MINIMUM AREA REQUIREMENTS OF POPULATIONS

One of the pioneers in developing practical guidelines for conservation managers and planners is Soulé (1987), who lead and edited the composition of a conservation classical, "Viable Populations for Conservation", the textual accumulation of a previous workshop in search of the minimum size a population should have to survive without human interference, referred to as "minimum viable population" (MVPs). The concept of thought regarding the limited life-times of populations have been heavily influenced by MacArthur and Wilson (1967) and Dobson (1970), who convincingly argue that insular populations of plants and animals undergo a continuous process of going extinct and being (re-)established by migrants from elsewhere and its consequences for nature reserves. Further underpinning of MVPs has taken place in various later publications of workshop presentations (Remmert 1994, Landweber and Dobson 1999) and reflects an on-going academic debate on the survival potential of species over certain periods of time. While most authors in those publications clearly approach the issue of species survival from an individual species viewpoint, their work is of vital importance for establishing minimum sizes of protected areas, as it allows for extrapolation of their findings for collective criteria on population size for all species in a given ecosystem. In this document, the survival criteria for individual populations will be used to argue the dimensions of ecosystems required for the survival of the vast majority of the species they contain and from there to the minimum sizes of protected areas.

In his book Soulé (1987) argues that while some conservationists resent the term "minimum", it is not practical to use the term optimal, as it is prone to widely vary among the users and with prevailing viewpoints in a society. Managers and policy makers need clear, understandable and defendable relatively fixed floors below which population levels should not drop. Soulé defines a MPV as a population that meets the minimum conditions for the long-term persistence and adaptation of a species or population in a given place. The theoretical conditions that need to be met for a population to be considered viable will be reviewed.

Conservationists – inter alia through the CBD – target to conserve biodiversity of species and ecosystems, or in other words, they try to prevent their extinction. By doing so, they have to deal with a time-horizon dilemma. Over a geological timescale, no species lives forever and natural ecosystems undergo continuous gradual alterations. The extinction of an established species is almost as common an event in the fossil records as the appearance of a new one. Conservationists think in a much more limited timescale. They observe that habitat destruction and impending climatic change have started to lead to massive loss of species which occurs at such a high speed, that they fear that the rate of extinction has become much higher than may be compensated by the rate of evolution of new species. They feel the need to prevent this for at least the duration of a period that is humanly somewhat comprehensible like centuries but not in terms of geological time horizons. To do so, population dynamists work with probabilistic models (Schaffer 1987) to predict what may happen with populations undergoing change. Soulé (1987) defines the "long-term" persistence of a species as follows: A species in any given ecosystem must have the capacity to maintain itself without significant demographic or genetic manipulation for the foreseeable ecological future - usually centuries - with a certain, agreed on, degree of certitude; he suggests several centuries with a degree of certitude of 95%.

Defining the horizon of duration in terms of time (a human perspective) instead of generations (a more biological perspective), tends to favor somewhat lower requirements for large organisms, as they usually live longer, and therefore, experience a slower population turnover, resulting in genetic processes and related risks proceeding at a slower pace (Korn 1994). This would mean that theoretically, long-living organisms - independently of external effects – should be able to survive more years at lower MVPs than smaller organisms during the same period of time. This benefit is important, as most species living at such low densities that the viability of their population becomes impaired, have long life-spans.

Many authors claim that a few hundred years is not enough. After all, we try to conserve those species – humanly speaking – forever. When set at survival targets for a thousand years and 99 percent certitude, the models that calculate the minimum population sizes for the survival of species with large territories predict that almost the entire earth will be required for conservation purposes (e.g. see Belovsky 1987).

Ciraicy-Wantrup and Phillips (1970) introduce the term "safe minimum standard of conservation", which they compare to "the objectives of an insurance policy against serious losses that resists quantitative measurement. Here the objective is not to maximise a quantitative net gain but to choose premium payments and benefits in such a way that maximum possible future losses are minimised". An insurance company cannot set its premiums to compensate for any and all possible future damages. The premiums would be so expensive that nobody would be able to afford to buy the insurance.

