NINE
Life History and Reproductive Ecology
CONTENTS
Life-History Patterns and the Allocation of Energy Resources
Age-Specific Survivorship and Reproduction
Offspring Number, Size, and Spawning Frequency
To Travel or Not: Spawning Migrations
Sex, Mating, and Investment in Offspring
THE PREVIOUS TWO CHAPTERS in this unit dealt with body form and function and included information on evolutionary trends in locomotion, effects of natural selection on body shape, basic jaw structure and function, and morphological specializations associated with prey capture. This chapter continues the study of form and function in relation to reproduction and the role of natural selection in shaping life-history patterns. In fact, evolution of body size and shape of fishes can often be best understood in terms of the selective pressures of life histories.
Fishes demonstrate a fascinating array of reproductive modes and life-history patterns and perhaps show the greatest range of variation among all vertebrates (Winemiller 1995). Also, in contrast to birds and mammals, most fishes continue to increase in body length and mass after reaching sexual maturity (termed indeterminate growth) so that there is a tradeoff between funneling production into somatic growth or into reproduction (Charnov et al. 2001). Although most fish species are gonochoristic (meaning that sexes are separate and sex is genetically fixed), there are various examples of unisexual species and species that change sex as a function of age, environment, or social conditions. Reproductive modes vary from fishes with external fertilization, where the eggs are fertilized and develop totally outside the body of the female (termed oviparity), to the opposite extreme of internal fertilization, with varying degrees of maternal nutrient contributions to the eggs (termed viviparity). Parental care of eggs and young spans the range from none to the extreme of viviparity, with numerous intermediate examples of nest construction, nest guarding, and oral brooding. Among these categories there is a huge range of variation among different taxa (reviewed by Balon 1975, 1981).
LIFE-HISTORY PATTERNS AND THE ALLOCATION OF ENERGY RESOURCES
Life-history studies of fishes are concerned with such variables as lengths of spawning seasons, single and annual clutch sizes, sizes of embryos and larvae, degree of parental care, age at first reproduction, reproductive lifetime, adult body size, ova diameter and energy storage, maximum longevity, the presence or absence of spawning migrations, and the relationship of adult versus larval and juvenile survival (Stearns 1976, 1977; Gotelli and Pyron 1991; Winemiller and Rose 1992). The actual life history shown by an organism is the distillation of competing demands or trade-offs on growth, survival, and reproduction. An ideal organism—a “Darwinian Demon”—would, in terms of reproduction, maximize reproductive performance throughout all life-history stages irrespective of any constraints (Partridge and Harvey 1988). However, in nature an organism must deal with the realities of historical (phylogenetic) constraints, physiological constraints, demographic constraints, and ecological constraints of obtaining sufficient food for growth and reproduction while minimizing exposure to predation (including parasitism), all accomplished within the context of long- and short-term environmental variation (Gotelli and Pyron 1991; Winemiller and Rose 1992). The outcome of these interactions is the development of a specific life history for an organism that includes traits shaped by natural selection in response to local ecological factors; however, as also pointed out in previous chapters, the actions of natural selection need not have occurred in the terminal taxon but could just as well have occurred earlier in the evolution of a lineage and represent synapomorphies (shared, derived characters) (Stearns 1976, 1977; Gotelli and Pyron 1991). Consequently, the life histories of many species can be expected to be a mosaic of contemporary responses to natural selection and historical traits that have been retained. For instance, among 21 species of North American minnows, one trait (length of spawning season) was significantly related to current ecological conditions (measured by latitude), but maximum body size had a strong historical component (Gotelli and Pyron 1991).
To elucidate this point further, some aspects of a species’ life history have evolved through physical or biotic selective forces originating outside of their gene pool, such as the effects of environmental factors shown previously. In contrast, other life-history aspects have evolved in response to the presence of other individuals of the same species or sex, such as frequency-dependent intraspecific competition. As shown later, for alternative reproductive strategies, the life histories of male fishes are often influenced by what other males are doing, whereas female life histories can be less influenced by frequency dependent selection and more so by factors originating outside of the gene pool (E. L. Charnov, pers. comm. 2011).
Living and Dying
Understanding trade-offs in life-history evolution as well as responses of fish populations to fishing pressure, or efforts at stock enhancement, requires knowing some basic population demography—the parameters dealing with age structure, mortality, and reproduction. A convenient way to organize age-specific survivorship and mortality data is with a life table. Depending on how the data are collected, life tables can be of two types, dynamic and static. Dynamic life tables, also referred to as horizontal or cohort life tables, are built on data obtained by following a single cohort (age-class) through time until all members have died. This is the preferable approach in terms of having few required assumptions, other than random sampling, but often not possible given the longevity of many species. In contrast, static life tables, also referred to as vertical, current, or time-specific, are usually based on a census taken over a short time period, such as a day or a season, and provide a snapshot of the age structure of a population. Static life tables, although the most common, are less desirable because of the number of necessary assumptions. The method assumes that samples are obtained randomly and that survivorship and birth rates are constant between cohorts, which would only occur if a population has a stable age distribution (Deevey 1947; Krebs 1985).
Construction of a life table requires knowing the number of animals of each age surviving or dying within a given time interval. Depending on the species and how short generation times might be, the time interval could be determined in days, months, or years. For most fishes, the time intervals used would be months or years. The type of input data depends on the particular system being studied. In an exploited population where fishing mortality is high relative to natural mortality, the number dying might be more readily obtained; in nonexploited populations the number of survivors in a given age class (i.e., the age structure of a population) is likely to be the starting point. Because columns in a life table are interconvertible, whether the initial data are based on numbers surviving or numbers dying does not matter.
Somewhat surprisingly, well-documented life tables, covering all life stages from fertilized eggs or newly hatched larvae to adults, are rare for North American freshwater fishes and for fishes in general—a fact pointed out by Deevey (1947), echoed by Weatherly (1972), and seemingly true today. This is not to say that extensive data on survivorship and mortality within certain segments of the life span are lacking. Examples include the detailed study of larval and adult (but not juvenile) survivorship of Pumpkinseed (Lepomis gibbosus) (Bertschy and Fox 1999), and numerous studies of fishes exploited in recreational or commercial fisheries. The difficulties in constructing complete life tables for fishes are due to the enormous ranges in body size and habitats occupied over a life span, contributing to greatly differing capture efficiencies over different life stages (i.e., random samples of the entire age structure are nearly impossible). In addition, most fish populations are open and subject to emigration and immigration, and there are often large swings in year-class strengths (Deevey 1947; Weatherly 1972). Even closed populations, such as farm ponds and small lakes, still suffer from the problem of unequal sampling efficiencies among life stages, leading Deevey (1947) to suggest that “a life table constructed from the available observations would certainly be lacking both head and tail.”
To provide an example of a life table, I combined data from several studies and time periods for the Fantail Darter (Etheostoma flabellare) (Table 9.1). A life table constructed in this manner is, of course, only an approximation of an actual life table that would present data on a single population. Survivorship from ripe, unfertilized eggs to age-1, information that is lacking in most studies of fishes, was 3.5% (Halyk and Balon 1983; Paine and Balon 1986). Age-1 to adult survival of 18.4% was based on population age structure determined from scale annuli by analyzing changes in the number of individuals in successive age groups (Box 9.1). This approach, termed a catch-curve, is useful over a wide range of age groups, from eggs and larvae to adults. Age-specific survival can also be obtained by direct observation and by mark-recapture studies, where different age groups are marked with a unique code or tag, released, and then recaptured on one or more later dates; however, this approach is better suited for larger fish where it is possible to apply external or internal tags.
BOX 9.1 • Catch Curves, Age Structure, and Survivorship
Early fisheries biologists devised various ways to estimate survivorship and mortality rates of populations from the catches of fish across uniform size groups or age classes. Such approaches are collectively referred to as catch-curve techniques. Various algorithms that incorporate changes in age-group size have been proposed, including those based on proportions of the relative abundance of consecutive age groups and those based on linear regression analysis (Ricker 1975; Miranda and Bettoli 2007). Both approaches require the assumption of constant recruitment and mortality rates over time and that catchability of age groups remains the same once they are vulnerable to the sampling gear.
To estimate the survival (S) of Fantail Darters, I used an approach modified from Stearns (1976) by Bertschy and Fox (1999) that weights within-group survival estimates by the number of fish in that interval, where
and nx = number of individuals in age group x, sx = the proportion of individuals surviving from age x to x+1, v = the first age group vulnerable to the sampling gear (in this case age 2), and ω = the oldest age group. Using the number of fish in successive age groups (based on Karr 1964), S is estimated as 26%, as shown in the following example.
For fishes with greater longevity, one of the most useful techniques for estimating survival is to plot the log of fish number (catch) versus age class and then determine the slope of the descending part of the curve. Only the descending arm is used in the calculations because the initial part of the catch curve shows size classes or ages that are not fully vulnerable to the sampling gear. Given that the previously mentioned assumptions are met, the slope is equal to the instantaneous mortality rate (Z), which is related to the finite mortality rate (A) as (1 − A) = S = e−z, where e is the base of natural logarithms (Ricker 1975). Age structure of Walleye (Sander vitreus), based on scale annuli, provides an example of such a catch curve (see the following figure). The slope of the regression line fitted to the natural logarithm of the number of individuals in the fully recruited age groups is −0.366, which is equivalent to 69% survival over the period from age 3 to age 12.
A catch curve based on the age structure of a population of Walleye (Sander vitreus) collected by experimental gill nets (i.e., variable mesh sizes) from Lake Sakakawea, North Dakota, and aged using scale annuli. Data from Isermann et al. (2003).
The life table for Fantail Darters shows a rapid decline with increasing age in both the numbers of individuals (nx) and proportions (lx) of individuals remaining. There is also a concomitant decline in the absolute number of individuals dying (dx) during successive year classes, simply because there are fewer individuals left to die, but a relatively constant finite mortality rate (qx) and an increase in age specific survivorship (sx) from age-0 to ages-1 and 2, before dropping to 0. Because values of qx are relative to population size, they are comparable among different populations, in contrast to dx values. Note that the terminology used in life tables and in treatments of mortality and survivorship varies within and among disciplines.
