Testing Weissman's Lineage Selection Model for the Maintenance of Sex: The Evolutionary Dynamics of Clam Shrimp Reproduction over Geologic Time

© 2020 Academia Sinica, Taiwan Open Access Special Issue: Fossil and Modern Clam Shrimp (Branchiopoda: Spinicaudata, Laevicaudata) Testing Weissman’s Lineage Selection Model for the Maintenance of Sex: The Evolutionary Dynamics of Clam Shrimp Reproduction over Geologic Time Timothy I. Astrop1,*, Lisa Park Boush2, and Stephen C. Weeks3 1Fossil Forest Project, Blast Road, Brymbo, Wales, United Kingdom, LI11 5BT. *Correspondence: E-mail: [email protected] (Astrop) 2Department of Geosciences, University of Connecticut, Storrs, CT 06269-1045, USA. E-mail: [email protected] (Park Boush) 3Department of Biology, The University of Akron, Akron, OH 44325-3908, USA. E-mail: [email protected] (Weeks) Received 11 June 2020 / Accepted 15 June 2020 / Published 5 August 2020 Special issue (articles 32-46) communicated by Thomas A. Hegna and D. Christopher Rogers One of the most perplexing questions within evolutionary biology is: “why are there so many methods of reproduction?” Contemporary theories assume that sexual reproduction should allow long term survival as dispersal and recombination of genetic material provides a population of organisms with the ability to adapt to environmental change. One of the most frustrating aspects of studying the evolution of reproductive systems is that we have not yet been able to utilize information locked within the fossil record to assess breeding system evolution in deep time. While the fossil record provides us with information on an organism’s living environment, as well as some aspects of its ecology, the preservation of biological interactions (reproduction, feeding, symbiosis, communication) is exceedingly rare. Using both information from extant taxa uncovered by a plethora of biological and ecological studies and the rich representation of the Spinicaudata (Branchiopoda: Crustacea) throughout the fossil record (from the Devonian to today), we address two hypotheses of reproductive evolutionary theory: (1) that unisexual species should be short lived and less speciose than their outcrossing counterparts and (2) that androdioecy (mixtures of males and hermaphrodites) is an unstable, transitionary system that should not persist over long periods of time. We find no evidence of all-unisexual spinicaudatan taxa (clam shrimp) in the fossil record, but do find evidence of both androdioecious and dioecious clam shrimp. We find that clades with many androdioecious species are less speciose but persist longer than their mostly dioecious counterparts. These data suggest that all-unisexual lineages likely do not persist long whereas mixtures of unisexual and sexual breeding can persist for evolutionarily long periods but tend to produce fewer species than mostly sexual breeding. Key words: Evolution of sex, Sexual dimorphism, Morphometrics, Androdioecy, Chonchostraca. Citation: Astrop TI, Park Boush L, Weeks SC. 2020. Testing Weissman’s lineage selection model for the maintenance of sex: the evolutionary dynamics of clam shrimp reproduction over geologic time. Zool Stud 59:34. doi:10.6620/ZS.2020.59-34. BACKGROUND The functional significance of outcrossing sexual reproduction has intrigued biologists from the very inception of evolutionary biology. Darwin (1859) mused that organisms that refrained from outcrossing sexual reproduction would “diminish vigour and fertility” and that “no organic being self-fertilises itself for an eternity of generations; but that a cross with another individual is occasionally—perhaps at very long intervals— indispensable.” Indeed, Weismann (1889) elaborated on this idea by suggesting that “all species with purely parthenogenetic reproduction are sure to die out; not, indeed, because of any failure in meeting the existing Zoological Studies 59:34 (2020) doi:10.6620/ZS.2020.59-34 1 © 2020 Academia Sinica, Taiwan conditions of life, but because they are incapable of transforming themselves into new species, or, in fact, of adapting themselves to any new conditions.” The notion that outcrossing sexual reproduction allowed species to persist in the face of changing environments persisted for more than a century before