener 2003) and interspeci®c variation of locomotion (e.g., Daniel 1983; Liu et al. 1996; Sane and in these traits may be substantial. Ascidians (Chordata: Uro- Dickinson 2002; McHenry et al. 2003). However, such idi- chordata) present an interesting case study of locomotor evo- ographic studies have limited applicability to evolutionary lution because the larvae of colonial species are similar in questions because they generally neglect the effects of intra- size and shape despite having evolved independently at least and interspeci®c variation in favor of understanding general three times among urochordates (Swalla et al. 2000). Through functional principles. Mechanistic studies that examine the the integration of biomechanics, phylogenetic comparative extreme differences among species within a group (e.g., analyses, and theoretical morphology, we examined how evo- Kingsolver and Koehl 1985; Crompton 1989; Emerson and lutionarily convergent colonial life histories have in¯uenced Koehl 1990; Drucker and Lauder 2001) have proven valuable the morphology and swimming performance of ascidian lar- for exploring the extremes of performance exhibited by a vae. group of species. However, general principles and knowledge of performance extremes alone are limited in their ability to Biomechanical, Comparative, and Theoretical Approaches inform our understanding of changes in function that result to the Evolution of Organismal Function from a sequence of historical transformations. Research on the evolution of organismal function may use Theoretical morphology provides the tools to examine his- extant species by either testing for correlations between traits torical transformations with mechanistic models toward the and performance or by investigating functional mechanisms ultimate goal of testing evolutionary hypotheses. This ana- (reviews include Wake and Roth 1989; Bennett and Huey lytical technique involves constructing a theoretical spectrum 1990; Garland and Carter 1994; Thomason 1995; Koehl 1996; of morphological parameters (a morphospace), the measure- Lauder 2003). A correlative approach explores natural co- ment of species distributions within that space, and using a variation between traits and performance (e.g., Arnold and mathematical model to determine the functional signi®cance Bennett 1988; Jayne and Bennett 1990; Losos 1990; Patek of morphologies that are both represented and absent among and Oakley 2003) to formulate mechanistic hypotheses (e.g., species (Raup and Michelson 1965; Raup 1967; McGhee Bennett et al. 1989; Jayne and Bennett 1989; Friedman et al. 1999). This model may be used to generate predictions of 1992) and to consider the in¯uence of shared ancestry on performance for each position in the morphospace and there- functional relationships (e.g., Losos 1990; Bauwens et al. by generate a performance landscape (also known as a per- 1995). Form-function relationships established with a strictly formance surface: Arnold 2003; a ®tness landscape: Gilchrist correlative approach (reviewed by Arnold 1983; Huey and and Kingsolver 2003; or a functional morphospace: Moore 1209 ᭧ 2004 The Society for the Study of Evolution. All rights reserved. 1210 M. J. MCHENRY AND S. N. PATEK and Ellers 1993). Performance landscapes have been used to (van Duyl et al. 1981; Durante 1991; Stoner 1994; Svane and test adaptive hypotheses that ammonite shell geometry gen- Dolmer 1995). Therefore, speed and cost of transport are two erates high locomotor stability (Raup 1967) and strength measures of swimming performance that may have important against hydrostatic pressure (Daniel et al. 1997), to determine consequences for dispersal distance and duration, microhab- whether morphological disparity among labrid ®shes facili- itat selection, larval survivorship, and, ultimately, to ®tness. tates disparity in function (Hulsey and Wainwright 2002), to Although ascidians exhibit tremendous diversity in life- identify developmental constraints in the body shape of sea history traits (reviewed by Svane and Young 1989), solitary urchins (Ellers 1993), and to investigate the effects of en- and colonial species represent the two major types of ascidian vironmental change on the macroevolution of vascular plants life-history strategies. Colonial (i.e., compound and social) (Niklas 1997). The models used to generate performance ascidians produce adult zooids by asexual reproduction and landscapes may come in the form of simple algebraic equa- they generally brood a relatively small number of large larvae tions (e.g., Moore and Ellers 1993) or elaborate computa- that spend a brief time in the (less than a few hours; tional simulations (e.g., Daniel et al. 1997), depending on Berrill 1935). Solitary species generally broadcast spawn the complexity of the functional system. their gametes and their numerous small larvae develop rap- The present study tested the accuracy of a mathematical idly in the plankton. Therefore, solitary species may provide model of swimming in ascidian larvae (McHenry et al. 2003) less material investment and protection for larvae than co- and used this model to construct landscapes of swimming lonial species, but they have higher fecundity (Svane and performance. According to this model, the hydrodynamics of Young 1989). The larvae of colonial species are so much swimming vary widely among ascidian species, which swim larger than solitary species that their trunks may be more at Reynolds numbers (Re ϭ␳UL/␮, where U is mean swim- than three orders of magnitude greater in volume (Cloney ming speed, L is body length, ␳ and ␮ are the density and 1978). viscosity of water; Lamb 1945) from 5 in Ciona intestinalis It remains unclear to what degree patterns of life-history (Bone 1992) to 100 in Distaplia occidentalis (McHenry traits and larval morphology are due to shared ancestry or 2001). At any Re value, the magnitude and direction of pro- convergent evolution. It has long been appreciated that both pulsive forces depends on the shape of the larval body and solitary and colonial species are distributed throughout as- the undulatory motion of the tail. cidian families (Berrill 1950) and recent phylogenetic studies suggest that coloniality has a number of independent origins Ascidian Larvae (Swalla et al. 2000). Finding the phylogenetic distribution of life-history strategy is requisite for understanding whether Ascidians are a large and diverse group of marine inver- the observed patterns in larval morphology are correlated tebrates with a complex life history characterized by a sessile with life history or due to shared ancestry. Using recent phy- adult and a pelagic larval stage. Comprised of more than 3000 logenetic systematic studies (e.g., Swalla et al. 2000; Stach species (Jeffery 1997), ascidians were included in the phylum and Turbeville 2002), we examined the phylogenetic distri- Chordata because their larvae possess a notochord (Kowa- bution of larval morphology among solitary and colonial spe- levsky 1866), an organ that plays a role in their undulatory cies. swimming (McHenry 2001). Recent phylogenetic studies The broad goals of the present study were to determine (e.g., Cameron et al. 2000; Swalla et al. 2000) support this the patterns of evolutionary change in larval morphology classi®cation and ®nd that the urochordata (which includes among ascidians and to understand how this change affected ascidians and the less speciose pelagic larvaceans and thal- swimming performance. We pursued these goals by focusing iaceans) is a monophyletic group that includes ascidians as on four questions: (1) Can a mathematical model accurately a polyphyletic assemblage united by the presence of a sessile predict swimming performance across