Back: Table of Contents | How is genetic diversity distributed in natural populations?

3. How do genetic questions enter into restoration?

The genetic principles outlined above have a direct bearing on the practice of ecological restoration (Falk and Holsinger 1991; Young and Clarke 2000). In this section we summarize how restoration practitioners and researchers can (and should) take population genetics into account.

3a. Accuracy and functionality of restored populations.

Sophisticated restoration practitioners recognize that the starting pool of genetic variation to be used is a critical element in design and implementation. However, opinions abound on how to select a suitable source pool for a restoration project (Landis and Thompson 1997).

The current debate often confuses two related, but distinct issues: the accuracy ("authenticity") of a restoration project, and its functionality (Clewell, 2000). A reintroduction project is genetically accurate ("authentic") if it replicates the original gene pool of the population it replaces. If the original population has been destroyed, then perfect accuracy is strictly speaking impossible, because some alleles were probably unique to that population (i.e., GST< 1). A restoration can be accurate and function poorly; it can also be inaccurate but functional.

Since no reintroduction is perfectly accurate genetically, the question then becomes: "How close is close enough?" To this there can be no absolute ecological answer, since not all variation is of adaptive significance. Moreover, variation among populations is a continuous variable: two populations can differ by one allele or thousands. If the motivation is an ethical commitment to fidelity to the historic distribution of genotypes, the line can be drawn anywhere. The best practice is to estimate and report the degree of genetic accuracy based on available information, so that others studying a project can take authenticity into account.

How well a reintroduced population functions is a very different matter. Here the concern is not so much whether specific genotypes have been re-established, but rather how well a restored population will work in terms of persistence, resilience, and stability. Since many adaptive traits have a genetic basis, reintroduction material that performs (survives, grows, reproduces) well may come from a site with similar ecological conditions, but not necessarily close geographically (Knapp and Rice 1998; Procaccini and Piazzi 2001). Moreover, if functionality is the main objective, then a range of genotypes can be introduced, allowing selection (differential survival and reproduction) to sort out the best for the site. If this approach is taken, there may be considerable attrition, which can be accounted for in initial sampling (§3.c).

3b. Geographic location of source material for (re)introduction.

Perhaps no single genetic issue is more intensely debated in restoration circles than the location of source populations: "Where should material for restoration come from?" Interest in population genetic variation often arises in the context of asking whence source material should come. In other words the restorationist is concerned not only with the degree of variability but its particular geographic distribution.

The most common approach is to specify a geographic (and often elevation) range, within which source material can be confidently collected. This is essentially a "space-for-genotype" substitution: we expect that populations near one another, and growing in similar conditions, will be more similar genetically due both to ecotypic variation and the effects of gene flow. Geographic distance is a reasonable first approximation of genetic distance, assuming that genetic diversity most likely obeys the "distance decay of similarity" in broad outline (Nekola and White 1999). In population genetic terms, this favors the "stepping stone" model (§2.b) of gene flow. On public lands in the western United States, commonly-applied guidelines suggest that material for outplanting be collected from within 1,000 ft. elevation bands and 100 miles lateral distance of the planting area (elevation bands vary from 500-3,000 ft for different agencies) (see § 4, Resources). As noted above (§ 2.c), climatic zones may also offer a better first approximation of genetic distance between populations than simple geographic distance (Knapp and Rice 1998; ONPS 2001).

Unfortunately, there are no simple distance rules that apply equally to all species, because species vary in gene flow among populations, population size, and the resulting distribution of diversity (GST). For some species (e.g., self-fertilizing plants in small, isolated patches of habitat, or fishes in isolated stream reaches), each site may reflect a unique local adaptation, and the geographic range of suitable genotypes can be very small (a few km2). Other species (for example, species with wind dispersed pollen and seeds) with higher rates of gene flow, and those with larger effective populations, are probably less differentiated over the landscape and can be collected from much wider ranges. Most species have dispersal curves that are leptokurtic (i.e., with long tails), meaning that a few seeds or offspring in each generation may travel far beyond the majority; while few in number, these long-range dispersers may play a critical role in helping species to adjust their ranges in periods of climate and vegetation change. For each species, the restorationist must ask: "How widely does this species disperse its genes under normal conditions, and what factors (distance, geographic barriers, habitat types) influence where genes can spread?" This uncertainty is reflected in the wide range of recommendations for suitable collecting zones, which range over more than three orders of magnitude from 100 meters to 100 miles.