being challenged as being a “group selection” argument by Williams (1966) in his seminal book. Since then, a plethora of “individual selection” hypotheses have been generated to replace Weismann’s hypothesis (Williams 1975; Bell 1982; Kondrashov 1993), most of which have not been successfully borne out (Otto 2009). The lack of definitive support for short-term benefits to outcrossing sexual reproduction has logically led to theoretical reexamination of Weissman’s ideas (termed the “lineage selection” model) as a mechanism to maintain sexual reproduction (Nunney 1989; Burt 2000; de Vienne et al. 2013). Even though the notion that unisexual lineages should be evolutionarily short-lived and less speciose than their outcrossing sexual counterparts (Weismann 1889; Fisher 1930; Muller 1932 1964) is indeed a “group selection” hypothesis, it may nevertheless either partially or wholly explain the predominance of sexual reproduction in the plant and animal kingdoms (Nunney 1989 1999; Burt 2000; de Vienne et al. 2013). Because of the long time frames dictated by the lineage selection model, empirical tests of this hypothesis have only been indirect; the scarcity of major clades of wholly or predominantly unisexual lineages—for example ostracods (Schön et al. 2009), oribatid mites (Norton et al. 1988), spinicaudatan ‘clam shrimp’(Weeks et al. 2009) and bdelloid rotifers (Arkhipova and Meselson 2000; Welch et al. 2004; Fontaneto et al. 2007 2012)— has been cited as indicative of the lineage-selection model (Bell 1982). In order to test such a temporally-dependent hypothesis, we would need a readily fossilized clade that is reproductively labile and from which breeding system type can be assessed. Unfortunately, to date, determination of reproductive mode of fossils has been problematic, and in those taxa that show sexual dimorphism [e.g., ammonites (Longridge et al. 2008; Zatoń 2008), ostracods (Ozawa 2013), and vertebrates (Klein et al. 2012)], reproductive mechanisms are often invariant, disallowing empirical comparison. Because of this, palaeontological tests of the long-term benefits of sexual reproduction in multicellular organisms (Weismann 1889; Fisher 1930; Muller 1932 1964) have been impossible. There is one clade—branchiopod crustaceans in the suborder Spinicaudata (Fig. 1)—that does fit the above criteria. These clam shrimp exhibit a diversity of reproductive systems: dioecy (males + females), androdioecy (males + hermaphrodites) and selfing hermaphroditism (Sassaman 1995; Brantner et al. 2013; Weeks et al. 2014). Additionally, unisexuality (i.e., selfing hermaphroditism) has independently evolved a minimum of four times from dioecious ancestors (Weeks et al. 2014). These crustaceans readily fossilize, and have a rich fossil record that is well established (Raymond 1946; Novojilov and Kapeljka 1960; Tasch and Shaffer 1964; Zhang et al. 1976; Tasch 1987; Gallego and Martins-Neto 2006; Kozur and Weems 2007; Astrop and Hegna 2015; Hethke et al. 2019). Recent methodological breakthroughs (Astrop et al. Fig. 1. The limnadiid spinicaudatan Calalimnadia mahei. b: Brood chamber with eggs, h: Head, p: Phyllopodous thoracic limbs, t: Telson. page 2 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan 2012) have allowed the extraction of sex ratio estimates from fossil clam shrimp to ascertain reproductive systems in fossil populations of Spinicaudata (Monferran et al. 2013; Stigall et al. 2014). The combination of these factors allows us a unique opportunity to test Weissman’s original hypothesis: that unisexual species should be short lived and less speciose than their outcrossing counterparts (Nunney 1989; de Vienne et al. 2013). Herein, we use Astrop et al.’s (2012) shape comparison methodology to assess sex ratios in fossil clam shrimp allowing us to assign mating systems to fossil species in a reproductively diverse taxon: the Spindicaudata (Weeks et al. 2008). We then use these analyses to directly address Weissman’s (1889) original predictions that unisexual species should be less speciose and shorter-lived than their dioecious counterparts, as well as to assess the prediction that mixtures of males and self-compatible hermaphrodites (androdioecy) should be short-lived (Charlesworth 1984). MATERIALS