TABLE 9.1 Approximate Life Table for Fantail Darter (Etheostoma flabellare) Based on Adult Survival Determined for an Iowa Population and Survival from Ripe Eggs to Age-1 for an Ontario, Canada, Population
In general, survival of eggs, embryos, and larvae, relative to the survival of juveniles and adults, is low for most freshwater fishes, especially for those having external fertilization and lacking parental care (Dahlberg 1979). Comparisons of survivorship and mortality among populations and species are difficult because of the great variation in the methods used among studies and the way in which the data are reported. For instance, some studies present survivorship data over short life-history intervals, such as yolk-sac larvae, whereas others might estimate survival from the time of spawning to young-of-year or age-1, and others present data based on a specific number of days or weeks. Because fishes are exposed to forces of mortality on a daily basis, the time over which survivorship is determined strongly influences the value. Consequently, depending on the question being asked, survivorship values can be compared across specific durations or life-history intervals or as instantaneous rates. The instantaneous mortality rate (Z; the mortality in an instant of time; see also Box 9.1) approaches the percent annual or seasonal mortality (A) as the time interval over which A is determined approaches zero. Instantaneous rates are advantageous because they can be added, multiplied, or divided and hence, if the time interval is known, allow conversion of annual or other time-specific rates to daily rates. For example, given a survivorship of yolksac larvae of 55% over a 10-day period, the daily survival rate of 94.2% is obtained by determining the instantaneous rate (Z) for the 10-day time interval:
As an example of changing survivorship values with life-history stage, daily percent survivorship of Walleye (Sander vitreus) gradually increases from 80% for eggs and yolk-sac larvae to 96.5% for post-yolk-sac larvae, and 99.9% for late juveniles and adults (Figure 9.1A). Although a daily survival rate of 80% might seem high, when carried over 24 days from spawning to larval emergence it results in an interval survivorship of only 0.5% (Figure 9.1B). By combining daily survival rates and the duration of each stage (assuming that survival from age-1 to age-3 is the same as survival from age-3 to age-12) and starting with an arbitrary initial number of 100,000 spawned eggs, three adults can be expected to reach sexual maturity and two adults can be expected to survive to age-12.
FIGURE 9.1. An example of how survival varies among life-history stages and the impact that this has on cohorts of eggs.
A. Daily percent survival values for six life-history stages of Walleye (Sander vitreus).
B. The impact of stage duration and mortality forces on the successive numbers of fish in each of the six life-history intervals (rectangles) of Walleye from fertilized eggs to age-12. The numbers are determined from interval survival probabilities (ovals) based on survival probabilities and stage duration, and assuming that spawning occurs in April and that first reproduction occurs at age-3. The data are from multiple northeastern localities (Dahlberg 1979; Henderson et al. 1996; He and Stewart 2001; Pratt and Fox 2002; and Isermann et al. 2003).
Survival probabilities from egg production to larval emergence tend to be lowest for fishes without any form of parental care and higher for those burying eggs or depositing them in nests, or building a nest and guarding eggs and larvae, although the available data that are directly comparable among these three life-history modes are extremely limited (Figure 9.2A). In contrast, when survival probabilities are compared among life-history categories, but over a longer time interval that extends well beyond any time of protection from a nest or parental guarding, there is no clear pattern and survival values are grouped much more closely together (Figure 9.2B). Fantail Darters, in which males locate nests beneath rocks and guard the eggs until they hatch (Page 1983; Page and Swofford 1984), do show the highest percent survival among the few species listed in Figure 9.2B, but other nest guarders have values less than species that only bury eggs or broadcast the eggs without any parental care. Even though the number of examples is extremely limited, the take-home message is that egg burying or nest guarding can improve survival probabilities, but mortality during later life-history stages can fully or partially offset early survival advantages (Bain and Helfrich 1983; Paine and Balon 1986).
Survivorship curves, constructed by plotting the log of the number of survivors (nx) against age, provide an efficient way to look at mortality and survivorship across year-classes and among species because a straight line indicates constant age-specific survival and mortality rates (Deevey 1947). The challenge, of course, is in the details of getting appropriate data. Three forms of survivorship schedules are usually recognized: type 1 has low mortality until when organisms approach their maximum life span, type 2 has a constant mortality rate over the entire life span, and type 3 has high mortality early in life followed by low mortality over the majority of life until senescence (Deevey 1947; Gotelli 2001). Type 1 is purported to represent most mammals, including humans, and some lizards. Although few organisms, including fishes, have constant survivorship throughout most of the life span, the type 2 pattern is approached by most birds and some lizards (Pianka 1988). Most fishes, as well as invertebrates and plants, fit a type 3 pattern and, as argued in the following section, most other organisms typically assigned to types 1 and 2 do as well. A survivorship curve for Fantail Darters shows an initially lower survivorship followed by higher survivorship of fish in ages-1–3 and thus falls between types 2 and 3. A major problem associated with the three types of survivorship curves is that almost all published survivorship curves for vertebrates (as well as other organisms) usually have the initial life-history stages missing, and in virtually all organisms, including humans, mortality is highest among early life-history stages (Levitis 2011). Consequently, type 1 and 2 survivorship curves are most likely curves that omit the life-history stages with the greatest mortality. In fact, when survivorship curves of fishes are based on incomplete life tables, for example, only juvenile and adult stages, they often approximate a type 2 schedule. Assuming that mortality of prereproductive individuals is not maladaptive, the general pattern of high mortality rates early in life is likely due to the early elimination of the most frail individuals, the progressive acquisition of greater robustness by surviving individuals, and the concentration of dangerous transitions early in life such as the change from endogenous or yolk feeding to exogenous feeding (Balon 1986; Levitis 2011).
FIGURE 9.2. Benefits of hiding or guarding eggs on survival from (A) egg stage to larval emergence of seven species of fishes, and (B) egg stage to age-1 of eight species of fishes. Pumpkinseed data are from three populations. Based on data from Dahlberg (1979), Paine and Balon (1986), and Bertschy and Fox (1999).
TABLE 9.2 Approximate Life Table for Fantail Darter (Etheostoma flabellare)
Age-Specific Survivorship and Reproduction
Thus far, the sample life table (Table 9.1) only includes age-specific survivorship and mortality data, and includes information on males and females. To be able to use life tables to provide the information necessary to further understand the evolution of life-history patterns, we must add age-specific natality (mx), the average number of age-0 female offspring produced by a female of age (x). This provides a way to estimate the per capita rate of increase (r) of a population and, more important, leads to an understanding of how gene substitutions affecting survivorship (lx), age-specific natality (mx), and generation time (T) interact in the evolution of life-history patterns. Age specific natality is used because age groups of fishes generally differ in their reproductive output. Fishes are not immediately sexually mature, but depending on the species, require weeks, months, or even years before maturity. In addition, once females become sexually mature, egg or embryo production is likely to vary with age and especially body size. The modified life table only represents females, assuming a 1:1 sex ratio (Table 9.2).
Fantail Darters mature at age-1 (Lake 1936) and females produce an average clutch size of 40 (Halyk and Balon 1983), of which 20 are assumed to be females. Large females may produce five clutches in a breeding season (Lake 1936); for purposes of illustration, I assumed that age-1 females produced one clutch and ages-2–3 produced five clutches, half of which are assumed to be females (Table 9.2). From the life table, the net reproductive rate, Ro, is estimated as follows:
Ro is interpreted as the number of female offspring produced by a female during her lifetime or as the average number of female offspring produced by the population per year, taking into account the probability of a female surviving from year to year (Wilson and Bossert 1971; Gotelli 2001). In a stable population, Ro would equal 1; values > 0 and < 1 indicate a declining population and those > 1 an increasing population. The net reproductive rate provides a measure of population increase per generation and is also a measure of fitness for nongrowing populations, albeit one that does not consider the importance of generation time (Gotelli 2001; Charnov and Zuo in press). Another measure of fitness includes the effect of generation time on reproductive output and is also scaled to absolute time rather than generation time. To do this, Ro is divided by an estimate of cohort generation time, Tc, where
Knowing both the net reproductive rate and the cohort generation time, the intrinsic rate of increase (r) is estimated as
For a population having a stable age distribution and exponential growth, the intrinsic rate of increase allows determination of population sizes over time, and can range from a negative value for a declining population, to zero for a stable population, to a positive value for an increasing population. For example, a Fantail Darter population with r = 0.36 and an initial size of 100, would reach 3,660 individuals in 10 years based on the following relationships:
There are more exact methods for determining r (see Wilson and Bossert 1971; and Gotelli 2001); however, the main interest here is to use the general formula for r to understand forces shaping life-history evolution.
Life-History Theory
Life-history theory attempts to predict how an organism might answer the challenges of survival and reproduction in a specific environment, or how fitness can be maximized for that environment. The answers to these challenges, such as how many offspring to produce annually and in a lifetime, is it better to produce a few large eggs or many small eggs, what should be the age at first reproduction, and how much energy should be allocated to parental care, represent trade-offs in response to environmental forcing functions such as habitat stability, annual temperature and rainfall patterns, resource availability, and predation levels (Stearns 1976; Charnov et al. 2001; Vila-Gispert et al. 2002).
Over the last 30–40 years, numerous hypotheses relative to life-history tactics have been proposed, and many have been eliminated or incorporated into other hypotheses as more information has become available (Stearns 1976; Winemiller 2005). For instance, the habitat template model (Chapter 4; Figure 4.4) is an example that includes life-history predictions of body size and longevity relative to habitat heterogeneity and the frequency of disturbance. One of the most influential models in the development of life-history theory was that of r- and K-selection, formally proposed by Robert MacArthur and Edward O. Wilson (1967) following earlier work by MacArthur and colleagues (Box 9.2) (Reznick et al. 2002). The model was originally developed to understand the colonization and persistence of organisms on islands, but was soon applied widely to other habitats (Pianka 1970). Although this model has been largely supplanted in life-history studies, it had a major role in stimulating extensive interest in the evolution of life histories (Reznick et al. 2002). The r-K model, which is a deterministic model because it does not consider fluctuations in variables such as adult and juvenile survival, has been replaced by stochastic models of life-history evolution that incorporate age-specific survival probabilities and consider multiple, and interacting, axes of selection (Stearns 1976, 1977; Reznick et al. 2002).
BOX 9.2 • Life-History Evolution and the Saga of r- and K-Selection
The understanding of how life-history traits are influenced by natural selection received a major impetus with the publication of the r- and K-selection model by MacArthur and Wilson (1967). The model was influenced by earlier work of Dobzhansky (1950) contrasting how mortality affected populations in temperate versus tropical zones. As proposed by MacArthur and Wilson, r- and K-selection represented the changing selection pressures of habitats as organisms first colonized islands (selection for rapid growth, high r) compared to organisms in saturated island habitats (selection for high competitive ability; high carrying capacity, K). Pianka (1970) greatly expanded the application of the r- and K-selection model to mainland systems and contributed greatly to its widespread acceptance by showing how differing selective regimes affected the evolution of different life-history strategies. As presented by Pianka, the r- and K-selection model can be viewed as two endpoints of life-history evolution that would follow from contrasting environmental conditions. As environments change (e.g., from the original concept of resource-rich, largely unoccupied habitats to resource-limited saturated habitats) then selection pressures would change, resulting in a change in the life-history strategy of an organism along the r-K continuum (Reznick et al. 2002). The r- and K-selection model became a paradigm, albeit relatively short lived, that provided major stimulation to the developing subdiscipline of life-history evolution (as shown in the table below) (Reznick et al. 2002).
