ARTICLE IN PRESS Journal of Theoretical Biology 233 (2005) 25–42 www.elsevier.com/locate/yjtbi Overgrowth competition, fragmentation and sex-ratio dynamics: a spatially explicit, sub-individual-based model Philip H. Crowley, Christopher R. Stieha, D. Nicholas McLetchie Department of Biology and Center for Ecology, Evolution and Behavior, University of Kentucky, 101 Morgan Building, Lexington, KY 40506-0225, USA Received 10 June 2003; received in revised form 10 September 2004; accepted 15 September 2004 Abstract Sessile organisms that compete for access to resources by overgrowing each other may risk the local elimination of one sex or the other, as frequently happens within clumps of the dioecious liverwort Marchantia inflexa. A multi-stage, spatially implicit differential-equation model of M. inflexa growing in an isolated patch, analysed in a previous study, indicated that long-term coexistence of the sexes within such patches may be only temporary. Here we derive a spatially explicit, sub-individual-based model to reconsider this interpretation when much more ecological realism is taken into account, including the process of fragmentation. The model tracks temporally discrete growth increments in continuous space, representing growth architecture and the overgrowth process in significant geometric detail. Results remain generally consistent with the absence of long-term coexistence of the sexes in individual patches of Marchantia. Dynamics of sex-specific growth qualitatively resemble those generated by differential-equation models, suggesting that this much simpler framework may be adequate for multi-patch metapopulation models. Direct competition between fragmenting and non-fragmenting clones demonstrates the importance of fragmentation in overgrowth competition. The results emphasize the need for empirical work on mechanisms of overgrowth and for modeling and empirical studies of life history tradeoffs and sex-ratio dynamics in multi-patch systems. r 2004 Elsevier Ltd. All rights reserved. Keywords: Clonal plant populations; Maintenance of sex; Liverworts; Marchantia inflexa; Simulation modeling 1. Introduction Organisms of many taxa in all five kingdoms compete for access to limiting resources through a type of interference known as overgrowth competition (e.g. Monera: Jarosz, 1996; Jarosz and Kania, 2000; Protista: Airoldi, 2000; Fungi: Gourbiere et al., 2001; Plants: Matlack, 2002; Lichens: Hestmark et al., 1997; Armstrong, 2002; Animals: Connell, 1961, Barnes and Dick, 2000; Coral–algal interactions: Lirman, 2001; McCook et al., 2001; Jompa and McCook, 2002). Overgrowth competition is the process of attempting to gain and hold space linked to limiting resources by growing in a Corresponding author. Tel.: +1 859 257 1996; fax: +1 859 257 1717. E-mail address: [email protected] (P.H. Crowley). 0022-5193/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.jtbi.2004.09.017 way that physically restricts a competitor’s access to its resource supply while increasing access by the growing individual. Here, space is taken to be a surrogate limiting resource that cannot be shared at any point, suggesting that coexistence of competitor clones or species may be difficult to maintain over a contiguous area (see Connell, 1961; Buss and Jackson, 1979; McLetchie et al., 2002). Organisms that frequently engage in overgrowth competition are generally capable of sub-dividing into fully functional, physiologically separate individuals through modular processes like budding, fission, or abscission—or through fragmentation (Armstrong, 1979; During, 1990). To date, little effort has been directed at understanding the population-level implications of fragmentation and its connection to competition for space. Fragmentation is itself a taxonomically ARTICLE IN PRESS 26 P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 widespread phenomenon (Protista: Amsler, 1984; Hooper et al., 1988; Ceccherelli and Cinelli, 1999; fungi: Leslie and Klein, 1996; lichens: Hale, 1974; Bailey, 1975; Armstrong, 1979; plants: Room, 1983; Outridge and Hutchinson, 1990; During, 1990; Ewanchuk and Williams, 1996; animals: Lasker, 1990; Bruno, 1998; Zakai et al., 2000) based on partitioning of individuals into totipotent parts that may vary greatly in size. Implications of this process for competitive effectiveness are difficult to determine, because the underlying mechanisms and spatial ramifications can be complex. Comparisons of fragmentation and overgrowth competition across taxa in relation to environments and life histories may be complicated by phylogenetic constraints. Here, we focus mainly on local populations of a particular organism, the thalloid liverwort Marchantia inflexa Nees et Mont, known to compete intensively for space within patches and to fragment when partially overgrown by other individuals or other parts of the same individual (H. During, personal communication; D.N. McLetchie, unpublished observations). Our previous attempt to model within-patch sex-ratio dynamics of this intriguing species led to the formulation of a multi-stage, linear, coupled ordinary differential-equation model, consistent with the simplifying assumption that each genotype (or sex in this model) could be represented as arbitrarily small, randomly distributed increments of biomass (McLetchie et al., 2002). This avoided explicit representation of fragmentation and of the structure of individual plants (or ramets) that actually tend to produce clumped rather than random distributions of genets within patches in nature. Results of this previous study indicated that overgrowth competition made coexistence impossible, at least across the range of biologically reasonable parameter values we explored. But we wondered whether our unrealistic assumptions about the geometry of these plants, their growth dynamics, and their spatial distribution might have influenced the model’s behavior and our conclusions. In the present study, to address spatial competition within individual patches of M. inflexa more realistically, we formulated a high-resolution simulation model focusing on competition between the sexes. Parameter values were mostly estimated from our own empirical observations. Our approach involved tracking discrete growth increments of each individual and two-dimensional spatial relationships among individuals and among their growth increments, resulting in a spatially explicit, sub-individual-based (SESIB) model. This work represents a contribution to our continuing studies of sex-ratio dynamics in M. inflexa metapopulations. If indeed overgrowth competition precludes coexistence of sexes in individual patches, then maintaining both sexes in nature seems to hinge on structure and dynamics at the metapopulation level (McLetchie et al., 2002). Multi-patch models and empirical studies of the M. inflexa system are in progress and will be reported elsewhere (also see Crowley and McLetchie, 2002). Our goals in the present study were to (1) derive and examine the behavior of a SESIB patch model (March) of M. inflexa, incorporating fragmentation; (2) compare sex-ratio dynamics in the March model with those obtained for simpler spatially implicit patch models; and (3) demonstrate and account for the importance of fragmentation in overgrowth competition. We begin the remainder of the presentation with a conceptual overview of the study system and its main components, describe some important details of the model’s structure and function, present results of the simulations, and consider how the results relate to the existing literature and possible future directions. 2. Conceptual overview of the study system M. inflexa is a New World hepatic bryophyte found in the eastern USA as far north as Tennessee and in and around the Caribbean basin as far south as Venezuela (Bischler, 1984). M. inflexa is usually attached to rock or to bark of fallen trees in moist environments, including on road-cuts in areas of frequent rainfall and along permanent streams. Because it tends to occupy discrete patches vulnerable to elimination by disturbances and loosely connected by dispersal, M. inflexa populations typically form metapopulations (McLetchie and Puterbaugh, 2000; McLetchie et al., 2002; see Hanski, 1998, 1999). M. inflexa grows by extending ribbon-like thalli horizontally from terminal mericells over its substrate, which may include other thalli of the same or other individuals (Fig. 1). (Bryophytes have the single mericell, in contrast to the multiple cells of angiosperms, at their meristematic loci (Newton and Mishler, 1994)). As with many bryophytes, the species is dioecious: individuals are unisexual and gender is chromosomally determined (Ramsay and Berrie, 1982; Bischler, 1986). In a particular season or seasons, both sexes reproduce asexually by forming cupules from mericells in connection with thallus growth. Cupules eventually produce and release gemmae, asexual propagules primarily dispersed by water. In another season or seasons, sexual reproduction is initiated via formation of umbrella-like sex structures by mericells, preventing any further terminal growth of the thallus. Males release spermladen fluids onto the upper surfaces of their sex structures (antheridiophores); sperm are dispersed by ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 27 Since the published reports indicate that this single surviving sex is most often females, which in Marchantia typically have slightly higher thallus extension rates than males (McLetchie et al., 2002; D.N. McLetchie, unpublished data), we suspect that overgrowth competition may at least partly account for many instances of single-sex patches and populations. Because of its strong association with overgrowth competition, the process of fragmentation is an integral part of the relatively