Glenn Summerhayes and colleagues
Data from the New Guinea Highlands (at an elevation of ~2000 meters) demonstrate the exploitation of the endemic nut Pandanus and yams in archaeological sites dated to 49,000 to 36,000 years ago, which are among the oldest human sites in this region. The sites also contain stone tools thought to be used to remove trees, which suggests that the early inhabitants cleared forest patches to promote the growth of useful plants.
The details of the assemblages are illuminating:
There are "waisted axes", large cutting tools with grooves on the sides for hafting onto wooden handles. They suggest, on ethnographic analogy, that these were used for forest clearing. I would imagine them useful for broader woodworking tasks, though, possibly including food extraction. The waisted axe artifacts here are not as extensively shaped as the later examples reported by Groube and colleagues
Groube:1986. The authors do not report on use wear for these.
Starch grains adhering to some of the stone tools indicate yam utilization, but yams live quite a bit lower than the site where these tools were found.
Lots of Pandanus nut roasting.
The dates don't make a huge impact on our understanding of the chronology. Almost 25 years ago, Groube and colleagues
The "Highland" aspect is more interesting, suggesting a fairly quick adaptability of early humans to a novel ecology. People had found the local plant foods in a unique ecology, they were exploiting a range of altitudes in their foraging activities, and possibly were altering their landscapes by forest clearing.
Or possibly, all this suggests that humans had already been in the area for a substantial length of time...
Or -- let me be even more subversive -- why is a New Guinea assemblage automatically assumed to be made by modern humans, when assemblages of equal (or greater!) technological sophistication on nearby Flores aren't?