Comments on: Life Curves http://bit-player.org/2008/life-curves An amateur's outlook on computation and mathematics. Thu, 17 May 2012 09:26:46 +0000 http://wordpress.org/?v=2.6.3 By: Jim Ward http://bit-player.org/2008/life-curves#comment-1768 Jim Ward Tue, 26 Aug 2008 15:06:06 +0000 http://bit-player.org/?p=160#comment-1768 I believe the Muller and Rohde technical supplement, mentioned in Brian's American Scientist article is at: http://www.nature.com/nature/journal/v434/n7030/extref/nature03339-s1.pdf I believe the Muller and Rohde technical supplement, mentioned in Brian’s American Scientist article is at:

http://www.nature.com/nature/journal/v434/n7030/extref/nature03339-s1.pdf

]]> By: brian http://bit-player.org/2008/life-curves#comment-1767 brian Tue, 26 Aug 2008 12:42:08 +0000 http://bit-player.org/?p=160#comment-1767 @Gerry: Good question, which I should have anticipated. There are at least two major reasons for the discrepancy, as I understand it. First, the Sepkoski data set includes roughly twice as many genera--about 36,000, compared with 18,700 in the Alroy et al. work. This isn't a disagreement about how many animals actually existed; it's more a matter of criteria for inclusion in the database. Sepkoski included anything for which he could find a literature reference giving stratigraphic dates; Alroy et al. include only references that list actual occurrences of fossils, with geographic coordinates, etc. They are gathering more information, but so far it covers fewer organisms. Second, the subsampling procedure in the Alroy graph is a filter that tends to reduce the height of the curve. The procedure for constructing the curve is to randomly select collections of fossils within each bin until a certain quota of specimens is achieved; then you count up the number of genera represented by those specimens. The specimen quota has to be set low enough that the bins with the smallest occupancy don't totally saturate during the sampling procedure. The number used for the published results was 16,200 specimens. (The curve shown is the average of 20 repetitions of this entire procedure; the gray error bars give a 95 percent confidence interval.) Thus the two curves really aren't comparable on a quantitative basis. But the actual counts of genera don't carry great meaning anyway. (Nobody believes we have even a rough estimate of the number of genera extant at any particular time, including now.) It's just the variations in diversity that matter -- the overall shape of the curve. Perhaps there is some normalization procedure that would put them on the same scale. All of the above is based on my reading of the Science article and supplemental material. If anyone who can comment more authoritatively on this matter happens to read this, please speak up! @Gerry: Good question, which I should have anticipated. There are at least two major reasons for the discrepancy, as I understand it.

First, the Sepkoski data set includes roughly twice as many genera–about 36,000, compared with 18,700 in the Alroy et al. work. This isn’t a disagreement about how many animals actually existed; it’s more a matter of criteria for inclusion in the database. Sepkoski included anything for which he could find a literature reference giving stratigraphic dates; Alroy et al. include only references that list actual occurrences of fossils, with geographic coordinates, etc. They are gathering more information, but so far it covers fewer organisms.

Second, the subsampling procedure in the Alroy graph is a filter that tends to reduce the height of the curve. The procedure for constructing the curve is to randomly select collections of fossils within each bin until a certain quota of specimens is achieved; then you count up the number of genera represented by those specimens. The specimen quota has to be set low enough that the bins with the smallest occupancy don’t totally saturate during the sampling procedure. The number used for the published results was 16,200 specimens. (The curve shown is the average of 20 repetitions of this entire procedure; the gray error bars give a 95 percent confidence interval.)

Thus the two curves really aren’t comparable on a quantitative basis. But the actual counts of genera don’t carry great meaning anyway. (Nobody believes we have even a rough estimate of the number of genera extant at any particular time, including now.) It’s just the variations in diversity that matter — the overall shape of the curve. Perhaps there is some normalization procedure that would put them on the same scale.

All of the above is based on my reading of the Science article and supplemental material. If anyone who can comment more authoritatively on this matter happens to read this, please speak up!

]]> By: Jim Ward http://bit-player.org/2008/life-curves#comment-1766 Jim Ward Tue, 26 Aug 2008 12:26:30 +0000 http://bit-player.org/?p=160#comment-1766 One of the articles: http://arxiv.org/vc/arxiv/papers/0807/0807.4729v1.pdf suggests that the cycles correspond with the rotation of our solar system to galactic north, and speculates that the Earth is exposed to high energy cosmic rays, causing DNA damage. Time to put on my tin foil underwear! One of the articles:

http://arxiv.org/vc/arxiv/papers/0807/0807.4729v1.pdf

suggests that the cycles correspond with the rotation of our solar system to galactic north, and speculates that the Earth is exposed to high energy cosmic rays, causing DNA damage. Time to put on my tin foil underwear!

]]> By: Gerry http://bit-player.org/2008/life-curves#comment-1765 Gerry Tue, 26 Aug 2008 02:47:50 +0000 http://bit-player.org/?p=160#comment-1765 The Seposki graph shows over 1000 genera for most of the last 450 million years, and over 5000 in recent years; the Alroy graph shows there have never been more than 700 genera. This strikes me (a complete outsider) as a huge discrepancy. Have they redefined the word genus? Did you label your axes wrong? What gives? The Seposki graph shows over 1000 genera for most of the last 450 million years, and over 5000 in recent years; the Alroy graph shows there have never been more than 700 genera. This strikes me (a complete outsider) as a huge discrepancy. Have they redefined the word genus? Did you label your axes wrong? What gives?

]]>