[samcday]

Not sure if you've seen the following statement from University of Sydney, but they're standing by the estimate: https://sydney.edu.au/news-opinion/news/2020/01/03/a-stateme...

In fact, they've pointed out the 500 million figure is a conservative estimate, and is likely to be much higher: "... The true loss of animal life is likely to be much higher than 480 million. ..."

[mistermann]

> The authors deliberately employed highly conservative estimates in making their calculations. The true mortality is likely to be substantially higher than those estimated.

It's a shame the University of Sydney doesn't have access to this study such that they could have included a link so skeptical people could see for themselves the specifics of this alleged conservatism. Not doing that causes my intuition to question the truthiness of these claims, so I will file this item under "Unknown-Suspicious".

[nl]

The report[1] (which is trivial to find) has a section on their estimation methodology and includes reasons why it is conservative (section 2.3):

Many species were excluded from estimates because there is insufficient information on their abundance. The largest omission is the entire NSW bat fauna (37 species), for which no information on density could be found.

Many small mammals in semi-arid and arid regions exhibit large fluctuations in density depending on the prevailing weather. For example, historical accounts of the long-haired or plague rat (Rattus villosissimus) suggest that densities well in excess of 1000 animals per hectare can be attained after years of good rain31, with the species virtually disappearing again during drought. For such eruptive species, only the low-density population estimates were used.

The densities of several common species, such as the brown antechinus (Antechinus stuartii), agile antechinus (A. agilis), yellow-footed antechinus (A. flavipes) and brushtail possum (Trichosurus vulpecula) have been measured in several studies, with most yielding low to moderate densities but small numbers of studies yielding very high estimates. To reduce bias arising from these rare high values, means for each species were first calculated as log- transformed densities and then back-transformed to produce normal values. This method was used also by Cogger et al.

As noted, the usually lower densities obtained from studies carried out in the tablelands, western slopes and plains were used in preference to the higher estimates obtained in surveys further east. my note: these bushfires are in the East, so the higher estimates would be more appropriate here

Several species were omitted from consideration due to uncertainty about how vegetation clearing would affect them.

[1] https://www.researchgate.net/publication/318029981_Impacts_o...