The issue of geographic range for source material cannot be separated from the distinction made earlier between historical accuracy and functionality. If historical accuracy (authenticity) sui generis is the primary concern, then the range of potential collecting sites is governed most strongly by historical patterns of dispersal, and the collecting radius will be very small. If population function is the primary concern, on the other hand, then the net may be cast over a wider area (although still focusing on similar ecological communities). This is because populations that are widely separated in space may nonetheless occupy similar ecological settings, and by selection may have developed genotypes that are similar in key ecological traits. If the main goal is population function, then it is probably more important to derive source material from large, diverse populations than from small, genetically depauperate sites, even if the latter are closer geographically. Of course, it is always possible to mix genotypes from different populations, and to let selection sort out the variation: this is, after all, exactly what nature does. Mixing of genotypes may be particularly suitable if existing "populations" are in fact isolated and reduced remnants of formerly widespread and interconnected groups. Re-combining fragmented populations genetically is appropriate as the geographic scale is ecologically realistic, and should not be done indiscriminately. Mixing too broadly (i.e., combining genotypes from very diverse ecological settings) can result in too many individuals that are poorly adapted for the new given environment (high genetic load). Some geneticists advocate the use of regional mixtures -- composite collections of genotypes, all of which are at least moderately well adapted to a given environment (Knapp and Dyer 1997; Lesica and Allendorf 1999).

There are valid arguments on both sides of the authentic vs. functionality debate. On one hand, there is little question that the gene pools of many remnant native populations have been seriously eroded, so that what occurs today is a small remnant of the original diversity. Small gene pools are more prone to inbreeding, as well as random genetic change from drift. Populations that formerly exchanged genes regularly may have also become genetically isolated by habitat fragmentation. In such cases, a good argument can be made to bring together genetic material from several nearby populations, in effect replicating the natural (but now disrupted) processes of gene flow. In addition, some restoration sites may be so heavily disturbed (i.e. mine tailing reclamation areas) that the most "local" population may no longer be the one best adapted to the growing environment. In such cases, a wide diversity of genotypes can increase the chances that at least some plants will have the genetic composition to survive.

It is possible, however, that genotypes from outside the apparent current range may perform too well -- that is, swamp the local genotype. If remnant native populations of the species being introduced already exist at a site, additional genetic considerations reinforce the importance of the scale of collection. Introduced populations may hybridize with the existing native population, introducing new genes (genetic pollution) and potentially negatively affecting genetic integrity (Rieseberg 1991; Glenne and Tepedino 2000). If a few poorly adapted individuals make it into the existing native population, it might be argued that natural selection will eventually remove the deleterious genes. However, the issue is partly one of numbers; with commercial seed production of many native species, restorationists now have the tools to dump large volumes of seed into ecosystems. If the number of poorly adapted, non-local propagules is large in relation to the number of local native types, the chance of matings between the well adapted and the poorly adapted plants will increase (especially in cross-pollinating species), thereby potentially swamping the native population and diluting the local genes. Progeny from such matings may additionally experience outbreeding depression (i.e. poor survival and growth in relation to the parents). Carefully chosen introductions, with genotypes similar to the existing native population, can avoid these negative impacts.

Take-home messages: Since no single rule applies to all situations, the most important recommendation is to use explicit and reasoned criteria in selecting source populations. To summarize the issues presented here:

  1. Species vary in their dispersal rates and distances, and hence in the degree of genetic differentiation among populations. These differences are often closely correlated with life-history attributes particular to each species that affect rates and patterns of gene flow. Therefore, guidelines are inherently species-specific.
  2. The restorationist must decide if historical accuracy or functionality is the primary objective. True historical accuracy can be difficult to determine if existing populations are remnants of a formerly widespread range. Moreover, most natural populations experience a degree of genetic change over time and space. However, even where the emphasis is on functionality, historical reference conditions are essential to "anchor" restoration work within a natural range of variability.
  3. Geographic distance (i.e., the "space-for-genotype" substitution) is a reasonable, but crude, substitute for patterns of gene flow among populations. However, in a disrupted landscape the most local remnant populations may be genetically reduced, and may not have genotypes that correspond to the conditions under restoration. Moreover, if populations are strongly selected to local habitat conditions, then habitat similarity may outweigh distance as a selection criterion. Local and regional climatic and soil zones can be useful criteria for obtaining material for reintroduction, since these will often represent plants adapted to the conditions at the restoration site.
  4. As a source of material for reintroduction, large, genetically diverse source populations are generally preferable to small, limited populations even when the latter are closer to the restoration site. It may be preferable to combine material from several suitable sites, to capture a wider array of genotypes that can succeed in the new location. Such "regional mixtures" should draw on populations within a similar ecological zone (climate, soil) to the restoration site, to avoid too large a proportion of individuals that may not be well adapted to the new conditions.
  5. Small local populations can be "swamped" by non-native genotypes if a species already exists at a restoration site. If existing populations are to be augmented, the number of individuals introduced from other locations should not be so large as to overwhelm the local gene pool, particularly if there is evidence of local adaptation.

3c. Sampling the diversity of source populations.