species? (2) How does adult stage (Swalla et al. 2000; Stach and Turbeville 2002). tail motion affect swimming performance? (3) Do colonial The larval stage provides an ascidian with its only oppor- ascidians have a convergent larval morphology? (4) How has tunity for dispersal by locomotion. Ascidian larvae have a evolutionary change in larval morphology affected swimming rigid globose trunk that is propelled through the water by its performance? ¯exible tail, thereby bearing a gross resemblance to an anuran tadpole. These tadpole larvae (sensu Brusca and Brusca 1990) MATERIALS AND METHODS do not feed and therefore rely on a ®xed storage of energy to survive through dispersal and metamorphosis (Burighel Biomechanics and Cloney 1997). This suggests that larvae that swim with The peripheral shape of the body and the tail motion of a relatively low energetic cost of transport (McMahon 1984) C. intestinalis and D. occidentalis were measured to model should have an improved chance of survival through the lar- the hydrodynamics of their swimming. These measurements val stage (as in bryozoans; Wendt 2000). Field observations were previously reported for D. occidentalis (McHenry 2001), suggest that larvae may enter into and exit from fast hori- and are newly presented for C. intestinalis. Adults of both zontal currents with vertically oriented swimming. Therefore, C. intestinalis and D. occidentalis were collected in northern dispersal distance may be in¯uenced by the speed with which California, USA, and held in a recirculating seawater tank at a larva traverses these currents (Young 1986; Bingham and 16ЊC. Larvae of C. instestinalis were cultured from the gam- Young 1991; Stoner 1992). Fast swimming may also shorten etes of adults using standard embryological techniques (see the duration of the dispersal phase and allow larvae greater Strathmann 1987). Colonies of D. occidentalis were exposed control over their selection of a microhabitat for settlement to bright incandescent light after being kept in darkness over- EVOLUTION OF ASCIDIAN LARVAE 1211 night to stimulate the release of brooded larvae (Cloney aging, San Diego, CA). Coordinates along the midline of the 1987). body were found from video images of larvae and trans- formed with respect to the frontal plane of the body using a Morphometrics of body shape custom computer program (for details see McHenry 2001). Undulatory kinematics were described by ®tting midline co- The peripheral shape of the bodies of larvae was measured ordinates to equations (explained in McHenry 2001) that de- using digital still images and approximated with a series of scribe changes in the angle between the trunk and tail and equations. Digital photographs (Coolpix 700, Nikon, Mel- the curvature of the tail with time. Neglecting any asymmetry ville, NY) of larvae from dorsal and lateral views (N ϭ 5 for in tail motion, these equations allowed the kinematics to be C. intestinalis, N ϭ 11 for D. occidentalis) were imported described completely by the maximum curvature of the tail, into Matlab (rel. 12 Mathworks, Natick, MA; on a Lifebook ␬ , the maximum angle between the trunk and the tail, ␪ , E, Fujitsu, Tokyo), where a custom program found the co- max max and the tail-beat frequency, f. The kinematics of species were ordinates describing the peripheral shape of the body. The compared (N ϭ 5inC. intestinalis and N ϭ 14 for D. oc- dorsal margin of the trunk was described by a half-ellipse cidentalis) by testing for signi®cant differences in these pa- having a major axis equal to half the length, l , and a minor max rameters using an unpaired Student's t-test (Sokal and Rohlf axis equal to the radius, w , of the trunk. The trunk radius max 1995). was measured as half of the mean of the maximum thickness of the trunk measured from lateral and dorsal views. The Mathematical modeling of swimming distance between the trunk midline and the peripheral margin, w, was de®ned relative to its position along the midline, l, We used a mathematical model of the biomechanics of with the following equation for an ellipse (Thomas and Fin- swimming that was developed by McHenry et al. (2003). This ney 1980): model approximates the quasi-steady ¯uid forces acting on the trunk and tail of a larva given a description of the body's 4(l Ϫ l /2)2 peripheral shape, tail kinematics, and mass. We solved the w(l) ϭ w2 1 Ϫ max , (1) max 2 equations describing these forces numerically using a vari- Ί []lmax able order Adams-Bashforth-Moulton solver programmed in where 0 Ͻ l Ͻ lmax. Half-ellipses were used in this way to Matlab (Shampine and Gordon 1975) to ®nd how the velocity, describe the dorsal, ventral, and lateral margins of the trunk. rate of rotation, position, and orientation of the body changed The trunk was assumed to be circular in cross section with with time in two dimensions during a swimming sequence. a radius equal to w. The cellular region of the tail was also The mean swimming speed and the hydrodynamic cost of assumed to be circular in cross-section (as in McHenry 2001; transport were calculated to assess the performance of each McHenry and Strother 2003), with the radius tapering pos- mathematical simulation. Each simulation lasted for a du- teriorly, as described by the following equation: ration of six tail beats and the results from unsteady swim- rs ming were discarded by removing calculations for the ®rst r(s) ϭϪmax ϩr (2) 0.85F max tail beat. The hydrodynamic cost of transport (COT) was calculated with the following equation (McHenry and Jed where rmax is the maximum measured radius, s is the position 2003): along the midline of the tail and 0 Ͻ s Ͻ 0.85F, where F is n the tail length. The distance between the dorsal margin of ͸ UTii⌬t the tail ®n and the tail midline, q, was described with the iϭ1 COT ϭ , (4) following function: mx  qmax at s Ͻ 0.2 where i is the index for each of the n instantaneous values  q(s) ϭ  (3) of speed, Ui, and thrust, Ti, and x is the net distance traversed 1.25qmax(1 Ϫ s/F)ats Ն 0.2, over the duration of a swimming sequence. This measure of energetic economy neglects internal costs and therefore pro- where q is the maximum height of the tail ®n and 0 Ͻ s max vides a minimum estimate of the metabolic cost of transport Ͻ F. Assuming dorsoventral symmetry, we used the same (Schmidt-Nielsen 1972). function to describe the ventral margin of the tail ®n. From The accuracy of the model was tested by comparing pre- these measurements of peripheral shape, we calculated the dictions of speed for C. intestinalis and D. occidentalis with body mass, m, center of mass, and the moment of inertia measurements. In simulations and experiments, average using a program written in Matlab (for details, see McHenry speed was calculated as the mean of instantaneous speed et al. 2003). We tested for signi®cant differences in mor- values. This measure of speed is not equivalent to the net phological parameters between D. occidentalis and C. intes- speed of swimming (i.e., total distance divided by duration), tinalis using an unpaired Student's t-test in Matlab (Sokal because of the meandrous path followed by models and lar- and Rohlf 1995). vae. For each larva that we measured tail kinematics and speed, a mathematical simulation was run with the same ki- Kinematics of swimming nematics, and the predicted and measured speeds were com- The three-dimensional tail kinematics of freely swimming pared with a paired Student's t-test for each species. larvae of both species were recorded with two high-speed The relative contributions of differences in morphological video cameras (Motionscope PCI Mono/1000S, Redlake Im- and kinematic parameters on the performance differences be- 1212 M. J. MCHENRY AND S. N. PATEK

