Climate variability and food security a New Zealand perspective

Abstract

The objective of this thesis was to develop a clearer understanding of inter-relationships between climate variability and food security in New Zealand. It was considered important to both clarify crop-climate relationships and possible response options available to regional planners and individuals. This is particularly relevant in the context of a probable global climate warming.

An empirical-statistical analysis of crop-climate interactions was carried out. This was followed by a more detailed agroclimate analysis of the Canterbury region, and an evaluation of one possible response option in the face of present, and possible future, climate variability and change. This involved a field based study of shelterbelt effects. The final part of the Thesis gives a tentative assessment of the possible impacts on agriculture in New Zealand of regional greenhouse warming scenarios.

Monthly rainfall and mean temperature data were used for the crop-climate analyses. Adjustments for site changes were made where necessary and missing values estimated. Trend removal was performed on temperate grain and pipfruit yield time series. Quadratic, and in the case of pears linear, trend lines were fitted to the yield data. Analyses were performed on the residuals. No trend removal was carried out on stonefruit data.

Principal component analysis, followed by stepwise multiple regression, showed the barley crop to be the most spatially responsive to climate of the three temperate grain crops examined. Wheat was intermediate in its response and oats the least spatially responsive. Autumn sown wheat showed a negative relationship with winter rainfall and spring temperature. The dominant result with oats and barley was a negative effect on yield of late spring to early summer temperatures.

Analyses of national stonefruit data met with mixed success. Peaches and nectarines, although of the same species, gave different significant predictors. This was attributed to differences in weighting on the climate data, related to geographic distribution. The susceptibility of apricots to late frosts appeared to show through with this crop. Cherries showed a strongly negative relationship with rainfall at blossom and harvest time. The most significant predictor with plums was a positive relationship with May temperature. This was treated with scepticism. These results highlight the need for further detailed analysis of these crops at the district level.

The limited district analyses of apple yields showed a graded response to rainfall. At the extremes, the wetter Auckland district showed predominantly negative responses and the drier Central Otago district showed positive responses. A negative relationship between apple yield and July temperature was found to be a climatic response from the Hawkes Bay district. This may be related to poor fruit set and flower quality, leading to low yields, as suggested by research in England. The dominant response apparent with pears, in earlier years, was a negative relationship with January temperature. This appears to have become less limiting and may be related to increased use of irrigation over this moisture sensitive period. As with stonefruit more detailed analyses, particularly at the district level are required.

A spatial analysis of Canterbury plains climate confirmed that the climate of this region is relatively homogenous. However a general north/south division was apparent with the Rakaia river as a general 'dividing line. Closer analysis of rainfall data revealed that this north/south contrast is predominantly between the Christchurch area, influenced by Banks Peninsula, and South Canterbury, influenced by the narrow coastal strip and proximity to the foothills. Differences between these two areas are greater than similarities. Principal component analysis of Canterbury county wheat yields confirmed results from the weighted national and Canterbury district analyses. Waimate, in South Canterbury, proved to be anomalous in its yield response, which was consistent with the spatial analyses of rainfall and deficit day data. Drought in Canterbury was shown to become a regional phenomenon in the driest of the dry years, and to be persistent in these years. Correlation analysis between deficit day data and time series of detrended Canterbury district yield data, for temperate grains and pipfruit, showed a significant negative relationship between agricultural drought and yield.

Analysis of shelter effects revealed a hierarchical classification of sites based on the site roughness parameter, z₀. The most exposed, reference, site was well representative of open plains conditions. The least exposed, highly sheltered, orchard site showed a high degree of "decoupling" from the regional environment. Mean temperatures were significantly higher in the three, more sheltered, remote sites as compared to the reference site. Maximum temperatures were significantly higher in the most sheltered sites. In all remote sites there were significant reductions in wind speed in relation to the reference site. Evapotranspiration, based on Penman estimates, was significantly lower in the two orchard sites. Priestley-Taylor estimates proved to be more conservative, attributed to the use of a constant not calibrated for different site conditions. It is speculated that these significant differences in site microclimate could lead to yield benefits in Canterbury, through greater water use efficiency. This could contribute significantly to mitigating the effects of non-periodic, but recurrent and persistent droughts in this region.

The tentative assessment of agricultural impacts of a greenhouse warming drew from past climate analogue scenarios. The temperate grains showed slight to moderate yield reductions.

It was suggested that Southland may increase in importance as a temperate grain growing district. Assessment of fruit crops was more speculative, as shown in the results. The east coast of the South Island and Central Otago could increase in importance for the growing of temperate fruit crops. From the Canterbury regional greenhouse scenario it would appear that in the future there will be greater potential for agricultural drought in this region. The results from the agroclimate analysis and field study of shelter effects are particularly relevant in this context.

This Thesis highlights the considerable uncertainty that exists in the field of crop-climate analyses. Data bases need to be consolidated and more critical analyses made of possible response options, particularly in the face of a probable global climate warming.