Models of life-history evolution can be used as a way to classify life histories, as a way to predict life-history traits given certain selective pressures, and as causal explanations of life histories (Stearns 1977). As such, the r- and K-selection model remains useful as a shorthand way of describing a suite of life-history tactics. The r- and K-selection model has suffered from criticisms of the theory and from criticisms of application, often in the form of unsubstantiated assumptions. Criticisms of the theory were related to the expanded r- and K-selection model based on Pianka and others, as well as the original, more limited model by Mac-Arthur and Wilson (1967).
Contrasts in Life-History Tactics as Predicted for r- and K-Selected Organisms
APPLES VERSUS ORANGES
Criticisms of the basic r-K continuum argue that r and K are not directly comparable. The intrinsic rate of increase (r) can be determined from basic life-history parameters: age-specific survivorship, age-specific natality, and generation time. In contrast, K is not determinable from basic life-history parameters but is a complex interaction of a population and its resource base (Stearns 1977). However, Ro, the number of female offspring produced by a female during her lifetime, can be substituted as a measure of fitness for K-selection (E. L. Charnov, pers. comm. 2011).
SINGLE VERSUS MULTIPLE CAUSATION
Criticisms of the expanded theory focused on the attempt to explain life-history evolution on the basis of one axis of selective pressure (density independent versus density dependent selection) while ignoring other selective pressures such as predation and age-specific mortality. As such, empirical tests of the theory were not necessarily supportive of the predictions (Stearns 1977). Finally, the r-K model was not derived from an age-structured population model (Gotelli 2001). More recent models of life-history evolution acknowledge multiple causation of life-history evolution by incorporating a broader range of selective agents, including age-specific mortality and fecundity, resource limitation, density dependence, density independence, environmental variation, and extrinsic mortality (Stearns 1977; Reznick et al. 2002).
A recent life-history model uses the three components of the estimator (Equation 9.6) for the per capita rate of increase, lx, mx, and Tc, to define its three endpoints (Figure 9.3A). The model is empirically based, using ordination and cluster analyses of life-history variables to identify patterns in North American freshwater and marine fishes and tropical freshwater fishes (Winemiller 1989, 1992, 2005; Winemiller and Rose 1992). The three life-history endpoints proposed by the model are different adaptive suites associated with maximizing the intrinsic rate of increase, r (Equation 9.6), and are opportunistic, periodic, and equilibrium.
Opportunistic fishes occur in areas with greater environmental disturbance and lower resource predictability and can rapidly recolonize disturbed habitats. In reference to Equation 9.6, higher fitness has resulted from short generation times (Tc) rather than large clutch sizes, although they do tend toward prolonged breeding seasons (another way of increasing mx). They also have small body sizes, short life spans, and high reproductive effort relative to body size, but have lower investment per offspring compared to equilibrium species. Opportunistic species generally have population levels well below the carrying capacity.
Periodic fishes occur in habitats with decreased environmental variation and moderate resource stability. In periodic species, selection has been in the direction of increasing fecundity (mx) and higher adult survival (lx). They have low parental investment per offspring, but increased fecundity is associated with larger body sizes and greater longevity and generation times. Such fishes tend to have short breeding seasons and one or a few reproductive events per season and also tend to undergo long spawning migrations. Age structure of periodic fish populations is strongly influenced by early life-history stages and the occurrence of occasional strong year classes. In addition, periodic fishes are disproportionately represented in recreational and commercial fisheries.
Equilibrium species are typical of habitats with low disturbance levels, high resource stability, and often high levels of competition and predation. In reference to Equation 9.6, equilibrium species have increased juvenile survival (lx), moderate to long generation times (Tc), and smaller clutch sizes (mx), but with greater parental investment per offspring, such as parental care and/or large eggs. This endpoint is largely consistent with K-selection of the r-K model, but differs in that equilibrium species often have small body sizes and short to moderate life spans. In fact, there is generally a trade-off between reproductive effort and the probability of future reproduction (Sargent and Gross 1993; Poizat et al. 1999; Charnov et al. 2001).
FIGURE 9.3. A. The three-endpoint model of life-history evolution, as defined by juvenile survivorship (lx), age specific fecundity (mx), and generation time (T) (inner arrows). Outer double arrows show abiotic or biotic forces that would select for different life-history patterns.
B. Approximate positions of some North American freshwater fishes in the three-dimensional life-history space. Based on Winemiller (1989, 1992, 1995, 2005), Winemiller and Rose (1992), and Mims et al. (2010).
More recent support for the three-endpoint model of life-history variation includes an analysis of North American freshwater fishes (excluding Mexico); a study of freshwater fishes of the continental United States; and a comparative study of freshwater fishes in Europe, South America, and North America, and marine fishes in North America (Vila-Gispert et al. 2002; Mims et al. 2010; Olden and Kennard 2010). The data also show a strong phylogenetic signal, emphasizing the importance of recognizing lineage divergences in understanding life-history patterns.
Because the three-endpoint pattern is based on the extremes of life-history differences, most fishes fall somewhere in between the endpoints. Large, long-lived fishes with no parental care and limited investment per offspring are grouped at or near the periodic endpoint (Figure 9.3B). Many of these fish groups, such as sturgeons, Threespine Sticklebacks (Gasterosteus aculeatus), large minnows (e.g., Pikeminnows), temperate bass, whitefish, cisco, and suckers (e.g., Razorback Sucker, Xyrauchen texanus; Redhorses) are migratory, moving between salt and fresh water, or showing extensive movements within freshwater (Lucas and Baras 2001). Some, such as sturgeon and Paddlefish (Polyodon spathula), are not considered extreme periodic strategists because of increased parental investment shown by large egg sizes. Trout and salmon tend to be intermediate between periodic and equilibrium endpoints because of the increased investment in offspring shown by large egg sizes, nest construction, and brood hiding, and the often great distances traveled to reach suitable spawning habitats. Fishes intermediate between opportunistic and equilibrium endpoints include most darters, madtom catfishes, freshwater populations of Three spine Sticklebacks, and sculpin. These fishes tend to have short life spans, short generation times, and smaller clutch sizes, but also show increased parental investment through nest construction, egg burying, or nest guarding. Fishes intermediate between opportunistic and periodic endpoints have small body sizes and short generation times but larger clutch sizes (Winemiller and Rose 1992).
Not surprising given the major geological and climatic events that have been a constant feature of the world’s continents and aquatic habitats (Part 2), there are differences in life-history patterns among geographical regions. South American freshwater fishes appear skewed toward opportunistic patterns whereas European and North American freshwater fishes are intermediate between opportunistic and periodic patterns, and North American freshwater fishes occupy reduced life-history space compared to North American marine fishes. A broad-based study of North American freshwater fishes (excluding Mexico and Alaska) also shows geographic differences in the occurrence of the three endpoints (Mims et al. 2010). Opportunistic species make up 69% of southeastern U.S. fish species, followed by periodic (19%) and equilibrium (12%) species (Olden and Kennard 2010). The southeastern region is dominated by minnows and darters and includes numerous topminnows and livebearers (Ross 2001; Boschung and Mayden 2004). Although the southeastern region was not glaciated during the Pleistocene, it was subjected to major changes in flow patterns, rainfall, river discharge, and temperature; coastal streams were also impacted by major changes in sea level (Chapter 3; Ross 2001). Especially for fishes in small streams, selection for rapid population growth and rapid colonization of habitats made available through increased flow and access to floodplains, or from the resumption of flow following droughts, would have favored opportunistic species (O’Connell 2003; Adams and Warren 2005; Olden and Kennard 2010).
Periodic species tend to occur more in northwestern and northern North America, and in the Rocky Mountain area—regions that were exposed to the full force of Pleistocene glaciation (Chapter 3). Because present-day fishes are generally those that migrated long distances from refugia, selection generally favored larger body size, migration ability, and a long reproductive life span to survive unfavorable reproductive conditions—all traits of periodic species (Mims et al. 2010). Equilibrium species are common along the Pacific Coast and along rivers draining northward from Canada, both faunas dominated by salmonids, and along drainages of the Texas Gulf coast, a fauna with strong representation of centrarchids and ictalurids. Inland from the Pacific Coast, especially along the Rocky Mountains, periodic species are common. Although there is considerable variation and numerous exceptions, opportunistic species tend to occur more in lower latitudes and periodic and equilibrium species at higher latitudes (Mims et al. 2010). At least some of the apparent differences in the patterns are likely caused by strong biases in the habitats and species studied and by hard-classifying families in one of the three endpoints (Winemiller and Rose 1992; Vila-Gispert et al. 2002; Mims et al. 2010).
Approximately 30% of the North American freshwater fish species are near the opportunistic end of the model, 54% are near the periodic endpoint, and 17% are near the equilibrium endpoint. However, opportunistic and equilibrium taxa compose more of the threatened and endangered species listed by the U.S. Fish and Wildlife Service than would be expected by chance. Opportunistic and equilibrium species make up 47% of the continental fauna but 64% of the U.S. threatened and endangered species (Winemiller 2005).
Models of life history, as with the Winemiller-Rose model, can start with a pattern and then infer process, or begin with a process, such as basic metabolism, and then predict a pattern (Mangel 1996). The Winemiller-Rose model is useful for visualizing different adaptive suites of life-history variation. A second approach is to derive a general growth model starting with first principles based on how metabolic energy is allocated between tissue maintenance and the growth of new tissue (i.e., increasing body size) (West et al. 2001). The model can then be used for developing various allometric relationships linking growth rates and the timing of lifehistory events, and the predictions of the model can then be tested against real data. By adding reproduction to the basic metabolic model, Charnov et al. (2001) developed a series of allometric scaling predictions relating age at maturity, average adult life span, and the relative body mass allocated annually to reproduction. One prediction that follows is that the logarithm of the average body mass allocated to annual reproduction is linearly related with a slope of approximately 0.75 to the logarithm of body mass at first reproduction (essentially a dimensionless invariant). The prediction has thus far been tested using the Winemiller and Rose (1992) data set of 139 North American marine and freshwater fishes, showing a very close fit (R2 = 0.74) (Figure 9.4). Basically, fishes that delay initial reproduction until very large body sizes devote a high level of annual mass to reproduction in contrast to fishes that mature at small body sizes. Although likely, the predictive relationship has not been tested for strictly North American freshwater fishes; however data from darters (Percidae) that were not included in the regression calculation also fall near the regression line.
FIGURE 9.4. The allometric relationship between the average annual body mass allocated to reproduction and the average body mass at first reproduction. The solid line is a regression (R2 = 0.74) based on 139 species of North American marine and freshwater fishes; the closed circle shows data from 48 species of darters not included in the regression calculation. Based on Charnov et al. (2001).
Offspring Number, Size, and Spawning Frequency
ANNUAL AND LIFETIME FECUNDITY AND REPRODUCTIVE EFFORT Annual fecundity, the number of mature ova produced within a single reproductive season, is a function of clutch size and the number of times a female spawns. The number of eggs is one part of a female’s reproductive output—the other is egg size (mass). For both marine and freshwater fishes, for any body size there usually is a strong negative correlation between clutch size and egg mass so that there is a trade-off between the two (Elgar 1990; Fleming and Gross 1990). The relationship of clutch size to egg mass is not always clear and can vary among individuals of the same species, as well as among species and genera (Marsh 1986).
