We have shown that the WPG during ENSO is a novel index for Last Millennium Walker Circulation strength, the latter defined by SST and SLP gradient indices across the tropical Pacific (Karnauskas et al., 2009; Tokinaga et al., 2012; Coats & Karnauskas, 2017; L’Heureux et al., 2013). For the 20th century, we also show that the WPG captured the SST and precipitation teleconnection patterns qualitatively similar to ENSO and Walker Circulation indices. Compared to other paleo-ENSO indices (Emile-Geay et al., 2013a, b; McGregor et al., 2010), the WPG showed a more stable relationship with the observed PWC indices (Coats and Karnasukas, 2017). Thus, the WPG is a key index for the PWC in the instrumental era and also for the past Millennium. We did not observe a clear linear trend in the WPG during the Last Millennium. Instead, both Last Millennium WPGs and ZGsst showed multi-decadal to multi-centennial variability with more negative WPG and La Niña or CP La Niña-like occurrences (positive ZGsst) in the late 11th, mid-late 12th, at the turn of the 14th century, between 1400-1600s, mid 1700s and since 1870 (West Pacific warmer than the central Pacific). Positive WPG and El Niño or CP El Niño-like occurrences (negative ZGsst) dominated the first century of the Last Millennium, the early 14th and short periods in the 11th and 18th centuries (West Pacific colder than central Pacific). The 20th century stands out as period with both strong positive and negative WPG (and ZGsst) occurrences and overall highest variability (running standard deviations) over the past Millennium. However, several studies caution oversimplified interpretations of variance changes due to diminishing proxy availability over time (Comboul et al., 2014; Dee et al., 2020). Nevertheless, our results are largely in agreement with other Niño3.4 indices in terms of multi-decadal oscillations and changing variance over time, despite vastly different methodologies (Abram et al., 2020; McGregor et al., 2010; Emile-Geay et al., 2013; Steiger et al., 2019). Clearly, the WPG could be an excellent index for CP ENSO or Modoki phases which often are characterized by strong SST gradients between the western and central Pacific, yet same sign anomalies in the EP and for the WP. Hoell and Funk (2013) showed evidence that most negative WPG phases since 1950 were associated with central Pacific or La Niña Modoki events while positive WPG phases had an affinity to occur predominantly during eastern Pacific or canonical ENSO events. However, a number of positive or negative WPG phases were associated with mixed type ENSO flavors. Thus, periods of extreme negative WPG in our Last Millennium reconstructions may represent La Niña or La Niña Modoki conditions, while positive WPG periods may indicate EP El Niño or mixed CP and EP El Niño flavors. To assess the relationship between ENSO flavors and the WPG, we turned to the NCT and NWP reconstructions based on Pacific coral proxy records (Freund et al., 2019; Figs. 3 and S8). The NWP and NCT study considered El Niño flavors only, yet the index also includes the La Niña flavors. Our hypothesis was that the NWP should predominatly co-vary with the WPG positive (negative) events when the latter would predominatly reflect Modoki or CP El Niño (CP La Niña) events. The comparison of the WPG indices with the NCT (EP) and NWP (CP) ENSO flavor indices revealed time varying relationships, especially between 1600 and 1850, yet strongest co-variability with both since the mid-19th century to the present (Freund et al., 2019; Fig. S8). The relationship during the mid/end-18th century revealed closer agreement with the NWP (CP flavor) while the early 18th century WPG indicated closer alignement with the NCT (EP flavors). However, the 18th century is also known for its reduced proxy data coverage in the Pacific Ocean, so should be viewed with caution (Freund et al., 2019). The enhanced CP to EP event ratio during the 17th century is less clearly reflected by the WPG’s. The WPG’s do, however, indicated extreme negative events (positive ZGsst) in the early and late 17th century (Figs. 2 and 3). However, when taking into account centennial mean state changes (without detrending), the WPG appears to reflect dominantly negative WPG or La Niña Modoki conditions with the exception of the 11th and 20th centuries, the early 14th and early 18th centuries and several short-lived periods over the past Millennium (Fig. 2). Thus, during the last Millennium the WPG mean state changes appeared to have skewed the WPG towards La Niña or La Niña Modoki type events for most of the time. Thus, there appears to a relationship between Pacific SST mean state, ZGsst and the WPG.
