By Rud Istvan,
This post is a lightly reworked last third of much longer essay No Bodies in ebook Blowing Smoke, foreword by Professor Judith Curry of Georgia Tech. It concerns the demonstrably misleading dishonesty of AR4 WG2 concerning climate change caused species extinctions, and the shoddy science behind the one paper AR4 ultimately relied upon. It is a saga not yet told at WUWT, which Charles the Moderator thought might be interesting since actually very science based in a perverse way.
One of the firmer catastrophic anthropogenic global warming (CAGW) predictions made by IPCC AR4 WG2 was an alarming increase in species extinctions. §4.4.11 Global synthesis including impacts on biodiversity was fairly specific. If warming reached 3°C above pre-industrial levels (projected absent serious mitigation) 21-52% of all species were committed to extinction (not necessarily yet extinct) by 2100. This official finding was based on 78 conclusions from 57 peer-reviewed papers on climate change biodiversity impacts, all listed in WG2 table 4.1. It appears to be supported by overwhelming scientific evidence. It isn’t.
How did IPCC AR4 WG2 reach its apparently scientific conclusion? A bit of forensic investigation starting with figure 4.4 shows how deceivingly dubious the IPCC process was.
The highlighted temperature rise of 1.7°C causing 9-31% species loss is supported by findings 5-7; 3°C causing 21-52% species loss is supported by findings 46-52. Most other findings discuss areal changes in habitat, or possible species impacts in specific ecosystems, for example the highlighted polar bears.
This official figure really says only 10 of the 78 references listed in table 4.1 actually support the IPCC global extinction estimates. The 10 references are:
Table 4.1 suggests deliberate IPCC intent to deceive in 4.4. The 1.7°C temperature estimate depends on only one reference, #1, not three. The 3°C global estimate also depends on only one reference, again #1, not the seven listed.[i] The entire IPCC AR4 global extinction estimate comes down to a single peer reviewed paper, #1 Thomas et. al. from 2004. [ii] In fact, 13 of the 78 IPCC 4.1 references are this single paper, including all numerical WG2 extinction estimates except freshwater fish. And fresh water fish weren’t globally risked in that sole other paper, as Table 4.1 #53 noted. [iii] This apparently deliberate IPCC deception merits closer scrutiny.
Thomas et. al. used a mathematical model of the exponential relationship between species number and species areal extent within ecosystem ‘climate envelopes’. Number of species S=c Az. Larger areas/envelopes A have more species; smaller areas/envelopes have fewer species. So as climate envelopes shrink with climate change, ‘extinctions’ rise since the equation says there will be fewer species. The exponent chosen by Thomas for final estimates for all species types in all regions was Z=0.25, which a paper footnote said was empirically appropriate for tropical birds given logging deforestation, citing a reference.[iv] That is a more than just dubious exponent value for all species including plants, and for climates outside the tropics.
A stronger methodological critique concluded this equation always overstates extinctions, no matter what exponent ‘Z’ is plugged in. [v]
HadCM2 was used to model temperature and rainfall change in seven regions for 2050, and hence future regional envelope areas A. Modeled regional climate change is dubious because of well-known problems with regionally downscaled GCMs. [vi] HadCM2 was also run “hot”. Supplemental information table 1 says a global mean anomaly of 2.6°C in 2050 for scenario SRES A2, ranging 2.1-3.9°C by modeled region. That is nearly twice what IPCC AR4 said. WG1 SPM Figure 5 scenario A2 ≈1.6°C. WG2 Figure 4.4 imaged above gives 1.5°C-1.7°C. Thomas modeled 2.6C for 2050 is absurdly inconsistent.
Using a dubious Z in a faulty equation with excessively hot climate model input, this paper then assessed 20% of Earth’s terrestrial surface in 7 specific regions, using seven groups of regionally endemic species including 112 mammals, 238 birds, 69 butterflies, and 607 plants. The Thomas supplemental information (SI) unfortunately does not identity the 1103 species assessed. Those endemic species results were extrapolated without modification to the world, a ‘hidden’ selection bias further overstating CAGW extinction risk.
(Note to WUWT readers. The earlier 2/3 of the ebook essay discussed species and extinction definitions in extensive detail, each with an illustrated example. An endemic species is one found only in some restricted, bounded singular ecosystem. The classic example is Costa Rica’s golden toad, found only in a 4km2 section of the Brilliante Ridge hiking trail of its famous MonteVerde Cloud Forest Reserve, The golden toad is now extinct because tourists hiking the Brilliante trail brought in chytridiomycosis fungal disease. Endemic species are automatically at much higher extinction risk by definition.)