TABLE 1. Morphometric data for comparative analyses.

Trunk Trunk Log10 Tail Body length radius trunk volume length length Species (mm) (mm) (mm3) (mm) (mm) Source Solitary Ascidia mentula 0.17 0.06 Ϫ2.97 0.47 0.64 Berrill 1931, Berrill 1950 Ascidiella aspersas 0.31 0.09 Ϫ2.30 0.87 1.19 Berrill 1929 Ascidiella scabra 0.30 Ð 0.60 0.90 Berrill 1931 Ciona intestinalis 0.30 0.06 Ϫ2.60 1.16 1.46 present study Corella in¯ata 0.20 0.08 Ϫ2.59 0.63 0.84 Young 2002, present study Dendrodoa grossularia 0.62 0.24 Ϫ1.12 2.07 2.68 Berrill 1929 Halocynthia roretzi 0.39 0.12 Ϫ1.90 1.38 1.77 Satoh 1994 Herdmania pallida 0.24 0.08 Ϫ2.48 0.91 1.15 Sebastian 1953 tubifera 0.17 Ð 0.50 0.67 Berrill 1931 Molgula citrina 0.39 0.12 Ϫ1.92 1.32 1.71 Grave 1926 0.11 0.04 Ϫ3.41 0.37 0.48 Jeffery and Swalla 1992 Molgula complanata 0.40 Ð 1.20 1.60 Berrill 1931 Molgula manhattensis 0.15 0.04 Ϫ3.33 0.50 0.65 Berrill 1950 ®brosa 0.20 0.11 Ϫ2.34 0.83 1.02 Berrill 1929 Pyura pachydermatina 0.31 0.11 Ϫ2.07 0.94 1.25 Anderson et al. 1976 Styela partita 0.22 0.06 Ϫ2.81 0.64 0.85 Berrill 1950 Colonial Aplidium constallatum 0.79 0.19 Ϫ1.24 1.46 2.25 Grave 1921 Aplidium punctum 0.53 0.19 Ϫ1.41 0.98 1.51 Berrill 1950 Botrylloides leachii 0.56 0.18 Ϫ1.44 1.04 1.60 Berrill 1935 Botrylloides sp. 1.08 0.28 Ϫ0.75 1.89 2.97 McHenry 2001, present study Botrylloides simodensis 0.47 0.18 Ϫ1.48 1.27 1.73 Mukai et al. 1987 Botryllus gigas 0.85 0.28 Ϫ0.86 2.08 2.93 Berrill 1935 Botryllus schlosseri 0.52 0.18 Ϫ1.43 1.15 1.67 Berrill 1950 Clavelina picta 1.08 0.32 Ϫ0.64 2.52 3.60 Berrill 1935 Didemnum paci®cum 0.53 0.19 Ϫ1.40 1.18 1.71 Tokioka 1953 Distaplia occidentalis 1.29 0.18 Ϫ1.04 2.27 3.55 present study 0.83 0.27 Ϫ0.90 1.72 2.55 Berrill 1950 argus 0.50 Ð 1.30 1.80 Berrill 1931 listeri 0.32 0.12 Ϫ2.01 0.58 0.90 Berrill 1950 Polyandrocarpa gravei 0.38 0.12 Ϫ1.93 1.45 1.83 Grave 1932 tween C. intestinalis and D. occidentalis were examined by partial cox 1 mitochondrial DNA sequences (®g. 4A in Stach running a series of simulations. Simulations were run with and Turbeville 2002) as tree D, and by strict consensus of parameter values of both species for tail length, tail height, amino acid sequences translated from partial cox 1 mito- trunk volume, trunk shape (the ratio of trunk radius to length), chondrial DNA sequences (®g. 4B in Stach and Turbeville and tail kinematic parameters (tail-beat frequency, maximum 2002) as tree E. curvature, and maximum trunk angle). For example, to ex- Life-history data were obtained from published sources amine the effect of tail length, we ®rst ran a simulation using (Berrill 1950; Burighel and Cloney 1997; Swalla et al. 2000; all the parameter values of C. intestinalis, and then ran a Cloney et al. 2002; Morris et al. 2002) for all taxa included simulation of the same model except that it used the tail in the Stach and Turbeville (2002) phylogenies. Analyses of length of D. occidentalis. The resulting differences in per- morphology were implemented using modi®ed trees that only formance were used to indicate the effect of tail length. included taxa with morphological data. For cases in which genera were monophyletic in all trees, we exchanged species Phylogenetic Comparative Analyses data within genera such that we maximized the amount of The evolutionary patterns of life-history strategy and larval comparative data to be used in the analyses. Due to these morphology were examined by mapping traits onto phylog- modi®cations of the tree structure, we were not able to in- enies and applying phylogenetically independent contrast corporate branch lengths into the trees when conducting in- analyses (Felsenstein 1985; Harvey and Pagel 1991). These dependent contrast tests. analyses were conducted using ®ve published phylogenies We reconstructed the phylogenetic pattern of life-history (®gs. 2, 4 from Stach and Turbeville 2002) and were coded strategy using MacClade's most parsimonious reconstruction into MacClade (ver. 4.05, Maddison and Maddison 2000). (Maddison and Maddison 2000) based on all ®ve trees under We will refer to the phylogeny found by strict consensus of consideration. Phylogenetic comparative tests of correlations 18S rDNA sequences (®g. 2A in Stach and Turbeville 2002) between traits were conducted using CAIC (ver. 2.6.9, Purvis as tree A, by strict consensus of 18S rDNA and morphological and Rambaut 1995). Brunch algorithms were used for the traits (®g. 2B in Stach and Turbeville 2002) as tree B, by categorical tests examining colonial/solitary transitions and maximum likelihood of 18S rDNA sequences (®g. 2C in Crunch algorithms were used for morphological traits. Mor- Stach and Turbeville 2002) as tree C, by strict consensus of phological traits were log-transformed to reduce the depen- EVOLUTION OF ASCIDIAN LARVAE 1213

FIG. 1. The body shape of larvae. The peripheral shape of the bodies of larvae of (A) Distaplia occidentalis (N ϭ 11) and (B) Ciona intestinalis (N ϭ 5) is shown from dorsal and lateral views, with mean values (horizontal hatches, Ϯ1 SD) at 70 positions down the length of the body. The black lines show the curves ®t to these data (eq. 1±3). The mean values (ϩ1 SD) for (C) linear body dimensions and (D) trunk volume for each species. Note that the error bars are too small to be visible in D. dence of the variance on the mean (Sokal and Rohlf 1995) of an exponential equation (F ϭ aVb; Huxley 1932). We and to examine scaling relationships (see below). Statistical tested whether colonial and solitary species scaled isomet- correlations of contrasts were calculated using a regression rically (F ϭ aV0.33) by calculating the pro®le likelihood con- line forced through the origin, as implemented by CAIC ®dence intervals for the regression slopes (JMP 5.0.1) and (Purvis and Rambaut 1995). observing whether the slope value of 0.33 fell within these Morphological traits were recorded from values reported con®dence intervals. We compared species and contrast val- in the literature (sources listed in Table 1), measured from ues of colonial and solitary species regression slopes and scans of published camera lucida drawings, or measured from intercepts using a t-test modi®ed for comparing linear re- digital photographs of larvae. The positions of morphological gressions (Zar 1999, ch. 18). landmarks were recorded from scans of drawings (Epson 3200 Photo, San Jose, CA) and digital photographs (Nikon Theoretical Morphology Coolpix 700) using a custom program in Matlab. We mea- sured trunk length and radius and tail length and calculated Theoretical morphology was used to examine the distri- the volume of the trunk by assuming an ellipsoidal shape bution of colonial and solitary species in a morphospace and (McHenry et al. 2003). Distaplia occidentalis was excluded performance landscape. We used a principal components in these comparisons because this species was not included analysis to de®ne areas of trunk volume/tail length mor- in the phylogenetic analysis of Stach and Turbeville (2002). phospace occupied by colonial and solitary species. The The scaling of body shape was examined in colonial and boundaries of these areas were de®ned by ellipses of 95% solitary species using both species and contrast values. The con®dence intervals for the major and minor axes of variation exponential relationship between trunk volume and tail length in each of these groups (Sokal and Rohlf 1995). We found was described by a scaling constant, a, and scaling factor, b, the speed and cost of transport predicted for positions 1214 M. J. MCHENRY AND S. N. PATEK

FIG. 2. The undulatory motion of the tail. The lateral motion of the midline of the tail can be seen as a time series from a dorsal view of a larva of (A) Distaplia occidentalis and (B) Ciona intestinalis. The mean values and standard deviations of the (C) maximum curvature of the tail, (D) the maximum trunk angle, and (E) tail-beat frequency for both species (N ϭ 5 for C. intestinalis and N ϭ 14 for D. occidentalis). throughout the morphospace by running simulations of our C. intestinalis in all linear dimensions measured (Fig. 1C). biomechanical model (see above) over a range of trunk vol- Furthermore, the mean trunk volume of D. occidentalis larvae ume and tail length values at 15 equal intervals, for a total (VÅ ϭ 3.35 ϫ 10Ϫ1 mm3) was more than 100 times greater of 225 simulations, which spanned beyond the range of mea- than that of C. intestinalis (VÅ ϭ 2.60 ϫ 10Ϫ3 mm3, Fig. 1D). sured values in each parameter measured. Simulations were The undulatory motions of both species were qualitatively run with the tail kinematics and body shape of C. intestinalis. similar (Fig. 2A, B) and the two species were indistinguish- For example, trunk volume was varied by altering the trunk able in their maximum curvature (P ϭ 0.75, Fig. 2C) and width and trunk length, but the ratio of these parameters maximum trunk angle (P ϭ 0.74, Fig. 2D). However, D. remained equal to that of C. intestinalis throughout all sim- occidentalis swam with a signi®cantly higher tail-beat fre- ulations. quency than C. intestinalis (P ϭ 0.04, Fig. 2E). Mathematical simulations of the biomechanics of swim- RESULTS ming predicted different trajectories and average swimming speeds for the two species (Fig. 3). Distaplia occidentalis was Morphology, Kinematics, and Performance of Ciona predicted to generate greater trunk rotation with each tail intestinalis and Distaplia occidentalis beat and thereby followed a relatively meandrous trajectory A comparison of morphology and kinematics between C. (Fig. 3A) compared to the swimming of C. intestinalis (Fig. intestinalis and D. occidentalis presents some of the differ- 3B). The average speed of swimming (mean of instantaneous ences and similarities between colonial and solitary species. values) in both species oscillated with time (Fig. 3C, D), but The peripheral shapes of the bodies of both species were well the average speed of swimming was more than an order of approximated by the equations and parameter values describ- magnitude greater in D. occidentalis than in C. intestinalis ing body shape (eq. 1±3; Fig. 1A, B). Distaplia occidentalis (Fig. 3E, F), a difference that was re¯ected in measurements was signi®cantly larger (unpaired t-test, P ϽϽ 0.001) than of freely swimming larvae. In both species, the mathematical EVOLUTION OF ASCIDIAN LARVAE 1215

