by Martin Jones
These bushes could also be constructed from the protein data that became available before the DNA evidence was sufficiently well studied. As that protein data accrued for different crops, so the pattern they produced repeatedly fell closer to Childe than to Higgs. The protein diversity among the wild plants was great, giving the phylogenetic bush a broad crown. When domesticates were added to the bush, their own diversity was seen to be much narrower. They sprang from a small part of that crown. Those patterns on the constructed bush could be transferred to a map, according to where the wild varieties grew. In the case of maize, for example, wild maize plants are found in various parts of Central America. The protein patterns in domesticated forms, however, relate them all to a single line, which grows today in the Jalisco Valley, not far from Mexico City.
As DNA methods came on stream, something similar could be done with the molecules at the heart of life. A recent example is the work of Manfredd Heun at the Agricultural University of Norway. He looked at the DNA of one of the most ancient wheat species around. It is a slender plant called einkorn wheat, which is nowadays rare, but was an important crop in early prehistory. From bio-archaeological evidence, it appears to be one of the first, perhaps the very first, plant species displaying physical evidence of domestication. Heun compared a range of modern wild and domesticated einkorn wheat samples using a method not dissimilar from the DNA fingerprinting employed by the police to catch criminals. This method is different from the DNA analyses featured in other chapters, that target and work with a single stretch of DNA sequence, bounded by carefully designed PCR primers. In Heun’s method, as in DNA fingerprinting, restriction enzymes are used to fragment the total body of DNA into many strands of various lengths. Any pattern within this constellation of lengths is a direct reflection of where along the DNA strands the chosen enzymes have been able to bind to and break them. A convenient way to reveal this pattern is to allow the lengths to migrate and separate along a suitable gel which, when stained, produces the characteristic ‘fingerprint’ by which one wheat variety, or indeed one criminal, is identified from many others.
The technical term for the patterns studied by. Heun is ‘amplified fragment length polymorphisms’, normally abbreviated to AFLP. Heun established the AFLP patterns for a range of stands of wild einkorn from Turkey, Syria and neighbouring countries. He then added selected populations of domesticated einkorn to the analysis. When the different patterns were grouped according to the respective similarities, whatever statistical method was used brought the domesticated einkorns very close to one particular population of wild einkorns.
This wild einkorn population was from the slopes of the Karagadag mountains in south-east Turkey. No more than 100 km from these hills, a Dutch archaeological team had been digging a very ancient settlement at a place called Cayonu, an early fixed settlement that had yielded some of the earliest carbonized einkorn grains. So here, within walking distance, was a contender for the earliest domesticated cereal deposit in the world, and a stand of wild cereals that could be the ancestor of all domesticated einkorn. Heun’s final words in a paper in Science (1997, 278: 1314) were suitably circumspect:
Localization of the precise domestication site of one primary crop does not imply that the human group living there at the end of the Palaeolithic played a role in establishing agriculture in the Near East. Nevertheless, it has been hypothesized that one single human group may have domesticated all primary crops of the region.
In the same issue of the journal, Jared Diamond was keen to push the boat out rather further:
[A] long straight line runs through world history, from those first domesticates at the Karagadag mountains and elsewhere in the Fertile Crescent, to the ‘guns, germs and steel’ by which European colonists in modern times destroyed so many native societies of other continents. (1997, 278: 1244)
Even The Economist (15 November 1997: 127) was moved by that ancient business opportunity emerging from the ruins of Cayonu and its neighbours:
[T]hese sites really are the earliest evidence of agriculture, rather than merely the earliest that archaeologists have yet discovered. It also suggests that the people who built them were the most important inventors in history. The West certainly owes its existence to them. So in all probability do the civilizations of India. Even China is suspected by some to have drawn its inspiration from what they started in the Middle East. And we do not even know their names.
Not just in the press, but also in the research community, the molecular evidence stood in contrast to Higgs’s idea of a widely dispersed adaptation to a radical event at a particular location, or at least a very small number of such locations across the world. Although dispersed evolution was more in keeping with modern ecological thinking, the idea of a great step forward in the human story was making a comeback. Furthermore, the various early European dates outside the Fertile Crescent had fallen by the wayside, victims either of flawed dating or of flawed identification. The acceptable carbon dates seemed instead to radiate out from the Crescent.
It seemed clear where the molecular evidence was going, but not as clear how much the archaeological evidence was in support. In general terms, the archaeological chronology worked. It had in any case been used in the genetic argument. However, the fine detail of the archaeological record revealed a picture that was more complex than the simple model of radical origin and dramatic spread. First of all, it proceeded at an imperceptible pace. A thousand years after the first domesticated crops appeared within the first generation of farming settlements, the Near Eastern landscape was far from transformed, and hunter-gatherer communities were still widespread. The new approach to food had not yet begun to spread across Europe. When it did, its spread was surprisingly discreet, with negligible impact upon the woodland covering much of Europe for several thousand years. Wild resources continued to be hunted and gathered even by those tending cereal plots. The whole process seemed slow, bringing to mind gradual adaptation rather than revolutionary change.