The same annual reproductive output (yearly egg number × egg mass) can be achieved by numerous small eggs or by fewer large eggs. Demersal spawning fishes produce fewer large eggs compared to pelagic spawning fishes, and (probably because demersal spawners are more common in freshwater compared to marine habitats) freshwater fishes produce fewer large eggs relative to body size compared to marine fishes, which produce more small eggs (Duarte and Alcaraz 1989; Elgar 1990). Egg number is positively correlated with body size whereas egg mass is not. Annual reproductive output also has a strong positive correlation with body size, much stronger than for egg number, for both freshwater and marine fishes; however, when annual reproductive output is adjusted for body size, there is no difference between marine and freshwater fishes—the difference is in the way reproductive energy is allocated (Wootton 1984; Elgar 1990). The same pattern is shown within benthic and limnetic morphotypes of Threespine Sticklebacks in Benka Lake, Alaska, where limnetic females produce a larger relative clutch mass comprising more, smaller eggs in contrast to benthic females, even though the total allocation to reproduction is the same for both (Baker et al. 2005). Based on a modeling exercise using pelagic marine eggs and larvae, food-rich environments favor the production of more, small eggs whereas food-poor environments favor the production of few, large eggs (Winemiller and Rose 1993). The same pattern may occur for freshwater fishes, although the production of larger eggs in benthic Threespine Sticklebacks is likely caused by selection for larger larvae. Other things being equal, sizes of newly hatched larvae decrease as developmental temperatures increase. Because benthic fish nest in warmer water than limnetic fish in Benka Lake, and thus would have smaller larvae, there apparently is selection for larger egg sizes, which contribute to larger larval sizes, to counteract the temperature effects during development (Baker et al. 2005).
Because of the positive relationship between clutch mass and the size of the female, age specific natality (mx) tends to increase with age. Fishes near the periodic endpoint (Figure 9.3) are characterized by larger body sizes as well as larger clutch sizes, although clutch sizes in some large migratory fishes are reduced from a theoretical maximum because of large ova.
Lifetime fecundity, the number of eggs or young produced over a female’s lifetime, is a function of age specific natality and the probability of survival to the next spawning season (lx).
CLUTCH PRODUCTION A single clutch comprises a cohort of propagules whose development is triggered by a single hormonal signal, that quickly increases in size through yolk deposition (vitellogenesis), and that is completely oviposited in a short time interval (Heins and Rabito 1986; Heins et al. 1992; Baker et al. 2008). Patterns of ovarian clutch production fall into three basic groups: synchronous, group synchronous, and asynchronous (Wallace and Selman 1981). In synchronous clutch production, all oocytes grow and ovulate from the ovary in unison, and there is no replenishment of one ovarian stage by an earlier stage. Synchronous clutch production is typified by fishes with single lifetime breeding periods (i.e., semelparous fishes) such as eels or salmon. In group-synchronous clutch production, there are at least two groups of oocytes developing in the ovary. Typically, there is one group of larger oocytes that will be released as a single clutch and a second group of smaller, more heterogeneous oocytes from which the large ovarian class is recruited. This is the most common pattern in fishes and is characteristic of those with multiple clutches and multiple breeding seasons (i.e., iteroparous fishes). Finally, in asynchronous clutch production, oocytes of all stages are present in the ovary without dominant size groups. This pattern is common in topminnows (Fundulus) (Selman and Wallace 1986).
SPAWNING WITHIN A SINGLE SEASON Species and individual populations within a species can vary in the number of times that spawning occurs in a single season or in an individual’s lifetime. Clearly, knowing whether fishes produce single or multiple clutches within a spawning season and whether egg mass varies within and among spawning seasons is crucial to an understanding of reproductive effort and age-specific natality (Heins and Rabito 1986; Heins et al. 2004).
Various North American fish groups are known to occasionally spawn a second time during the breeding season, such as catfishes (Least Madtom, Noturus hildebrandi) and pickerel (Chain Pickerel, Esox niger) (J. G. Miller 1962; Baker and Heins 1994). This occasional repeat spawning is not the same physiologically, or in terms of its life-history implications, as the normal production of multiple clutches produced at short intervals throughout the breeding season. Although the production of multiple clutches in freshwater fishes has been known since the 1950s through work by Clark Hubbs and his students on darter (Etheostoma) species, the full appreciation of this on estimates of annual fecundity has only occurred in the last several decades. In aquaria, both Greenthroat Darters (Etheostoma lepidum) and Rio Grande Darters (E. grahami) produced clutches of mature eggs every 4–10 days (Strawn and Hubbs 1956; Hubbs and Strawn 1957). More recent studies have shown that the ovaries of many species are extremely dynamic and that a sexually mature female can cycle eggs through the various stages quite rapidly (Box 9.3). Although the list is no doubt incomplete, multiple spawning by an individual female in a single spawning season is documented (or strongly suspected) in 8 families, 13 genera, and 23 species of North American freshwater fishes (Table 9.3). Multiple clutches may occur in North American populations of the tetras (Characidae), but this remains to be documented.
BOX 9.3 • Gonadal Stages in Teleosts
Gonadal development can be described by using external features or by histological procedures, and both approaches are useful depending on the questions being asked and the resources available. Examples of ovarian and oocyte staging, using histological techniques, were shown by Selman and Wallace (1986); Grier (1981) provided an example of histological staging of teleost testes. An improved terminology for histological staging in fishes, applicable to both sexes of elasmobranchs and teleosts, includes the following six categories (Brown-Peterson et al. 2007).
IMMATURE Never spawned
DEVELOPING Gonads starting to develop
SPAWNING CAPABLE Spawning possible within current breeding season
ACTIVELY SPAWNING Spawning currently happening, just completed, or will happen soon
REGRESSING Spawning activity completed for season
REGENERATING Sexually mature but not currently reproductively active
External staging, especially of ovaries, can be useful because it can be rapidly accomplished in the field as well as in the laboratory (Ricker 1968). A useful system of external staging of ovaries is based on six categories and was developed for minnows and darters but is applicable to other North American fish families as well (Heins and Rabito 1986; Heins and Baker 1993).
EXTERNAL STAGING OF OVARIES
LATENT (LA) The transparent to slightly translucent ovaries are very small and thin. The maturing oocytes may be without visible yolk or with some yolk present (vitellogenic) but with the nucleus still visible.
EARLY MATURING (EM) The translucent to opaque ovaries are small to moderate in size. The maturing oocytes are small or of moderate size, translucent to opaque, and with the nucleus obscured by yolk.
LATE MATURING (LM) The white to cream ovaries are small to greatly enlarged and may occupy much of the space in the body cavity. The maturing oocytes are of moderate to large sizes and white to cream or yellow.
MATURE (MA) The cream to yellow ovaries are moderate in size to greatly enlarged and may occupy a large portion of the body cavity. There are two separate groups of follicular oocytes, a group of smaller maturing oocytes that are translucent to opaque and a group of larger mature oocytes that are usually opaque and cream to yellow but without the vitelline membrane (the membrane surrounding the yolk) separated from the yolk.
RIPENING (MR) The cream to yellow ovaries are moderate sized to greatly enlarged. There are two distinct groups of follicular oocytes, a group of small oocytes and a group of larger ripening oocytes that are usually translucent (sometimes transparent) with the vitelline membrane obviously separated from the yolk.
RIPE (RE) The cream to yellow ovaries are moderate sized to greatly enlarged and may occupy much of the body cavity. There are two groups of relatively large oocytes, one group includes white-to cream-colored maturing follicular oocytes of moderate to large size, and a second group of translucent to transparent ripe ova concentrated in the lumen of the ovary with the vitelline membrane separated from the yolk. These ovulated oocytes are ready to be oviposited and fertilized (in species with external fertilization).
The production of multiple clutches is generally viewed as an adaptation for producing more clutches per spawning season, and thus increasing age-specific natality, without increasing body or ovary sizes (Burt et al. 1988). Fishes with multiple spawning typically are small-bodied, are generally not of recreational or commercial importance, and have parental care only by the male or lack parental care altogether so that multiple clutch production is not constrained by time spent with parental care. Pumpkinseed are an apparent exception to this general pattern because they have a larger body size and, at least in an Ontario, Canada, study population, the quantity of eggs produced by females over an average of 2.1 spawning events per season was small enough to be produced in a single clutch. Instead, multiple spawning in this population of Pumpkinseed is perhaps due to bet-hedging selection in a variable environment (Fox and Crivelli 1998). Fox and Crivelli also suggested that parental care might constrain repeated spawning by females, but in Pumpkinseed, as in other centrarchid species (with the exception of a southeastern population of Largemouth Bass, Micropterus salmoides), parental care is only provided by the male and, at least in some species, a male may attract and mate with more than one female during a single nesting interval (Cooke et al. 2006; Warren 2009).
Various approaches have been used to determine the presence of multiple spawning in fishes. For instance, based on laboratory studies, female Bannerfin Shiner (Cyprinella leedsi), which are batch synchronous, can produce egg clutches every 3–10 days (mean = 4.6) (Figure 9.5). The spawned ova, represented in the top panel, were already oviposited and were recovered from crevices of the spawning substrata in the aquaria. The intermediate size group of ripe, ovulated oocytes were found in the lumen of the ovary, and the smallest size group were all intrafollicular oocytes (Heins and Rabito 1986). In reference to ovarian categories of Box 9.3, females sacrificed before spawning were MA, those during spawning were RE, and those after spawning were LM. In Bannerfin Shiner, 26–228 mature ova are deposited per clutch. To maintain the cycle, there is continual recruitment from the oogonia, so that the eggs spawned during a season would be much more than the complement of oogonia and oocytes present at the start of the season.
TABLE 9.3 Examples of North American Freshwater Fishes Showing Multiple Clutch Production by a Female within a Single Reproductive Season
FIGURE 9.5. Ovarian cycling, illustrated by the Bannerfin Shiner (Cyprinella leedsi). The graphs show the percent of ova in 0.05 mm size classes in prespawning, spawning, and postspawning females. The average time of the cycle, from spawning to spawning, is 4.6 days. The data are an average of three fish in each category. Based on Heins and Rabito (1986).
Another approach to determining multiple clutch production is to use outdoor enclosures, where a male-female pair is placed in a cage that contains spawning substrata. By daily examination of the substrata, the occurrence of multiple spawning under natural temperature and light regimes can be determined. Using plastic pools (1.1 or 1.5 m diameter) suspended in a larger outside pool, 11 pairs of Bluntnose Minnows (Pimephales notatus) produced from 7 to 19 clutches over a three-month period, with an average spawning interval of 5.3 days and an average clutch size of 93 to 239 (Gale 1983). Over the three months, each pair produced from 1,112 to 4,195 eggs (average = 2,396). Impressively, the total volume of water-hardened eggs over three months was 1.6 times the volume of a single female. Fathead Minnows (Pimephales promelas) produce an even greater volume of eggs relative to the female’s body size during a spawning season, with egg volumes 3.8–6.8 times greater than the female’s volume (Gale and Buynak 1982).