AND METHODS Sampling for this study was conducted at multiple museums and repositories across the world (Tables 1 and 2). Specimens were processed using a ‘portable imaging station’ which comprised a Nikon D3000, macro-lens, tripod, lighting, laptop computer and image capture/editing software. Morphometric Protocol The outlines of individual carapaces were digitized using tpsDig v2.10 (Rohlf 2006) and then subjected to standard eigenshape analysis. The protocol and proof of concept utilized in this study is covered in depth in Astrop et al. (2012). A brief description of the methodology follows. Eigenshape analyses (sensu MacLeod 1999) operate via the conversion of the digitized outline of an individual specimen into equidistant, Cartesian (xy) coordinates. These coordinates are subjected to a generalized Procrustes analysis (GPA sensu Bookstein 1996 1997) in order to remove the effect of size, location and rotation and allow the data to projected into a two dimensional space. The Procrustes-aligned coordinates are then transformed into a shape function as angular deviations (phi function: φ; Zahn and Roskies 1972) from the previous step (coordinate) in order to describe the shape of the curve. This description is derived from a set of empirical, orthogonal shape functions via an eigenfunction analysis of a matrix of correlations between shapes. Eigenshape ‘scores’ can be then used to project individual specimens into a multidimensional morphospace that allows the visualization of individual vectors of shape change and highlight whether particular vectors of deviation from the ‘mean shape’ are characteristic of a particular group. Digitized outline data was then processed using modified versions of the Eigenshape v2.6 & Guide to Models v0.7 Mathematica notebooks available via the morphotools site (http://www.morpho-tools.net). The analysis interpolates and standardizes the raw Cartesian data before performing a singular value decomposition to produce eigenvalues, eigenscores and eigenshapes that describe variation of shape within the dataset. Size is removed from the analysis as eigenshape axis one which is manually discarded and the second eigenshape reported by the analysis is treated as the ‘true’ first eigenshape (ES1) describing shape change. The eigenshapes produced by the analysis describe two-dimensional axes of shape change that can be Table 1. Number of fossil species represented by adequate numbers to be of use in this study from visited institutions Collection # of viable species CONICET 7 NHM 5 SMNH 8 PIN 2 NIGPAS 4 AMNH 3 Institution abbreviations: CONICET, National Research Council Scientific and Technical, Corrientes, Argentina. NHM, Natural History Museum, London, UK. SMNH, Smithsonian Museum of Natural History, DC, USA. PIN, Paleontological Institute, Russian Academy of Sciences, Moscow. NIGPAS, Nanjing Institute of Geology and Palaeontology Chinese Academy of Sciences, China. AMNH, Australian Museum of Natural History, Sydney, Australia. page 3 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan used to construct morphospaces that specimens may be projected onto, allowing trends in shape variation to be observed. The eigenscores can then be used in a simple cluster analysis to evaluate the existence of morphotypes that should correspond to sex. The current study employed some changes to the protocol outlined in Astrop et al. (2012). Astrop (2014) found that using 10 rather than 500 equidistant points reduced the likelihood that taphonomic and/or human error would enter the analysis while extracting the same level of useful shape information from the carapaces of extant Spinicaudata. Thus, we used this less “noisy” method herein. Evolutionary Context In order to provide a framework for interpreting the evolutionary dynamics of sexual systems over geologic time, hypothesized relationships between extinct taxa were based on Zhang et al. (1976). Trees were constructed manually in Mesquite (V2.75) based on existing literature (e.g., Novojilov 1961; Zhang et al. 1976; Chen and Hudson 1991) to produce files in a nexus format that were manageable by the R language environment and associated phylotools package (Revell 2012). Unfortunately, most phylogenetic methods and reconstructions do not