highresolution SESIB model presented here. Consider a single colonist ramet—a physiologically independent individual successfully germinating within a patch. Several key processes and variables interact to determine contributions to the expected lifetime reproductive success or fitness of the ramet’s clone or genet in an ephemeral patch system (Fig. 2; see Crowley and McLetchie, 2002). ‘‘Area held’’ is a state variable PHYSICAL DISTURBANCE + OVERGROWTH BY OTHER GENETS - - D AREA HELD + FITNESS A - - + FRAGMENTATION Fig. 1. The thalloid liverwort M. inflexa in a tropical rainforest in Trinidad. (a) A rock, mostly covered by a dense growth of M. inflexa, adjacent to a stream. White areas are dead tissue, probably killed by desiccation resulting from intense insolation during a dry interval. Assemblages of such semi-isolated clumps of M. inflexa on rocks in and near streams constitute metapopulations (Crowley and McLetchie 2002, McLetchie et al., 2002). (b) Close-up view of overlapping thalli representing many different independent individuals or ramets. MORTALITY + + + NET PRODUCTIVITY + + THALLUS EXTENSION REPRODUCTION (SEASONAL) TRADE-OFF + + + BIFURCATION B & SPROUTING + C - -( SEX) water droplets, resulting in possible contact with female sex structures (archegoniophores) and subsequent fertilization. Fertilized archegoniophores eventually release wind-dispersed spores, apparently at approximately a 1:1 sex ratio, and these can colonize available substrate. Some clumps of M. inflexa contain individuals of only one sex (McLetchie and Puterbaugh, 2000), either because of (1) unisexual colonization, (2) a disturbance that happened to remove one sex but not the other, or (3) the elimination of one sex by the other via overgrowth competition. In fact, some entire (meta)populations of Marchantia and other bryophytes consist only of a single sex (Schuster, 1992; Bowker et al., 2000). NUMBER OF GROWTH TIPS Fig. 2. Interactions among key features of individual ramets and their environment that determine the potential for long-term expansion (fitness) of their genotype or genet. Positive and negative effects and the growth-reproduction trade-off in the expending of net productivity are indicated. Growth, resulting in greater area coverage and more numerous growth tips, helps generate two positive feedback loops A and B. The negative feedback loop C and opposing effects of area held and growth-tip number on splitting and sprouting stabilize the area held per growth tip. Overgrowth by other genets helps create the positive feedback loop D. The predominance of these positive feedback loops underscores the unstable nature of competitive interactions in this system. ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 28 ramet or individual growth increments. Overgrowth by other ramets can be inhibited by rapid thallus extension, according to geometric relationships described in the next section. Overgrowth triggers fragmentation, as covered tissue (i.e. one or more growth increments) that falls below the energetic compensation point is eliminated, breaking the ramet into physiologically independent parts (i.e. into new ramets). Fragmentation in turn can reduce the current rate of thallus extension but helps protect the ramet as a whole from falling below the compensation point and becoming moribund, as happens when the model is modified so that overgrowth no longer elicits fragmentation (see below). The March model builds thalli as a series of linked rectangular growth increments of constant width (Fig. 3), extending from the terminal mericell of the previous rectangle to a length determined by the thallus extension rate and the time step duration. Associated with each increment are state variables and nondynamic (fixed) variables. An increment’s present status, shown on each increment in the example thallus of Fig. 3, indicates whether it can grow (if located at the end of the thallus, not overgrown at the tip, and alive, having not produced a sex structure), engage in asexual or sexual reproduction (if already in this same reproductive status—or else as for growth, except with seasonal restrictions and time limitations), become static (if it cannot grow but is alive and not engaged in reproduction), or is dead (killed by disturbance or by negative energy balance). The number of living specifying the total amount of ramet surface exposed to light (i.e. not overgrown), under the plausible assumption that M. inflexa is light-limited in the field. To keep things manageable, we ignore spatial heterogeneity of light availability (but see Caldwell and Pearcy, 1994; Oborny and Kun, 2002). Gross productivity is assumed proportional to area held; when decremented by maintenance costs, the resulting net productivity represents the energy available for partitioning into growth and reproduction. The ‘‘number of growth tips’’ is another state variable indicating the number of thallusterminal mericells available for growth. Under the assumption that growth is equally supported at all of a ramet’s growth tips (as in Oborny and Kun, 2002; see also Caraco and Kelly, 1991), this support is expressed by dividing the area held, decremented for metabolic and reproductive costs, by the number of growth tips. The resulting quotient is here termed the support ratio s; which figures prominently in determining details of the ramet’s demography, described in the next section. Growth tips, used up through the formation of sex structures by terminal mericells, are replenished through bifurcation of mericells and thalli (note the Y-shaped thallus segments in Fig. 1b) or by occasional sprouting from thallus segments without terminal mericells. In the model, as in nature, death of entire ramets can result from physical disturbances like flooding or desiccation or from complete overgrowth by other ramets. More localized physical disturbances and overgrowth may instead simply reduce the area held by a 9 GROW 8 DATA LINKED TO EACH GROWTH INCREMENT GROW STATE VARIABLES 6 STATIC 7 SEX 4 5 ASEX ASEX 3 DEAD 2 STATIC • • • • STATUS (GROW, ASEX, SEX, STATIC, DEAD) NUMBER OF LIVING PROGENY INCREMENTS AREA HELD (REDUCED WHEN OVERGROWN) TIMER (CONTROLS TIMED PROCESSES) FIXED VARIABLES • • • • IDENTIFICATION NUMBER (ID) ID NUMBER OF PROGENITOR INCREMENT LOCATION COORDINATES GENOTYPE / GENDER 1 STATIC Fig. 3. An example cartoon thallus built from rectangular growth increments as if by the March model, with the current status of each increment indicated. Black dots at the distal ends of the increments indicate the locations of active mericells (increments 8 and 9), a sexual-reproductive structure (an antheridiophore or archegoniophore—7), asexual-reproductive structures (cupules—4 and 5), and former growth points (1, 2, 3, and 6), including two bifurcating points (3 and 6). This geometry implies that increments 4 and 5 were formed by bifurcation of the mericell of increment 3 in one time step. In the next step, increments 6 and 7 grew from their respective progenitors, forming cupules at the former mericell loci of increments 4 and 5. In the final step indicated by the diagram, increments 8 and 9 formed by bifurcation from 6, while increment 7 produced a sex structure from its mericell. Note that increment 3 is now dead, either because of disturbance or extensive overgrowth; its elimination will fragment the original ramet into three independent ramets. The data linked to each growth increment that are needed to track the fate of these increments and their ramets are summarized in the box. ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 3. Model details In this study, we make the simplifying assumption that productivity per unit area held remains constant over time and space. Temporal constancy may be a rough approximation for systems like tropical rainforest metapopulations in a model with multi-day time steps that effectively average over time. Spatial constancy is also imposed primarily for tractability, but we note that other modeling studies have addressed the role of spatial resource heterogeneity in clonal plants (e.g. see Oborny, 1991; Cain et al., 1996). 3.1. Vital rates, growth, and overgrowth pe = P r Slo Extension Rate, E (cm2/tip/day) We express thallus extension rate E as an increasing decelerated function of the support ratio s; asymptotic E= Pσ 1 + Pσ / r 0 Support Ratio, σ to its maximum at r (Fig. 4a). Thus more support for a growth tip at low s generates a proportional increase in E, where the productivity P is the constant of proportionality, but diminishing returns through physiological constraints on growth at higher s causes E to flatten toward the asymptote (see Caraco and Kelly, 1991). Dividing the extension rate by the support ratio then generates the per-capita growth rate G, a declining function of s asymptotic to zero (Fig. 4b). We assume that the remainder of the productivity that is not committed to growth is used in reproduction, whenever this is seasonally appropriate. To generate the additional growth tips needed to accommodate growth at increasing ramet sizes and to replace growth tips lost to sexual reproduction, we make the simplifying assumption that the bifurcation rate is proportional to the support ratio. The intersecting growth and bifurcation curves have the effect of stabilizing the ramet’s support ratio (cf. Sánchez et al., 2004). Thalli stop growing whenever their tips are overgrown, are removed by disturbance, encounter the patch boundary, or produce a sex structure. For circumstances in which s exceeds a threshold value, thalli unable to grow (except those with an active sex structure) are assumed to produce a new mericell and thallus by sprouting (consistent with unpublished observations of sprouting by D.N. McLetchie). We assume that thallus growth in the patch can be represented as two-dimensional and that encounters between thalli result in one growing over the other (Fig. 5) but without explicit representation of the third dimension. Using this two-dimensional approximation, the area of overlap between growth increments is subtracted from the area held by the overgrown increment. Since rectangular growth increments can potentially overlap each other in many different ways, the algorithm to calculate the area of overlap, though Per-Capita Rates (1/day) ‘‘progeny’’—surviving increments that grew directly from the progenitor increment—and the identity number of the given increment’s direct progenitor must be retained so that changing characteristics of ramets can be tracked and used to drive the dynamics. Area held by each increment is continually reduced by partial or complete overgrowth. A timer tracks duration-limited physiological processes, such as gemma production by a cupule, located at the site of a former mericell. To summarize the March model’s depiction of Marchantia demography, thallus interactions and environmental constraints are expressed through the states and behavior of growth increments, which, taken together, determine the characteristics and behavior of individual ramets and entire genets. The most important of the relationships underlying these main features are described in more detail in the next section. 