Based on the behavior of nine females held in instream cages that were provided with spawning substrata, Kentucky Snubnose Darter (Etheostoma rafinesquei) can produce a clutch on average every 3.7 days and can spawn seven or more times in 26 days (Weddle and Burr 1991). Females also can spawn partial clutches over a period of several days.
In addition to laboratory studies, multiple clutch production can be determined from histological studies of wild-caught specimens. For example, based on the presence of postovulatory follicles, two Alaskan populations of Threespine Stickleback had spawning intervals of 2.2–7.8 days. The spawning interval increased as the rate of clutch production slowed with the advance of the breeding season and fish became more energy limited as body condition declined (Bagamian et al. 2004; Brown-Peterson and Heins 2009). In fact, a major controller of the interval between clutches in Threespine Stickle back is food availability (Ali and Wootton 1999). Female Sticklebacks can also retain a clutch of ovulated eggs up to 48 h while they search for a suitable mate (Brown-Peterson and Heins 2009).
The appreciation of multiple production of clutches during a single spawning season in the two most speciose families of North American freshwater fishes, in addition to other families (Table 9.3), has necessitated a reevaluation of annual fecundity estimates for many species (Conover 1985). Because of this, estimates of annual fecundity based on preseason counts of vitellogenic follicles are likely to greatly underestimate age-specific natality. Estimates of annual fecundity and reproductive effort are further complicated because of the potential for changes in clutch size and/or ovum size over the spawning season, with a general negative relationship between ovum size and clutch size (Heins et al. 2004). Ovum size is often inversely related to temperature, so that larger ova tend to be produced earlier in the breeding season (Marsh 1984, 1986; Heins et al. 2004). This is perhaps an adaptation for providing greater energy stores for developing embryos and larvae at a time when food availability might be lower (parental investment hypothesis), or providing fewer, high-quality eggs at a time when environmental uncertainty might be greater (bet hedging hypothesis) (Heins et al. 2004). Changes in clutch size, adjusted for female body size, over a breeding season of North American freshwater fishes support three of four basic patterns, and patterns may differ even within the same genus. In one pattern, clutch size peaks midway through the breeding season. This is shown by Etheostoma rafinesquei (Weddle and Burr 1991), E. caeruleum (Heins et al. 1996), and Menidia menidia (Conover 1985). In a second pattern, clutch size declines throughout the breeding season as shown by Fundulus heteroclitus (Kneib and Stiven 1978). In a third pattern, clutch size increases throughout the breeding season. This is shown by Etheostoma lynceum (Heins and Baker 1993; Heins et al. 2004) and Percina vigil (for one of two years; Heins and Baker 1989). A fourth pattern, in which clutch size is relatively invariant over the breeding season, has not been linked with North American freshwater fishes, but is shown by marine species and may be most likely under relatively constant food availability and environmental conditions (Conover 1985). Overall, changes in clutch size seem to be proximally driven by variation in temperature or nutrient availability, or from selection resulting in the greatest offspring production when the probability of survival is greatest (Conover 1985).
SPAWNING WITHIN A LIFETIME: SEMELPARITY VERSUS ITEROPARITY A key to understanding how reproductive effort is allocated over time is the relationship between adult and larval/juvenile survival, which can be affected by environmental variation such as floods or droughts, size-specific predation that might impact one life-history stage more than another, and the cost to the adults, including survival costs, of producing offspring (Partridge and Harvey 1988). As the probability of adult survival increases relative to survival of larvae/juveniles, iteroparity (reproducing over multiple seasons) should evolve along with the reduced cost of annual reproductive output (Periodic Strategy of Figure 9.3) (Partridge and Harvey 1988). Iteroparity can represent bet hedging and be a form of resilience in response to environmental variation (Chapter 6), and is predicted to be favored as environmental uncertainty increases (Orzack and Tuljakarpur 1989). When there is low environmental variation, so that the survival probability of young is likely to be high, coupled with extensive parental investment such as numerous large eggs, nest construction, and risky, long-distance migration (intermediate between Periodic and Equilibrium of Figure 9.3), selection should favor semelparity—“big-bang” reproduction. This is a single lifetime breeding event with maximal egg production at the expense of future survival (Winemiller 1992; Crespi and Teo 2002). Semelparous North American freshwater fishes include some populations of American Shad (Alosa sapidissima), most populations and species of Pacific Salmon (Oncorhynchus spp., excluding nonmigratory Cutthroat and Rainbow trout), American Eels (Anguilla rostrata), and all Lampreys (Hardisty and Potter 1971; Leggett and Carscadden 1978; Haro et al. 2000).
American Shad populations along the Atlantic coast of North America vary in the extent of repeat breeding, with populations in lower latitudes (< 32 N), where moderate temperatures make riverine spawning habitats more predictable, being totally semelparous. American Shad have a narrow temperature range for spawning and egg development of 13–19° C, and optimal growth of larvae and juveniles occurs at 20–25° C (Leach and Houde 1999; Bilkovic et al. 2002). At higher latitudes, the increased temperature variation reduces the probability of juvenile survival, relative to that of adults, and the proportion of repeat breeding increases, reaching 60–80% in New Brunswick populations. The southern semelparous populations put more energy into reproduction, in contrast to more northern populations, and have greater relative and absolute fecundities. This results in an inverse relationship between the expected lifetime fecundity, which is greatest in low latitudes, and the frequency of repeat breeding, which is greatest in high latitudes (Leggett and Carscadden 1978).
Females of all five North American Pacific Salmon species (Coho, Oncorhynchus kisutch; Chinook, O. tshawytscha; Sockeye, O. nerka; Pink, O. gorbuscha; and Chum, O. keta) are semelparous, as are most males of these species. There are several instances, however, of freshwater-resident males of Chinook, Coho, and Sockeye salmon that sometimes will breed more than once (Crespi and Teo 2002). Most other salmonid species, including Steelhead, are iteroparous, although there are exceptions.
In salmonids, the semelparous life cycle evolved from an iteroparous life cycle in association with a larger body size, greater ovum mass, anadromy, and strong development of secondary sex characteristics. In addition to being able to produce more eggs, larger fish are also more efficient swimmers (Chapter 7), having a lower mass-specific cost of locomotion, and thus are capable of more extensive migrations. Larger eggs result in larger fry and juveniles, with concomitantly greater survivorship, and are thus a form of increased parental care (Crespi and Teo 2002). Semelparous salmonids, the five species of Pacific Salmon, are also unique among the subfamily Salmoninae in showing nest guarding once the nest is completed and in continuing nest guarding until they die (Fleming 1998). Finally, semelparous salmonids take parental care to yet another level because the carcasses of adults increase the productivity of streams and thus help to increase the survivorship and growth of young salmonids (Crespi and Teo 2002).
Semelparous salmonids generally show greater reproductive costs (as percent energy loss) for males and females compared to iteroparous salmonids. Energy losses in anadromous populations are determined from the time of entry into coastal areas or into freshwater to the time of postspawning or death (in the case of semelparous species), or between pre- and postspawning in freshwater resident populations. Thus determined, the costs of reproduction do not include those of migration, which would be partially compensated by feeding in the ocean (until feeding ceases in fresh water), but do include those of territory defense, courtship, secondary sex characteristics (primarily in males), nest construction (females), nest guarding (females), and gamete production (Hendry and Berg 1999; Fleming and Reynolds 2004). Although quite variable, average energy losses of females exceed those of males by 13% (for semelparous populations) to 24% (for iteroparous populations). Atlantic Salmon (Salmo salar), which have relatively higher energy losses during reproduction than other iteroparous salmonids, actually have a low level of iteroparity, with repeat breeding on the order of only 10% in most populations (Fleming and Reynolds 2004).
In spite of the overall greater energy loss at reproduction for semelparous versus iteroparous populations, the gonadal-somatic index (GSI) does not differ between the two life-history patterns (Figure 9.6A). Consequently, semelparous salmonids invest more in other aspects of reproduction, including migration, nest preparation, nest guarding, courtship, and sexual dimorphism (Fleming 1998). Although there is again considerable overlap, greater differences in female GSI are shown between resident and anadromous populations of salmonids, reflecting the greater investment in current reproduction given the reduced chance of breeding again in anadromous populations (which is even further reduced by the current increased reproductive investment of migration) (Figure 9.6B; Fleming and Reynolds 2004). At least in marine fish populations, and likely so for other fish populations, GSI is strongly and positively correlated with the adult instantaneous mortality rate (Gunderson 1997).
To Travel or Not: Spawning Migrations
Fishes near the periodic endpoint of Figure 9.3, or those intermediate between periodic and equilibrium endpoints, often undertake long-distance spawning migrations. Spawning migrations, as discussed previously, can represent a major parental investment in offspring, where adults move from feeding or resting areas to areas that are thought to be more suitable for egg, larval, and juvenile survival (see Chapter 5; Figure 5.5). The majority of migratory fishes of inland habitats restrict their movements to fresh water, termed potamodromous, whereas relatively few species show some form of periodic, physiologically mediated movement between the sea and fresh water as part of their life cycle (termed diadromous). Two subgroups of diadromy shown by North American fishes have different biomes for late-juvenile and adult feeding and for reproduction (Myers 1949; McDowall 1997). Anadromous fishes have the major feeding and growth areas at sea and move from the sea to fresh water for purposes of spawning, with generally little or no feeding once adult fish enter fresh water. Catadromous fishes use fresh water as the major feeding and growth biome and move from fresh water to the sea as adults for spawning, with little or no feeding in the sea (Figure 9.7) (McDowall 1988, 1992, 1997).