The initial gene pool for a restoration project is inevitably limited by the diversity of the original sample. While other genes may enter the project area over time (for instance, by migration of individuals, dispersal of gametes, or additional reintroduction measures), the starting pool of genetic diversity will govern the performance of a reintroduced population for a long time. Hence, the diversity of the original source collection is a critical consideration in restoration.

A variety of guidelines have been developed for sampling wild populations of plants and animals for breeding and reintroduction (Center for Plant Conservation 1991; Guerrant 1992; BGCI 1994; Guerrant 1996; IUCN/SSC 1998). These guidelines address four fundamental sampling issues:

  1. How many populations will be sampled to create the source pool?
  2. How many individuals will be sampled from each population?
  3. Will samples be collected all at once, or over a period of years?
  4. What is the probability of a collected sample surviving to establishment?

Number of populations: In most species, the cumulative amount of genetic variation captured increases as successive populations are added to the sample. However, since populations have some measure of similarity (0 < GST< 1), each additional population added to a sample collects some alleles that are new to the sample, and some that are already present from previous samples. As the number of populations sampled increases, the marginal diversity rate decreases (that is, fewer and fewer novel alleles are captured), and the cumulative diversity function approaches an asymptote. For a pool of populations sampled at random, there comes a point at which further sampling across populations provides little or no additional genetic benefit (Falk 1991; Falk and Gibbs In prep.).

The number of populations at which this occurs is related strongly to the measure of differentiation among populations, GST, which we have already encountered. When GST is high, populations are markedly different from one another, and more should be sampled to capture the maximum total diversity. When GST is low, most populations are similar, and sampling from only a few will capture most of the diversity that exists(provided that the statistic was generated using populations across the full range of the species). Of course, it is always worth remembering that a significant fraction of most genomes is monomorphic, and is therefore captured in the first population, if not the first individual.

There can be no hard and fast rules for the number of populations to be sampled across all taxa. In a recent survey of 13 plant species, the sample for some species was saturated for common isozyme alleles after just 3-5 populations (20% of the total), while in other species the cumulative diversity curve was still rising after all populations (as many as 15-30) were included in the sample (recall, however, that isozyme loci may be less variable among populations than other loci under strong selection) (Falk and Gibbs In prep.). Rare or uncommon alleles accumulate in the sample at a slower rate than do common alleles, so if it is suspected that important adaptations occur at low frequency (for instance, because of recent environmental change), then it is advisable to sample a larger number of sites. Further research is ongoing in this area; in the meantime, GST or some comparable measure of population differentiation is a suitable guide for the sampling strategy.

One exception to this general null model is for populations that are strongly differentiated along habitat lines (e.g. ecotypes). If the reintroduction area is unusual habitat for a species, then it is worth sampling any populations that occur in similar conditions, to increase the likelihood of capturing alleles that offer adaptive benefit.

Number of individuals to sample within populations. The underlying theoretical basis for sampling multiple individuals within a population is that populations are rarely truly panmictic (that is, with completely randomized breeding). In plants, a large proportion of mating occurs between neighboring individuals, even when pollination occurs via an animal vector. In animal populations, a wide range of behavioral adaptations exists that commonly concentrates breeding success in a few individuals at any given time. The result is that populations are not genetically homogeneous; to capture their genetic diversity adequately, multiple individuals need to be sampled.

Again, a number of general guidelines have been released on this topic, and again, there can be no hard and fast "magic numbers" that apply across taxa (Brown et al. 1990; Brown and Briggs 1991). Plant studies evaluated by the Center for Plant Conservation (Center for Plant Conservation 1991) and its colleague institutions suggest that anywhere from 10-50 individuals should be sampled per population, with the understanding that there is bound to be considerable redundancy among these samples. Sampling of propagules (seeds, vegetative shoots) has less impact on the source population, although there is often lower success associated with these life stages compared to outplanting established plants (Guerrant 1992; Guerrant 1996).

Number of years during which samples will be collected. For many species, an adequate sample can be collected in a single year. However, for many others, collecting may need to be distributed over a number of years. Reasons for multiple-year collecting include:

Probability of a collected sample surviving to establishment. In the end, what counts in a reintroduction is the number of individuals in the new population as well as their diversity (Menges 1991). However, less than 100% of the samples (seeds, cutting, eggs, adults) collected in the field will survive to establishment. Attrition occurs at every step along the way: during the collecting process, transportation, storage, propagation/curation, and outplanting/release. High initial mortality rates are often observed in reintroduced populations, often continuing for several years (Brown and Briggs 1991).

A simple calculation can help to account for attrition of collected samples. Let Ps represent the survival probability of a collected sample s, and N the number of individuals desired in the final restored population. Then we account for attrition during the reintroduction process by collecting samples in the field. For example, suppose that trees dug for transplant have an ultimate survival rate of 40% from initial collecting to the third year of a reintroduction project, and that we are trying to establish a population of 50 trees. Correcting for attrition, an initial collection of (50/.40), or 125 individuals, will give us a good chance of ending up with the final population size we wish (Brown and Briggs 1991).