29 BIFURCATION P REPRODUCTION G =E / σ GROWTH P = 1 + Pσ / r 0 Support Ratio, σ Fig. 4. Key rates for ramets and their dependence on the support ratio s (cost-decremented area held per growth tip). (a) Higher support ratios increase the extension rate E of thalli at each growth tip. At low support ratios, E is approximately proportional to s, with slope determined by the productivity parameter P; toward higher support ratios, E increases with diminished returns toward an asymptotic maximum at r. The particular function used to represent these relationships is commonly employed to express response rates that depend on resource abundance, as in Michaelis–Menten enzyme kinetics (Michaelis and Menten, 1913) and the type 2 functional response of predators to prey density (Holling, 1959). (b) Per-capita growth rate G decreases with s as ramet spreading becomes growth-tip limited, and the remainder of the productivity is expended on reproduction, when this is seasonally appropriate. The rate at which growth tips bifurcate is taken to be directly proportional to the support ratio. The declining per-capita growth rate and increasing per-capita bifurcation rate tend to stabilize the support ratio. ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 30 I 3.2. Reproduction L2 II D2 L1 D1 III IV D2 L1 L2 L1 D1 ∋ θ L2 L2 V θ D2 L1 D1 Fig. 5. Determining which of two growing thalli overgrows the other, depending on the spatial relationship of their rectangular growth increments. Growth increments are based on a fixed width w but may vary in length. For each increment here, each thallus begins growing at the beginning of the time step along the bold line (the ‘‘base’’ of the increment) and reaches the other end of the rectangular increment at the end of the time step. Extension rate during the step is assumed to be constant, and the thallus that winds up on top is taken to be the one growing fastest in the direction of the other at the point of contact. This hinges in some cases on increment lengths L1 and L2, distances from base to intersection point D1 and D2, and an angular measure of relative growth direction y. There are five different categories of overgrowth geometry (identified using Roman numerals in the figure, next to an example) that can result, and each of these can arise in a number of different ways (see Appendix A). straightforward in principle, is rather complex to implement as a computer program. Determining which of two thalli overgrows the other can be more challenging. We assume that the thallus growth increment advancing faster across the other’s boundary at the initial point of contact obtains the top position (see Buss and Jackson, 1979). This immediately implies that a growing thallus always overtops a non-growing increment of another thallus. But if two different growing tips encounter each other, then relative overgrowth potential at the contact point depends on the absolute growth rates, growth directions, and other geometrical details (Fig. 5 and Appendix A; see Barnes and Dick, 2000 on the relation between the geometry of encounters and overgrowth). These calculations to identify overgrower and overgrown require the majority of the processing time when March is implemented as a computer program. During the part of the year in which asexual reproductive structures (cupules or ‘‘cups’’) can be formed (i.e. during the ‘‘asex season’’), the energy budget allocated to reproduction is used to produce as many new cups as possible from growing tips, except those that bifurcate to form two tips. Cup formation is linked to thallus extension (Parihar, 1956; Hollensen, 1981); in the model, a cup may form at the end of a previous growth increment where a new increment is being generated. There is a lag from cup formation until the asexual propagules (gemmae) are ready for release, and an interval and rate of release by the cup (taken here to be constant), after which the cup becomes inactive. During each time step that the cup is releasing gemmae, a fixed number (decremented to reflect the small chance of successful germination) are dispersed in a random direction for a normally distributed distance, having its mode at zero and mean at 0.798 times the standard deviation of the dispersal distance (i.e. at the mean distance from the mode of the normal distribution). Gemmae that land within the patch on unoccupied substrate are assumed to germinate. If not overgrown before achieving the minimum size for ‘‘emergence’’, a gemma is then recognized as a ramet having the same genotype as its progenitor and is tracked accordingly thereafter. During the sexually reproductive season, the part of the energy budget not allocated to growth, metabolic costs, or ongoing reproduction is available to form new sex structures at tips with viable mericells. There is a lag between structure formation and maturity, and what follows depends on the ramet’s gender. Mature male structures begin producing and dispersing sperm at a constant rate over a fixed interval, after which the structure becomes inactive. Within each time step, for each receptive female structure, sperm-producing male structures are randomly ordered. Each sperm-producing male structure in turn then has a probability of fertilizing the female structure that is a Gaussian function of the distance between the two structures, with a maximum at distance zero. This sequence of fertilization attempts continues until the female structure is fertilized or until all male structures have taken a turn. Female structures are receptive to fertilization over a fixed number of time steps, after which, if not fertilized, they become inactive. When fertilized, a maturation interval begins, after which the female structure releases a fixed number of spores during a single time step and becomes inactive. Spore dispersal is normally distributed with modal distance at zero and thus analogous to gemma dispersal, except that the mean dispersal distance is much longer for spores. Dispersing spores ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 are equally likely to be male or female and are assumed to initiate ramets of unique genotypes (genets) at emergence. 3.3. Mortality Individual increments and entire ramets can die in either of two ways in the March model—through a physical disturbance eliminating all increments within a certain area of the patch, or by falling into energetic deficit by being almost entirely covered. There is no mortality directly linked to increment age. Disturbances are an attempt to represent regions of local desiccation or scouring by flood water as a circle centered at a random point in the patch, with the diameter drawn from the positive half of a normal distribution centered at zero. Disturbances are also assumed to be temporally independent. Space freed by disturbance is then immediately available for occupancy by adjacent thalli and germinating propagules. We assume a fixed metabolic cost per unit area of living tissue, whether or not this tissue is covered by other tissue. This implies that once a growth increment becomes sufficiently covered, its area held becomes inadequate to compensate for the metabolic cost. In most of the runs considered here, we assume that increments in energetic deficit die and are eliminated from the model at the end of the step in which they fall into deficit. Since there seems to be a lag of days to weeks before tissue death from energetic deficit in M. inflexa (D.N. McLetchie, unpublished observations), immediate death is a simplification that may slightly speed up the competitive dynamics in these runs, but presumably without altering the outcome. When competing clones differ in the time lag from energetic deficit to death, however, there can be substantial, ecologically meaningful implications, as demonstrated for one extreme case below. Mortality resulting from disturbances and especially from overgrowth helps prevent a continual increase to unmanageable levels in the number of growth increments in the model, each with its substantial amount of associated data (Fig. 3). A complication that arises with mortality of growth increments, however, concerns the need to keep track of which growth increments constitute part of a single physiologically integrated ramet, even as ramets fragment into smaller ramets. For example in Fig. 3, the death of increment 3 creates three new ramets—one containing increments 4, 6, 8, and 9; another containing five and seven; and the other with one and two. Recording which increments are tips and which is the ‘‘base’’ of each intact ramet, along with the data retained in association with each increment (Fig. 3), permits the newly formed ramets and their tips and bases to be determined by ‘‘tip tracing’’. When an 31 increment dies within a ramet, the ramet is tip-traced by stepping systematically from each tip toward the base, identifying dead increments and defining new bases above them and tips below them. This flexible approach avoids the necessity of retaining considerably more information in association with each growth increment and further complicating the redefinition of ramets resulting from fragmentation (cf. Oborny and Kun, 2002). 