We also established the hydroclimatic relationships of the WPG with the three drought atlases which were largely in agreement with assessments of ENSO impacts on drought and pluvial conditions. In accordance with the findings of Steiger et al. (2019), megadroughts in southwestern North America were found to be associated with an extremely negative WPG (positive ZGsst) and colder values of the Niño3.4 index in PHYDA. The North American megadroughts around 1400 and the turn of the 16th to 17th century during the Little Ice Age stand out as periods with persistent negative WPG and Niño3.4 anomalies (positive ZGsst). The latter is consistent with Dee et al. (2020) and Steiger et al. (2019) who showed evidence for enhanced North America ENSO teleconnections during the Little Ice Age. The latter was found to hold for both El Niño and La Niña events and was shown to be unaffected by proxy availability over time. In other words, we find more drying in the North American Southwest in more negative WPG years (positive ZGsst) during periods of enhanced ENSO teleconnections. Steiger et al. (2019) pointed out that the megadroughts pre-1600 were likely radiatively unforced and most strongly associated with unusually cold SST in the central and eastern Pacific and warm SST in the Atlantic. Yet, they also showed evidence for a potential role of local radiative forcing in excerbating regional drought in the North American Southwest. The western Pacific region of the WPG and MC are oceanic regions of high sensitivity to radiative forcing (Funk & Hoell, 2015; Chung et al., 2019; Seager et al., 2019). Thus, changes to the WPG might bear a footprint of local radiative forcing coupled with slow decadal oceanic feedbacks that affect the SST patterns in the Pacific and associated teleconnections (Luo et al., 2017). Thus, the SST patterns across the western Pacific appear to play a crucial role in the drought inducing teleconnections in addition to those in the central and eastern Pacific.
ENSO was also shown to play an important role in South American hydroclimate next to SAM and Atlantic influences (Morales et al., 2020). Wet conditions were found during positive Niño3.4 anomalies in southeastern South America and central Chile while dry conditions prevailed during negative Niño3.4. The PHYDA WPG reconstructions revealed that wet conditions persisted in central Chile during a positive WPG while dry conditions prevailed during the negative WPG phases. This is largely in agreement with ENSO teleconnections based on the SADA (Morales et al., 2020). Our study suggested a weaker influence of the WPG on central Chile in parts of 18th and 19th centuries. Here, influences from the SAM or Atlantic could have a played a larger role.
Eastern Australia was shown to experience wetter condition during La Niña phases and drier conditions during El Niño phases amplified by negative and positive Interdecadal Pacific Oscillation (IPO) phases, respectively (Palmer et al., 2015). Our WPG reconstruction revealed three periods of negative correlations with the Eastern Australia region in ANZDA during the turn of 16th to 17th century, between 1700–1800 and the 20th century. Weaker relationships with the WPG were observed in between. The WPG also showed the typical anti-correlation between Eastern Australia and southern New Zealand PDSI in ANZDA between 1500 and 2000 AD (Palmer et al., 2015; Fig. S13). The relationship during the 20th century indicated that teleconnections with the WPG were strongest when ENSO variability was higher between 1880–1920 and since 1960. The latter is consistent with abundant evidence from previous studies of enhanced ENSO teleconnections during strong ENSO phases (Torrence and Compo, 1998; Mestas-Nunez and Enfield, 2001; Wang et al., 2019). ENSO modulation by low-frequency oscillations of the IPO can modulate ENSO impacts on Australia (Verdon et al., 2004; McGovan et al., 2009). The latter may also hold for relationships between Eastern Australia PDSI in ANZDA and the WPG.
In summary, the new PHYDA WPG reconstructions indicated robust hydroclimate teleconnections with regions sensitive to ENSO and WPG drought and pluvial conditions across the Pacific and on its rims. The late 16th century was confirmed as period of strong negative WPG, positive ZGsst and Niño3.4 occurrences with megadroughts in North America and South America and pluvials in Australia. This illustrates that the WPG may serve as a powerful index together with ENSO indices to reveal the impact of drought-inducing climate teleconnections that arose from changes in Pacific SST patterns. We call for a concerted effort of generating new proxy data from this key region in the western Pacific which could dramatically improve the knowledge base for centennial and decadal PWC changes.