take into account terminal taxa becoming extinct before the present or the sampling error intrinsic to palaeontological data. Thus, in these analyses, the R package paleotree (Bapst 2012) was utilized, which allowed for time-scaling of branches in the tree and testing for serious issues in assuming the data collected are representative of the actual diversity of the fossil group. Statistical tests regarding the distribution and duration of sexual systems in fossil groups were performed in R and PAST (Hammer et al. 2001) RESULTS Monomorphism vs. Dimorphism In order to establish the presence of different reproductive phenotypes, we must first establish that the methods outlined in Astrop et al. (2012) can effectively discriminate monomorphism (i.e., parthenogenesis Table 2. Metadata of fossil material used in this study. Institution abbreviations same as in table 1 Species Familial affiliation Collection Specimen # Age Useful Eigenshapes % variance captured Carapacestheria disgregaris Eosestheriidae (Shen 1994) NHM London, Ohio University (OU), SMNH NHM it2566-81 Jurassic 1,2 72% Martinestheria (Lioestheria) codoensis Antronestheriidae (Gallego et al. 2013) Argentina Uncurated Lower Cretaceous 1,2,3 49% Challaolimnadiopsis mendozaensis Eosestheriidae (Sensu Zhang et al., 1976)1 Argentina Uncurated Triassic 1,2,3 72% Wolfestheria smekali Fushunograptidae (Wang) in Hong et al. 1974 Argentina Uncurated Upper Jurassic 1,2,3 36% Menucoestheria wichmanni Eosestheriidae (Zhang et al., 1976) Argentina Uncurated lower Upper Triassic 2,3 20% Leaia gondwanella Leaiidae (Raymond 1946) SMNH usnm426155 Mid-Upper Permian 1,2 34% Estheria forbesi (all) Eosestheriidae (Sensu Zhang et al., 1976)1 NHM London, Argentina NHM in44340 - 51883, ARG “New stuff in tissue”, TA1-TA7 Triassic 1,2 59% Cyzicus (Euestheria) crustapatulis Euestheriidae1SMNH usnm427800/06 & usnm427807 Lower Jurassic 1,2 30% Eosolimnadiopsis santacrucensis Eosestheriidae (Sensu Zhang et al., 1976)1 Argentina Uncurated Jurassic 1,2 67% Lioestheria malacaraensis Fushunograptidae (Gallego et al. 2011) Argentina, SMNH usnm427989 Jurassic 1,2,3,4 90% Euestheria taschi Euestheriidae (Monferran et al., 2013) Argentina 5718 middle Late-Jurassic 1,2,3,4 91% Estheria mangliensis Euestheriidae1NHM London NHM in4961 - 35274 Upper-Triassic 1,2 68% Euestheria mangliensis (?) Euestheriidae1Argentina Uncurated middle Late-Jurassic 1,2 71% Estheria mangaliensis Euestheriidae1SMNH Uncurated Upper-Triassic 1,2 54% Triassoglypta sp. 3 Loxomegaglyptidae (Novojilov 1958) Argentina Uncurated Late Triassic 1,2,3 82% Estheria middendorffi Euestheriidae1NHM London NHM in9262 - un-cataloged Upper Cretaceous 1,2 62% Leaia leidyi Leaiidae (Raymond 1946) NHM London NHM in3088-3114 Lower Carboniferous 1,2 61% page 4 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan Species Familial affiliation Collection Specimen # Age Useful Eigenshapes % variance captured Cyzicus (Euestheria) formavariabalis Eosestheriidae (Sensu Zhang et al., 1976)1 SMNH usnm426198 Lower Jurassic 1,2 68% Cyzicus (Euestheria) crustabundis Eosestheriidae (Sensu Zhang et al., 1976)1 SMNH usnm427901/4+985 Lower Jurassic 1,2 59% Cyzicus (Lioestheria) antarctis Eosestheriidae (Sensu Zhang et al., 1976)1 SMNH usnm426177 Lower Jurassic 1,2 58% Perilimnadia sp. Perilimnadiidae (Sensu Zhang et al., 1976)1 AMNH Tray L34-CO1 Upper Permian 1,2 59% Hemicyclolaeia mitchelli/ discoidea Leaiidae (Raymond 1946) AMNH Tray L34-CO3 Upper Permian 1,2 62% Cyzicus (Lioestheria) branchocarus Euestheriidae1AMNH Tray L34-CO5 Cretaceous 1,2 58% Estheria simoni Euestheriidae1PIN Uncurated Upper Carboniferous 1,2 73% Limnadia volgaica Palaeolimnadiidae (Sensu Tasch 1956)1 PIN 2141/1 Upper Permian 1,2 61% Eosestheria luanpingensis Eosestheriidae (Zhang et al., 1976) NIGPAS 97438-57 Early Cretaceous 1,2 76% Neodiestheria changmaensis Diestheriidae (Chen) in Zhang et al. 1976 NIGPAS 45564-45566 Early Cretaceous 1,2,3 71% Dictyestheria elongata/ ovata Halysestheriidae (Zhang et al., 1976) NIGPAS Uncurated Upper Cretaceous 1,2,3 69% Halysestheria yui Halysestheriidae (Zhang et al., 1976) NIGPAS Uncurated Upper Cretaceous 1,2 63% Species Tot N N = M1 %M1 N = M2 %M2 Predicted sexual system2 Carapacestheria disgregaris 34 17 50.0 17 50.0 D Martinestheria (Lioestheria) codoensis 15 7 46.7 8 53.3 D Challaolimnadiopsis mendozaensis 14 7 50.0 7 50.0 D Wolfestheria smekali 33 14 42.4 19 57.6 D Menucoestheria wichmanni 23 9 39.1 14 60.9 D Leaia gondwanella 16 6 37.5 10 62.5 D Estheria forbesi (all) 116 63 54.3 54 46.6 D Cyzicus (Euestheria) crustapatulis 33 10 30.3 23 69.7 A Eosolimnadiopsis santacrucensis 48 17 35.4 31 64.6 A Lioestheria malacaraensis 55 38 69.1 17 30.9 A Euestheria taschi 20 9 45.0 11 55.0 D Estheria mangliensis 38 13 34.2 25 65.8 A Euestheria mangliensis (?) 