4. Simulations The March model was implemented as a computer program in MATLAB 6 (see the flow chart, Fig. 6). The program can illustrate spreading, overgrowth, and reproduction by clones and disturbances in the patch as a cartoon and can produce graphs summarizing the dynamics. Because of the complexity of the geometry and extensive bookkeeping required to implement the model, runs typically lasted for hours to days on 2 GHz microprocessors. Run time increases very rapidly with patch size—as a power of the patch diameter between 3 and 4. The standard or default parameter set used in the runs considered in the present analysis is presented in Table 1. Four analyses were conducted to address the goals of the study: five runs to characterize the default condition and consider whether eliminating the disturbance regime or changing relative growth rates alters the dynamics of competition between sexes; a series of runs constituting a preliminary sensitivity analysis for some key parameters; two runs and related results from a much simpler model to examine the March model’s relationship to classical differential-equation models of competition; and a run to evaluate the role of fragmentation in overgrowth competition. 5. Results Fig. 7 shows an example growth pattern produced at day 225 from a default run of the March model, with male and female genets and their reproductive structures color-coded. Unlike in Fig. 1, disturbances remove killed increments entirely, but otherwise the simulated growth pattern is generally similar to the natural ones. The thallus colors show that the genet distributions, begun with 10 randomly distributed emerging ramets per genet, were by this point strongly patchy yet extensively interwoven. Runs of 10 years or longer were usually required for one sex (consistently females) to eliminate the other sex under default conditions, with the 0.1 m2 patch size. In the default run of Fig. 8, only a tiny part of the patch area is still occupied by the soon-to-be ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 32 INITIALIZE TALLY & UPDATE INCREMEN TS NEXT STEP PROPAGULE EMERGENCE ASSESS COSTS OF REPROD’N DISTURBANCE GROW & ADD REPRODUCTION GROWTH & REPRODUCTION LAST RAMET? NEXT RAMET RAMET LOOP START NO YES TIME-STEP LOOP MORTALITY NO DEAD INT’S? NEXT RAMET YES TIP-TRACE & FIND RAMETS PROPAGULE GERMINATION LAST RAMET? RAMET LOOP OVERLAP ADJUSTMENTS NO YES FERTILIZATION DRAW PATCH NO LAST STEP? YES STOP OUTPUT DATA Fig. 6. Flow diagram of the MATLAB computer program used to implement and explore the March model. The simulations are based on a series of time steps, during each of which new growth increments may be formed, others may be eliminated, and the demography of the entire patch is tracked. The scheme on the right provides more detail on the growth and reproduction, and mortality components of the basic within-step sequence on the left. extinguished males after 10 years. Note here that females had already gained a considerable advantage in the first year as the patch filled, and the advantage gradually widened, despite some abrupt declines in areas held that are attributable to disturbances. The parameter values for growth and for reproductive costs of males and females were set to be identical in the default run, except that unfertilized female sex structures were taken to be much less expensive to maintain than male sex structures, which while viable are always actively producing sperm (Stark et al., 2000). Removing this cost difference in other runs (not shown) removed most or all of this otherwise consistent advantage that females had in competition with males. In the middle panel of Fig. 8, note the large number of generally small ramets formed by this intensive overgrowth and fragmentation process. In the bottom panel are the numbers of propagules released (regardless of where they landed). Asexual propagules (gemmae) were released in rough proportion to area held by each genet, whereas spore production by the patch may have been as high or slightly higher with a moderately femalebiased sex ratio. Thus sperm were apparently dispersed widely enough to be in adequate supply, even with less area held by males and presumably fewer male than female sex structures. Our preliminary sensitivity analysis evaluated effects of parameter changes by factors of two on the sex ratio ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 33 Table 1 Parameters of the model Definition Default magnitude Units Basic spatial and temporal features Time step duration Length and width of (square) patch Days from July 1 until seasonal sex-to-growth transition Days from July 1 until seasonal growth-to-asex transition Days from July 1 until seasonal asex-to-sex transition 15a 31.6b 0c 75c 210c Days/step cm Days Days Days Initial abundances Number of male genets at the beginning of each run Number of female genets at the beginning of each run Number of emerging ramets/genet to start each run 1d 1d 10d Genotype Genotype Ramets/genet Growth, metabolism, and death Maximum thallus extension rate, growth season Maximum thallus extension rate, a sex season Maximum thallus extension rate, sex season Gross productivity coefficient Split rate per support ratio Standard deviation of Gaussian shift in growth direction Thallus width Baseline metabolic cost rate Angle between thalli following split Minimum support ratio for sprouting Lag between overgrowth and death of a thallus module 0.020e 0.016f 0.012f 0.200g 0.040g 0.250g 0.500h 0.005i 0.698j 1g 0k cm2/day/tip cm2/day/tip cm2/day/tip 1/day Tips/cm2/day Radians cm 1/day Radians cm2/tip Days Reproduction (general) Cost (as lost support area) per active sex structure (~ and #) Germination-to-emergence lag for sex propagule (~ and #) Area of growing thallus at emergence of ( via sex or asex) Minimum area of a ramet for reproduction (sex or asex) 0.015l 45e 0.250m 0.250e cm2/day Days cm2 cm2 Asexual reproduction Cost (as lost support area) per asex structure Lag from germination to emergence of asexual propagule Standard deviation of Gaussian gemma dispersal distance Duration of asexual propagule release from a cupule Lag from cup formation to first asexual propagule release Expected number of asex propagules germinating per cup 0.0015l 15e 30n 60e 0e 1o cm2/day Days cm Days Days Propagule Sexual reproduction (females and spores) Cost (as lost support area) per unfertilized ~ sex structure Standard deviation of Gaussian spore dispersal distance Interval when a female sex structure is fertilizable Time from formation until ~ sex structure is fertilizable Duration of spore release from fertilized ~ sex structure Time from ~ structure fertilization until first spore release Expected number of spores germinating per ~ structure 0.003l 300o 15e 0e 15e 60e 1o cm2/day cm Days Days Days Days Spore Sexual reproduction (males and sperm) Interval of sperm release from male sex structure Time from formation until # sex structure releases sperm Chance that sperm from adjacent # would fertilize ~ Maximum standard deviation of sperm dispersal distance Fraction of sx reached in 7 days 30e 15e 0.900e 2.00e 0.400p Days Days Dimensionless cm Dimensionless Disturbance Time over which the chance of a disturbance is 1/2 Standard deviation of Gaussian disturbance radius 30p 4p Days cm a Chosen arbitrarily; if modified, then s should be modified in the same direction to maintain a similar growth pattern. Corresponds to a patch area of approximately 0.1 m2, the maximum size at which individual runs can be completed in a few hours on a microcomputer. c Based on field observations for Marchantia inflexa in Trinidad (see McLetchie et al. 2002). d Arbitrary but consistent with expectation of a few colonizing propagules at roughly 1:1 offspring sex ratio. e Rough estimate based on greenhouse data (D.N. McLetchie, unpublished) and, where necessary, discretized to fit a 15-day time step (see also Ramsay and Berrie 1982; McLetchie 1996). f Rough guess based on the estimate of the growth-season maximum thallus extension rate. g Chosen to generate reasonable growth dynamics generally consistent with observed patterns. h Direct measurement; thallus width is consistently within the range 0.45–0.55 cm in this species in the greenhouse and the field (D.N. McLetchie, unpublished). i Rough guess consistent with the estimate of the gross productivity coefficient. j Directly measured; mean of 124 split angles in the greenhouse (C. Stieha, unpublished). k In the absence of data, setting this to zero allows the computer program to run faster and has little or no effect on the outcome, relative to any other fixed magnitude. l Rough guesses based on plausible relative magnitudes. m Arbitrarily chosen size above which origin from sexual vs. asexual propagules is ignored. n Rough guess, based on water transport of these hydrophobic asexual propagules. o Rough guess (cf. McLetchie et al., 2002). p Rough guess (D.N. McLetchie and P.H. Crowley, unpublished). b ARTICLE IN PRESS 34 P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 Fig. 7. Results of a default simulation in a 0.1 m2 patch observed at