Evaluation of yield-density relationships and optimization of intercrop compositions of field-grown pea-oat intercrops using the replacement series and the response surface design

In a 2-year field experiment (2002/2003) on a loess soil near Göttingen/Germany, pea (Pisum sativum L.) and oat (Avena sativa L.) were grown alone and intercropped at a range of densities. Shoot biomass, grain yields and amount of N in grain were evaluated and optimized using two different replacement series and a hyperbolic yield-density equation describing a response surface to address the following questions: (i) what is the optimal composition of the pea-oat intercrop with regard to maximum yields, (ii) which intercropping design is most suitable to describe competition effects in pea-oat intercrops and the optimal intercrop compositions and (iii) which intercropping design is best suited for the evaluation of field data. For (i), the optimal intercrop compositions varied depending on the growth conditions for the crops. Furthermore, optimal intercrop compositions were found above the recommended sole crop densities. The density of oat had to be reduced more than that of pea, especially when optimal grain-N yields were desired and soil-N content was high. For maximum grain-N yields, pea could be sown at high densities in combination with 5-50% of the recommended density of oat. Thus, density can be used as a yield regulator for specific purposes such as a high N yield. The effects of competition at final harvest were described equally by both designs (ii). Oat was the clearly stronger and pea the inferior competitor. In contrast to the replacement series design, the hyperbolic yield-density equation was capable of adding valuable information about the extent of intra and interspecific competition. As intraspecific competition was consistently more important than interspecific competition, resource complementarity could be hold responsible for intercrop advantages. The highest intercrop advantage was found when total intraspecific competition was low, as shown by the relative yield total (RYT) and niche differentiation index (NDI) values >1. However, due to the RYT dependence on sole crops and total densities, the replacement series design led to misleading interpretations of the yield advantages. Both experimental designs were able to describe the field-data reliably (iii), but the response surface design had the advantage of being unaffected by insufficient field emergences, as it is not based on total densities. Numbers of plantsm⁻² instead of seedsm⁻² can be used for the evaluation. Data from sole crops are not needed for the response surface design and thus the feared high experimental effort of this design can be reduced. However, when using the replacement series design, experimental effort should be greater than normal, as different sole crop densities and more intercrop compositions within a replacement series can lead to a more precise interpretation of the competition effects.