FIG. 3. Typical results from the mathematical model of swimming biomechanics. Points show examples of the movement predicted for the center of mass over a duration of seven tail-beats at intervals of 2 msec, starting at the arrow, for (A) Distaplia occidentalis and (B) Ciona intestinalis. The predicted average speed (mean of instantaneous values) for the (C) D. occidentalis model in A and (D) the C. intestinalis model shown in B. The mean (ϩ1 SD) swimming speed predicted and measured for (E) D. occidentalis and (F) C. intestinalis larvae. model predicted speeds that were statistically indistinguish- lidae ϩ Pyuridae includes both solitary and colonial species able from these measurements of speed (paired t-test, P ϭ and suggests at least one origin of coloniality. All Aplou- 0.07 in D. occidentalis, Fig. 3E; P ϭ 0.28 in C. intestinalis, sobranchiata species were colonial, which implies an origin Fig. 3F). of coloniality prior to their most recent common ancestor. The effect of individual morphometric and kinematic dif- Perophora japonica is a colonial member of the largely sol- ferences on the performance differences between D. occi- itary Phlebobranchiata clade. Coloniality appears to have ei- dentalis and C. intestinalis were evaluated with the results of ther evolved in the lineage leading to P. japonica (as in trees a series of simulations (a±f in Fig. 4). Increasing the tail A and C), or possibly prior to the common ancestor to the length and height from the size of C. intestinalis (a in Fig. Phlebobranchiata and Thaliacea (equivocal in tree B). The 4) to that of D. occidentalis (c in Fig. 4) resulted in an 18% origin of coloniality arising prior to the common ancestor of decrease in speed and a 27% increase in the cost of transport. the Thaliacea was not considered in our analysis of larval Increasing the trunk volume to the size of D. occidentalis (d morphology due to a lack of larval data for this group. As a in Fig. 4) resulted in swimming that was nearly twice the result of these patterns of life-history evolution, independent speed, but had a lower cost of transport. A further decrease contrast analyses based on each of these trees yielded either in the cost of transport was achieved by changing the trunk three or four contrasts for comparison of larval morphology shape (the ratio of trunk radius to length) from that of C. between species of each life-history strategy (Table 2). intestinalis (d in Fig. 4) to that of D. occidentalis (e in Fig. Phylogenetically independent contrast analyses were used 4). However, these effects of morphology were small relative to examine whether the observed patterns of larval mor- to the effect of tail kinematics. A model having the body phology in colonial species were the result of shared ancestry shape of D. occidentalis, but the tail kinematics of C. intes- or evolutionary convergence (Fig. 6). We found that colonial tinalis (e in Fig. 4) moved with a speed that was just 14% ascidians have signi®cantly greater trunk volume than the the speed and 3% the cost of transport of the same model larvae of solitary species (PÅ ϭ 0.035, Table 2) and that tail animated with the kinematics of D. occidentalis (f in Fig. 4). length was signi®cantly correlated with trunk volume (PÅ ϭ 0.004, Table 2, Fig. 7A, B). However, colonial species were Phylogeny and Independent Contrasts statistically indistinguishable from solitary species in tail Parsimony reconstructions of ascidian life history sug- length (PÅ ϭ 0.515). Although the absolute values for tail gested that the common ancestor to the urochordates was length were indistinguishable between colonial and solitary solitary, and that coloniality evolved independently at least species, the scaling of tail length relative to trunk volume three times among urochordates (Fig. 5). The clade of Stye- was different in these groups (see below). 1216 M. J. MCHENRY AND S. N. PATEK

and tail length morphospace occupied by ascidian larvae (Fig. 7C). The species values for Herdmania pallida approximated the mean trunk volume and tail length of solitary species (V ϭ 8.2 ϫ 10Ϫ4 mm3, F ϭ 0.91 mm) and the mean values for colonial species were approximated by Aplidium constellatum (V ϭ 1.4 ϫ 10Ϫ2 mm3, F ϭ 1.46 mm). Therefore, we con- sidered these species to be representative of larvae produced by the species of their respective life history strategy. The results of our mathematical simulations allowed us to examine the effects of trunk volume and tail length on swim- ming performance given the same tail kinematics (Fig. 8). The fastest swimming was predicted for larvae having rel- atively long tails (F Ͼ 2.5 mm) and either low (V Ͻ 10Ϫ3 mm3) or intermediate (10Ϫ2 mm3 Ͻ V Ͻ 10Ϫ1 mm3) trunk volume (Fig. 8A). However, the cost of transport was lowest in larvae having relatively small tails (F Ͻ 1.6 mm) and large trunks (V Ͼ 10Ϫ3 mm3), and greatest in larvae with large tails (F Ͼ 1.6 mm) and small trunks (V Ͻ 10Ϫ3 mm3, Fig. 8B). The morphospace occupied by colonial species was char- acterized by slightly slower speed, but a lower cost of trans- port than solitary species. For example, the solitary H. pallida moved 69 % faster than colonial A. constellatum (12.3 mm secϪ1 compared to 8.5 mm secϪ1), but with a 44% greater cost of transport (2.5 J kgϪ1 mϪ1 compared to 1.1 J kgϪ1 mϪ1, Fig. 8C). A model larva having the proportionate tail length of H. pallida, but the trunk volume of A. constellatum moved faster than both species, but with an intermediate cost of transport (13.6 mm secϪ1,2.2JkgϪ1 mϪ1, Fig. 8C). Neither group of species occupies regions of morphospace that result in extremely fast or energetically costly swimming. Further- more, the in¯uence of morphological variation on swimming performance is subtle compared to the effect of tail motion (Fig. 4, described above).

DISCUSSION

FIG. 4. The effect of morphology and kinematics on swimming Can a mathematical model accurately predict swimming performance. The parameter values and predicted performance are performance across species? aligned in columns for each of six simulations (a±f). Filled squares denote which of the listed parameters have the value for Distaplia The accuracy of a mathematical model of swimming was occidentalis and open squares denote which have the value for Ciona tested by comparing predictions of speed with measurements intestinalis. The resulting performance for each simulation is pre- in C. intestinalis and D. occidentalis. These species are rep- sented in the bar charts for the (A) speed and (B) cost of transport resentative of many of the differences between solitary and predicted for larvae. colonial species (Table 1, Fig. 6). For example, D. occiden- talis has a trunk volume that is more than two orders of Both colonial and solitary species were found to scale iso- magnitude greater than that of C. intestinalis (Fig. 1D). There- metrically. For all regressions of log-transformed values of fore, our ®nding that the predicted speeds were indistinguish- trunk volume and tail length, the slope of 0.33 fell within able from measurements in both species (Fig. 3) suggests that the limits of the con®dence intervals for both independent the model accurately characterizes the dynamics of swim- contrast and species values (Table 3). Scaling factors did not ming in a diversity of ascidian larvae. differ signi®cantly between solitary and colonial species. Although mathematical models of organismal function However the scaling constant (i.e., the intercept of the re- may be based on extrapolations from ®rst principles, complex gression, Fig. 7C) was signi®cantly different between colo- models are inevitably dependent on numerous simplifying nial and solitary species values (Table 3). The lower intercept assumptions. A model's predictions may vary tremendously of colonial ascidians (Table 3, Fig. 7C) means that these depending on the assumptions used by the investigator. For species have a tail length that is smaller in proportion to example, both Daniel et al. (1997) and Hassan et al. (2002) trunk volume than solitary species. created sophisticated ®nite- element models of the shells of extinct ammonites to test how septal complexity affects shell Morphospace and Performance Landscapes strength. Daniel et al.'s ®nding that septal complexity reduces The 95% con®dence intervals of principal components for shell strength was explicitly refuted by the model of Hassan solitary and colonial species de®ned areas of trunk volume et al. Without a validation of either model with measurements EVOLUTION OF ASCIDIAN LARVAE 1217