day 225 of year 1, early in the sexual reproductive season. Male ramets are shown in blue and females in green. The black diamond-shaped dots represent former sites of asexual reproduction; purple dots are sites of active asexual reproduction continuing from the recently ended asexual reproductive season. Black stars are male sexual structures; red asterisks are unfertilized female sexual structures; and yellow asterisks are fertilized female sexual structures. There are small areas of substrate not yet occupied by ramets at the lower corners of the patch. The light circular areas of substrate are sites of disturbances that are now partially overgrown. Though there is no reproduction in progress in Fig. 1, there are obvious similarities between that figure and this one. after 10 years. We only sketch these results here for the strongest responses, as difficulties with the long run times precluded replication. Sex ratio was considered sensitive to a factor-of-two parameter shift if the increase or the decrease (but not both) yielded a malebiased sex ratio (cf. the strongly female-biased default run in Fig. 8). These shifts were implemented simultaneously for both sexes except as noted. Dividing all seasonal maximum thallus extension rates (i.e. the asymptote r in Fig. 4a) by two led to the elimination of females within 10 years; this parameter shift corresponds to growth rate inhibition attributable to a reduction in limiting resources, but field data published to date fail to show more males at lower light irradiance (Fuselier and McLetchie, in press), as might be expected from this result. Moreover, males are strongly favored when the gross productivity coefficient (P, Fig. 4), expressing the efficiency with which a given support ratio produces thallus growth, is increased; increasing this coefficient shifts more of the available energy toward reproduction relative to growth, helping to relieve high male costs of sexual reproduction. Though sex ratio appears to be insensitive to costs of baseline metabolism or of gamete-producing sex structures, the sex ratio favors males when the cost of asexual reproduction is halved. This result is consistent with the greater investment in asexual reproduction and thus the greater benefit from reduced cost for males than for females (McLetchie and Puterbaugh, 2000); but the Fig. 8. Results of a default run for a 0.1 m2 patch (female=green, male=blue) and based on the parameter values in Table 1. Females gain a coverage advantage during the first year (i.e. the first 24 time steps) as the patch fills up. Males are almost completely eliminated after 10 years, the end of the run. Note the impact of disturbances in reducing area held at various points. As overgrowth intensifies early in the run, the total number of ramets increases sharply and then fluctuates around 1500. Annual cycles of asexual propagule (gemma) production by each sex alternate with sexual propagule (spore) production by females, shown in red. Gemma production by each sex is in general accord with relative coverage; because of relatively efficient sperm transfer and fertilization in this small patch, spore production remains fairly high until males near extinction. ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 model was not parameterized to include a difference in investment between sexes, and this result must therefore be considered with caution. Finally, it may initially seem surprising that reducing the cost of unfertilized female sex structures generates a male-biased sex ratio; here, the lower cost induces females to produce more of these structures and thus ultimately incur higher reproductive costs at the expense of growth when they are fertilized. Implications of some additional deviations from the default parameter set are explored in Fig. 9. The top panel shows that removing disturbance has very little effect on the outcome or even in the basic dynamics. In small patches with small-scale disturbances, like those considered in this analysis, disturbances actually provide little or no opportunity for propagules to emerge and survive the onslaught from peripheral thalli (cf. McLetchie et al., 2002; see also Watson, 1981; Kimmerer, 1991). Because of this, the reduction in 35 maximum thallus extension rate of males depicted in the second panel, though permitting additional asexual reproduction in exchange for reduced growth, further weakens the males’ ability to compete for space. If instead, males are given a slight growth advantage during the season of sexual reproduction (third panel), this was enough to shift the balance completely in their favor. Combining the parameter shifts from the second and third panels in the run depicted in the bottom panel generated a more balanced competition, though the trend over the final 4 years of the run seems clearly to favor females. The basic patterns observed in Figs. 8 and 9 generally resemble those resulting from two-species competition equations similar to the classical Volterra equations in the absence of stable coexistence (Volterra, 1926; Hutchinson, 1978). The failure of a Volterra-type model incorporating both seasonality and disturbances to yield Fig. 9. Area held vs. time in four different simulations of the March model for a 0.1 m2 patch containing two genets (female=green, male=blue). In all cases, runs are based on the default parameter set (Table 1), except as indicated below. (a) Here there were no environmental disturbances, resulting in smoother graphs but with trends similar to those in Fig. 8, in which disturbances were present. (b) In this run, the maximum thallus extension rate for males during the asexual reproductive season was reduced from 0.016 to 0.015 cm2/day/tip, slightly reducing thallus growth but allowing a slight compensatory increase in asexual reproduction. Here, males were completely eliminated during year 7. (c) When the maximum thallus extension rate for males during the sexual reproductive season was increased from 0.012 to 0.013 cm2/day/tip, males soon dominated the patch, and females appeared likely to face eventual elimination. (d) When the maximum thallus extension rate for males was both reduced from 0.016 to 0.015 cm2/day/tip in the asexual reproductive season and increased from 0.012 to 0.013 cm2/day/tip in the sexual reproductive season, combining the changes made in (b) and (c), results were somewhat more ambiguous, though females seemed to be gaining the advantage by the end of the run. ARTICLE IN PRESS 36 P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 coexistence in previous work (McLetchie et al., 2002) had led us to expect this outcome, particularly in patches too small for frequent propagule germination. The resemblance to the outcome for the simple two-species system is considered more systematically in Fig. 10. Handicapping males by reducing their growth rates slightly made the patterns from 2-year March runs somewhat easier to visualize (first two panels). This disadvantage is obviously accelerating the elimination of males relative to default dynamics (Fig. 8). The more erratic pattern in the top panel of Fig. 10 was smoothed out considerably in the middle panel, an otherwise identical run with disturbances and reproductive seasons eliminated. These smoothed curves generate a highly repeatable pattern in which overgrowth proceeds more rapidly and the males are almost eliminated at the end of the run. The bottom panel was obtained by constructing coupled linear ordinary differential equations mathematically similar to the Volterra equations, but based on the specifics of space-limited growth (see the legend of Fig. 10 and Crowley et al., in press for details; also see McLetchie et al., 2002). Dynamics of the coupled linear differential equations, with parameters chosen to match those of the non-seasonal, no-disturbance March model Fig. 10. Comparing clonal dynamics in the March model to a linear model of two-clone overgrowth competition that is mathematically equivalent to a two-species competition model closely related to the classical Volterra model. (a) Results of a 2-year March run under default conditions, except that the maximum thallus extension rates for males (blue line) were reduced to 80% of their default values. (b) Another 2-year March run identical to the one above, except that there are no disturbances or reproductive seasons. (c) Results of simulating two linear, coupled ordinary differential equations, namely df 1 =f 1 dt ¼ r1 ð1 f 1 f 2 Þ þ ðr1 r2 Þf 2 d 1 and df 2 =f 2 dt ¼ r2 ð1 f 1 f 2 Þ þ ðr2 r1 Þf 1 d 2 where fi is the fraction of the patch area held by clone i, ri is the per-capita expansion rate, and di is the per-capita death rate. In each equation, the first term on the right-hand side represents expansion into unoccupied space, the second term is net overgrowth, and the third term is disturbance-induced mortality (Crowley et al., in press). For this simulation, r1=0.3, r2=0.33, d1=d2=0 (i.e. no disturbances), f1 and f2 were initialized at 0.004 (corresponding to starting conditions for the March runs), and both areas held and time were rescaled to match the March runs. The ri values were chosen to produce a pattern resembling the previous panel; their absolute magnitudes are arbitrary under temporal scaling, and their relative magnitudes are consistent with plausible relative per-capita increase rates in corresponding March run. ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 as closely as possible, did achieve a strong resemblance, though with noticeable differences. The most obvious differences were that the sexes in the March model diverged more slowly in coverage as the patch filled but then closed in more rapidly on extinction of the loser— in comparison to results for the differential-equation model parameterized to mimic the initial growth phase of the March run. We provide an interpretation of this result in the Discussion. To test directly whether fragmentation was advantageous in overgrowth competition as represented by the March model, we conducted a run that included a fragmenting genet (blue) and a non-fragmenting genet (green) in the absence of reproductive seasons and disturbance (Fig. 11). In the top panel, the nonfragmenter gained an advantage in area held as the patch filled during the first year, at least partly by not eliminating increments that were mostly but not entirely overgrown. (Other factors contributing to this difference are the focus of work in progress based on this model. During the second year (i.e. between time steps 24 and 49), the non-fragmenting ramets apparently fell below the compensation point and were completely overgrown by ramets from the fragmenting clone. The bottom panel confirms the absence of fragmentation in the nonfragmenting clone (green) and the presence of fragmentation, with subsequent accumulation of large numbers of ramets, in the fragmenting clone (blue). The number of fragmenter ramets increased rapidly around the time 37 of patch filling but soon leveled off during an interval in which new ramets were being overgrown about as rapidly as they were formed. In the second year, the number of ramets from the fragmenting clone began to increase again and overwhelm the increasingly moribund non-fragmenting ramets. 6. Discussion We have formulated and implemented the SESIB patch model March and explored some of its major features. Under standard (default) conditions that did not include any growth-rate advantage per se, female M. inflexa consistently vanquished males in overgrowth competition and dominated the 0.1 m2 patches convincingly within a decade. This is consistent with infrequent-disturbance results obtained by McLetchie et al. (2002) for somewhat larger patches using a spatially implicit model of patch dynamics. As previously postulated (McLetchie et al., 2002), the female advantage seems to result primarily from the higher cost of mature male sex structures than mature but unfertilized female sex structures. Results for sexual investment in a moss are generally consistent with this interpretation (Stark et al., 2000); other recent estimates of the costs of sexual reproduction in bryophytes (Bisang and Ehrlen, 2002; Rydgren and Okland, 2002) unfortunately focus exclusively on females. Fig. 11. Competition between a clone (blue) that fragmented by immediately eliminating its overgrown increments falling below the compensation point and a clone (green) that did not fragment. These results are for a 0.1 m2 patch in the absence of disturbances or reproductive seasons, with all other parameters at default values. ARTICLE IN PRESS 38 P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 In the small patches largely invulnerable to intrapatch propagule establishment that are the focus of the present analysis, spatially and temporally random disturbances have little effect on the dynamics or outcome of competitive interactions between genders, other than the stochastic variation introduced by these occasional die-backs. In contrast, even shifts of a few percent in relative growth rates between sexes in any of the non-reproductive or reproductive growth seasons were enough to shift the advantage decisively. The male advantage found by McLetchie et al. (2002) with a spatially implicit model at high-disturbance frequencies reflected increased emergence success of asexual propagules produced in larger numbers by males. But when the high vulnerability of these new ramets, obvious in the spatially explicit March model, is taken into account, this advantage largely disappears. Large disturbances in especially large patches (say X10 m2) may provide an exception not investigated here, in which reproductive propagules can at least occasionally germinate and establish sufficiently to withstand overgrowth pressure from the disturbance periphery. In the absence of disturbance and seasons, the dynamics of space coverage by male and female genets in the March model were fairly well mimicked by a pair of coupled linear ordinary differential equations that are mathematically similar to the classical Volterra equations. These equations amount to a greatly simplified version of the McLetchie et al. (2002) model—without disturbance, reproduction, or stage structure (see Crowley et al., in press). But the rate of overgrowth is slower in the March model while the two sexes are roughly equal in coverage than in the differentialequation models, because the genet patchiness inherent in the March model (and presumably in Marchantia distributions in nature) generates more within-genet overgrowth. Once one genet is heavily dominant, however, the other genet is likely to be fragmented into small, weakly growing ramets surrounded by more vigorous competitors that may overwhelm them relatively rapidly. These more subtle spatial effects are difficult or impossible to capture realistically in an analytical model, providing one rationale for spatially explicit models like March. Nevertheless, the basic similarities in dynamics resulting from the two approaches suggest that linear differential-equation systems may be adequate to represent within-patch dynamics in the context of multi-patch metapopulation models. The key conclusion of the McLetchie et al. (2002) analysis, that coexistence of male and female M. inflexa in individual patches is only temporary, is consistently upheld by results for the much more detailed and realistic representation of overgrowth competition in the March model. Work in progress focuses more inten- sively on the coupled two-equation system and some ecologically meaningful relatives to probe the boundaries of this generalization (Crowley et al., in press). Other studies are addressing multi-patch systems to account for long-term coexistence of the sexes through metapopulation structure and dynamics (Garcı́aRamos et al., in preparation; see also Crowley and McLetchie, 2002). The direct competition between fragmenting and nonfragmenting genets demonstrated that non-fragmenters may achieve a temporary advantage with initial patch filling (see also Fig. 2 in Oborny et al., 2000), but the near certainty that non-fragmenting ramets must eventually fall into energetic deficit by being overgrown (as in the example run) dooms them to highly probable elimination. This means that species engaging commonly in overgrowth competition should be capable of fragmentation, modular abscission, or sloughing of overgrown tissue. In fact, the March model can be used to understand the fragmentation process more thoroughly, accounting for the consistent delay in the sloughing of overgrown tissue by Marchantia and for the initial advantage of the non-fragmenter in competition with the fragmenter. Harper and Bell (1979); Bell (1984), and Sutherland (1990) noted the considerable but largely untapped potential for simulation models to improve our understanding of the spatial distribution and dynamics of clonal plant populations. By spreading across space and separating into independent but genetically identical parts through asexual propagules or via abscission or fragmentation, these intriguing organisms raise important questions about spatial ecology and evolution. Some of the relevant modeling work to date has used a deterministic (e.g. Room, 1983; Callaghan et al., 1990) or stochastic (e.g. Remphrey et al., 1983; Callaghan et al., 1990; Kron and Stewart, 1994) approach to characterize branching architecture and speculate about how particular growth ‘‘rules’’ and geometries might be adaptive. Other studies have specifically addressed growth form and dynamics as a solution to the problem of plant foraging across a spatially heterogeneous resource distribution (e.g. Sutherland and Stillman, 1988; Oborny, 1994; Cain et al., 1996) or of contending with disturbances distributed randomly in space and time (Inghe, 1989). More recent work has begun to evaluate the adaptive significance of clonal integration (Caraco and Kelly, 1991; Oborny and Kun, 2002) and the implications of overgrowth competition (McLetchie et al., 2002; Crowley and McLetchie, 2002). Spatially implicit (or pseudospatial) models have permitted spatially distributed dynamics to be tractably approximated in some cases (Cain, 1994; McLetchie et al., 2002; Crowley and McLetchie, 2002). Two-dimensional cellular automata methodology (Wolfram, 1984), based on variable-state ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 nodes in discrete time and space, have been used successfully to evaluate spatial strategies (e.g. Inghe, 1989; Oborny and Kun, 2002) from a spatially explicit perspective. Relative to these previous studies and approaches, the March model provides a particularly high level of biological realism and spatial resolution as a spatially continuous representation of local ecological dynamics in a clonal system. Competition for space, particularly where holding space provides access to some demonstrably limiting resource, is obviously common in nature. Overgrowth competition is a type of interference that in effect maintains distinct boundaries separating regions held by one sessile competitor from those held by others. Moreover, the means of shifting these boundaries involves the physical expansion of one competitor such that the other is locally excluded from the resource supply. This is a process of overtopping in the typical case when the resource is light or food particles streaming down from above, though other geometries having the same basic effect can be imagined, such as mussels and barnacles that gradually exert enough lateral force to crush and remove immediate neighbors (Connell, 1961). In the absence of spaceclearing