The scale and complexity at which many biological processes operate create hurdles for researchers; ultimately, undermining the development of mechanistic understanding of their causes and limiting generalizations of experimental findings across systems. More so than experiments in natural landscapes, experiments in managed ecosystems have potential to overcome these issues. In managed ecosystems, researchers can replicate experimental units at the appropriate scale and manipulate systemic complexity. Here, I apply this methodology by using a pistachio orchard to (1) test hypotheses of the proximate causes of masting and (2) examine the effects of spatial autocorrelation on the results of field trials.

Masting, the intermittent and synchronous production of seeds by perennial plants, is commonplace in temperate trees; nevertheless, the proximate causes of this phenomenon are unclear. It is thought that masting results from carbon dynamics, environmental entrainment, or pollen exchange. In contrast to previous studies, the two studies performed here conduct spatiotemporal and permutation tests on seed production of a large sample of individual trees (n=4,288) over six years under manipulated growing conditions. Results demonstrated that fruiting patterns were largely regulated by carbon dynamics. However, climatic events affecting flowering and the distance between reproducing partners exerted a significant impact on synchrony. Such findings suggest that masting in pistachio results from complex interactions of all three mechanisms. These results elucidate obstacles in orchard management and horticultural research and provide insights into the effects of climate change on masting species.

Collection of this tree-level dataset, also, provided an opportunity to test the effects of spatial autocorrelation on the precision of numerous experimental designs within a horticultural system. Results supported that spatial autocorrelation decreases statistical precision; thus, calling into question the results of some previous research. The spatial structure of the autocorrelated variable, yield, changed yearly and this significantly influenced the precision of each design. In general, small experimental blocks should be located as far from each other as possible to account the level of spatial autocorrelation in these data. More study is needed to determine the best method to perform replicated experiments in systems where spatial autocorrelation changes through time.