Mixed diversity of shifting IOD and El Niño dominates the location of Maritime Continent autumn drought

Summary The Maritime Continent is a huge heat source region over the Indo-Pacific warm pool and it plays a key role in global weather/climate variations. The locations of Maritime Continent autumn droughts, linked to frequent rampant forest wildfires, are closely related to the mixed diversity of El Niño and Indian Ocean Dipole events.

The Maritime Continent is a huge heat source region over the Indo-Pacific warm pool and it plays a key role in global weather/climate variations. The locations of Maritime Continent autumn droughts, linked to frequent rampant forest wildfires, are closely related to the mixed diversity of El Niño and Indian Ocean Dipole events.
Climate-change impact evaluation on Maritime Continent (MC) land precipitation is becoming an important research arena, given the severe forest fires induced by frequent autumn droughts under greenhouse warming [1,2] and the importance of latent heat released from MC-rainfall processes for local and global atmospheric circulations [1]. It is widely accepted that MCprecipitation activities, via changing Walker circulation [3], are indirectly modulated by El Niño/Southern Oscillation (ENSO) [1,4] and Indian Ocean Dipole (IOD) [5], which are both high-impact ocean-atmosphere coupled phenomena with a global 'footprint' on interannual timescales. Thus, their interaction attracts considerable scientific attention [3].
However, understanding of the changing impacts from shifting ENSO and IOD diversities on the MC land precipitation remains insufficient. Projection of the effects of ENSO-IOD change on the MC-drought positions is therefore inherently uncertain and is a subject of profound scientific interest in the present study. Here, the results show that, for the satellite era of 1979-2016, two distinct MC-drought positions are significantly modulated by different ENSO-IOD flavors during boreal autumn (i.e. the seasonal mean of September-November, SON).
Considering that the MC land precipitation possesses large dry-wet annual cycle and significant local features/variances due to the unique geographic location [1], here we apply a rotated empirical orthogonal function (REOF; see Supplementary Data and Methods for more details) to capture the interannual leading modes of the normalized-and-detrended NOAA (i.e. the National Oceanic and Atmospheric Administration) land-precipitation anomalies in the MC region (95 • E-145 • E, 11 • S-9 • N), which (i.e. the REOF1 and REOF2 shown in Fig. 1a and b, and see Supplementary Fig. 1 for details) highlight the rainfall deficit over the western MC (WMC) and the eastern MC (EMC), respectively. The corresponding principal components (i.e. RPC1 and RPC2) of MC land precipitation, with clear interannual variations, are plotted in Fig. 1a and b (black lines), which explain 34.3% and 27.9% of the total normalized variances (see Supplementary Note 1 and Supplementary Fig. 1 for details). It is worth noting that similar results can be reproduced with another three different sets of higher-resolution data ( Supplementary Figs 2-4), suggesting that such REOF results are significant and independent of data choice.
The changes of Indo-Pacific seasurface temperature (SST) related to    Relevant SST regions are outlined in Fig. 1c and d. Their correlations are shown in Supplementary Fig. 6. As expected, their time series are strongly in phase, with high correlations (Fig. 1a and b) up to 0.71 (0.68) between RPC1 and EP-IOD (between RPC2 and CP-SIOD), statistically exceeding the 99.9% confidence level. Relevant SST modes are shown in Fig. 1c and d, which perfectly mirror the SST patterns (Supplementary Fig. 5) associated with RPC1 and RPC2, respectively. Two such types of combined Indo-Pacific SST modes would lead to different changes in (i) the large-scale Rossby waves ( Fig. 1e and f) via upper-level divergence perturbed by tropical convective activities [7,8] and (ii) the low-level winds and convergence via forcing different surface-pressure gradients [9] (Supplementary Fig. 7a and b), which together result in distinct Walkercirculation anomalies with different ascending and sinking motions over the Indo-Pacific Ocean and the WMC/EMC ( Supplementary Fig. 8), respectively.
In comparison with Fig. 1e and Supplementary Fig. 7a for EP-IOD events, Fig. 1f and Supplementary Fig. 7b reveal that the resultant upper-level convergent winds and low-level divergent winds are shifted from WMC to EMC during CP-SIOD events (referring to the orange circles shown in Fig. 1e-h), corresponding to the location changes in the sinking motions of the Walker circulation (see the red boxes shown in Supplementary Fig. 8a and b). Then, the resultant water-vapor-flux responses (see the orange circles shown in Fig. 1g and h and Supplementary Fig. 7c and d) contribute to the location shift of MC drought (Fig. 1g and h and Supplementary Fig.  7c and d). Such phenomena indicate the geographical adaptability of MC precipitation/drought to ENSO-IOD diversity ( Supplementary Fig. 5). Of note is that Fig. 1c-f and Supplementary Figs 7 and 8 are obtained independently from the REOF analysis, except for the SST-box choices used for defining the EP-IOD and CP-SIOD indices, suggesting that the above results are reliable and robust.
In summary, the nature of ENSO-IOD combinations and the changes in their performances, including the modulation of tropical waves, convections and Walker-cell patterns, make good sense to understand MC-rainfall and forest-fire activities. Additionally, a higher prevalence of extreme El Niño and IOD events is anticipated in future climate scenarios [10,11] and that the spurious IOD as well as the Modoki El Niño are mysteriously changing under greenhouse warming [3,8,12], indicating that climate change and variability may exert more severe impacts on the MC autumn drought than previously thought. More importantly, our results highlight that, once the precursor signals of variant ENSO-IOD combinations were monitored or predicted, it would provide early warning to relevant policymakers to plan and act effectively to minimize forest-fire losses (including homes and crops destroyed, fisheries ruined, etc.) and shelter air quality and life safety during the dry-season months for MC countries. The disposable soma hypothesis explanation of the effects of caloric restriction (CR) on lifespan fails to explain why CR generates negative impacts alongside the positive effects and does not work in all species. I propose here a novel idea called the clean cupboards hypothesis which overcomes these problems.

