Director Christine Shaw
It is self-evident to say that the output measurements should be set to match the objectives of any test ! However, it is always a temptation to include additional measurements "in case they may be useful". There is no harm in this, provided it is not used as a substitute for thinking about what is really needed ! Any additional measurements to those necessary should, nevetheless, be selected based on some concept of the need for them or the purpose to which they might be put.
Another cardinal rule is that the likely type of analysis of the data should be considered at the outset, so that the data may be collected in such a way as to be appropriate. This particularly includes the methods of sampling, sub-sampling (if necessary) and the numbers of replicates and duplicates (they are different !), which may be appropriate.
Since this page is under the Yields of Cereals banner, it is clear that measurements relating to yield are immediately relevant. While "crude" yield has a value, clearly, survivability, reliability, ability to compete with weeds, storage qualities and other characteristics are all matters that require attention.
However, research into the performance of cereals "of the Iron Age" is clearly intended to help answer other questions, such as how did each perform in different environments, would the farmer choose one genotype compared with another for different climates, soils or needs, what made the farmer switch from one type to the next, as it appeared by natural genetic modification and probably many others.
Throughout this text it is taken for granted that the criteria ultimately chosen will all be reflected in the experiment duration. It should already be more than clear from other pages that crop trials must extend over many years for any meaningfulness to be obtained, especially allowing for the fact that such data is ultimately destined for extrapolation into the long distant past.
It is implicit in long term research that the seed failure rate is one of the most important criteria that any farming community will be aware of in its crop selection strategy. Correspondingly, quite independent of the impact of the weather cycle on output data there are questions relating to the germinability of the seed over the years and from successsive years of yield. It seems to be a generality for crop cereals that the germinability decays over a period of about 7 years to a value that is effectively zero. It is interesting to reflect that this may explain the significance of seven years of famine in the bible, since the implication is that seed resources for future planting are diminished to the point of requiring an external supply to replenish stocks.
This factor alone sets a period of seven to ten years of any meaningful data set from which speculation about farmers' choice of cereal might be made. The climatic effects are normally regarded as extending the experiment duration to twenty years as discussed elsewhere (The Weather Station).
The next factor worthy of contemplation is the time of emergence which relates to minimum ground temperature at which a given cereal will germinate. Naturally, too early a germination may expose the tender young plant to subsequent frost damage, so early emergence is not the be-all and end-all it might seem, so that one cereal may be favoured over another according to the local climatic pattern. The conventional input variable measures used are the ground surface temperature and the soil temperature at both 50 mm and 100 mm depth.
The exact time of emergence will fluctuate from year to year and over longer term climatic cycles but in the UK one suggestion has been that the time of germination correlates with the date when the sum of the excess degree days from mid-February exceeds 43 degrees Fahrenheit.
Further issues of plant survivability and ability to compete with weeds merit study before one may be able to test meaningful propositions about choice in a given climatic and soil zone.
An obvious question relates to the behaviour of a cereal if animal cropping occurs despite efforts through the use of field systems to avoid it. The relevant property is that of tillering, the tendency or ability to throw up side shoots separate from the principal germination stem(s). The same property may enhance the response of a given cereal to early drought and allow recovery in the event of subsequent rainfall. Einkorn has proved particularly resilient in such circumstances and still exists as a catch-crop, mainly in the Eastern Mediterranean, albeit largely for animal feed. A word of caution for the researcher is that it is an exceedingly difficult characteristic for which to define a recording protocol and exceedingly time consuming in practice.
This property is related to the tendency for a grass or cereal to exhibit a a tufted or clumped habit. Where fertility is low and single stem plants may be well spaced, then the tendency to tiller may minimise lodging (laying flat) in wet and/or windy conditions. While not so critical with modern harvesting methods, lodging is a severe impediment to the use of tools such as sickles and promotes damage by mould etc.. A general record of the crop area subject to lodging should be considered.
There are several other measures after emergence and ripening that are relevant to the cereal's behaviour under different conditions. It will be recognised that many of them are inter-related, so that a clear purpose in collecting each element of the data should be set at the outset.
Stand height, which varies in response to the input variables of weather, especially rainfall and temperature, and soil fertility and manuring, as a minimum, is also a character of a given cereal type. Greater height may be regarded as a competitive advantage, so long as other plants, especially weeds, can be outgrown, since a better access to light should ensue. However, weeds themselves may respond differently in different circumstances, so it is relative response that is crucial. To this end it is necessary to check the relevant input variable by undertaking a survey of the background weed community on an annual basis.
There is an ancillary issue that relates to plant height wherein modern demands differ from the past. Straw has many uses, but especially for thatching. Local demands in the past might have put a premium on this quality, thereby biassing the decision to use one cereal against another. Nowadays, straw is regarded as a waste product and there has been considerable success in developing short stem varieties, to be grown using wide-spectrum herbicides. These varieties do not thrive under the competitive conditions of the trial schemes used for evaluating crops of the past.