disturbances, this scenario boils things down to which competitor can advance the frontier in its favor, once the patch fills up, a recipe for the eventual elimination of one or the other in the absence of frequency dependence. Disturbances, though, allow for colonization by propagules and uncontested growth into cleared spaces (spatial exploitation) to become important (Airoldi, 2000, and references therein). We emphasize that the patch model of McLetchie et al. (2002), the March model, and the differential equations of the Fig. 10 legend all assume that both uncontested growth and overgrowth are based on the same growth rates (the ri of the Fig. 10 legend). Since modifying that assumption in particular ways can permit stable coexistence of competitors (Crowley et al., in press), we are currently planning empirical work to determine whether this mechanism might apply to the M. inflexa system. M. inflexa, along with other taxa with separate sexes that engage in intense clonal overgrowth competition, provide an unusually clear-cut and extreme example of male–female conflict (Parker, 1979; Westneat and Sargent, 1996). Here, natural selection seems to overwhelm sexual selection: the complete loss of access to the other sex as a result of its local extirpation prevents sexual reproduction, perhaps (as in Marchantia) without even eliminating the significant costs of sexual expression. Under the assumptions of the March model, male–female conflict is ameliorated during the sex season and even during the asex season, because growth is assumed to be reduced somewhat to permit significant amounts of energy to be expended on reproduction. 39 Nevertheless, for one sex or the other (usually males in M. inflexa) this seems to confer only a temporary stay of execution, and the fugitive sex seems ultimately to depend on metapopulation dynamics for its long-term persistence (McLetchie et al., 2002). Since metapopulations generally contain patches at different stages of colonization, substrate coverage, and disturbance, even a few sexually reproducing 2-sex patches may provide enough spores for both sexes to reach unoccupied habitat and thus persist. Gender-specific detection ability has to our knowledge not been demonstrated in bryophytes. But if males and females could detect each other on thallus contact, avoiding overgrowth of the opposite sex would nevertheless seem unlikely to be evolutionarily stable, since a mutant without such avoidance would benefit in competition for space. If males could detect each other on thallus contact, both natural and sexual selection would favor an enhanced overgrowth effort or other interference effect, but this could also prove to be evolutionarily unstable if males could arise that were able to foil detection by mimicking females. More plausibly, if ramets were able to detect the absence of the opposite sex (e.g. females cued hormonally that sex structures remain unfertilized), at least some expenditures could be shifted away from sex expression to asex (as an alternative means of colonizing other patches) or to growth. Of course, even without any such detection ability, natural selection should act to reduce sexual expression in favor of growth at the patch level; at the metapopulation level, however, where spore dispersal and patch colonization are paramount, natural selection must favor sexuality. An interesting implication worth pursuing is that the loss of sex by populations may result primarily from infrequent disturbances or insufficiently severe disturbances; a benign environment may not conduce to the long-term maintenance of sexual reproduction. The results of this analysis have suggested a number of issues already under study, including more thorough consideration of fragmentation, detailed analysis of simple differential-equation models of overgrowth, and multi-patch metapopulation models of sex-ratio dynamics. Others worth addressing in future work are empirical studies of thallus overgrowth and of the relationship between extension rate and support ratio. Also, the tradeoffs among growth, asex, and sex in the March model need to be thoroughly investigated for larger patches and related to our own empirical results (McLetchie and Puterbaugh, 2000; D.N. McLetchie, in preparation). Ultimately, understanding sex-ratio dynamics will hinge on being able to account for how these life history tradeoffs play out in the multipatch systems characteristic of nature (see the conceptual approaches of Ronce et al., 2000 and Crowley and McLetchie, 2002). ARTICLE IN PRESS 40 P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 Acknowledgements We thank Heinjo During and the UK Wort Group (particularly Linda Fuselier, Gisela Garcı́a-Ramos, Charles Richardson, Carey Snyder, and Nicole Sudler) for helpful suggestions during this study and Gisela Garcı́a-Ramos for insightful comments on an earlier draft of the manuscript. PHC gratefully acknowledges the hospitality of Isabelle Olivieri, Ophelie Ronce and colleagues for hosting his visit to the Institute of Evolution at the Université de Montpellier, France, during May, 2002, where this work and many related ideas were discussed to good effect. We acknowledge support for this project from NSF grant DEB 9974086. Appendix A. Which thallus overgrows the other? Fig. 5 illustrates five general categories of overgrowth geometry for increments of fixed width w and constant extension rate during the interval of increment formation. There are several different ways or ‘‘cases’’ by which each geometric category can arise. Distinguishing the categories and determining the outcome requires the algorithm to establish and compare growth orientations (via the metric y) and lengths L1 and L2 of growth increments. When sides of the two increments intersect, the distances D1 and D2 from an intersection point to the nearest base point of each increment may become essential to this determination as well. (When there is more than one side–side intersection, the one associated with the earliest point of contact during the time step is the relevant one.) To identify the two thalli and their growth increments here, we arbitrarily refer to the increment oriented directly toward the top of the figure as 1, and the other increment as 2. When the two thalli are growing in directions within 901 of each other, then category I (20 distinguishable geometric patterns or ‘‘cases’’) or II (eight cases) applies. In this situation, if one of the bases is intersected (indicative of category I), then the increment having the intersected base is overgrown (two in the example) and the other (1) is on top. In this particular example, this is because 1 necessarily reaches the base of 2 at some point during the time step and can simply overgrow it without opposition. For category II (growth directions within 901 but neither base is crossed), whichever thallus reaches the initial intersection point first will be overgrown by the later arriving thallus. If D1 =L1 oD2 =L2 ; then 1 must reach this point first and be overgrown, because a smaller fraction of the time step would have passed for 1 to reach the point than for 2 to reach it; otherwise, ignoring the special case in which the two fractions are exactly equal, 1 overgrows 2. When the two thalli are growing in directions that are not within 901 of each other, then category III (20 cases), IV (five cases), or V (18 cases) applies. In this situation, the absence of side–side intersection of the growth increments implies category III, as does a side–side intersection in which there is overlap of growth increments between both closest base corners and the intersection point (as in the example). Here, an increment in position 1 would overgrow 2 if and only if L1 cos y4L2 : This is because, at the point of first contact during the step, 2 is growing straight over 1, whereas the vector component of 1’s growth in the direction of 2 is diminished by the multiplicative factor cos y. Category IV results from a side–side intersection in which there is overlap between one closest base point (on 1 in the example) and the intersection point—but not between the other base point (on 2) and the intersection point. In the diagram, 1 overgrows 2 if and only if D1 =L1 4D2 =L2 and L1 cos y4L2 ; following logic similar to that for categories II and III. Finally, category V involves a side–side intersection in which there is overlap between neither closest base corner and the intersection point. Here, there are two ways 1 could wind up on top: either {D1 =L1 4D2 =L2 and L1 4L2 cos y} or {D2 =L2 4D1 =L1 and L1 cos y4L2 }. Otherwise, 2 is on top. References Airoldi, L., 2000. Effects of disturbance, life histories, and overgrowth on coexistence of algal crusts and turfs. Ecology 81, 798–814. Amsler, C.D., 1984. Culture and field studies of Acinetospora crinita (Carmichael) Sauvageau (Ectocarpaceae, Phaeophyceae) in North Carolina, USA. Phycologia 23, 377–382. Armstrong, R.A., 1979. Growth and regeneration of lichen thalli with the central portions artificially removed. Environ. Exp. Bot. 19, 175–178. Armstrong, R., 2002. The effect of rock surface aspect on growth, size structure and competition in the lichen Rhizocarpon geographicum. Environ. Exp. Bot. 48, 187–194. Bailey, R.H., 1975. Ecological aspects of dispersal and establishment in lichens. In: Brown, D.H., Hawksworth, D.L., Bailey, R.H. (Eds.), Lichenology: Progress and Problems. Academic Press, London, UK. Barnes, D.K.A., Dick, M.H., 2000. Overgrowth competition in encrusting bryozoan assemblages of the intertidal and infralittoral zones of Alaska. Mar. Biol. 136, 813–822. Bell, A.D., 1984. Dynamic morphology: a contribution to plant population ecology. In: Dirzo, R., Shanikhan, J. (Eds.), Perspectives on Plant Population Ecology. Sinauer Associates, Sunderland, MA. Bisang, I., Ehrlen, J., 