BACKGROUND
The effects of caloric restriction (CR) on longevity were discovered 100 years ago [1]. Since then the effect has been replicated in a wide variety of animals (reviewed in [2]). More recent work in non-human primates provides a complex picture but also indicates that there are some beneficial impacts on both healthspan and lifespan [3][4][5][6]. Impacts on human lifespan will likely never be systematically studied, but a short-term randomised controlled trial produced indicative changes in important biomarkers of ageing [7,8]. There are also suggestions that naturally restricted human populations show increased longevity [9].
The question of why CR has these effects can be answered at several different levels. For example, there is a debate over whether the effect is due to a deficit in calories or protein [10,11]. There is also the issue of the exact molecular mechanism. Finally, we may consider the evolutionary function of the effect: why did the responses to CR evolve? In this perspective I will address this latter issue and propose a novel hypothesis for why CR affects lifespan. I have called the hypothesis the 'clean cupboards hypothesis' for reasons that will become apparent later. This is an important question to answer because our understanding of why CR influences life and healthspan has ramifications for whether we expect CR to have similar effects in humans. Since, as noted above, a direct study of the impact of CR on human lifespan is unlikely to ever be conducted, we are reliant on this theoretical understanding of the function of CR to make inferences about whether it is worth pursuing as a human lifespan intervention, although if it leads to other benefits it may be worth pursuing anyway.
The main evolutionary hypothesis for the effect of CR on lifespan is derived from the disposable soma hypothesis (DSH) [12][13][14]. The DSH is a classical trade-off model that posits energy resources are limited and hence animals must make an evolutionary decision about how to use them. There are two main uses: somatic maintenance and reproduction. If an animal invests in somatic maintenance it improves survival probability, but it does so at the cost of reproduction. Alternatively, investing heavily in reproduction can only occur at the costs of somatic maintenance and hence survival. The DSH therefore explains the phylogenetic inverse correlation between reproductive output and lifespan [15]. This theory also provides a potential evolutionary explanation of what is happening during CR. It is presumed that wild animals would only experience CR as a temporary phenomenon. Since attempts to reproduce under such limited energy supply would likely fail, animals are better served by switching off reproductive investment completely and diverting all their resources into somatic maintenance. This would maximise their chances of surviving the period of restricted energy supply. This resource allocation model is illustrated in Fig. 1. In the laboratory the lean period never ends, and the animals keep the somatic maintenance activities switched on indefinitely, leading to the lifespan increase [14]. This idea is consistent with direct observations that exposure to CR reduces reproductive investment [16], and potentially explains why CR may be less effective in males, which expend less energy on reproduction ([17] but see review [11] suggesting no sex difference in the response).
This DSH interpretation of the evolutionary background to the function of CR leads to the prediction that because the costs of reproduction in small animals are substantially higher (relative to baseline non-reproductive costs) than the reproductive costs in larger animals, the savings that can be made by switching off reproduction and diverting energy to somatic maintenance are considerably larger in smaller animals. Hence it is argued we observe a large effect of CR in small mammals like mice [18], rats [19] and very small primates [6], but in larger animals the impact is attenuated