38 13 34.2 25 65.8 A Estheria mangaliensis 61 33 54.1 28 45.9 D Triassoglypta sp. 3 28 11 39.3 17 60.7 D Estheria middendorffi 34 11 32.4 23 67.6 A Leaia leidyi 30 10 33.3 20 66.7 A Cyzicus (Euestheria) formavariabalis 13 / / / / N/A Cyzicus (Euestheria) crustabundis 17 / / / / N/A Cyzicus (Lioestheria) antarctis 17 8 47.1 9 D Perilimnadia sp. 28 / / / / N/A Hemicyclolaeia mitchelli/discoidea 45 12 26.7 33 73.3 A Cyzicus (Lioestheria) branchocarus 41 21 51.2 20 48.8 D Estheria simoni 17 7 41.2 10 58.8 D Limnadia volgaica 19 9 47.4 10 52.6 D Eosestheria luanpingensis 37 17 20 54.1 D Neodiestheria changmaensis 60 28 46.7 32 53.3 D Dictyestheria elongata/ovata 99 N/A Halysestheria yui 69 33 47.8 36 52.2 D 1suggested change. 2A = androdioecy; D = dioecy. Table 2. (Continued) page 5 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan or all hermaphrodites) from dimorphism (dioecy or androdioecy). Thus, we began this study by subjecting known sexes of differing combinations to analysis: a sample of 15 males and 15 hermaphrodites of the extant androdioecious Eulimnadia texana Packard, 1871 was used as the “dimorphic” population and 15 hermaphrodites as the “monomorphic” population. The first two eigenshape axes contained 82% of the variation for the dimorphic dataset and 57% of the variation observed in the monomorphic dataset. Hierarchical cluster analyses of these respective datasets (Fig. 2) revealed two groups separated by long branch lengths (relative to disparity between either cluster’s eigenshape scores) in the dimorphic dataset (Fig. 2B) whereas in the monomorphic dataset (Fig. 2A) branch lengths were considerably lower and did not show the distinct grouping seen in the dimorphic dataset. Thus, the method implemented by Astrop et al. (2012) can successfully distinguish between a monomorphic vs. a dimorphic data set. Fossil Comparisons A total of 29 species of fossil Spinicaudata represented by between 30–200 individuals were used in these analyses (Tables 1–3). Individual fossil specimens from collections were deemed viable if there was little to no visible taphonomic interference in the outline of the preserved carapace valve (approximately 30% of observed specimens were of suitable preservation). Small shape variations in individual specimens are described by the eigenshape analysis as ‘non-affine’ or non-uniform. This non-uniform variation is likely to be relegated to lower eigenshapes as ‘noise’ whereas more uniform or ‘affine’ shape change, that is, trends in shape change seen across specimens in the dataset, comprised the majority of variance captured by higher eigenshapes. A total of 1,098 specimens from 29 species (Table 2) were analyzed using the morphometric protocol outlined above. Observing the branching patterns in the extant (Fig. 2) examples and fossil examples (Fig. 3), similarities and differences are clear. The fossil taxon Lioestheria malacaraensis Tasch 1987 (Fig. 3B) definitively displays a strong basal dichotomy in shape variation of a magnitude similar to that seen in the dimorphic dataset of the extant Eulimnadia texana (Fig. 2B). Conversely Palaeolimnadia sp. (Fig. 3A) exhibits no clear clusters, reminiscent of the monomorphic dataset of E. texana (Fig. 2A). A major difference between the patterns seen in Palaeolimnadia sp. versus that seen in the monomorphic E. texana data is the size of the Euclidean distance between specimens. This distance measure is an effective way of discerning groups because data contained in the vectors are all in the same physical units (a measure of disparity in shape, with size, scaling and rotation removed). The distance between specimens in the Palaeolimnadia sp. dataset (Fig. 3A) is of an order of magnitude higher than