To bring the explanations coming from archaeology and genetics closer together, various researchers have attempted to look at the genetics of archaeological specimens, by amplifying ancient DNA from the battered plant fragments that were increasingly being recovered from archaeological sites.
the ancient molecules
It still surprises me how uninquisitive we can be about aspects of what we observe that lie beyond the realm of our immediate questions. By the time Terry Brown and I had been drawn together by our common curiosity about the survival of DNA in ancient plants, I had been studying blackened fragments under the microscope for almost two decades, drawing them, measuring and counting them and recording cells, hairs and other features. When Terry posed an innocent and perfectly reasonable question for a molecular biologist–what are they made of?–I surprised myself by how speedily I exhausted my rather flimsy knowledge. I suggested to him that they were largely composed of carbon, that they had been exposed to a reducing fire, and that the steam released had generated the bubbly, honeycomb structure and the black residue. However, I knew full well that so-called ‘carbonized seeds’ come in various shades of dark brown and black, and by no means all display the honeycomb structure. We archaeobotanists had shown little interest in even the most basic of chemical assays to check this out, and remained happy to continue observing, drawing, counting and so on. Much the same was true of other forms of seed preservation. Seeds from both waterlogged and desiccated deposits survived by virtue of the cessation of biological activity in the absence of air or water–another over-simplification. Mineralized seeds were a kind of fast fossil, in which a calcium salt in the surrounding matrix had replaced much of the decaying plant tissue.
It is now much clearer how we have over-simplified each of these categories. Like other bio-archaeological materials, the majority of ancient plant remains persist after the routes to decay and breakdown have been partially, rather than fully, arrested. It is not easy to completely reduce a plant fragment to
carbon. Charcoal burners build very specific fires to achieve this. Nor is it easy fully to exclude either oxygen or water. The oxygen in a waterlogged deposit may be extremely low, but such deposits rarely offer a complete barrier to diffusion. Similarly, a ‘desiccated’ seed may retain a surprisingly large quantity of water in its make-up. Mineralization is by its very nature a partial process. So we have many plant remains in which molecular breakdown has been considerably reduced rather than stopped, and sufficiently so for their visual form to be studied in detail. Their molecular preservation was likely to be varied, and their DNA content was difficult to predict.
We assembled ancient cereal grains in a variety of different preservation contexts. Desiccated wheat grains excavated from the Jordanian desert seemed the best bet, and some of these were also partly mineralized. We were assembling these soon after the publication of Miocene Magnolia and the problems of water were not widely appreciated. So the waterlogged wheat grains we recovered from a mediaeval well also seemed a likely source. Finally, the most ubiquitous but least promising category, carbonized material, was an unlikely source. Even if they were only partially exposed to fire, the heat would surely destroy the DNA. However, these blackened seeds had emerged as the core of the fast-growing archaeobotanical database around the world. Brown built a probe that was specific to wheat DNA, and which carried a radioactive marker. If any of the sample retained wheat DNA, the marker would bind and emit sufficient radiation to mark a photographic film with a spot. Both the desiccated and waterlogged wheats came up positive; some of their DNA had survived. We were taken by surprise by the carbonized seeds, which also came up positive.
For our sample of carbonized seeds, we had chosen a rich cache of cereals from an underground pit at the prehistoric hillfort of Danebury in southern England. By this time, we had many specimens from which to choose. It was a chance selection of numerous assemblages that was taking up rows and rows of storage in our Cambridge archive. With hindsight, we can estimate that at least 80 percent of the assemblages in that archive would have completely lost their DNA. Purely by chance, we chose one that had not. It would have been perfectly in order to discard carbonized seeds after that first negative in the pilot study, but it just so happened we had chosen the one assemblage in six that encouraged us to continue. It was now possible to go to the heart of the domestication issue in search of DNA.
a suitable blueprint
The ancient DNA research on humans had made much use of Anderson’s sequencing of the human mitochondrial genome. Food plants also have mitochondrial genomes, but the same intense effort that has been applied to sequencing our own DNA has not been lavished on the very many plant species we use for food. There are now some detailed mitochondrial maps for plants, which may figure more prominently in future studies. In the meantime, there is another structure within the plant cell that can be explored in a similar way. While mitochondria are the cell’s powerhouses, burning up energy to fuel the cell’s activities, the task of the chloroplasts is to capture that energy from the sun. Just as the cell has several mitochondria, it also has several of these bright green chloroplasts. Just like the mitochondria, they also carry a circular DNA molecule, and are normally inherited through the maternal line.
The chloroplast genome in plants is considerably larger than that of the mammalian mitochondrion, and is highly conserved across the plant kingdom. While it has been used to explore the ancient evolutionary origin of the plant kingdom as a whole, it has been less widely used in the kind of micro-evolutionary story of interest in the human past. For that, a greater focus has been placed upon the chromosomes within the cell’s nucleus. In most plant species, the nuclear sequence is only patchily known, but certain genetic regions are very well charted. These tend to be around genes with some economic significance, such as those affecting yield, disease resistance or cooking qualities. At either end of the genes are the long non-coding regions that occupy so much of the nuclear genome, the so-called spacer regions. The punctuation marks within the DNA sequence that signal the start and finish of the reading of an individual gene are referred to as the promoter and terminator sequences. The non-coding DNA is not confined to the stretches that lie outside the genes themselves. At various points on a gene, lengths of DNA sequence are apparently skipped over or edited out when it comes to actually assembling the sequence of amino acids within a protein. They are called ‘introns’.
