Ecological Genetics and Evolution: Essays in Honour of E. B. Ford (1971) pp. 34-43
Plant Evolution in Extreme Environments
A. D. Bradshaw


Another very distinctive environment is the derelict areas produced by mining for metals such as copper, lead and zinc. These mine workings are scattered all over Britain. The heaps of waste, called tailings, assault the eye because of their desolate bareness. Many of them have been in existence for over 100 years, yet they have few, if any, plants growing on them. This is because the waste material still contains quite appreciable quantities of metals, up to 1 per cent. The metal is usually in the form of a sulphide which is almost completely insoluble. But the sulphide weathers very easily to more soluble compounds which are extremely toxic in concentrations as low as i part per million for copper or 10 parts per million for lead and zinc. The heaps are also grossly deficient in major plant nutrients, nitrogen, phosphorus and potassium, and have a very poor physical texture. No wonder therefore very few plants can be found growing on them.

But it is interesting that a few very characteristic plants do occur and may grow quite happily. In Britain the commonest are species of grass, Festuca and Agrostis in particular. In other parts of the world a whole range of different species can be found. The flora of these metal contaminated sites is so distinctive that in the past prospectors have used the plants to help their exploration for new ore bodies (see Antonovics, Bradshaw and Turner 1971).

All the species that grow on metal contaminated soils also grow on normal soils where they are mixed with a lot of other species. How is it possible for a few species to grow in this very distinctive extreme environment when others are not able to do so? There are two alternative possibilities: firstly, that these few species possess throughout their range the ability to tolerate the metal toxicity, or secondly, that they possess the ability to tolerate metal toxicity only where they are growing on toxic soils. The second possibility implies that the species can evolve tolerance under the influence of natural selection. In 1934 Prat showed that the second alternative was true for Melandrium sylvestre growing on copper mine workings in Central Europe. Since then all the cases that have been investigated have shown the same thing.

PLATE 2.1. Root growth of copper mine (left) and pasture (right) populations of Agrostis tenuis in 0.5 ppm copper solution.

An elegant technique for testing the tolerance of mine populations has been developed. It was found that the major effect of metals is on root growth and that tolerant mine populations possess the ability to continue rooting in much higher concentrations of metal than normal populations (Bradshaw 1952). This difference can readily be demonstrated in solution (see plate 2.1), and tolerance expressed as the ratio of root growth in toxic solution to growth in normal solution (Wilkins 1957). There are many different mines containing different metals either singly or in combinations. Tolerance is specific for individual metals so that a population that grows on a copper soil is tolerant to copper only while a population that grows on a soil containing copper and nickel is tolerant to both copper and nickel (Jowett 1958, Gregory & Bradshaw 1965) (fig. 2.9).

FIGURE   2.9. The indices of tolerance to different metals of four populations of Agrostis tenuis from areas contaminated by different metals (from Jowett 1958) (1, Parys Mountain, Anglesey; 2, Black Forest, Germany; 3, Goginan, Cardiganshire; 4, Capel Bangor, Cardiganshire).

What is the genetic basis of this character? All gradations of tolerance can be found. Some of the variation in tolerance could be environmental in origin, but critical examination of a range of individuals shows little sign of discontinuity of variation. This is particularly so in A. tenuis (McNeilly & Bradshaw 1968). In F. ovina some discontinuity is found (Wilkins 1960). More work is necessary but at the moment it appears that several genes are involved in whatever species metal tolerance is found.

The selective factors evoking metal tolerance are very easily demonstrable. If normal seed of A. tenuis is sown on metal contaminated soil it germinates, produces a coleoptile but no roots. After a few weeks the seedlings die. Tolerant material however germinates and establishes normally. Material of intermediate tolerance behaves in an intermediate fashion. Selection for tolerance is therefore absolute and the factors maintaining tolerance are very clear cut. However in the reverse situation the performance of tolerant seed on normal soils is not the reverse. Both tolerant and normal seed germinate and grow very successfully on ordinary soils in most species. Subsequently there may be some difference in performance. In A. tenuis when spaced plants are examined there is very little difference in the performance of tolerant and non-tolerant material but in Anthoxanthum  odoratum tolerant material is at a distinct disadvantage (about 40 per cent) (Jain & Bradshaw 1966). But fitnesses determined by examining single spaced plants are of limited use. In natural non-toxic situations there will always be very severe competition from other species and it is therefore necessary to determine fitness under the full impact of competition. When this is done quite different values of fitness may be obtained (McNeilly 1968) (Table 2.2).