Anadromous fishes include representatives within the Salmonidae, Osmeridae, Clupeidae, Acipenseridae, and Moronidae (McDowall 1988, 1997). Catadromous fishes are less common in North America, with only three primary examples, Striped Mullet (Mugil cephalus, Mugilidae), American Eel (Anguillidae), and Hogchoker (Trinectes maculatus, Achiridae) (Dovel et al. 1969; De Silva 1980; T. L. Peterson 1996; McDowall 1988, 2007). Worldwide, only about 0.9% of fishes are diadromous, of which 48% are anadromous and 25% are catadromous (McDowall 1997; Nelson 2006). The remainder are amphidromous, a third type of diadromy first described by Myers (1949) and then refined by McDowall (1987, 2007), in which the reproductive and major growth biomes are the same. Amphidromous fishes can be either fresh water or marine, although the former appears much more common. Larvae of fresh water, amphidromous fishes migrate to the sea soon after hatching, where early feeding and growth occur for a few weeks to several months, followed by a return migration to fresh water while still small juveniles. Most feeding and growth occur within fresh water, as do maturation and spawning. In contrast to anadromy and catadromy, the return migration to fresh water has a trophic base in contrast to a gametic base (McDowall 1997, 2007). Application of the term amphidromy to North American fishes has been complicated by various usages; the term does not refer to aperiodic movement between the sea and fresh water but to an actual migration (see Chapter 5). Although amphidromy is more common in the tropics and subtropics, there are North American examples including some populations of three cottid species, Prickly Sculpin (Cottus asper), Coastrange Sculpin (C. aleuticus), and Pacific Staghorn Sculpin (Leptocottus armatus) on the Pacific Coast (McDowall 1988; Brown and Moyle 1997; Moyle 2002). Some populations of the Mountain Mullet (Agonostomus monticola), which occur along the southern Atlantic and Gulf coasts likely also are amphidromous (Matamoros et al. 2009). Other Gulf and Atlantic species that may have amphidromous populations include the Bigmouth Sleeper (Gobiomorus dormitor, Eleotridae) and the River Goby (Awaous banana, Gobiidae) (Musick et al. 2000).
FIGURE 9.6. Female GSI [(gonad weight/somatic weight) × 100] compared between (A) semelparous or iteroparous and (B) anadromous or resident populations of salmonids. FW = freshwater populations. Based on Fleming (1998).
FIGURE 9.7. Migratory patterns of fishes. In potamodromy, migration takes place entirely within fresh water, such as moving from a lake or large river where adult feeding and growth occur to a smaller stream for spawning, hatching, and early growth of larvae and juveniles. In anadromy, movement is from the adult feeding and growth area at sea into fresh water for spawning, hatching, and early growth of larvae and juveniles, followed by the return migration to the sea. The situation is reversed in catadromy, where most growth occurs in fresh water, and spawning, hatching, and early development of young occur at sea. Amphidromous fishes grow, mature, and spawn in the same biome, either fresh water or marine. The size of the closed boxes and circles indicates the relative amount of feeding and growth in a particular habitat. Based in part on Gross (1987) and McDowall (1988, 1992, 2007).
The evolution of migratory behavior in fishes, including those restricted to fresh water and those that show diadromy, can be understood by Ro, the simple model of fitness introduced earlier in this chapter, where
Age-specific natality (mx) is influenced by such things as differences in productivity and temperature of a habitat, and survivorship (lx) is influenced by such things as predation and disease levels and environmental stress. Migratory fishes, moving from one habitat to another at different stages in their life cycle, incur costs and benefits that change values of lx and mx and thus alter fitness (Figure 9.7). For instance, in a potamodromous fish, migration from the adult feeding ground, such as a lake, into a tributary stream for spawning, incurs some costs to the adult fish that could lower adult survivorship. This can be balanced by increased hatching success and better larval and juvenile growth and survival in the small stream as a consequence of greater food availability, lowered predation or disease, or less environmental stress. In addition, assuming the adult feeding ground allows faster growth and thus larger body size, the number of eggs spawned would be increased, thus raising age-specific natality. Selection for migration would occur if fitness (Ro) increases as a result of such behavior.
Among diadromous fishes, anadromy is most common in temperate to polar regions of both the Northern and Southern Hemispheres, whereas catadromy is greatest in the tropics. The pattern seems to be driven by the relationship of freshwater to marine productivity, referred to as the food availability hypothesis (FAH), which emphasizes selection for adult growth (Gross et al. 1988; Maekawa and Nakano 2002). When ocean productivity exceeds that of fresh water, as occurs in temperate to polar regions, anadromy is more common; when freshwater productivity exceeds marine productivity, as occurs in the tropics, catadromy is more common. Although the FAH has been criticized for being overly simplistic by ignoring selection on survival and growth of young, and because of questions concerning the classification of certain fish groups as catadromous or anadromous and concerns about the relationship of productivity measures to actual food availability (Dodson 1997), the hypothesis is supported by work on growth and body size in resident freshwater and anadromous populations of Dolly Varden (Salvelinus malma) (Maekawa and Nakano 2002).
Even though diadromy or potamodromy benefited fitness at some point in the evolutionary history of a group, changing environmental conditions, including those caused by humans, might reduce or eliminate the benefit, resulting in the decline or extirpation of populations/species showing a particular migratory behavior (Gross 1987). Consequently, a life history dependent on multiple habitats and access to them can be more fragile than a simple life history (Gross 1987). For example, consider the Gulf Sturgeon (Acipenser oxyrinchus desotoi), an anadromous fish that has three principal habitats required in the life history, in addition to requiring access to the habitats (i.e., no barriers along rivers): a specific spawning ground that can be 100–200 km upstream, a summer-fall holding area located 40–60 km upstream, and a marine feeding area located primarily in barrier island passes of coastal waters (Heise et al. 2004, 2005; Ross et al. 2009). For purpose of example, assume that there is a 90% probability that each habitat remains suitable and a 10% chance that each will not. This scenario corresponds to a 73% chance (0.903) that all three habitats remain suitable and a 27% chance that all three will not remain suitable. Thus, all else being equal, compared to a single habitat, diadromy, in this case, lowers the chance of survival by 17% (27%−10%). Not surprisingly, many North American diadromous fishes are considered endangered, threatened, or of special concern, a pattern also true worldwide (Williams et al. 1989; McDowall 1999; Warren et al. 2000).
SEX, MATING, AND INVESTMENT IN OFFSPRING
As with other aspects of their life history, fishes show amazing variation in male behavior and mating tactics. In species where male reproductive success is dependent on male-male competition and aggression, and where there is extensive male parental care, there may be several types of males (Dominey 1981; Gross 1982). Male reproductive strategies are context dependent and are solutions of how to maximize fitness with given amounts of energy and probabilities of future survival (Dominey 1984). Often the trade-off is between energy spent on somatic growth for larger body size, secondary sex characteristics, and female attraction behaviors, versus energy spent on gonadal growth.
Mating Tactics
Fishes in the family Centrarchidae are known for nest construction and defense, and in the genus Lepomis, only the male is involved in these behaviors (Warren 2009). After nest construction, the male attracts females to his nest for spawning through visual displays in conjunction with well-developed secondary sex characteristics. These parental males (also referred to as bourgeois males) are highly aggressive, have a relatively large body size, and a light body color with a dark yellow to orange breast (Gross and Charnov 1980; Dominey 1981; Gross 1982; Avise et al. 2002). After spawning, the male guards the nest for 1–2 weeks against potential predators on the developing embryos and early stage larvae, and such male care is critical for survival of young (Gross and MacMillan 1981). Parental males spend energy on somatic growth associated with female attraction and securing mates, and relegate relatively less energy to gonadal development. In Bluegill (Lepomis macrochirus), parental male gonad weights average 1.1–1.9% of body weight (Dominey 1980; Gross 1982). Parental males occasionally enter the nests of other parental males and fertilize eggs of females that they are courting, perhaps made easier by habituation of males to other males in nearby nests of breeding colonies (Avise et al. 2002; Jennings and Philipp 2002; Mackiewicz et al. 2002).
Male mating tactics and parentage of young have been studied thus far in six sunfish species. In some populations of at least four species, Bluegill, Pumpkinseed, Longear (L. megalotis), and Spotted Sunfish (L. punctatus), there are two mutually exclusive mating strategies: parentals and peripherals (Box 9.4), which are defined by male size and behavior (Gross 1982; Warren 2009). Other sunfish species, especially those occurring in nesting colonies such as Redear Sunfish (L. microlophus), are likely to have similar alternative reproductive strategies.
BOX 9.4 • Cuckoos, Stolen Fertilizations, and Deception: Peripheral Males, Cuckoldry, and Kleptogamy
From the fourteenth century, cuckoldry has been used to describe the situation where a husband (the cuckold) unknowingly supported offspring resulting from adulterous behavior by his wife (Power et al. 1981). The original term derives from the habit of Cuckoos depositing their eggs in the nests of other birds and then leaving eggs and young to be supported by the unsuspecting parents, even though in the case of the cuckoo, both the male and female of the host nest provide care to young that are not their own. In biology, cuckoldry has generally been used to describe instances where a male provides parental care to offspring that are not his own, usually through stolen fertilizations by another male (the cuckolder) (Power et al. 1981). Although the term cuckold, as historically used, follows the concept of a male raising offspring that are not his own, it suffers from the human connotation of the female’s unfaithfulness. To correct for this, the term kleptogamy has been proposed as a more neutral, as well as more specific, term for stolen fertilizations (Gowaty 1984). Peripheral male is another, even more general, descriptor that has been used to describe the same breeding behavior in fishes (Blanchfield and Ridgway 1999). For example, sneaker and satellite males in sunfishes are both peripheral males engaging in cuckoldry through kleptogamy. Peripheral males among salmonids, where females build and defend nests, engage in kleptogamy but not cuckoldry.
Peripheral males can sometimes achieve similar male fitness as parentals through cuckoldry (or kleptogamy; Box 9.4) (Neff and Clare 2008). By stealing fertilizations from the parental male, the cuckolder is acting as a reproductive parasite. Two forms of cuckolders shown within sunfishes are sneakers and satellites (Gross and Charnov 1980; Gross 1982). Sneaker males, which generally are smaller than parental males, dart in between a spawning pair just at the moment of egg deposition and attempt to fertilize the female’s eggs. The sneakers are chased by the larger, nest-defending male and often show missing scales, wounds, and damaged fins at the end of the breeding season. Compared to parental males, sneakers pay nothing for nest construction, female attraction, and defense of young, instead putting much more energy into sperm production, with the gonad weight in Bluegill sneakers averaging 4.6% of body weight (Gross 1982). In addition to Bluegill, sneaker males are documented in some populations of Pumpkinseed, Spotted, and Longear sunfishes (Jennings and Philipp 1992; Warren 2009).
Another approach to mating is shown by satellite males that hover in the water column above a parental male’s nest (hence satellite). Satellite males have only been observed for Bluegill (Warren 2009). Satellite males reduce, but do not eliminate, attacks by the defending parental male by mimicking the dark vertical bars, dark background, and dark eyes of females and by descending slowly into a nest—normally while a female is paired with the male. Satellite males attempt to position themselves between parental male and female and thus fertilize the eggs. Although less common, Bluegill satellites will also engage in apparent homosexual behavior by pairing with a parental male in the absence of a female (Dominey 1980, 1981). Again, because they are not investing in somatic growth associated with nest construction and defense and mate attraction, more reproductive energy is allocated to gonadal development, with gonad weight averaging 3.3–4.2% of body weight (Dominey 1980; Gross 1982). The evolution of this system may be asymmetric, with satellites gaining matings by deceiving parental males as to the sex of female mimics, or symmetrical, where the parental male benefits from the nest site being more attractive to a female because a “female” is already on the site, and the satellite male benefits by having access to females. Disadvantages to the parental male of sharing incoming females with satellite males may be balanced with the advantages of producing a spawning center (Dominey 1980).