We must accept that we cannot look into the future forever. If mankind finds ways to somehow redistribute wealth and well-being more equitatively, peace and conservation may both benefit and grow, and conservation may remain a human concern. If this optimistic scenario cannot be achieved, mankind may altogether lose its interest in conservation, in which case, anything we do now, will be in vain. Assuming the positive scenario, the possibilities for conservation may change dramatically, for the worse or for the better from an ecological point of few. For instance, worldwide, we can observe a tremendous shift from rural dwellings to city dwelling, as can already be seen in quite a few protected areas in Mexico, where the younger members of some villages have mostly migrated to the cities as could be observed in the context of several GEF project evaluations, leaving their native mountain regions abandoned. If the trends of the native populations of the wealthy countries are any indication, the world population may start to decrease in by the end of this century, and some poorly performing agro-production lands may be converted again into ecosystems suitable for large animals. On the other hand, a shortage of fuel may generate tremendous pressure on protected areas for the expansion of agricultural land to produce biofuels. In short, there are so many factors of uncertainty, that it is not realistic to set targets for eternity while current needs for land are so pressing. If today, we can create the conditions for the larger species to hold on to life for a few more centuries, we must count on future generations to find ways to extent that period to the next millennium and beyond. Building on such uncertainties, conservationists may agree to settle for a somewhat lower certitude than 99 percent and a horizon of no more than a century or two. In the following paragraphs we review the five factors that potentially determine the sizes of MVP.

Inbreeding depression is the exposure of the individuals in a population to the effects of deleterious recessive genes through mating between close relatives. Experience of animal breeders indicates that rapid inbreeding in a very small population recently founded from a large one produces substantial decreases in body size, viability, and fecundity and frequently leads to the extinction of the population (e.g. Dobson 1996, Lande et al. 1994, Ryan and Siegfried 1994). A good example of this phenomenon come from butterfly farms that need to restock their breeding populations as fertility and size decrease after a certain number of generations (J. Meerman pers. com.). This is due to the fact that for a given locus, some alleles will confer more fitness on an individual than other ones. Within the other class of alleles are rare deleterious recessive alleles, which, when appearing as a homozygous genotype in an individual, greatly reduces the fitness. Deleterious alleles arise constantly through mutation, so they are always present in a population at low frequencies (e.g. Lynch 1995). The slower the rate of inbreeding, or, in the present context, the larger the effective population (consisting of members that effectively reproduce, often symbolised by "Ne") size immediately after a population crash, the greater the opportunity for selection to eliminate recessive deleterious mutations, and consequently, the less inbreeding depression is manifested. It has been suggested that inbreeding is a problem only when Ne is less than 50. Ryan and Siegfried (1994) give a variety of examples of birds in which some degree of inbreeding could be expected but apparently does not occur, but they don’t suggest a minimum population size.

On the other hand, many examples of survival from very low population numbers can be given, and it must be concluded that deleterious effects of inbreeding may not occur in genetically carefully bred populations. The somewhat conflicting evidence on inbreeding depression indicates that this may be a less straight-forward problem than is so commonly believed (Ryan and Siegfried 1994). Some species apparently can tolerate high levels of inbreeding.

With countless cases of global populations having narrowly escaped extinction, it is now clear that in general, populations can at least temporarily survive extreme constrictions in number at least temporarily, and some over longer periods of time. Still, one wonders what happens genetically to populations that have undergone such dramatic constrictions in their population size. There is no doubt that there is considerable permanent loss of genetic diversity when it passes through a "bottleneck", particularly when recovery is slow or bottlenecks are repeated. Nevertheless, Korn (1994) writes that going through a "bottleneck" once does not necessarily mean that a great percentage of the heterozygosity is lost, as long as the population is expanded rapidly afterwards. Skilful breeding programs in zoos may effectively reduce some of the losses by selectively breeding back rare surviving traits in the population. In the wild - in absence of computerised mate selection - genetic variation in very small populations may be corrected over time if more viable population levels can be restored and genetic variation can be regenerated by mutation. The increased population would again undergo adaptive evolution, particularly if (re-)introduced in the wild, where natural selection may further enhance variation. In fact, Korn (1994) finds that genetic variation in a founder population rapidly approaches that in a wild source population, once the effective size exceeds 25. Mathematical models of quantitative genetic variation suggest that at equilibrium, Ne = 500 is sufficient for mutation to counter losses resulting from genetic drift (Lande et al. 1987). Ample details on models are provided in aforementioned works and different authors write about the consequences of each variant. However, Ne normally does not reach census numbers (Korn 1994). This will usually translate into a total size of "several times" that number, when taking into account the factors that determine the participation in reproduction, like age, ratio of breeding adults, variance in family size and fluctuations in populations size (Soulé 1987).