that seen in the monomorphic E. texana dataset (Fig. 2A) and is very similar to distance measures in other dimorphic taxa studied (Astrop et al. 2012). This can be simply interpreted as there being very little difference in shape between individuals in the monomorphic E. texana data set and differences in shape between multiple individuals in the Palaeolimnadia sp. data set of a magnitude similar to Fig. 2. Cluster analyses of a monomorphic (hermaphrodites only) sample of Eulimnadia texana (A) and a dimorphic (males + hermaphrodites) sample (B) based on scores of individuals along the first four eigenshape axes. Note the differing distances along the Y-axes in the two graphs. page 6 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan that seen in the dimorphic data. Therefore, we interpret these patterns in Palaeolimnadia sp. (Fig. 3A) as either reflective of the presence of multiple species in a collection labeled as a single species, or as being caused by severe taphonomic interference in these specimens. We found these patterns in four of the 29 (~14%) taxa examined: Dictyestheria elongata / D. ovata Chang and Chen 1963, Palaeolimnadia sp., Cyzicus (Euestheria) Table 3. Fossil clam shrimp measured Name Reference Carapacestheria disgregaris Tasch 1987 Challaolimnadiopsis mendozaensis Shen et al. 2001 Cyzicus (Euestheria) crustabundis Tasch 1987 Cyzicus (Euestheria) crustapatulis Tasch 1987 Cyzicus (Euestheria) formavariabalis Tasch 1987 Cyzicus (Lioestheria) antarctis Tasch 1987 Cyzicus (Lioestheria) branchocarus Talent 1965 Dictyestheria elongata/ovato Chang and Chen 1964 Euestheria luanpingensis Zhang et al. 1990 Eosolimnadiopsis santacrucensis Gallego 1994 Estheria forbesi all Jones 1862 Estheria mangaliensis L3 Jones 1862 Estheria mangliensis L1 Jones 1862 Estheria middendorffi Jones 1862 Estheria simoni Pruvost 1911 Euestheria mangliensis L2 Jones 1862 Euestheria taschi Vallati 1986 Halysestheria yui Chang 1957 Hemicylcolaeia discoidea/mitchelli Mitchell 1925; Etheridge 1892 Leaia gondwanella Tasch 1987 Leaia leidyi Lea 1855 Limnadia volgaica Novojilov 1970 Lioestheria malacaraensis Tasch 1987 Martinsestheria codoensis Cardoso 1962 Menucoestheria wichmanni Gallego 2010 Neodiestheria changmaensis Shen and Chen 1982 Paleolimnadia sp. Tasch and Oesterlen 1977 Triassoglypta sp. 3 Gallego 2005 Wolfestheria smekali Mongerran et al. 2013 Fig. 3. A, Cluster analysis of Palaeolimnadia sp. based on informative eigenshapes (ES1 & 2); no discernible major groupings and multiple long branches of similar length interfere with deducing sexual system based on a clear morphotype ratio. B, Cluster analysis of Lioestheria malacaraensis based on informative eigenshapes (ES1 & 2) exhibiting a clear basal dichotomy with branch lengths much larger than any subsequent groupings. page 7 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan formavariabalis Tasch 1987 and Cyzicus (Euestheria) crustabundis Tasch 1987. Thus, for these taxa, no sexual system could be inferred. For the remaining taxa, morphotypes were assigned when cluster analysis of the informative eigenshape scores either produced two distinct groups (such as in Figs. 2B and 3B) or a single grouping (such as in Fig. 2A). We found that none of the remaining 25 taxa showed a pattern indicative of a single sex (i.e., as in Fig. 2A). Instead, all 25 taxa had two distinct groupings separated by large Euclidian distances, as seen in figures 2B and 3B. Thus, these analyses resulted in two sexes being present among the samples of each of these 25 taxa. Actual sexes may be assumed for the two distinct morphotypes in two ways. First, although naturally occurring extant populations often display fluctuating frequencies of sexes (often seasonally), the average ratio over the season remains indicative of the sexual system employed by the species (50:50—dioecious; 30:70 male: “female”—androdioecious). This has been recorded in extant, wild populations of dioecious species such as Cyzicus tetracerus (Krynicki 1830) (Popović and Gottstein-Matočec 2006), Leptestheria nobilis Sars, 1900 (Karande and Inamdar 1959) and the androdioecious species E. texana (Strenth 1977). When considering ecological observations, alongside the fact that fossil-bearing strata usually