In Bluegill, but not necessarily other Lepomis species, whether an alternative male is a parental or a cuckolder seems to be the result of a genetically controlled polymorphism—the same male does not practice both behaviors, and parentals and cuckolders have different growth trajectories (Figure 9.8). Parental males are larger, mature later, and have lower mortality rates compared to sneakers and satellites. In contrast to the discrete reproductive pathways of parentals and cuckolders, sneakers and satellites are an ontogenetic progression. Although body size generally forms a progression of increasing size from sneaker to satellite to parental, in Cazenovia Lake, New York, parentals and satellites had similar age distributions (Dominey 1980; Gross 1982). In Bluegill populations, parental males are generally less common than cuckolders, although the frequency of cuckoldry decreases with decreases in overall population density (Gross 1982). In Lake Opinicon, Ontario, offspring in nests of Bluegill paternal males were sired by the paternal male an average of 76.5% (range 14–100%), with the level of paternity highest early and late in the breeding season, and lower during midseason when reproductive activity in the breeding colony was greatest and paternal male condition was lowest because of increased nest defense (Neff and Clare 2008). Even within Lake Opinicon, different breeding colonies varied in the level of cuckoldry (0–59%) based on the percentage of young sired by nonparental males (Philipp and Gross 1994). In general, a single parental male Bluegill will spawn with several females over the course of the reproductive season (Mackiewicz et al. 2002).
FIGURE 9.8. Male mating systems in Bluegill showing two discrete pathways of cuckolders and parental males, and the ontogenetic pathway of peripheral males from sneaker to satellite. Percents of peripheral and parental males are for a Lake Opinicon, Ontario, population. Based on Gross (1982).
Only sneaker males have been documented in Pumpkinseed and Spotted Sunfish, the two other sunfish species known for having peripheral and parental males (DeWoody et al. 2000a; DeWoody and Avise 2001; Neff and Clare 2008). In Pumpkinseed, young in nests were sired by the parental male an average of 62% (range 0–100%). In Spotted Sunfish, young sired by the parental male made up 98% of the young in a nest, but almost half of the nests contained at least a few young sired by peripheral males. Whether or not the two male morphs are behaviorally mediated or the result of a genetic polymorphism, as is thought to be the case in Bluegill, is not known (DeWoody et al. 2000a).
Redbreast Sunfish (L. auritus) and Dollar Sunfish (L. marginatus) apparently lack peripheral males, although field observations suggest that sneaker males might be present in a Virginia population of Redbreast Sunfish (Lukas and Orth 1993). Stolen fertilizations appear to be rare, with the parental male generally siring 95% of offspring in the nests of both parental species (DeWoody et al. 1998; Mackiewicz et al. 2002). Stolen fertilizations that do occur are caused by parental males spawning with females in the nests of other nearby parental males.
Several females normally spawn in the nests of a parental sunfish male over the course of a single reproductive season. For instance, young in Dollar Sunfish nests were from an average of 2.5 females, young in nests of male Redbreast Sunfish were from an average of at least 3.7 females, and young in nests of male Spotted Sunfish were from at least 4.4 females (DeWoody et al. 1998; DeWoody et al. 2000a; Mackiewicz et al. 2002).
In addition to sunfish, peripheral males have been genetically documented in Percidae in the Tessellated Darter (Etheostoma olmstedi), although parental males sired an average of 86% of young in their nests, and observed in Orangefin Darter (E. bellum) (Fisher 1990; DeWoody et al. 2000b). Other North American freshwater fishes showing types of alternative mating systems in at least some populations are found in the families Cyprinidae (Mayden and Simons 2002), Cyprinodontidae (Kodric-Brown 1986), Gasterosteidae (Blais et al. 2004), Poeciliidae (Snelson 1985; Plath et al. 2007), and Salmonidae (Blanchfield and Ridgway 1999).
Sex and Mating Systems
The great majority of fishes (roughly 88%) are gonochoristic, having separate sexes with sex determined genetically and/or environmentally, and the sex of an individual does not change once maturation has occurred (Avise et al. 2002; Patzner 2008). Although rare, environmental sex determination (ESD) has been documented in Atlantic Silversides (Menidia menidia) and in laboratory populations of the Clearfin Livebearer (Poeciliopsis lucida) (Sullivan and Schultz 1986; Conover and Heins 1987). In these fishes, more females are produced under cooler temperatures and more males under warmer conditions. This is presumably adaptive in allowing females more time for development of larger body sizes and consequently greater fecundity, although other factors may also be involved. In both species, the degree of ESD is apparently controlled by a genetic polymorphism.
Although gonochorism is the most common, and is the ancestral condition among fishes, some fishes are hermaphroditic and a few are even unisexual (Schultz 1977; Avise et al. 2002; Avise and Mank 2009). Unisexuality is rare among animal species, and among vertebrates there are only 22 known genera. Over a third of these (eight) are fishes and most occur in freshwater habitats of North America, where at least 17 unisexual forms have been discovered (Table 9.4) (Schultz 1977; Vrijenhoek et al. 1989; Kraus 1995). In fact, the first known unisexual vertebrate was the Amazon Molly (Poecilia formosa), documented as an all-female taxon by Carl and Laura Hubbs (Hubbs and Hubbs 1932, 1946). All unisexual fishes are of hybrid origin between two gonochoristic species (Vrijenhoek 1994). However, this is not to say that these clones are short lived, at least in an ecological sense. Poeciliopsis monachaoccidentalis, for instance, has apparently survived for some 60,000 years (Avise et al. 1992), although the gonochoristic progenitors have existed for much longer, perhaps 4–12 million years (Vrijenhoek 1994).
In diploid unisexuals of Poeciliopsis, reproduction occurs by hybridogenesis, where a female mates with a male congeneric and only female offspring develop. The male genome is expressed in the offspring but is eliminated during meiosis, whereas the female genome passes intact to the haploid egg, creating a hemiclonal lineage. Hybridogenesis does not form true clones because paternal genomes are added afresh each generation (Vrijenhoek 1979, 1994; Moore 1984). The remaining unisexual fishes, including some Poeciliopsis, are typically triploid and reproduce through gynogenesis, where sperm from a congeneric is needed for egg activation, but the male genome is not incorporated and the eggs develop matroclinously.
TABLE 9.4 Genera of Freshwater Unisexual Fishes Occurring in North America Family Genus Number of forms
Gynogenetic and hybridogenetic populations form ecologically fascinating systems because the unisexual clones, although likely interacting competitively with gonochoristic, congeneric species, are also dependent on heterosexual males for reproduction, even though the males’ genomes are not incorporated in the offspring’s genome (gynogenesis) or are incorporated but not transmitted between generations (hybridogenesis) (Vrijenhoek 1979). Assuming equal fecundity, an all-female species can achieve twice the population growth rate of a gonochoristic species, at the expense of donor males from conspecific populations (Moore 1984). Also, in environments that are intermediate and relatively unchanging, sex is costly, as only 50% of one’s genome is transferred to offspring. Consequently, unisexuality and clonal populations might be expected to be successful in such habitats. A critical question, therefore, is how do clones keep from out-competing and eliminating their gonochoristic host species? Although numerous hypotheses have been proposed, three nonmutually exclusive hypotheses relating to this are the behavioral regulation/sperm limitation hypothesis, the resource partitioning hypothesis, and the Red Queen hypothesis (Van Valen 1973; Moore 1984; Vrijenhoek 1994). The behavioral regulation hypothesis involves mate recognition and the tendency of heterosexual males to prefer female conspecifics over clones. At low population densities of sexually reproducing fish, clones would be sperm limited. As population sizes of the sexual fish increase, breeding males form dominance hierarchies and subordinate males tend to mate more often with the clones. In support of this hypothesis, the occurrence of male preference for conspecifics has been widely documented (Moore 1984).
There is also some support for versions of the resource partitioning hypothesis. One form of this hypothesis is the “frozen niche-variation model,” where each clone has a fixed adaptive complex that includes a small portion of the niche space occupied, but underutilized, by nonclonal gonochoristic species (Vrijenhoek 1979). Predictions of this model have been supported by coastal populations of the Texas Silverside (Menidia clarkhubbsi), a clonal, allfemale species inhabiting coastal areas from Alabama to Texas, and by all-female clones of Poeciliopsis occurring in streams of northern Mexico (Vrijenhoek 1979; Echelle and Echelle 1997). Coexistence of these clones with each other and with the gonochoristic species is perhaps facilitated by trophic resource partitioning. For instance, during the dry season, one member of a pair of P. monacha-lucida clones specialized as a scraper feeding on organic detritus derived from leaf litter in shaded, rocky pools. The lower jaw dentition consisted of numerous, small tricuspid teeth. The other clone had fewer, larger teeth in the lower jaw and browsed on floating mats of organic detritus in nutrientrich, unshaded, shallow pools (Vrijenhoek 1978, 1979). However, the extent of resource partitioning between the clones has been questioned because stomach contents showed high overlap (Moore 1984).
Why sexual reproduction should be favored over asexual reproduction is addressed by Van Valen’s Red Queen hypothesis (Van Valen 1973). The term comes from Lewis Carroll’s Alice in Wonderland, where the Red Queen shows Alice that she must run fast to stay in the same place. Biologically, the Red Queen hypothesis predicts that the increased genetic diversity, and potentially more rapid evolutionary change, allowed by sexual reproduction compensates for any genetic or ecological disadvantage of sex (Salathé et al. 2008; Morran et al. 2011). Consequently, unisexual organisms should be at a disadvantage in terms of responding over the long-term to parasites, predators, or competitors. The prediction that clonal forms should have greater parasite loads in contrast to sexual forms has been tested with two clones of Poeciliopsis monachalucida and the syntopic, sexually reproducing Headwater Livebearer, Poeciliopsis monacha. The most common clone in a stream pool showed significantly greater parasite burdens compared to the less common clone or the sexual population, except where a sexual population had low genetic diversity because of a recent founder effect caused by complete drying and then recolonization of a pool (Lively et al. 1990). When a few Headwater Livebearer females were added to the pool after 15 generations, the genetic diversity again increased with a concomitant drop in parasite load (Vrijenhoek 1994).
At least 12 families of teleostean fishes, in nine orders, are hermaphroditic (C. L. Smith 1967; Avise and Mank 2009). Three forms of hermaphroditism are protogynous hermaphrodites, in which an individual is first a female and then becomes a male; protandrous hermaphrodites, in which an individual is first a male and then becomes a female; and synchronous hermaphrodites, in which functional male and female cells are present in the gonads at the same time. Hermaphroditic fishes are almost all marine species and most are not self-fertilizing (C. L. Smith 1967). An exception is the Mangrove Rivulus (Kryptolebias marmoratus, Aplocheilidae), a small cyprinodontiform occurring in fresh to brackish water in tropical and subtropical areas of the Gulf, Atlantic, and Caribbean, including southern Florida (Kallman and Harrington 1964; Miller 2005). The Mangrove Rivulus is a synchronous hermaphrodite and most populations are obligate, self-fertilizing clones (Kallman and Harrington 1964). However, some populations do include functional males and thus gain increased genetic variation through outcrossing. The presence of males is apparently triggered by some unknown ecological factor (Weibel et al. 1999). In theory, synchronous hermaphroditism would allow for the most rapid rate of population increase compared to other forms of hermaphroditism and to gonochoristic species—if longevities were equivalent, which is not often the case (C. L. Smith 1967). Synchronous hermaphroditism would also be favored where populations are small and isolated so that finding mates is difficult. In this case, synchronous hermaphroditism provides “reproductive assurance” because reproduction can occur without another individual (Charnov et al. 1976; Weeks et al. 2006; Avise and Mank 2009).