For mammals the total population may be as much as 4 – 10 times higher. This varies greatly per species and in lower organisms the ratio may be thousands times higher during reproduction season when including eggs or seeds in the population. As recent off-spring thins out in the course of a season, the relationship changes. Also of relevance is how social behavior may influence the participation of individuals in the reproduction process. In some species many males are permanently or partially excluded from mating by dominant males, while in other species sexual maturity may be delayed considerably, thus influencing the ration between the reproducing and total population.

Demographic stochasticity consists of individual variation in fecundity, longevity, accidents, sex ratio of the offspring, etc. In general, this rarely leads to extinction, unless the population size is very small, generally under 40, which is somewhat subjected to the population growth rate of each species (Ryan and Siegfried 1994). These numbers are well under the MVP requirements of previous factors. For very small populations (less than a few dozen), the chance that all (or most) individuals are of the same sex is "rather large" (Wissel et al. 1994), but those risks diminish rapidly with increasing numbers. A number of rodents, such as rabbits and hares, are subjected to large swings of their population sizes (Korn 1994), and such species probably have MVPs at an order of magnitude higher (Soulé 1987). But then, such species have relatively high population densities and are usually found in numbers far above MVP levels. With species numbers in the low thousands, demographic stochasticity can be ignored for long living large animals.

Ryan and Siegfried (1994) define environmental stochasticity to encompass a continuum of unfavorable conditions ranging from short-term fluctuations (particularly weather) to long-term variation (like prolonged draughts), to catastrophes (like fires, hurricanes or floods). Earlier theoreticians (Shaffer 1987) were inclined to distinguish between the effects of stochastically occurring in unfavorable environmental conditions and disasters. Disasters are different in the sense that they may wipe out an entire population all at once, and in that sense, may be regarded as independent of population size. Ryan and Siegfried argue, that catastrophes are no more than extreme environmental conditions, whose impact may largely depend on the scale of an organism and the survival strategy of each species Although Ryan and Siegfried’s (1994) viewpoint is logical, the risk of full blown disasters requires special attention in risk abatement strategies, which will be dealt with in the paragraph on spreading of risks. In general environmental stochasticity influences the net survival of a population, but does not lead to general rules for minimum population sizes.

Many species are patchily distributed over a grid work of their acceptable habitats (Gilpin 1987). For the more suitable parts, the densities are much higher and are likely to have larger populations, healthier individuals and greater emigration than the less suitable parts. According to the theory, sub-populations or metapopulations may occasionally go extinct, but as some individuals disperse from other sub-populations, formerly populated patches may be re-colonised, or genetically depleted sub-populations may be enriched and numerically strengthened. Particularly among birds with their high mobility, this rescue effect is not uncommon, even over considerable distances (Bezzel 1994).

Theoretically, this has consequences for the minimum sizes of populations, as the effects of inbreeding repression and genetic drift is more severe in smaller populations than in bigger ones. On the other hand, partial isolation may also have its advantages, particularly in the case of some types of disasters, such as the outbreaks of highly infectious epidemic diseases, hurricanes or large fires. Gilpin (1987) reviews many consequences of the existence of meta-populations, but does not come with numerical estimates of the consequences and Soulé did not take them in consideration at all in his overall evaluation. Assuming survival advantages and disadvantages, no numerical consequences are generalised in this review from the phenomenon of metapopulations. For particularly vulnerable species, the meta-population specifics may need to be considered, which may lead to specific management recommendations. However, Hanski and Simberloff (1997) have downgraded the relevance of metapopulations at least in the context of MVP rules due to the observation that most species in nature are not as structured as metapopulations in the original sense and the document will not further take this phenomenon in consideration.

