“…For the reasons outlined in Section 3.1, many studies take the oldest boulder age-within a cluster of boulder ages for a given moraine and after the exclusion of outliers attributed to nuclide inheritance-to be the closest to the timing of the glacier advance (e.g., Zech et al, 2006Zech et al, , 2007aZech et al, ,b, 2009bZech et al, , 2010Hall et al, 2009;May et al, 2011). On the contrary, some studies instead (i) calculate mean values of all boulder ages collected on a moraine (Licciardi et al, 2009;Shakun et al, 2015b;Stansell et al, 2015Stansell et al, , 2017Bromley et al, 2016;Martini et al, 2017); (ii) calculate average boulder ages using frequency density plots (Smith and Rodbell, 2010;Jomelli et al, 2011;Smith et al, 2011;Ward et al, 2015); (iii) organise ages chronologically and take the plateau or modal age (Smith et al, 2005a,b) or (iv) combine these various approaches . As our results for the Sierra de Aconquija demonstrate (see Section 4.1), these different approaches could, in some cases, result in substantially different moraine ages (and consequently palaeoclimate interpretations) for an identical set of boulder ages.…”