3d. Genetic diversity within reintroduced populations.

As discussed above, breeding among closely related individuals can lead to expression of unfavorable traits that compromise an organism's ability to survive and reproduce. Individuals resulting from inbreeding in populations where this did not occur naturally from highly inbred lines can have stunted growth, altered behavior, deformed morphology, poor physiological function; such individuals are also more likely to be reproductively sterile. Hence, the genetic origin of individuals used in a restoration project can influence whether a project succeeds or fails. If source collections are made from diverse, naturally occurring populations, the probability of breeding among close relatives is reduced. A more diverse population may include individuals that can tolerate a wider range of conditions. If, on the other hand, all individuals in a restored population are genetically similar, and the total population size is not large (fewer than 100 individuals), the probability is great that their offspring will be homozygous at some loci, and thus have reduced fitness. A genetically narrow population may be able to survive only in a narrow range of conditions. While genetic diversity is not always associated with ecological amplitude, in general the correlation is positive (Huenneke 1991).

In addition to affecting how well a restored population succeeds in current environmental conditions, genetic diversity may affect performance over time. Important ecological traits, such as tolerance of disturbance or climatic extremes often have a strong genetic basis. Such traits need to be considered not only in terms of conditions at the time of reintroduction, but in light of the natural range of variability over many years or generations. If the goal of restoration is to establish self-sustaining populations (Pavlik 1996), then the range of conditions to which the population and its offspring will be exposed is a vital consideration. Genetically uniform populations may do well one year when conditions are favorable for their genotype, and then fail the next, even though the range of environmental variation is within normal limits. Such failures are even more likely in the case of less common extreme events (Gaines and Denny 1993). Recalling from § 1.b and 1.d that environmental conditions vary within an envelope of natural variation, it is safe to assume that most reintroduced populations will be exposed to a wide range of conditions over time. Hence, the genetic diversity of reintroduced populations is a non-trivial consideration for their long-term persistence (Montalvo et al. 1997).

It is not uncommon in restoration work for large numbers of individuals to be released or outplanted, a large proportion of which fail to survive beyond 2-3 years. This means that restoration releases are subject both to the founder effect (the result of the initially limited gene pool) and then further genetic bottlenecks as effective population size diminishes due to mortality (Robichaux, Friar, and Mount 1997). If the initial outplanting or release is itself genetically uniform (e.g., hundreds or thousands of clonally-produced plants), the resulting gene pool in the field can be quite narrow. Such apparently large populations are, in genetic terms, very small (Brown and Briggs 1991).

Of course, genetically narrow populations occur in nature, often resulting from similar forces (founder events and bottlenecks). For example, aspen (Populus tremuloides) and many species in the Iridaceae (Iris, Lilium) are often found in large clumps of many ramets of very few genets, or even just one. Natural populations of species where sexual (as opposed to vegetative or parthenogenic) reproduction dominates are less likely to be genetically homogeneous (Meffe and Carroll 1994).

The effects of genetic homogeneity in a reintroduced population may not be immediately evident, but over a period of years the population may have lower rates of growth, survival and reproduction, and may persist less successfully through periods of natural environmental variability.

What you can do: Restorationists should be aware of the level of genetic diversity that they are working with. Above all, restoration planners and managers should understand how the plants and animals they use were generated. Ask your nursery or whoever collected and propagated your seed source (or your breeder, in the case of animals) what methods were used, and what steps were taken to assure the presence of a suitably wide range of genotypes. While clonally-produced flats of thousands of identical plants may appear to offer short-term advantages (as they do for agricultural crops) of predictable response to current growing conditions, in the long run (again as with crops) such populations may be less likely to persist and succeed in the face of disease, competition, and climate variability. Of course, knowing the methods by which individuals or material was produced provides genetic information only by inference; it is unfortunately not common to have good genetic data for reintroduction efforts.

3e. Genetic diversity among reintroduced populations.

The same principles apply to diversity among population diversity. Most species have a degree of differentiation among populations (i.e., GST > 0). Population geneticists debate whether these differences have adaptive value in any given case - that is, whether they are the result of selection - or whether they represent the effects of sampling error (founder effects, drift, mutation). Nonetheless, a reasonable precautionary position is to mimic the degree of genetic difference found in natural populations of a species when (re)introducing populations to the landscape.

Most species, however, have little or no published data on their among-population genetic diversity. The few surveys that have been done are invaluable sources of information, and restorationists should read and cite them whenever possible. In some cases studies of congeners can be used, although with the caveat that species that are closely related phylogenetically are not always similar ecologically.

Next: Resources