2002. Reproductive effort and cost of sexual reproduction in female Dicranum polysetum. Bryologist 105, 384–397. Bischler, H., 1984. Marchantia L. The New World species. Bryophytorum Biobliotheca 26. Bischler, H., 1986. Marchantia polymorpha L. s. lat. Karyotype analysis. J. Hattori Bot. Lab. 60, 105–117. Bowker, M.A., Stark, L.R., McLetchie, D.N., Mishler, B.D., 2000. Sex expression, skewed sex ratios, and microhabitat distribution in the dioecious desert moss Syntrichia caninervis (Pottiaceae). Am. J. Bot. 87, 517–526. ARTICLE IN PRESS P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 Bruno, J.F., 1998. Fragmentation in Madracis mirabilis (Duchassaing and Michelotti): how common is size-specific fragment survivorship in corals? J. Exp. Mar. Biol. Ecol. 230, 169–181. Buss, L.W., Jackson, J.B.C., 1979. Competitive networks: nontransitive competitive relationships in cryptic coral reef environments. Am. Nat. 113, 223–234. Cain, M.L., 1994. Consequences of foraging in clonal plant species. Ecology 75, 933–944. Cain, M.L., Dudle, D.A., Evans, J.P., 1996. Spatial models of foraging in clonal plant species. Am. J. Bot. 83, 76–85. Caldwell, M.M., Pearcy, R.W. (Eds.), 1994, Exploitation of Environmental Heterogeneity by Plants. Academic Press, Inc., San Diego, CA. Callaghan, T.V., Svensson, B.M., Bowman, H., Lindley, D.K., Carlsson, B.A., 1990. Models of clonal plant growth based on population dynamics and architecture. Oikos 57, 257–269. Caraco, T., Kelly, C.K., 1991. On the adaptive value of physiological integration in clonal plants. Ecology 72, 81–93. Ceccherelli, G., Cinelli, F., 1999. The role of vegetative fragmentation in dispersal of the invasive alga Caulerpa taxifolia in the Mediterranean. Marine Ecol. Progr. Ser. 182, 299–303. Connell, J.H., 1961. The influence of interspecific competition and other factors on the distribution of the barnacle Chthalamus stellatus. Ecology 42, 710–723. Crowley, P.H., Davis, H.M., Ensminger, A.L., Fuselier, L.C., Jackson, J.K., McLetchie, N.M., 2005. A general model of local competition for space. J. Theor. Biol., in press. Crowley, P.H., McLetchie, D.N., 2002. Trade-offs and spatial lifehistory strategies in classical metapopulations. Am. Nat. 159, 190–208. During, H.J., 1990. Clonal growth patterns among bryophytes. In: van Groenendael, J.deKroon,H. (Ed.), Clonal Growth in Plants. SPB Academic Publishing, The Hague, The Netherlands, pp. 153–176. Ewanchuk, P.J., Williams, S.L., 1996. Survival and re-establishment of vegetative fragments of eelgrass (Zostera marina). Can. J. Bot. 74, 1584–1590. Fuselier, L., McLetchie, D.N., 2004. Microhabitat and sex distribution in Marchantia inflexa, a dioicous liverwort. Bryologist 107, 345–356. Gourbiere, F., van Maanen, A., Debouzie, D., 2001. Associations between three fungi on pine needles and their variation along a climatic gradient. Mycol. Res. 105, 1101–1109. Hale, M.E., 1974. The Biology of Lichens. Arnold, London, UK. Hanski, I., 1998. Metapopulation dynamics. Nature 396, 41–49. Hanski, I., 1999. Metapopulation Ecology. Oxford University Press, Oxford, UK. Harper, J.L., Bell, A.D., 1979. The population dynamics of growth form in organisms with modular construction. In: Anderson, R.M., Turner, B.D., Taylor, L.R. (Eds.), Population Dynamics. Blackwell, Oxford, UK. Hestmark, G., Schroeter, B., Kappen, L., 1997. Intrathalline and sizedependent patterns of activity in Lasallia pustulata and their possible consequences for competitive interactions. Funct. Ecol. 11, 318–322. Hollensen, R.H., 1981. A gemmiparous population of Marchantia polymorpha var. aquatica in Cheboygan county, Michigan. Michigan Bot. 20, 189–191. Holling, C.S., 1959. Some characteristics of simple types of predation and parasitism. Can. Entomol. 91, 385–398. Hooper, R.G., Henry, E.C., Kuhlenkamp, R., 1988. Phaeosiphoniella cryophila gen. et sp. nov., a third member of the Tilopteridales (Phaeophyceae). Phycologia 27, 395–404. Hutchinson, G.E., 1978. An Introduction to Population Ecology. Yale University Press, New Haven, CN. 41 Inghe, O., 1989. Genet and ramet survivorship under different mortality regimes—a cellular automata model. J. Theor. Biol. 138, 257–270. Jarosz, J., 1996. Do antibodies and compounds produced in vitro by Xenorhabdus nematophilus minimize the secondary invasion of insect carcasses by contaminating bacteria? Nematologica 42, 367–377. Jarosz, J., Kania, G., 2000. The question of whether gut microflora of the millipede Ommatoiulus sabulosus could function as a threshold to food infections. Pedobiologia 44, 705–708. Jompa, J., McCook, L.J., 2002. Effects of competition and herbivory on interactions between a hard coral and a brown alga. J. Exp. Mar. Biol. Ecol. 271, 25–39. Kimmerer, R.W., 1991. Reproductive ecology of Tetraphis pellucida II. Differential fitness of sexual and asexual propagules. Bryologist 94, 284–288. Kron, P., Stewart, S.C., 1994. Variability in the expression of a rhizome architecture model in a natural population of Iris versicolor (Iridaceae). Am. J. Bot. 81, 1128–1138. Lasker, H.R., 1990. Clonal propagation and population dynamics of a gorgonian coral. Ecology 71, 1578–1589. Leslie, J.F., Klein, K.K., 1996. Female fertility and mating type effects on effective population size and evolution in filamentous fungi. Genetics 144, 557–567. Lirman, D., 2001. Competition between macroalgae and corals: effects of herbivore exclusion and increased algal biomass on coral survivorship and growth. Coral Reefs 19, 392–399. Matlack, G.R., 2002. Exotic plant species in Mississippi, USA: critical issues in management and research. Nat. Areas J. 22, 241–247. McCook, L.J., Jompa, J., Diaz-Pulido, G., 2001. Competition between corals and algae on coral reefs: a review of evidence and mechanisms. Coral Reefs 19, 400–417. McLetchie, D.N., 1996. Sperm limitation and genetic effects on fecundity in the dioecious liverwort Sphaerocarpos texanus. Sex. Plant Reprod. 9, 87–92. McLetchie, D.N., Puterbaugh, M.N., 2000. Population sex ratios, sexspecific clonal traits and tradeoffs among these traits in the liverwort, Marchantia inflexa. Oikos 90, 227–237. McLetchie, D.N., Garcı́a-Ramos, G., Crowley, P.H., 2002. Local sexratio dynamics: a model for the dioecious liverwort Marchantia inflexa. Evol. Ecol. 15, 231–254. Michaelis, L., Menten, M.L., 1913. Die Kinetik der Invertinwirkung. Biochem. Z. 49, 333–369. Newton, A.E., Mishler, B.D., 1994. The evolutionary significance of asexual reproduction in mosses. J. Hattori Bot. Lab. 76, 127–145. Oborny, B., 1991. Criticisms on optimal foraging in plants: a review. Abstr. Bot. 15, 67–76. Oborny, B., 1994. Growth rules in clonal plants and predictability of the environment: a simulation study. J. Ecol. 76, 807–825. Oborny, B., Kun, A., 2002. Fragmentation of clones: how does it influence dispersal and competitive ability? Evol. Ecol. 15, 319–346. Oborny, B., Kun, A., Czaran, T., Bokros, S., 2000. The effect of clonal integration on plant competition for mosaic habitat space. Ecology 81, 3291–3304. Outridge, P.M., Hutchinson, T.C., 1990. Effects of cadmium on integration and resource allocation in the clonal fern Salvinia molesta. Oecologia 84, 215–223. Parihar, N.S., 1956. Bryophyta. Indian Universities Press, Allahabad, India. Parker, G.A., 1979. Sexual selection and sexual conflict. In: Blum, M.S., Blum, N.A. (Eds.), Sexual Selection and Reproductive Competition in Insects. Academic Press, New York, NY, pp. 123–166. Ramsay, H.P., Berrie, G.K., 1982. Sex determination in bryophytes. J. Hattori Bot. Lab. 52, 255–274. ARTICLE IN PRESS 42 P.H. Crowley et al. / Journal of Theoretical Biology 233 (2005) 25–42 Remphrey, W.R., Neal, B.R., Steeves, T.A., 1983. The morphology and growth of Arctostaphylos uva-ursi (bearberry): an architectural model simulating colonizing growth. Can. J. Bot. 61, 2451–2458. Ronce, O., Perret, F., Olivieri, I., 2000. Landscape dynamics and evolution of colonist syndromes: interactions between reproductive effort and dispersal in a metapopulation. Evol. Ecol. 14, 233–260. Room, P.M., 1983. ‘Falling apart’ as a lifestyle: the rhizome architecture and population growth of Salvinia molesta. J. Ecol. 71, 349–365. Rydgren, K., Okland, R.H., 2002. Ultimate cost of sporophyte production in the clonal moss Hylocomium splendens. Ecology 83, 1573–1579. Sánchez, J.A., Lasker, H.R., Nepomuceno, E.G., Sánchez, J.D., Woldenberg, M.J., 2004. Branching and self-organization in marine modular colonial organisms: a model. Am. Nat. 163, E24–E39. Schuster, R.M., 1992. Volume VI. The Hepaticae and Anthocerotae of North America. Field Museum of Naural History, Chicago, IL. Stark, L.R., Mishler, B.D., McLetchie, D.N., 2000. The cost of realized sexual reproduction: assessing patterns of reproductive allocation and sporophyte abortion in a desert moss. Am. J. Bot. 87, 1599–1608. Sutherland, W.J., 1990. The response of plants to patchy environments. In: Shorrocks, B., Swingland, I.R. (Eds.), Living in a Patchy Environment. Oxford University Press, Oxford, UK, pp. 45–61. Sutherland, W.J., Stillman, R.A., 1988. The foraging tactics of plants. Oikos 52, 239–244. Volterra, V., 1926. Variazioni e fluttuazioni del numero d’individui in specie animali conviventi. Mem. R. Acad. Naz. dei Lincei (ser. 6) 2, 31–113. Watson, M.A., 1981. Chemically mediated interactions among juvenile mosses as possible determinants of their community structure. J. Chem. Ecol. 7, 367–376. Westneat, D.F., Sargent, R.C., 1996. Sex and parenting: the effects of sexual conflict and parentage on parental strategies. TREE 11, 87–91. Wolfram, S., 1984. Universality and complexity in cellular automata. Physica D 10, 1. Zakai, D., Levy, O., Chawick-Furman, N.E., 2000. Experimental fragmentation reduces sexual reproductive output by the reef-building coral Pocillopora damicornis. Coral Reefs 19, 185–188.
© Copyright 2026 Paperzz