represent multiple generations living and dying over time, obtaining morphotype (or ‘sex’) ratios for a fossil taxon from such a deposit should reflect the ratio representative of the sexual system of that taxon. Second, rare instances of soft part preservation have allowed the matching of claspers (male copulatory appendages) to specific carapace shapes; for instance, soft part preservation in the Jurassic euestheriid Euestheria luanpingensis (Zhang et al. 1990), where claspers are associated with more elongated sub-quadrate carapace shapes and eggs are preserved within sub-spherical carapaces. Fossil taxa were interpreted as displaying a sex ratio indicative of androdioecy if one morphotype comprised less than 35% of the sample. This percentage was used because it is close to the 30/70 ratio (male/ female respectively) that is observed in most extant androdioecious species (Weeks et al. 2008) but leaves some room for sampling error. Examples of morphotypes in fossil species can be seen in figure 4. Seven of the 25 fossil taxa included in this analysis exhibited distinct morphotypes with a skewed frequency where the less common morphotype made up 35% of the sample or less. These taxa occurred in three of the nine families studied (Fig. 5): the Leaiidae, Fushunograptidae and Euestheriidae. Two of three taxa in the Leadiidae, one of two taxa in the Fushunograptidae and four of eight taxa in the Euestheriidae exhibited androdioecious sex ratios. Interestingly, where androdioecy was suggested in a fossil family, it seemed to occur in at least half of the species sampled in that family (Fig. 5). Polytomies in the tree presented in figure 5 originate from uncertain intra-familial relationships inferred by Zhang et al. (1976) and revised by Astrop and Hegna (2015). However, by time-calibrating the tree using the software package Paleotree (Bapst 2012), it was possible to bound first occurrences in the fossil record to branches and resolve polytomies according to (in this case) the range of geologic stages through which the genera occur (Fig. 6). This revised analysis adds information that would otherwise be lost and that is often ignored in modern phylogenetic studies that incorporate extinct taxa. Time-scaling the tree shows that androdioecious lineages have occurred multiple times since the Devonian. The durations of the branches (Fig. 6) are reflective of the first and last known occurrences of species within that genus in the fossil record. It is clear that the fossil clam shrimp have two distinct clades both originating ~300 mya (Fig. 6). In this case, clade A has 15 species and only a single case of a skewed sex ratio (Lioestheria malacaraensis). The breeding system of three of the species in this clade were not determined (Dictyestheria elongata / D. ovata, Cyzicus (Eustheria) formavariabalis and Cyzicus (Eustheria) crustabundis), but the remaining 11 species were determined to be dioecious. Thus, the range of dioecy possible for this clade is ~73–93% dioecious. Clade A is determined to be a primarily dioecious (PD) clade. Clade B has only seven species, of which four are androdioecious. Thus, this clade is 57% androdioecious. Clade B is considered the primarily androdioecious (PA) clade. We can compare the PD to the PA clade in two ways: number of species per clade and average species duration in the fossil record. For the former, we used a binomial test to assess the likelihood of equal numbers of species being distributed in the two clades because the clades appear to be approximately equally old (300–320 my). The likelihood that the two clades are actually equally speciose is 0.041, and thus the PD clade has significantly more species than the PA clade. Using the species duration estimates from figure 6, we found that the PA average duration (57.4 ± 8.7 MY) was significantly longer (F1,20 = 11.45; P < 0.0030) than the PD average duration (21.9 ± 5.9 MY) indicating that the PD clade has more, shorter-lived species than the PA clade (Fig. 6). page 8 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan DISCUSSION Although the theory that sexual reproduction has been maintained because it allows organisms to adapt to evolutionarily rapid changes in the environment (Weismann 1889) has been largely discounted