Despite the missing SI species information, some things can be inferred about the Thomas paper’s species selections using birds (since they are passionately well studied by ‘birders’, and the basis for its ‘Z’). ‘Small’ Endemic Bird Areas (EBAs) account for about 2500 of the known 10518 avian species, and 93% of all range restricted birds. EBAs are intensely monitored by birders since half of the EBA species are already classified as threatened (CR, EN, VU per IUCN). Thomas Table 1 says endemic birds were assessed in Queensland Australia, South Africa, Europe, and Mexico. Queensland has 630 bird species of which 20-22 are endemic; table 1 says 13 were used. South Africa has 843 bird species of which only 19 are endemic; 5 were used. Europe has about 800 species of which only about 10 are endemic; 34 were used. Mexico has 1054 species, 145 EBA’s, and 98 endemic species; 186 were used. The paper selected 238 out of 3019 bird species known in the studied regions, 8%. The selection is weighted at least (149/238) 63% EBA birds—half already listed as threatened. That is a strong selection bias. It is not subtle, merely hidden.
Queensland can be scrutinized even more closely. All 20-22 endemic bird species (sources vary as to how many are just in Queensland) have limited ranges and specialized habitats. For example, Victoria’s Riflebird (the Duwuduwu) is a bird of paradise found only in Atherton Tableland (Plateau) remnant cool rainforests (the rest of those originally extensive cool rainforests having been logged and converted to farmland). Half (11) of Queensland’s endemic birds live only in those high cool rainforest remnants, as viewed from the Mt. Bartle Frere hiking trailhead above present Atherton Tableland farms.
Select only range limited endemic birds surviving only in a remnant cool rainforest ecosystem, regionally downsize HadCM2 to model excessive overheating and reduced rainfall in it, and voila! Avian climate change extinction risk increases—a lot. Queensland birds were the highest risked of all animal species/regions in the Thomas paper’s Table 1 assessments. Even if it were true (it obviously isn’t), this finding would not extrapolate to the other 610 Australian nonendemic birds inhabiting a much broader range than a single remnant ecosystem. By limiting assessed avians to endemic/ near endemic species, results were inherently (but silently) biased toward extinction risk. The paper’s bird result is methodologically overstated even for EBA species. It does not apply at all to the other three fourths of bird species. Yet it was. Then that overstatement was compounded into a simple averaged high end IPCC AR4 WG2 estimated extinction risk for all species.
Two further examples of Thomas et. al. selection bias can be identified for plants. The World Wildlife Federation estimates there are 80,000 plant species in the Amazon region. Thomas et. al. selected 9 (but did not say which 0.01%), then modeled a 69-87% extinction risk for those. Those estimates became part of the simple averages for all plants. Thomas et. al. selected 243 out of roughly 400 species of South African Proteaceae (afrikanns ‘suikerbossie’ or sugarbushes). The Proteaceae family comprises 80 generi and roughly 1600 species worldwide including the familiar macadamia nut. South African Proteaceae grow mainly in a small Cape Hope ecosystem known as the Fynbos, which covers just 6.7% of South Africa. The Australian Atherton Tableland EBA ‘trick’ was deliberately repeated with South African Fynbos Proteacease plants.
IPCC AR4 WG2 provably relied on nothing but this one grievously flawed paper for its oft-repeated extinction estimates.
[i] AR4 WG2 T 4.1 Fn 30 is Najjar 2000. Raymond Najjar is a well-known ecologist at Penn State University. A complete list of his publications is available at PSU. In 2000 he published 6 papers, four on ocean chemistry and two on the US mid-Atlantic seacoast. Nothing about butterflies, extinctions, or Australia.
[ii] Thomas et. al., Extinction Risk from Climate Change, Nature 427: 145-148 (2004)
[iii] The fish estimate was 4-22% in just 66 out of 165 rivers (AR4 WG2 table 4.1 #53 said >10%) from a single paper Xenopoulos et. al., Scenarios of freshwater fish extinction from climate change and water withdrawal, Global Change Biology 11: 1557-1564 (2005). This paper took IPCC AR4 drought predictions, modeled additional irrigation water withdrawals, and then modeled fish extinctions as river flows declined. In 99 out of 165 rivers there were none.
[iv] Brooks et. al., Deforestation predicts the number of threatened birds in insular Southeast Asia, Conserv. Biol. 11: 382-394 (1997) and Deforestation and Bird Extinction in tropical forest fragments, Conserv. Biol. 13: 1140-1150 (1999)
[v] He and Hubbell, Species-area relationships always overstate extinction risk from habitat loss, Nature 473: 368-371 (2011)
[vi] Pielke and Wilby, Regional Climate Downscaling: whats the point?, EOS 93: 52-53 (2012)