Research to support the development of organic food and farming - Organic Agriculture: A Global Perspective - page 369

horticulture or agriculture. Few studies compare systems of perennial cropping, one exception
is the Washington State apple production systems trial (Reganold et al . 2001). Within Table
15.3, treatments have been classified into organic, biodynamic, low input or integrated and con-
ventional on the basis of the author's descriptions. One difficulty in interpreting these experi-
ments, and indeed making comparisons between them, is the lack of definition of the terms of
low input and integrated farming. Furthermore, there is huge variation in farming systems
across relatively small geographical regions (Trewavas 2004). There is also a lack of information
available in the public domain on actual practices on organic farms, although this information
may be available within certification organisations. Such comparisons can provide useful infor-
mation when the purpose of the study is clearly defined (Spedding 1975) and the basis of the
comparison is fair. Defining starting points, boundaries and time scale are important in this
respect. Lampkin (1994) points out that the fundamental issue is the comparison of systems
rather than modification to individual management practices. This perhaps means it is more
difficult to make valid and useful comparisons of biophysical properties than economic ones. It
is important to separate out those aspects of the system that need to be assessed at the whole
systems level (i.e. those which are dominated by interactions or large-scale ecological processes,
and those which can be compared at the small plot scale) (Watson and Atkinson 2000).
In the trials listed in Table 15.3 and Table 15.4 the study areas vary from relatively small
plots, such as the DOK (bioDynamic-Organic-Konventional) trial (Mäder et al . 2002), to
several hectares (e.g. Leake 1996). One complicating factor in interpreting results of these
trials is whether they truly compare farming systems or simply rotations. Factorial crop
rotation experiments (e.g. Mäder et al . 2002, Watson et al . 1999) and field-scale testing of crop
rotations (e.g. Cormack 1999) that allow factorial experiments within them, contribute to dif-
ferent aspects of the understanding of how crop rotations function. As soon as the crops or
even varieties within a rotation are changed the effect of that rotation both in terms of yield
and productivity as well as soil structure and environmental impact will change, regardless of
the production system. However, under given soil and climatic constraints the most produc-
tive choice of crops and varieties in a rotation will differ depending on whether the system is
managed conventionally or organically. Thus, are any differences between the biophysical
aspects of the rotation due to the system or the rotation? The DOK trial in Switzerland has
compared the same crop rotation under different systems of manuring and pest management
since 1978 (Mäder et al . 2002). Although this trial has provided a wealth of interesting infor-
mation on soil properties and crop protection and production but the question remains as to
how applicable this information is in the context of practical farming. Despite the reliance on
forage legumes for fertility building in many organic systems, surprisingly few trials include
grazing livestock. Many trials utilise livestock manure to mimic whole systems, but these can
never truly represent realistic grazing situations where there is constant interaction between
soils and plant and animal production.
Lack of replication is a drawback of many of the trials in Table 15.3 and Table 15.4 (e.g. two
replicates in the SAC crop rotations trials) (Watson et al . 1999). Lack of replication and the
non-use of livestock perhaps both ref lect the costs and funding commitment needed to run
trials of this type. Some of the difficulties of running rotation trials are discussed in Taylor et
al . (2002). Olesen (1999) recommends the involvement of experts from outside individual
research groups in the design of long-term experiments. This would increase the rigour of the
experimental design by drawing on wide experience from past experimentation. It can also
help to add value to experiments by ensuring that design factors such as plot size are appropri-
ate for all the parameters likely to be studied in the experiment.
Where studies are carried out within organic systems (e.g. Table 15.4), it is particularly
important that background information on aspects like time since conversion and the particular