FIG. 5. The phylogenetic pattern of life-history strategy among urochordates. Colonial (®lled circles), solitary (open circles), and equivocal (half-®lled circles) states are mapped onto the phylogenetic relationships proposed by Stach and Turbeville (2002) using parsimony reconstruction (MacClade ver. 4.05, Maddison and Maddison 2000). The major urochordate clades are denoted with shaded boxes with names given to the right and the full species names are given in only tree A. See Materials and Methods for details. 1218 M. J. MCHENRY AND S. N. PATEK

TABLE 2. Phylogenetically independent contrast analyses. the same or reduced energetic cost may act on morphological traits such as trunk volume. Number of Comparisons contrasts Slope r2 P Do colonial ascidians have a convergent larval Log trunk volume versus life history morphology? Tree A 3 0.43 0.926 0.038 Tree B 3 0.33 0.901 0.051 Our phylogenetic analysis supports prior evidence that the Tree C 3 0.34 0.970 0.015 common ancestor to the urochordates was solitary and that Tree D 4 0.45 0.838 0.029 Tree E 3 0.59 0.915 0.044 coloniality had multiple independent origins (Fig. 5). These Mean 0.43 0.910 0.035 results are consistent with traditional , which large- Log tail length versus life history ly ignored coloniality as a trait for classi®cation and consid- Tree A 3 0.03 0.156 0.606 ered both life-history strategies to be represented in many Tree B 3 0.03 0.205 0.547 ascidian families (Berrill 1950). Although urochordate phy- Tree C 3 0.04 0.156 0.605 logenetics remains a debated topic, there is no evidence (e.g., Tree D 4 0.04 0.310 0.330 Cameron et al. 2000; Swalla 2001) that coloniality, evolved Tree E 3 0.05 0.261 0.489 Mean 0.04 0.217 0.515 in a single evolutionary event among ascidian species. All trees considered presently agreed on at least three indepen- Log trunk volume versus log tail length dent origins of coloniality, and this robust feature was the Tree A 10 0.26 0.674 0.002 most important to our comparative analysis because each Tree B 13 0.30 0.839 Ͻ0.001 Tree C 15 0.29 0.698 Ͻ0.001 origin provided the opportunity to compare the larval traits Tree D 12 0.26 0.697 Ͻ0.001 of one life-history strategy against another. Tree E 6 0.22 0.719 0.016 The results of our independent contrast analyses suggest Mean 0.26 0.726 0.004 that colonial ascidians have a convergent larval morphology. Although both species and contrast values scaled isometri- cally among both groups of species (Fig. 7), colonial ascid- from related extant species, it is dif®cult to evaluate which ians had a signi®cantly greater trunk volume (Table 2; Figs. investigators more accurately replicated the biomechanics of 6, 7) and shorter tails in proportion to the trunk volume (as ammonoid shells. In the present study, model veri®cation was shown by their lower scaling constant, Fig. 7, Table 3). These an essential ®rst step toward applying the biomechanics of morphological differences have implications for both the lo- ascidian larvae (McHenry et al. 2003) to tests of evolutionary comotion and life-history strategy of these groups of species. hypotheses of larval morphology and performance. For example, a larger material investment in the larvae of colonial species (re¯ected by trunk volume) requires that How does tail motion affect swimming performance? adults either have lower fecundity or higher total reproductive investment than solitary species and a greater investment may Our results suggest that although the body shape of a larva have an adverse effect on adult growth or survivorship has important functional consequences (Fig. 8), interspeci®c (Stearns 1992). In fact, colonial ascidians likely use a similar differences in swimming speed may largely be explained by total reproductive investment as solitary species because they differences in tail-beat frequency (Figs. 2, 4). Our simulation generally produce fewer of their relatively large larvae (Svane results found a sevenfold increase in swimming speed when and Young 1989). The large investment that colonial species a model was run with the high-frequency kinematics of D. provide for each larva may be necessary to create a trunk occidentalis (f in Fig. 4) compared to the same model having volume with the capacity to accommodate the differentiated the low-frequency kinematics of C. intestinalis (e in Fig. 4), adult organs carried during dispersal that solitary larvae typ- whereas manipulations of morphology resulted in much ically do not carry (Berrill 1975). Therefore, the large trunk smaller changes in speed (Figs. 4, 8). This suggests that in- volume of colonial species affects not only the performance terspeci®c variation in speed may largely be explained by of larval locomotion (see below), but may have a cost in differences in tail-beat frequency and that selection for in- terms of adult fecundity and a bene®t in allowing larvae to creased speed could result in an evolutionary increase in tail- carry adult organs. beat frequency. Swimming with higher tail-beat frequency may allow a How has evolutionary change in larval morphology affected larva to move faster, but this high performance comes at a swimming performance? great energetic cost. The simulation moving at the rapid tail- beat frequency of D. occidentalis (f in Fig. 4) was over 32 Using a mathematical model of swimming biomechanics times more energetically costly than swimming at the slow allowed an examination of how morphology in¯uences swim- frequency of C. occidentalis (e in Fig. 4). However, it is ming performance in the absence of the potentially confound- possible that small increases in speed may be achieved with- ing effects of kinematic variation. We investigated the in- out an energetic cost over the course of evolution from chang- dividual effects of trunk volume and tail length on swimming es in larval morphology. For example, increasing the trunk performance by running a series of simulations with model volume from that of C. intestinalis (c in Fig. 4) to that of D. larva that differed only in these two parameters. Simulation occidentalis (d in Fig. 4) resulted in swimming that was 2.2 results suggest that the evolutionary convergence of colonial times faster, but the cost of transport was reduced by 58% species toward larger trunk volume and proportionately short- (Fig. 4). This suggests that selection for faster swimming at er tail length resulted in swimming that was slower, but with EVOLUTION OF ASCIDIAN LARVAE 1219

FIG. 6. The phylogenetic distribution of larval morphology. The silhouettes illustrate the proportions of tail length, trunk radius, and trunk length for representative solitary (open circles) and colonial (®lled circles) species. The phylogenetic relationships in this case were derived from tree C. a lower cost of transport, than solitary species (e.g., a in Fig. faster running with the same stamina as young produced from 8) or larvae the size of colonial species with the proportionate smaller eggs. However, larger lizard eggs come at the cost tail length of solitary species (e.g., c in Fig. 8). of fecundity (Sinervo and Licht 1991). The position of a larva in morphospace may affect both The effects of larval morphology on multiple aspects of locomotor performance and life-history traits that affect ®t- performance may be considered with a multidimensional per- ness. Among solitary ascidians, larger species are predicted formance landscape. This landscape is distinct from Raup's to swim faster (Fig. 8A) but have virtually the same cost of concept of a multidimensional morphospace (see Raup and transport as smaller species (Fig. 8B). More rapid swimming Michelson 1965) because it is the performance variables, not may allow larvae to make their dispersal phase more brief the morphometric parameters, that create deviation from a (which improves ®tness in other marine invertebrates, e.g., three-dimensional surface. Figure 9 illustrates such a land- bryozoans, Wendt 1996), and may enhance control over dis- scape in trunk volume/tail length morphospace using arrows persal distance and habitat selection (van Duyl et al. 1981; that point in the direction of increasing performance. Under Durante 1991; Stoner 1994; Svane and Dolmer 1995). This the assumption of a ®xed total reproductive investment, in- suggests that natural selection should favor larger, solitary creased fecundity is directed toward smaller body size along larvae. However, assuming a ®xed total reproductive in- the axis of isometric scaling (Fig. 9A). The results of our vestment, larger larvae come at the cost of adult fecundity. mathematical simulations (Figs. 8A, 8B) indicate the direc- Similarly, Sinervo and Huey (1990) experimentally demon- tions of increasing speed and energetic economy (i.e., re- strated that larger eggs in lizards create young capable of duced cost of transport, Fig. 9A). 1220 M. J. MCHENRY AND S. N. PATEK