Parental Care
Parental care of young can include prefertilization activities, such as substratum cleaning and nest construction, and postfertilization activities, such as egg hiding or egg burying; parental defense of nest and offspring; caring for eggs by fanning, cleaning, and removing dead or diseased embryos; mouth brooding; pouch brooding; and internal gestation (Baylis 1981; Blumer 1982; Sargent et al. 1987). Overall, approximately 22% of actinopterygian families show parental care, which is represented in basal fishes such as Bowfin (Amia calva) as well as in derived lineages such as darters, sunfishes, and livebearers (Blumer 1982; Gross and Sargent 1985; Mank et al. 2005). The evolution of parental care is a function of the relative maturity and survival rates of young and the survival rates of adults. Parental care is favored when the survival rate of eggs and juveniles is low in the absence of care, the adult death rate is relatively high, the egg maturity rate is relatively low, and the duration of the juvenile stage is relatively short (Klug and Bonsall 2009). In general, fishes with parental care are near the equilibrium end of the three-endpoint life-history model, characterized by smaller clutch sizes and increased parental investment in offspring (Figure 9.3). Egg size is often positively correlated with the quality (i.e., intensity and duration) of parental care in fishes, and a life-history model predicts that increased egg size is favored when parental care reduces the instantaneous mortality rate of eggs (Sargent et al. 1987).
Among families of bony fishes showing parental care, male-only care is most common (∼50%), followed by female-only care (∼32%) and biparental care (∼18%) (Figure 9.9A). This is quite different from the majority of vertebrate taxa where there is internal fertilization and the care of young is provided by the female. However, internal fertilization has evolved in at least 21 families of bony fishes and 14 of these families have internal gestation of embryos—an extreme form of parental care (Mank et al. 2005). Parental care is more prevalent in freshwater compared to marine fishes, occurring in 60% of freshwater fish families versus only 16% of marine families (Baylis 1981).
The increased frequency of parental care in freshwater habitats compared to marine habitats, and the preponderance of male care over female care, are both explained by a model based on the differential survival of zygotes and the rates of gametogenesis males and females (Baylis 1981). Freshwater environments tend to show greater heterogeneity in physical and chemical factors per unit distance compared to marine environments. Demersal and/or adhesive eggs are likely favored in fresh water because pelagic eggs would run the risk of being swept out of favorable conditions—in fact, the only North American freshwater fish groups with pelagic eggs are those in large lakes or those that migrate extensive distances upstream to spawn (Balon 1975). In contrast, pelagic eggs in a marine environment have a much greater chance of remaining within the same water mass during their developmental period, and even in coastal species tidal influences would put demersal/adhesive eggs at risk because of periodic changes in the water mass (Baylis 1981). Male residency is favored when suitable spawning sites are limited because a male is able to produce fertile gametes more rapidly than a female and can leave more offspring than a female through the potential of multiple matings over a short time period (Baylis 1981).
The kind of parental care is strongly influenced by whether fertilization is internal or external. In fishes with external fertilization, nest construction is a preadaptation for maleonly parental care; close to 80% of such families have care of young provided solely by the male (Gross and Sargent 1985; Mank et al. 2005). Internal fertilization is a preadaptation for internal gestation, and 90% of families with internal fertilization show female-only parental care, most through internal gestation but several with external guarding (Mank et al. 2005). Although various scenarios have been proposed for the evolutionary sequence of parental care (e.g., C. Smith and Wootton 1995), a recent phylogenetically based approach suggests that the three forms of parental care (male-only, femaleonly, and biparental) have arisen independently from an externally fertilizing ancestor without parental care (Figure 9.9A) (Mank et al. 2005; Mank and Avise 2006).
Parental care has been documented in 54% of North American freshwater fish families (27 of 50), with male-only care shown in 63% of the families, female-only care in 52%, and biparental care in 33% (Table 9.5). In some cases, multiple types of care are shown by different taxa in the same families. Within the Centrarchidae, male-only parental care occurs in most species and both sexes are usually polygamous (Cooke et al. 2008; Warren 2009). Attendant males remain on the nest for only 1–7 days (Sacramento Perch, Archoplites interruptus; Pumpkinseed; and Bluegill) to nearly a month (Smallmouth, Micropterus dolomieu; and Largemouth bass), depending on the species (Gross and MacMillan 1981; Sargent et al. 1987; Cook et al. 2006). However, in at least one population of Largemouth Bass, located in the Savannah River drainage of South Carolina, males and females are mostly monogamous and females also participate in providing care (DeWoody et al. 2000c).
TABLE 9.5 Types of Parental Care Shown by North American Freshwater Fish Families
Families are listed in decreasing order of species richness
For a downloadable PDF of all tables, go to ucpress.edu/go/northamericanfishes
FIGURE 9.9. Parental care in fishes.
A. Evolutionary pathways of parental care in bony fishes, showing the three independent origins of the three forms of parental care (solid ovals) from an external fertilizing ancestor lacking parental care. Percents show the families exhibiting each form of care. Based on Blumer (1979), Sargent and Gross (1993), Mank et al. (2005), and Mank and Avise (2006).
B. Types of parental care shown by families of North American freshwater fishes. Data from Table 9.5.
The most common form of parental care is egg guarding, followed by nest construction, fanning eggs with the fins or by expelling water from the mouth or opercular chambers, removing dead or diseased eggs from the nest, guarding larvae against predators, and internal gestation of the embryos by the female (Figure 9.9B). Pipefishes, family Syngnathidae, provide the only North American example of parental care being provided by placing embryos in a brood pouch.
SUMMARY
Life-history patterns of North American freshwater fishes are the distillation of numerous competing demands of growth, survival, and reproduction. Although life histories are shaped by natural selection, any life history is a mosaic of traits shaped by contemporary selection and traits shaped by selection at various stages in the evolution of the lineage. Life tables are a convenient way to organize the basic parameters of life histories that include age-specific survivorship, age-specific natality, and cohort generation time. Survivorship usually increases with age and is lowest in the egg and early larval stages. As a consequence, survivorship curves for fishes are generally closest to a type 3 shape.
A three-endpoint model of life histories effectively captures the variation in life histories among freshwater and marine fishes. The model is based on how fitness can be maximized by increased juvenile survival and generation times and small clutch sizes (equilibrium species), shorter generation times and extended reproductive seasons (opportunistic species), or increasing age-specific natality and adult survival (periodic species). In general, opportunistic species are more common at lower latitudes whereas periodic and equilibrium species are more common at higher latitudes.
A model derived from basic metabolic principles predicts that the annual mass devoted to reproduction scales allometrically with body mass at first reproduction, with strong empirical support from a data set including North American freshwater and marine fishes. The body mass devoted to reproduction can be packaged in single large clutches, multiple clutches produced within the same reproductive season, or single clutches per reproductive season produced over several or more years. An increasing number of fishes are now known to produce multiple clutches within a single spawning season so that single measurements of fecundity underestimate annual egg production. Iteroparity is more common in fishes that occupy varying environments and that have high adult (relative to young) survivorship. In contrast, semelparity occurs in fishes that experience lower environmental variation and that have high reproductive investment, including long-distance migrations. Spawning migrations are common in fishes intermediate between periodic and equilibrium life histories. Among fishes that show regular migrations, movement within fresh water (potamodromy) is most common among North American freshwater fishes, followed by anadromy, amphidromy, and catadromy. Although complex migratory behaviors evolved because of the increased fitness realized by migration, changes in habitats, including those caused by humans, often put such lifestyles at greater contemporary risk.
Male fishes of numerous species often show a trade-off in how reproductive energy is allocated. Parental males expend high amounts of energy in such things as territory defense, nest construction, and mate attraction, and relatively less on actual gonadal size. In contrast, peripheral males shunt most of their reproductive energy into gonadal growth and attempt to steal fertilizations from the parental males. Although less common, some fishes are hermaphrodites and others are unisexual. Unisexuality is shown within several groups of North American freshwater fishes, especially the Poeciliidae and the Atherinopsidae. Hermaphrodites are more common in marine fishes and only one species, the Mangrove Rivulus, occurs in North American fresh water.
Parental care is documented in over half of North American freshwater fish families and the most common type of care is egg guarding. Care is most often provided by the male; biparental care is least common. The kind of parental care is strongly influenced by whether fertilization is external or internal. In the latter case, care is generally provided only by the female.
SUPPLEMENTAL READING
Avise, J. C., A. G. Jones, D. Walker, J. A. DeWoody, et seq. 2002. Genetic mating systems and reproductive natural histories of fishes: Lessons for ecology and evolution. Annual Review of Genetics 36:19–45. A thorough review of mating systems in fishes.
Balon, E. K. 1975. Reproductive guilds of fishes: A proposal and definition. Journal of the Fisheries Research Board of Canada 32:821–64. A detailed classification of reproductive guilds of fishes.
Balon, E. K. 1981. Additions and amendments to the classification of reproductive styles in fishes. Environmental Biology of Fishes 6:37789. Further clarification of reproductive guilds.
Gotelli, N. J. 2001. Age-structured population growth, 50–80. In A primer of ecology. Sinauer Associates, Sunderland, Massachusetts. A lucid treatment of growth in natural populations of organisms.
Miranda, L. E., and P. W. Bettoli. 2007. Mortality, 229–77. In Analysis and interpretation of freshwater fisheries data. C. S. Guy and M. L. Brown (eds.). American Fisheries Society, Bethesda, Maryland. A thorough coverage of how mortality is determined in fish populations.
Winemiller, K. O., and K. A. Rose. 1992. Patterns of life-history diversification in North American fishes: implications for population regulation. Canadian Journal of Fisheries and Aquatic Sciences 49:2196–218. One of the foundation papers on the three-endpoint model of fish life history patterns.
WEB SOURCES
Brown-Peterson, N., S. Lowerre-Barbieri, B. Macewicz, F. Saborido-Rey, J. Tomkiewicz, and D. Wyanski. 2007. An improved and simplified terminology for reproductive classification in fishes. http://digital.csic.es/handle/10261/11844.
Frimpong, E. A., and P. L. Angermeier 2009. Fish-Traits database. http://www.cnr.vt.edu/fisheries/fishtraits.