In literature, almost all considerations of MVPs are heavily focussed on animal populations (e.g. Soulé et al. 1987, Remmert et al. 1994, Dobson 1996, Landweber and Dobson 1999, Holsinger 2001). It makes sense to wonder if special considerations must be made for plants. Willmanns (1984) argues that while the "Gesetzmässigkeiten der Polulationsentwicklung" (the laws of population development) are primarily derived from animal species, that the same principles also apply to plant communities. Stacy (1997) did a study on "Mating Patterns in Low-Density Populations of Neotropical Trees" on Barro Colorado Island in Panamá on low-density tree populations. Within the study area, the density of reproductive adults (the effective population) for the three species under study, Calophyllum longifolium, Spondias mombin, and Turpinia occidentalis, ranged from one tree per 6.3 ha to one tree per 10 ha. She found that all three species were essentially 100% outcrossed, and that mating in each population involved some percentage of pollen flow over long distances. Where flowering adults were clumped, the majority of matings were among near neighbors with some small fraction of successful pollen originating from outside the clump. In contrast, where flowering adults were more evenly spaced, a large fraction of effective pollen dispersed 200 to 300 m, or farther, and well beyond the nearest reproductive neighbors. These findings of appreciable levels of moderate- to long-distance pollen movement in all three populations suggest that small Neotropical insects, which likely pollinate a large fraction of neotropical tree species, are effective in transferring viable pollen among widely dispersed flowering conspecifics.

Based on the mating patterns observed for each species, she estimated the smallest area required for a natural breeding unit. This was defined as the minimum area in which 95% of the pollen received by a centrally-located adult originates. Using Calophyllum longifolium as a model of an evenly dispersed population with a low density of reproductive adults, she suggests that a natural breeding unit would extend a minimum of 60 ha. For populations characterised by clumping of reproductive trees (e.g., Spondias mombin and Turpinia occidentalis), Stacy suggests that a natural breeding unit would need to occupy at least 40 ha. This is a requirement far below the minimum ecosystem size suggested later in this chapter.

An average distance apart of 1,000 m or 1 adult per 100 ha, would require an area of 50,000 ha to maintain an effective population of 500 individuals. A limiting factor could be posed by pollinators, mostly flying insects in the humid tropics. Domestic bees, Apis mellifera, regularly fly 3 km for feeding, but beyond that, the chance of encountering a certain condition – in this case an individual of the same tree species, decreases rapidly, with the increasing surface of the flight radius. If the fertile individuals of a certain tree species would live 3 km apart, a MVP would require 450,000 ha. Honey bees, however, are very powerful pollinators, compared to many other insects, and it is more likely that most tree species depend for their pollination on less powerful insects, and therefore, must live closer together. One would be inclined to think that relatively few if any evenly distributed tree species live at densities of less than 1 fertile tree per 1000 ha, but one must be alert for exceptions and individual cases must be treated with appropriate care. Until contrary indications emerge, probably no special considerations for minimal area size are required for tree species or any other plants beyond those of animals, but specific consultation is required on the matter among tropical taxonomists. Another factor in favor of the survival of tree species is the greater longevity of trees, which makes them less vulnerable to extinction than most animal populations during a given period.

Considering aforementioned stochastic processes that work at populations, minimum areas must be set to allow the selection of protected areas systems that may warrant the survival of the majority of the species that live in them, including the ones requiring large territories. Soulé (1987) argues that from a genetic variability point of view, the effective MVP sizes should be 500 reproducing animals or larger, which translates into census populations of a few thousand. Environmental stochasticity in a stable environment also requires census population sizes of "a few thousand" individuals, but obviously, no survival guidelines can be given to buffer against all disasters.