TABLE 3. Scaling of log-transformed trunk volume versus tail length. b, scale factor; L1, lower 95% con®dence interval, L2 upper 95% con®dence interval; N, sample size.

bL1 L2 N Contrast values of colonial species Tree B 0.42 0.22 0.65 5 Tree C 0.42 0.22 0.62 6 Contrast values of solitary species Tree B 0.35 0.25 0.46 4 Tree C 0.34 0.17 0.51 4 Species values Colonial 0.34 0.18 0.49 13 Solitary 0.31 0.23 0.39 13 Comparison P df Colonial versus scaling factor Species values 0.75 23 Tree B 0.46 6 Tree C 0.59 6 Colonial versus scaling constant Species values 0.01 22

This multidimensional performance landscape may be used to predict evolutionary change under selective conditions. For example, if natural selection strongly favored increased fe- cundity regardless of the cost to locomotor performance, the larvae of a solitary species would be predicted to isometri- cally decrease in tail length and trunk volume (a in Fig. 9B). Under strong selection for faster swimming, tail length would increase over the course of evolution at an allometric rate outpacing increases in trunk volume (b in Fig. 9B). In con- trast, selection for improved energetic economy would result in an allometric reduction in tail length and an increase in trunk volume (c in Fig. 9B). However, the observed pattern of evolutionary change does not follow any of these patterns, which suggests a more complex evolutionary scenario (Fig. 9C). The evolutionary convergence of larval morphology in co- lonial ascidians may have resulted from trade-offs between the swimming speed and energetic economy of larvae and the fecundity of adults. The relatively low energetic cost of swimming with a proportionately short tail supports the hy- pothesis that the tail length of colonial ascidians evolved as an adaptation for high energetic economy. The relatively low speed that accompanies this disproportionately short tail is partially offset by the larger size of colonial species. The

← isometry. (A) Dark lines show linear regressions for contrasts be- tween solitary and between colonial species based on tree C. (B) The dark line illustrates the linear regression for all contrasts, in- cluding contrasts between solitary and colonial species (®lled squares) and ambiguous nodes (half-®lled circles), based on tree D. Tree D most closely represents the independent contrasts results from the average across all trees. (C) Ellipses of 95% con®dence FIG. 7. Trunk volume and tail length among ascidian larvae. (A, intervals for solitary (thin line) and colonial (heavy line) species B) These representative contrast values for trunk volume and tail are drawn around species values in this log-transformed morphos- length were log-transformed. Open circles represent comparisons pace of trunk volume and tail length. The straight lines show the between solitary species, closed circles are for comparisons between least-squares scaling relationship for solitary (thin line) and colonial colonial species, and the gray line shows the scaling predicted by (heavy line) species. EVOLUTION OF ASCIDIAN LARVAE 1221

FIG. 8. Swimming performance predicted by mathematical simulations using the mean tail kinematic parameters of Ciona intestinalis. The effects of trunk volume and tail length on the swimming speed (A) and cost of transport (B) are shown by contour lines and a gradient of gray values. The 95% con®dence intervals for the distribution of solitary (thin line) and colonial (thick line) species (as in Fig. 7C) are overlaid on the contour map. Italicized letters denote the positions of individual species: (a) Herdmania pallida,(b) Aplidium constellatum, and (c) a scaled-up model of larva H. pallida having the trunk volume of A. constellatum. (C) The speed (gray bars, left axis) and cost of transport (white bars, right axis) predicted for larvae a±c. 1222 M. J. MCHENRY AND S. N. PATEK

evolutionary increase in larval size in colonial species may have been favored by selection for faster swimming. How- ever, further increases in size may have resulted in a penalty to fecundity that exceeded the bene®ts of even faster swim- ming. This consideration of evolutionary forces has necessarily made simplifying assumptions, so it is important to consider the potential for more complex dynamics in the evolution of ascidian larvae. For example, ascidian larvae may be con- strained from occupying regions of morphospace that are not occupied by extant species. The size of larvae may in¯uence the dynamics of ascidian life-history evolution beyond the fecundity of the adult stage (Roff 1992; Stearns 1992). Fur- thermore, the evolution of swimming performance is a func- tion of changes in both tail motion and morphology. Our ®nding that swimming performance is strongly affected by tail-beat frequency (described above) suggests that evolu- tionary changes in performance my largely be attributed to change in this kinematic trait. The present study conducted an intensive three-dimensional kinematic analysis that would be prohibitively labor intensive to conduct on the numerous species necessarily for a broadly comparative study. How- ever, we found that the only signi®cant kinematic difference between C. intestinalis and D. occidentalis was in their tail- beat frequency, a parameter that is relatively easy to measure from video recordings of swimming. It would therefore be tractable and interesting to examine how changes in tail-beat frequency have in¯uenced the evolution of swimming speed in ascidian larvae in a future study. In summary, the present study ®nds that colonial ascidians have independently evolved a larval morphology with a larger trunk volume and proportionately smaller tail length than solitary species. Our mathematical modeling suggests that this evolutionary convergence resulted in slower swimming with improved energetic economy. However, the larger size of colonial larvae may have required a reduction in adult fecundity. These results were found through the use of bio- mechanical techniques that veri®ed the accuracy of a math- ematical model, phylogenetic comparative analyses that rig- orously demonstrated the convergence of trunk volume, and theoretical morphology to examine the functional signi®- cance of convergent morphology in the absence of the con- founding effects of behavioral variation. This integration of approaches holds potential for understanding the evolution of organismal function in a diversity of complex organismal systems.

ACKNOWLEDGMENTS We thank M. Koehl, G. Lauder, and B. Swalla for their advice. Two anonymous reviewers provided valuable sug-