Not all the grains that exist at
ripening will reach the point at which the final gross
yield is determined i.e. ready for storage. For example,
the tendency of seeds to ripen at different times along the
length of the head will determine the tendency for seeds to drop
to the ground, whenever harvest may be delayed. In some plants,
such as tobacco, where leaves ripen successively, harvesting is
sequential, starting from the base of the plant and working up,
in rotation, to reflect this. This strategy could be adopted in
grain harvesting by hand, though it is not known to occur widely.
The tightness and toughness of the seed heads may reflect an
ability to resist loss to birds and small mammals. These
tendencies are not easily measured directly, so that indirect
measures may be adopted. These include the following measurements
on samples of individual ears:
spikelet length: number of seed pockets: number of
pockets full or empty: weight of individual seeds.
Finally, it is often instructive to incorporate the issues of nutrition or food value. The commonest measure is protein.
For the dedicated student, who wishes to assess their absorption of the lessons in these pages, the following excercise is offered.
Tutorial excercise for students.
Students who have sought to digest the preceding pages under the "Yield of Cereals" banner may like to test themselves by reflecting on the following questions relating to a paper by van der Veen and Palmer (1997).
The data used to claim the conclusions given in the paper was derived from plots which were only 1 metre square. Students should re-read Genotypes to recognise the serious consequences of this in relation to the validity or otherwise of the data resulting. Students with a wider interest, in the relationship of research plot size to the greater areas to which land management practices are applied, might like to read Andersen (1997). Dr Andersen "recently" (1999) gave three lectures in the Distinguished Ecologist Lecture Series at the Annual Symposium of the Australian Institute of Biology, Darwin, Australia. He has written three books and has 87 refereed papers published. Although the paper quoted from here emanates from his principal specialisation, he deals with other fields, especially fire management (one of the oldest of man's approaches to agriculture and still applied today) and, by allusion, with grazing and logging research. He is particularly clear "that many research studies are undertaken on plots inadequately sized to guide the understanding of land management issues on a wider scale."
Students might like to reflect on the review process of the publishing Journal. Would the same paper have been published at all had it been submitted to a journal specialising in agricultural science?
The process of peer review must necessarily remain anonymous or there would be even more concern about the way this is undertaken, than is sometimes the case now. Do you think the journal editors gave adequate consideration to the necessary scientific background to any reviewers selected or do you think that they may have overlooked any criticisms received?
The summary states that crops were grown on 22 sites. Careful reading of the text is required to show that this number was reduced by "data loss". Three sites gave no data at all. It requires even more work to construct the table of actual data gathered site by site, species by species, using the various Figures presented. The resulting tabulation suggests that in Figure 4 of the paper, there is a "rogue" point for site 8. The authors themselves acknowledge in their text that site 4 gave an "outlier" result but have clearly not fully excluded the data from the work, as one might reasonably expect. Final inspection suggests only 16 sites gave a complete 3 year data set for all species. One might reasonably suppose that, in view of the type of data analysis carried out and the sweeping conclusion given in the summary, only the data for these sites be used. Does it give a proper reflection of the potential worth of the paper to use the number of sites as 22 in the summary?
The principal author declined specific answers to these and a number of other queries put to her.
We now turn to the data analysis. One may wonder at a procedure using some of the most powerful statistical approaches available, when the value of the data employed is at the least questionable, at the worst valueless (stemming from a plot size that would be regarded as meaningless in any mainstream agricultural research, down to the fact that Dr van der Veen gave no assurance that her "volunteers" - presumably not trained researchers - had been given written protocols). Does this highlight the weakness of multivariate statistical methods in throwing up apparently meaningful effects of variables when there may be none?
The principal author declined to reply to the request to discuss this with the statistical advisers used.
Even if by some chance the short term conclusions from this data analysis actually stand up in comparison with fuller work, is there any basis to expect that extrapolation to the past, incorporating all possible locations and weather variations, can support the extravagant claim that ".. the switch from emmer to spelt wheat may, instead, have been the result of changes in cultivation regime during the period in question"? Did the authors consider the consequences of crop failure or germinability of the cereals tested and whether their test duration had any relevance to the many issues that would determine such choices as implied in claiming a "switch" from one crop to another?
Students may conclude that the publication of this paper does a disservice to experimental archaeology.
References
Andersen A.N. 1997 "Using ants as bioindicators: Multiscale
issues in ant community ecology." Conservation Ecology
[online] 1 (1) 8. http://www.consecol.org/vol1/iss1/art8
An extensive list of references is
included.
Marijke van der Veen and Carol Palmer, "Environmental Factors and the Yield Potential of Ancient Wheat Crops", Journal of Archaeological Science (1997) 24, 163-182.
Created 01 August 2001 - Updated 21 January 2002