Multi-decadal variability and trends in the El Nino-Southern Oscillation and tropical Pacific fisheries implications

Extremes of the El Nino-Southern Oscillation (ENSO) are known to have various socio-economic impacts, including effects on several Pacific fisheries. The 137-year-long record of Darwin sea-level pressure offers a uniquely long-term perspective on ENSO and provides important insight into various aspects of interannual to century-scale variability that affects these fisheries. One particular issue of interest is whether there is a centennial-scale (or longer) trend that can be expected to alter the future distributions of these fisheries. Since most tropical Pacific fishery records are no longer than a few decades, another issue is the extent to which trends over these recent decades are a good basis for detecting the presence of long-term (e.g., centennial-scale) deterministic changes, and perhaps thereby projecting future conditions. We find that the full 137-yr trend cannot be distinguished from zero with 95% confidence, and also that the ENSO variance in recent decades is very similar to that of the early decades of the record, suggesting that ENSO has not fundamentally changed over the period of large increase in atmospheric CO2. However, the strong multi-decadal variability in ENSO is reflected in decades with quite different levels of ENSO effects on the ecosystem. Many multi-decadal subsets of the full record have statistically significant trends, using standard analysis techniques. These multi-decadal trends are not; however, representative of the record-length trend, nor are they a useful basis for projecting conditions in subsequent decades. Trend statistical significance is not a robust foundation for speculation about the future. We illustrate how the difficulties involved in determining whether a trend is statistically significant or not mean that, even after careful consideration, an unexpectedly large number of trends may reach standard statistical significance levels over the time spans for which many newer records are available, but still not continue into future decades or be indicative of deterministic changes to the system. Analysis of the Southern Oscillation Index, another common ENSO index, but one that has been directly measured for fewer years than has Darwin, yields similar results. (C) 2014 Elsevier Ltd. All rights reserved.