Hence we have a number of different kinds of sequence within and around the gene itself–spacer region, promoter and terminator sequence, intron–that are separated from the eventual design of proteins. We assume therefore that they are also separate from the direct impact of the forces of external, whole-organism selection. Instead, they serve as valuable accumulators of neutral evolutionary mutation, recorders of evolutionary history. Various of these have been brought together to probe into the ancestry of nuclear genes.
agriculture in the new world
It was back in Allan Wilson’s Berkeley lab, the home of so many innovations in this field, that the first attempt to amplify ancient DNA from a crop plant was under way in the early 1990s. The crop in question was the prime example of the transformations that prehistoric breeders could effect. The bulky maize cobs that end up on our table bear little obvious resemblance to the slender-headed wild ancestor from which they were bred. In genetic terms, however, domesticated maize is so close to its wild ancestor, ‘teosinte’, that many experts regard them as members of a single species. Yet in a few thousand years, domestication has turned a modest grass head into the high-yielding monster we feed on today.
Maize is a plant of enormous diversity. The cobs and kernels each vary considerably in size, shape, colour and texture. In the modern period, North American maize has been modernized and streamlined into a select suite of successful varieties, but it is still possible in the rural parts of South America to find an extensive and colourful diversity of forms. This variability has led some to query the notion that in the New World, like the Old, crop domestication had a point of origin in space and time. Perhaps a scattered origin would better account for the regional diversity, especially in the south. However, a mounting body of genetic and archaeological evidence has been assembled to support a single point of origin. The genetic evidence related in the first instance to isozymes. The protein variation among domesticated maize was narrower than among wild teosinte, and tallied with particular populations of teosinte growing today in the Jalisco Valley on the west coast of central Mexico.
At roughly the same latitude on the east coast of Mexico, Richard MacNeish mounted one of the pioneering archaeological explorations of early agriculture in central Mexico. His team worked through the deep deposits of the rock shelters visited in prehistory along the Tehuacan valley. Within these were some of the earliest levels containing maize cobs. A whole series of cobs in different shapes and sizes was recovered, ranging from diminutive forms much closer in scale to that of the wild teosinte, through to the substantial form of cob more familiar today. At first, the time scale of agricultural beginnings in the New World appeared to have an antiquity similar to the Old, but refined carbon dating has brought the dates forward by several thousand years. Nevertheless, a time sequence could still be constructed beginning in central Mexico, close to the most similar populations of wild teosinte today. Here domestication began and successive generations of cob swelled to their familiar enlarged size. The carbon dates then trace the spread of maize farming north and south, where, alongside beans and squash, it went on to fuel some of the best-known urban civilizations of pre-contact America. The single-origin model seemed to fit, but not completely. A few genetic points needed tidying up.
The diversity of maize was not just related to physical appearance. If it were, it would amount to a weak source of evidence, as so little is known about the relationship between physical and genetic diversity. It also related to the DNA sequence. Some sequences that had been studied argued against a bottleneck in the last few thous
and years. One such sequence was the promoter region of gene associated with the metabolism of alcohol within the plant, a gene labelled adhi. Enough is known about these genes across the grass family to make an estimate of mutation rates within, in other words to calibrate its molecular clock. For this sequence, the average divergence rate of two individual plants was calculated at 1.6 percent per million years. To put this another way, imagine gathering together all the modern progeny from the very first kernel planted by the very first American farmer. Among the vast assemblage of modern kernels so gathered, the sequence of any selected pair of kernels would deviate from one another along this sequence by an average of 0.008 percent. The deviation observed by the Berkeley team was greater, by several orders of magnitude. They measured it at 2.2 percent. Even if we allow a few more farmers to scatter several kernels, the breadth of genetic variation seemed large for a single event.
As the genetic consensus was for a single domestication event, some geneticists had suggested this might have to do with a molecular clock that accelerated after domestication. The estimate of a convergence well over a million years back needed to be reduced accordingly. Even if this were far-fetched, we have already seen how the precision of the clock is being sorely pushed at this level of resolution, and is in need of corroboration by the far more secure dates from archaeological specimens.
Pierre Goloubinoff, a graduate student of Wilson and Pääbo, set about assembling some suitable archaeological specimens of early maize, with a particular emphasis on South America, where such striking diversity is seen today. One group of kernels was from the north coast of Chile. They had been added to a human burial 1,500 years ago, and survived in a state of desiccation. Another group came from Christine Hastorf’s excavations of a Wanka settlement in the mountains of Peru. They were from around the time of European contact, and were carbonized rather than desiccated. A much older cob, whose mode of preservation was not specified, came from the north coast of Peru. At 5,000 years old, it was significantly closer to the earliest maize cultivation than to the modern epoch.