←

FIG. 9. Multidimensional performance landscape. Arrows denote metric scaling. (B) Predicted evolutionary change from the position the direction of increasing performance in speed (green), energetic of a typical solitary species (®lled circle) is illustrated in the di- economy (orange), and fecundity (violet) for ascidian larvae. (A) rection of heavy black arrows under different selective conditions: The direction of increasing speed was determined from the perfor- (a) selection for increased fecundity, (b) selection for faster swim- mance landscape shown in Figure 8A and the direction of higher ming, and (c) selection for improved energetic economy. (C) El- energetic economy (i.e., lower cost of transport) was found from lipses of 95% con®dence intervals approximate the distribution of Figure 8B. Increased fecundity is directed along the axis of iso- solitary (thin gray line) and colonial (heavy gray line) species. EVOLUTION OF ASCIDIAN LARVAE 1223 gestions for the manuscript. Larvae and adults of C. intes- Cloney, R. A. 1978. Ascidian metamorphosis: review and analysis. tinalis were provided by the Levine lab at the University of Pp. 255±282 in F. S. Chia and M. E. Rice, eds. Settlement and metamorphosis of marine invertebrate larvae. Elsevier-North California at Berkeley. This work was supported with a Na- Holland Biomedical Press, Amsterdam. tional Science Foundation predoctoral fellowship and grants- ÐÐÐ. 1987. Phylum Urochordata, Class . Pp. 607±646 in-aid of research from the American Society of Biome- in M. F. Strathmann, ed. Reproduction and development of ma- chanics, Sigma Xi, the Department of Integrative Biology rine invertebrates of the northern Paci®c Coast. Univ. of Wash- ington Press, Seattle. (University of California Berkeley) and the Society for In- Cloney, R. A., C. M. Young, and I. B. Svane. 2002. Phylum Chor- tegrative and Comparative Biology to MJM, the Miller In- data: Urochordata. Pp. 607±646 in C. M. Young, ed. Atlas of stitute for Basic Research in Science to SNP, and grants from marine invertebrate larvae. Academic Press, San Diego, CA. the National Science Foundation (OCE-9907120) and the Of- Crompton, A. W. 1989. The evolution of mammalian mastication. ®ce of Naval Research (AASERT N00014-97-1-0726) to M. Pp. 23±40 in D. B. Wake and G. Roth, eds. Complex organismal functions: integration and evolution in vertebrates. John Wiley Koehl. and Sons, New York. Daniel, T. L. 1983. Mechanics and energetics of medusan jet pro- LITERATURE CITED pulsion. Can. J. Zool. 61:1406±1420. Daniel, T. L., B. S. Helmuth, W. B. Saunders, and P. D. Ward. Alexander, R. M. 2003. Principles of animal locomotion. Princeton 1997. Septal complexity in ammonoid cephalopods increased Univ. Press, Princeton, NJ. mechanical risk and limited depth. Paleobiology 23:470±481. Anderson, D. T., B. M. White, and E. A. Egan. 1976. The larval Drucker, E. G., and G. V. Lauder. 2001. Locomotor function of the development and metamorphosis of the ascidians Pyura prae- dorsal ®n in teleost ®shes: experimental analysis of wake forces putialis (Heller) and Pyura pachydermatina (Herdman) (Pleu- in sun®sh. J. Exp. Biol. 204:2943±2958. rogona, family Pyuridae). Proc. Linn. Soc. N.S.W. 100:205±217. Durante, K. M. 1991. Larval behavior, settlement preference, and Arnold, S. J. 1983. Morphology, performance and ®tness. Am. Zool. induction of metamorphosis in the temperate solitary ascidian 23:347±361. Molgula citrina Alder and Hancock. J. Exp. Mar. Biol. Ecol. ÐÐÐ. 2003. Performance surfaces and adaptive landscapes. In- 145:175±188. tegr. Comp. Biol. 43:367±375. Ellers, O. 1993. A mechanical model of growth in regular sea- Arnold, S. J., and A. F. Bennett. 1988. Behavioral variation in urchins: predictions of shape and a developmental morphospace. natural populations. V. Morphological correlations of locomo- Proc. R. Soc. Lond. B 254:123±129. tion in the garter snake, Thamnophis radix. Biol. J. Linn. Soc. Emerson, S. B., and M. A. R. Koehl. 1990. The interaction of 34:175±190. behavioral and morphological change in the evolution of a novel Bauwens, D., T. Garland, A. M. Castilla, and R. van Damme. 1995. locomotor type. Evolution 44:1931±1946. Evolution of sprint speed in lacertid lizards: morphological, Felsenstein, J. 1985. Phylogenies and the comparative method. Am. physiological, and behavioral covariation. Evolution 49: Nat. 125:1±15. 848±863. Friedman, W. A., T. Garland, and M. R. Dohm. 1992. Individual Bennett, A. F., and R. B. Huey. 1990. Studying the evolution of variation in locomotor behavior and maximal oxygen consump- physiological performance. Pp. 251±284 in D. Futuyma and J. tion in mice. Physiol. Behav. 52:97±104. Antonovics, eds. Oxford surveys in evolutionary biology Vol. Garland, T., and P. A. Carter. 1994. Evolutionary physiology. Annu. 7. Oxford Univ. Press, London. Rev. Physiol. 56:579±621. Bennett, A. F., T. Garland, and P. L. Else. 1989. Individual cor- Gilchrist, G. W., and J. G. Kingsolver. 2003. Is optimality over the relation of morphology, muscle mechanics, and locomotion in hill? Pp. 219±241 in S. Orzack and E. Sober, eds. Adaptationism a salamander. Am. J. Physiol. 256:R1200±R1208. and optimality. Cambridge Univ. Press, Cambridge, U.K. Berrill, N. J. 1929. Studies in development. I. General Grave, C. 1921. Amaroucium constellatum (Verrill). II. The structure physiology of development of simple ascidians. Philos. Trans. and organization of the tadpole larva. J. Morphol. 36:71±101. R. Soc. Lond. B 218:37±78. ÐÐÐ. 1926. Molgula citrina (Alder and Hancock) activities and ÐÐÐ. 1931. Studies in tunicate development. II. Abbreviation of structure of the free-swimming larva. J. Morphol. 42:453±471. development in the Mogulidae. Philos. Trans. R. Soc. Lond. B ÐÐÐ. 1932. The Botryllus type of ascidian larva. Papers from the 219:281±346. Tortugas Laboratory 28:145±156. ÐÐÐ. 1935. Studies in tunicate development. III. Differential Harvey, P. H., and M. D. Pagel. 1991. The comparative method in retardation and acceleration. Philso. Trans. R. Soc. Lond. B 219: evolutionary biology. Oxford Univ. Press, New York. 281±346. Hassan, M. A., G. E. G. Westermann, R. A. Hewitt, and M. A. ÐÐÐ. 1950. The Tunicata. The Ray Society, London. Dokainish. 2002. Finite-element analysis of simulated ammo- ÐÐÐ. 1975. Chordata: Tunicata. Pp. 241±282 in A. C. Geise and noid septa (extinct Cephalopoda): septal and sutural complexi- J. S. Pearse, eds. Reproduction of marine invertebrates. Vol. II. ties do not reduce strength. Paleobiology 28:113±126. Academic Press, New York. Huey, R. B., and A. F. Bennett. 1986. A comparative approach to Biewener, A. A. 2003. Animal locomotion. Oxford Univ. Press, ®eld and laboratory studies in evolutionary biology. Pp. 82±98 London. in M. E. Feder and G. V. Lauder, eds. Predator-prey relation- Bingham, B. L., and C. M. Young. 1991. Larval behavior of the ships. Univ. of Chicago Press, Chicago. ascidian Ecteinascidia turbinata Herdman: an in situ experimen- Hulsey, C. D., and P. C. Wainwright. 2002. Projecting mechanics tal study of the effects of swimming on dispersal. J. Exp. Mar. into morphospace: disparity in the feeding of labrid ®shes. Proc. Biol. Ecol. 145:189±204. R. Soc. Lond. B 269:317±326. Bone, Q. 1992. On the locomotion of ascidian tadpole larvae. J. Huxley, J. S. 1932. Problems of relative growth. Dial Press, New Mar. Biol. Assoc. U.K. 72:161±186. York. Brusca, R. C., and G. J. Brusca. 1990. The invertebrates. Sinauer Jayne, B. C., and A. F. Bennett. 1989. The effect of tail morphology Associates, Sunderland, MA. on locomotor performance of snakes: a comparison of experi- Burighel, P., and R. A. Cloney. 1997. Urochordata: Ascdiacea. Pp. mental and correlative methods. J. Exp. Zool. 252:126±133. 221±347 in F. W. Harrison, ed. Hemichordata, Chaetognatha, ÐÐÐ. 1990. Selection on locomotor performance capacity in a and the invertebrate . John Wiley and Sons, New York. natural population of garter snakes. Evolution 44:1204±1229. Cameron, C. B., J. R. Garey, and B. J. Swalla. 2000. Evolution of Jeffery, W. R. 1997. Evolution of ascidian development. Bioscience the body plan: new insights from phylogenetic analyses 47:417±425. of deuterostome phyla. Proc. Natl. Acad. Sci. USA 97: Jeffery, W. R., and B. J. Swalla. 1992. Evolution of alternate modes 4469±4474. of development in ascidians. Bioessays 14:219±226. 1224 M. J. MCHENRY AND S. N. PATEK