TABLE 2.2 Effect of competition between a mine and a pasture population of Agrostis tenuis on their fitness, measured as dry wt. yield per plant, grown on two soils (McNeilly 1968).

Populations Parys Mountain (mine) Llandegfan (pasture)
In monoculture In mixture In monoculture In mixture
On acid soil 1.44 0.59 1.42 1.37
On basic soil 0.46 0.24 0.46 0.54
  l.s.d. 5% prob. 0.28

The operation of these selective factors in nature can readily be seen in mine populations when the tolerance of adult material is compared with the tolerance of seed material produced by those adults when on the mine and when in isolation. Isolation seed shows the same mean as adult material, but a greater variance. Seed produced on the mine shows a distinctly tower mean which must be due to gene flow from neighbouring non-tolerant populations. Since the adult population represents a stable situation, selection must be acting between seed and adult in a directional and stabilizing manner to maintain the characteristics of the adult population (McNeilly 1966) (fig. 2.10).  The power of selection is even more clearly seen when looked at in a whole series of populations being subject to gene flow (see fig. 2.12).

FIGURE 2.10. Histograms of copper tolerance of a mine population sampled as adults, and as seed collected in situ and in isolation (from McNeilly 1966).

Mine habitats are usually extremely restricted. Even large mine workings are seldom more than 500 metres across, many mine workings spread for only 100 metres and some may consist only of a dump of material a few metres across. They will always be surrounded by normal habitats very often containing extensive amounts of the species which has evolved tolerant populations on the mine material. Despite this, tolerant populations can be found on extremely small areas of mine waste. If transects are made across the boundary between contaminated and normal areas the tolerance of the populations sampled is found to change precisely at the boundary. There is very little sign of gradation on either side of the boundary. This is very clear in A. odoratum (fig. 2.11). In many ways such a steep cline, if it can be called this at all, is surprising, because we are dealing here with a plant which is outbreeding and wind pollinated. Gene flow by pollen movement would certainly be expected to blurr the distinction of the populations on the two sides of the boundary. The patterns of differentiation that can be expected with different degrees of gene flow and selection have recently been examined (Jain & Bradshaw 1966). Very sharp clines are to be expected with normal amounts of gene flow and selective advantages on each side of the boundary of about 50 per cent. But the sharpness of the transition shown by A. odoratum is remarkable. It is possible the gene flow is not as high as expected because of perenniality, but this does not seem to be so since the average life of Anthoxanthum individuals is only about two years (Antonovics, in Harper 1967).

FIGURE 2.11. The zinc tolerance of populations of Anthoxanthum odoratum at the boundary of Trelogan mine (from Putwain, in Jain & Bradshaw 1966).

What then can be happening to allow such localized differentiation of populations? There will be selection for any mechanism which reduces gene flow across the boundary. One major way that this can be achieved is by differences in flowering time. In an extreme environment such as a mine habitat there is likely to be selection for earliness of flowering so that seed production is completed before severe summer conditions occur. There is plenty of evidence of evolution of such differences for ecological reasons, differences which will have the incidental effect of reducing gene flow. But there is also evidence of enhancement of flowering time differences at the boundary of mine populations which can best be interpreted as arising as a result of selection specifically for isolation (McNeilly & Antonovics 1968). Another way in which gene flow can be reduced is by self-fertilization. It is very interesting that in Anthoxanthum and Agrostis much higher levels of self-fertility are found in mine populations than in normal populations (Antonovics 1968). In metal contaminated sites elsewhere in the world there are suggestions of a variety of ways in which reduced gene flow has been achieved (Antonovics, Bradshaw & Turner 1971). Although the colonization of an extreme habitat does not demand the isolation of its emergent population from the original populations surrounding it, there will be selection for isolation mechanisms, and it looks as though we can find this isolation in progress of evolution in metal contaminated habitats.

If gene flow is likely to be occurring between mine and normal populations we should be able to pick it up quite easily, particularly since the populations are so distinctive and we have an easy way of measuring the character even if its genetics is not clear yet. A small mine population surrounded by normal populations is an ideal situation in which to see what is going on. The mine at Drws-y-coed in Caernarvonshire is of this sort, and it has the added advantage that it is situated in a U-shaped valley in which the wind is highly polarized and normally blows from the west. The gene flow has been investigated by examining the difference between the tolerance of adult populations and the seed they produce in situ (McNeilly 1968). Two transects were made, one in a predominantly upwind and the other in a predominantly downwind direction. In the upwind direction the adults show the sort of very sharp transition in tolerance at the boundary that we have already seen for Anthoxanthum, but in the down wind direction the transition is by no means so sharp and populations some distance away from the mine in completely normal pasture have a quite considerable level of tolerance (fig. 2.12). The difference between the two transects is surprising. It certainly cannot be explained by differences in the sharpness of the environmental boundary:

FIGURE 2.12. Copper tolerance of populations of Agrostis tenuis across Drws-y-eoed mine sampled as adults and as seed (from McNeilly 1968).

the transition from mine to normal conditions occurs over the same very short distance in both transects. The explanation appears when we look at the tolerance of the seed populations. In the upwind transect the seed populations on the mine are less tolerant than the adult populations that gave rise to them. In the down wind transect the seed populations in the pasture are much more tolerant than the adult populations that gave rise to them. This fits in with the suggestion that the gene flow is highly polarized by the wind direction. In the up wind transect where there is flow of non-tolerant genes into tolerant populations this gene flow is kept in check by the very severe selection occurring on the mine. But in the down wind transect the flow of tolerant genes into the non-tolerant populations is not kept in check since the selection pressures here are not so severe, as we have already seen. The result is that genes move out of the mine populations more readily than they move in.

Selection retains the upper hand and, despite the gene flow, the distinctiveness of the mine population is maintained. The situation is therefore very similar to the one that we have already seen in Agrostis stolonifera. An extreme habitat, by the very fact that it generates high selection pressures, can maintain a distinctive population even if there is gene flow into it. The reverse gene flow from the mine populations into surrounding normal populations is not very great and is kept in check by selection. But it does occur and it must be remembered that gene flow in a plant such as Agrostis is markedly leptokurtic: a small amount of pollen is carried a large distance. Low frequencies of tolerant individuals in Agrostis populations can be determined by sowing seed samples on appropriately toxic soils: tolerant individuals survive, non-tolerant ones do not. When this was done for a series of seed samples taken at distances of several kilometres round the large mine at Parys Mountain in Anglesey, tolerant individuals were found in frequencies above average in populations a long distance away from the mine, particularly down wind (Khan 1969) (fig. 2.13). The down wind end of the transect was limited by the sea, but the results are sufficient to show that gene flow at an extremely low level can carry genes a very long way away from the original habitat. Genes can therefore migrate long distances and be available for selection in new extreme habitats when they arise.

FIGURE    2.13. Frequency of copper tolerant individuals in seed samples of populations of Agrostis tenuis around Parys Mountain copper mine (measured as number of survivors on copper soil slightly diluted with ordinary soil) (from Khan 1969).

In A. stolonifera we saw that it was possible for a highly localized population to arise consisting of only a few individuals with very distinctive characteristics. If this can happen in a cliff situation where there is selection for morphological characters, it can surely happen in metal contaminated areas where the selection is probably even more severe. There is evidence of this from the remarkable observation of Snaydon (Bradshaw, McNeilly & Gregory 1965) that the plants of Festuca ovina and A. canina immediately underneath a zinc coated iron fence were significantly more zinc tolerant than those about 30 cm away. Here, as for the A. stolonifera growing in the stream, we can hardly consider that there is an independent population, but high selection pressures, together with vegetative propagation, has enabled this extremely localized habitat to be colonized by a few adapted individuals.

FIGURE 2.14. Height of seedlings of a copper mine and a pasture population of Agrostis tenuis (grown for three months on copper soil slightly diluted with ordinary soil) (from Khan 1969).

What are the origins of metal tolerance? Mine workings are often not very old, so that it is tempting to believe that the mine populations are themselves not very old. But many mine workings were preceded by ore bodies exposed at the surface giving toxic soils which could well have been in existence since the glacial period. This is very clear in other parts of the world where the ore bodies have not been disturbed (Nicolls et al. 1965). So natural mine populations could have a very long history. But if non-tolerant seed is sown on to metal contaminated soil, although nearly all the seedlings die, one or two per thousand survive and grow very well and a few more grow rather more weakly (Khan 1969). A histogram of the size of the plants in such an experiment is given in fig. 2.14. The few large individuals are the only survivors, the rest are dead. If survivors of varying heights are grown on and tested for tolerance it is found that there is an extremely good correlation between size and tolerance. The few large individuals have tolerances similar to that of plants from mine populations. Metal tolerant individuals in Agrostis are therefore to be found in normal non-tolerant populations, although at very low frequency. With very severe selection a tolerant population can therefore be produced from a non-tolerant one in the course of only one or two generations. The time scale for the evolution of individuals to colonize this very extreme environment is therefore very short.