because it was determined to be “group-selectionist” (Williams 1966; Maynard Smith 1971 1978), an effective empirical test of this theory in metazoans has been impossible until now. Herein, we have made the first such test, using fossil species from the reproductively labile clam shrimp as our study organism. From these comparisons, we have made two important discoveries: (1) all-female/ hermaphrodite species were not observed in the fossil data studied (2) there were several fossil lineages with sex ratios indicative of androdioecy, and those species were non-randomly divided into two clades that differed in average species duration and species number. Fig. 4. A sample of the fossil taxa studied, their diagnosed sexual systems and overlaid mean-shapes of the detected morphotypes (M1 = Morphotype 1; M2 = Morphotype 2). page 9 of 16Zoological Studies 59:34 (2020) © 2020 Academia Sinica, Taiwan Namibia. Palaeontology 57:127–140. doi:10.1111/pala.12055. Strenth NE. 1977. Successional variation of sex ratios in Eulimnadia texana Packard (Crustacea, Conchostraca). Southwest Nat 22:205–212. Talent JA. 1965. A new species of conchostracan from the Lower Cretaceous of Victoria. Proc Roy Soc Victoria 79:197–203. Tasch P. 1956. Three general principles for a system classification of fossil chonchostracans. J Paleontol 30:1248–1257. Tasch P. 1987. Fossil Conchostraca of the southern hemisphere and continental drift. Paleontology, biostratigraphy, and dispersal. Geological Society of America, Boulder, CO, USA. Tasch P, Oesterlen PM. 1977. New data on the “Phyllopod beds” (Karroo System) Northern Angola. South Central Geological Society America, Annual Meeting, El Paso Abstracts with Program: 77. Tasch P, Shaffer BL. 1964. Conchostracans: Living and fossil from Chihuahua and Sonora, Mexico. Science 143:806–807. Vallati P. 1986. Conchostracos jurásicos de la Provincia de Chubut, Argentina. In: Congreso Argentino de Paleontología y Bioestratigrafía, No. 4, Actas 4:29–38. Weeks SC, Brantner JS, Astrop TI, Ott DW, Rabet N. 2014. The evolution of hermaphroditism from dioecy in crustaceans: Selfing hermaphroditism described in a fourth spinicaudatan genus. Evol Biol 41:251–261. doi:10.1007/s11692-013-9265-0. Weeks SC, Chapman EG, Rogers DC, Senyo DM, Hoeh WR. 2009. Evolutionary transitions among dioecy, androdioecy and hermaphroditism in limnadiid clam shrimp (Branchiopoda: Spinicaudata). J Evol Biol 22:1781–1799. doi:10.1111/j.14209101.2009.01813.x. Weeks SC, Posgai RT, Cesari M, Scanabissi F. 2005. Androdioecy inferred in the clam shrimp Eulimnadia agassizii (Spinicaudata: Limnadiidae). J Crustacean Biol 25:323–328. doi:10.1651/ C-2555. Weeks SC, Reed SK, Cesari M, Scanabissi F. 2006. Production of intersexes and the evolution of androdioecy in the clam shrimp Eulimnadia texana (Crustacea, Branchiopoda, Spinicaudata). Invertebr Reprod Dev 49:113–119. doi:10.1080/07924259.2006. 9652200. Weeks SC, Sanderson TF, Zofkova M, Knott B. 2008. Breeding systems in the clam shrimp family Limnadiidae (Branchiopoda, Spinicaudata). Invert Biol 127:336–349. doi:10.1111/j.17447410.2008.00130.x. Weismann A. 1889. The significance of sexual reproduction in the theory of natural selection. In: Poulton EB, Schonland S, Shipley AE (eds) Essays Heredity Kind. Clarendon, Oxford. Welch JLM, Welch DBM, Meselson M. 2004. Cytogenetic evidence for asexual evolution of bdelloid rotifers. Proc Natl Acad Sci USA 101:1618–1621. doi:10.1073/pnas.0307677100. Williams GC. 1966. Adaptation and Natural Selection. A Critique of Some Current Evolutionary Thought. Princeton University Press, Princeton, NJ, USA. Williams GC. 1975. Sex and Evolution. Princeton University Press, Princeton, NJ, USA. Zahn CT, Roskies RZ. 1972. Fourier descriptors for plane closed curves. IEEE T Comput 100:269–281. Zatoń M. 2008. Taxonomy and palaeobiology of the Bathonian (Middle Jurassic) tulitid ammonite Morrisiceras. Geobios 41:699–717. Zhang W, Shen Y, Niu S. 1990. Discovery of Jurassic conchostracans with well-preserved soft parts and notes on its biological significance. Palaeontol Cathayana 5:311–351. doi:10.1007/9783-662-12662-2_14. Zhang WT, Chen PJ, Shen YB. 1976. Fossil Conchostraca of China. Science Press, Beijing, China. page 16 of 16Zoological Studies 59:34 (2020)