Kingsolver, J. G., and M. A. R. Koehl. 1985. Aerodynamics, ther- Schmidt-Nielsen, K. 1972. Locomotion: energy cost of swimming, moregulation and the evolution of insect wings: differential scal- ¯ying, and running. Science 177:222±228. ing and evolutionary change. Evolution 39:488±504. Sebastian, V. O. 1953. The development of Herdmania pallida Koehl, M. A. R. 1996. When does morphology matter? Annu. Rev. (Heller). Proc. Ind. Acad. Sci. 37:174±187. Ecol. Syst. 27:501±542. Shampine, L. F., and M. K. Gordon. 1975. Computer solution of Kowalevsky, A. 1866. Enwickelungsgeschichte der einfachen As- ordinary differential equations: the initial value problem. W. H. cidien. Mem. Acad. Imp. Sci. St. Petersb. 7:1±19. Freeman, San Francisco, CA. Lamb, H. 1945. Hydrodynamics. Dover, New York. Sinervo, B., and R. B. Huey. 1990. Allometric engineering: an Lauder, G. V. 2003. The intellectual challenge of biomechanics and experimental test of the causes of interpopulation differences in evolution. Pp. 319±325 in V. L. Bels, J. P. Gasc, and A. Casinos, performance. Science 248:1106±1109. eds. Vertebrate biomechanics and evolution. BIOS Scienti®c Sinervo, B., and P. Licht. 1991. Proximate constraints on the evo- Publishers Ltd., Oxford, U.K. lution of egg size, number and total clutch mass in lizards. Sci- Liu, H., R. J. Wassersug, and K. Kawchi. 1996. A computational ence 252:1300±1302. ¯uid dynamics study of tadpole swimming. J. Exp. Biol. 199: Sokal, R. R., and F. J. Rohlf. 1995. Biometry. W. H. Freeman, New 1245±1260. York. Losos, J. B. 1990. The evolution of form and function: morphology Stach, T., and J. M. Turbeville. 2002. Phylogeny of Tunicata in- and locomotor performance in West Indian Anolis lizards. Evo- ferred from molecular and morphological characters. Mol. Phy- lution 44:1189±1203. logenet. Evol. 25:408±428. Maddison, D. R., and W. P. Maddison. 2000. MacClade 4: analysis Stearns, S. C. 1992. The evolution of life histories. Oxford Univ. of phylogeny and character evolution. Sinauer Associates, Sun- Press, Oxford. derland, MA. Stoner, D. S. 1992. Vertical distribution of a colonial ascidian on McGhee, G. R. 1999. Theoretical morphology. Columbia Univ. a coral reef: the roles of larval dispersal and life-history vari- Press, New York. ation. Am. Nat. 139:802±824. McHenry, M. J. 2001. Mechanisms of helical swimming: asym- ÐÐÐ. 1994. Larvae of a colonial ascidian use a non-contact mode metries in the morphology, movement and mechanics of larvae of substratum selection on a coral reef. Mar. Biol. 121:319±326. of the ascidian Distaplia occidentalis. J. Exp. Biol. 204: Strathmann, M. F. 1987. Reproduction and development of marine 2959±2973. invertebrates of the northern Paci®c Coast. Univ. of Washington McHenry, M. J., and J. Jed. 2003. The ontogenetic scaling of hy- Press, Seattle. drodynamics and swimming performance in jelly®sh (Aurelia Svane, I., and P. Dolmer. 1995. Perception of light at settlement: aurita). J. Exp. Biol. 206:4125±4137. a comparative study of two invertebrate larvae, a scyphozoan McHenry, M. J., and J. A. Strother. 2003. The kinematics of pho- planula and a simple ascidian tadpole. J. Exp. Mar. Biol. Ecol. totaxis in larvae of the ascidian Aplidium constellatum. Mar. Biol. 187:51±61. 142:173±184. Svane, I. B., and C. M. Young. 1989. The ecology and behavior of McHenry, M. J., E. Azizi, and J. A. Strother. 2003. The hydro- ascidian larvae. Oceanogr. Mar. Biol. Annu. Rev. 27:45±90. dynamics of undulatory swimming at intermediate Reynolds Swalla, B. J. 2001. Phylogeny of urochordates: implications for numbers in ascidian larvae (Botrylloides sp.). J. Exp. Biol. 206: chordate evolution. Pp. 219±224 in H. Sawada, H. Yokosawa, 327±343. and C. Lambert, eds. The biology of ascidians. Springer Verlag, McMahon, T. A. 1984. Muscles, re¯exes, and locomotion. Princeton Tokyo. Univ. Press, Princeton, NJ. Swalla, B. J., C. B. Cameron, L. S. Corley, and J. R. Garey. 2000. Moore, A. M. F., and O. Ellers. 1993. A functional morphospace, Urochordates are monophyletic within the deuterostomes. Syst. based on dimensionless numbers, for a circumferential, calcite, Biol. 49:52±64. stabilizing structure in sand dollars. J. Theor. Biol. 162:253±266. Thomas, G. B., and R. L. Finney. 1980. Calculus and analytic ge- Morris, R. H., D. P. Abbott, and E. C. Haderlie. 2002. Intertidal ometry. Addison-Wesley, Reading, MA. invertebrates of California. Stanford Univ. Press, Stanford, CA. Thomason, J. J. 1995. Functional morphology in vertebrate pale- Mukai, H., Y. Saito, and H. Wantanabe. 1987. Viviparous devel- ontology. Cambridge Univ. Press, Cambridge. opment in Botrylloides (Compound Ascidians). J. Morph. 193: Tokioka, T. D. 1953. Ascidians of Sagami Bay. Iwanami Shoten, 263±276. Tokyo. Niklas, K. J. 1997. Effects of hypothetical developmental barriers van Duyl, F. C., R. P. M. Bak, and J. Sybesma. 1981. The ecology and abrupt environmental changes on adaptive walks in a com- of the tropical compound ascidian Trididemnum solidum. I. Re- puter-generated domain for early vascular land plants. Paleo- productive strategy and larval behavior. Mar. Ecol. Prog. Ser. biology 23:63±76. 6:35±42. Patek, S. N., and T. H. Oakley. 2003. Comparative tests of evo- Wake, D. B., and G. Roth. 1989. Complex organismal functions: lutionary trade-offs in a palinurid lobster acoustic system. Evo- integration and evolution in vertebrates. Wiley and Sons, New lution 57:2082±2100. York. Purvis, A., and A. Rambaut. 1995. Comparative analysis by inde- Wendt, D. E. 1996. Effect of larval swimming duration on success pendent contrasts (CAIC): an Apple Macintosh application for of metamorphosis and size of the ancestrular lophophore in Bug- analysing comparative data. Comput. Appl. Biosci. 11:247±251. ula neritina (Bryosoa). Biol. Bull. 191:224±233. Raup, D. M. 1967. Geometrical analysis of shell coiling: coiling in ÐÐÐ. 2000. Energetics of larval swimming and metamorphosis ammonoids. J. Paleontol. 41:43±65. in four species of Bugula (Bryozoa). Biol. Bull 198:346±356. Raup, D. M., and A. Michelson. 1965. Theoretical morphology of Young, C. M. 1986. Direct observations of ®eld swimming behavior the coiled shell. Science 147:1294±1295. in larvae of the colonial ascidian Ecteinascidia turbinata. Bull. Roff, D. A. 1992. The evolution of life histories: theory and anal- Mar. Sci. 39:279±289. ysis. Chapman and Hall, New York. ÐÐÐ. 2002. Atlas of marine invertebrate larvae. Academic Press, Sane, S. P., and M. H. Dickinson. 2002. The aerodynamic effects New York. of wing rotation and a revised quasi-steady model of ¯apping Zar, J. H. 1999. Biostatistical analysis. Prentice-Hall, Upper Saddle ¯ight. J. Exp. Biol. 205:1087±1096. River, NJ. Satoh, N. 1994. Developmental biology of ascidians. Cambridge Univ. Press, New York. Corresponding Editor: P. Wainwright