From RationalWiki
Jump to: navigation, search

This is an archive page, last updated 23 October 2021. Please do not make edits to this page.
Archives for this talk page: <1>, <2>, <3>, <4>, <5>, <6>, <7>, <8>, (new)(back)

Misinformation and fake references[edit]

Edit warring[edit]

Why is this article a lighting rod for edit warring and trolling? Tabula Rasa (talk) 06:30, 28 December 2018 (UTC)

"This talk page has been vandalised by Mikemikev on countless sockpuppets for years; don't feed the troll." Troll[edit]

Perhaps useful material[edit]

I debated NotPoliticallyCorrect last year. Here's a response I gave him:

What is a subspecies?

A subspecies (synonymous with geographical race) is a natural [1] group of populations isolated from each other by allopatry or parapatry (Smith et al. 1997). One of the earliest definitions of geographical races made it clear: "no other geographical race of the same species occurs within the same range" (Rensch, 1934). However, the zoologist Ernst Mayr (1942: 106) found this definition too strict, and wanted to redefine geographical races to include groups of animal populations that are spatially contiguous/overlapping (sympatric). Mayr was criticised for making classification arbitrary, since contiguous populations don't have discrete (non-overlapping) geographical boundaries, e.g. Edwards (1954, 1956) and Pimentel (1959) argued to standardise subspecies by limiting the term only to "allopatric groups of populations". [2] This concept is used in entomology (Van Son, 1955; Askew, 1970), ornithology (Monroe, 1982; Rising, 2007) and mammology (Wang et al. 2015), etc.

Interestingly, Mayr during his later career realised the difficulties of sympatry and came to accept the denial of sympatric geographical races is reasonable because arbitrary groupings of contiguous populations cannot be phylogenetic units as they aren't isolates (Mayr, 1970: 210, 1982). In the 1990s Mayr restricted subspecies to groups of populations which occupy non-overlapping geographical ranges i.e. allopatry (O'Brien & Mayr, 1991) after consensus was reached subspecies are not useful as a mere arbitrary pigeon-holing device (Cracraft, 1992: 101). Subspecies are geographically separated by prezygotic isolation mechanisms (Fig. 1, allopatry/parapatry [3]) and given enough time will be postzygotic reproductively isolated. Geographical races are genetically discrete and/or distinct infraspecific lineages (Templeton, 1998) and/or clades (Andreasen, 2004), i.e. have independent evolutionary trajectories and are widely considered incipient species/semi-species: "populations partway through the evolutionary divergence towards full speciation" (Frankham et al. 2010: 369).

(Figure 1, modified from Mason et al. 2018): Three bird subspecies of Sporophila torqueola, showing one geographical race is allopatric, while the other two parapatric. Humans in contrast are sympatric and therefore don't have any subspecies.

It is therefore to be expected subspecies are significantly different from each other as a result of absent or very restricted gene flow (genetic divergence). Prior to whole genome sequencing and advanced multivariate statistical procedures in physical anthropology that can determine overall genetic similarity, zoologists faced the problem of what characters to select to distinguish subspecies: "different types of characters... may lead to rather different grouping" (Mayr, 1981). As an example certain characters when analysed individually or in few numbers, often vary non-concordantly (Wilson & Brown, 1953) and make different classifications, especially if the pattern of variation is the result of selection (Diamond, 1994; Brace, 2000a: 293-294). This problem is now avoidable by discarding non-neutral traits that are poorly informative about ancestry, or maximising the number of characters and/or variables. [4] However, merely distinguishing two or more groups of populations doesn't mean there are always subspecies – differences must be taxonomically meaningful.

Many zoologists and biologists impose quantitative thresholds (on genetic/morphological characters) as a criteria to diagnose subspecies (Maglo et al. 2016), e.g. an Fst value of 0.25 (Templeton, 1998, 2013) based on Sewall Wright's (1978: 85) calculation 0.25 > Fst means "very great" infraspecific differentiation. And if two or more groups of populations are to a massive degree genetically distinctive – this justifies subspecies as phylogenetic units (i.e. taxonomic meaningfulness) despite an Fst threshold is subjective. A different, but objective criteria is to identify geographical races based on qualitative (present/absent) characters, rather than quantity or frequency (proportion of variation). This is achieved by identifying uniquely derived traits (autapomorphies) with best phylogenetic resolution that are exclusive to and typically appear in a single subspecies because of divergent mutations and don't appear in any other subspecies because of geographical isolation: "an identifiable evolutionary lineage is a qualitative state thereby avoids... arbitrary quantitative threshold" (Templeton, 2003: 239).

Geographical races don't exist today, but once did...

There are no geographically isolated human groups of populations alive today, [5] so there are no living human subspecies, but geographical races once existed thousands of years ago (Smith, 2010: 373; Wolpoff & Caspari, 2013: 332) to a time when there was no vehicle technology and humans were a lot less mobile, also predating large-scale migrations that started with the spread of agriculture (10,000-5,000 years ago). As Smith (2010: 373) puts it: "to see true racial variation in humans, one has to go to the fossil record". The so-called Neandertals were likely a parapatric subspecies and are sometimes identified as a subspecies by palaeontologists based on their autapomorphic non-metric skeletal characters e.g. suprainiac fossa (Balzeau & Rougier, 2010).

There is far less population structure in humans today than the distant past. Neandertal fossils show polarity to "Anatomically modern humans"'s in their average frequency of non-metric cranial characters (Trinkaus, 2006) and mean cranial measurements (Howells, 1989: 66; Harvati, 2003), while ancient DNA reveal discontinuities (Caramelli et al. 2003; Briggs et al. 2009; Prufer et al. 2014). Krings et al. (1997) found the difference between Neandertal/AMH mtDNA is three times larger than mean variation among living humans. Neandertals and "AMH" fossils have autapomorphic traits (Santa-Luca, 1978) that make palaeontologists able to distinguish or sort individual fossil specimens – with a rate of 100% accuracy. In sharp contrast, "genetic studies of [living] human populations have revealed that there are no alleles distinctive of this race or of that race" (Kitcher, 2007) [7] and forensic scientists today have a fairly lower success rate at determining biogeographical ancestry of human remains because (recurrent) gene flow between populations has blurred genetic boundaries.

Arguments against racialism

When scientists analyse population structure today there is no inter-regional discontinuity in neutral genetic markers, but rather smooth gradients in average allele frequencies across space (Barbujani et al. 1997; Serre & Paabo, 2004; Glasgow, 2009: 105-106; Barbujani & Colonna, 2011: 105; Morning, 2014a; Maglo et al. 2016) and gradual spatial variation in morphometric means (Relethford, 2004, 2009) and non-metric trait averages (Hanihara et al. 2003, 2008) explained by isolation-by-distance (Fig. 2): "genetic similarity is inversely correlated with geographic distance" (Bolnick, 2008: 72). Cavalli-Sforza et al. (1994: 121) note that until two centuries ago: "most movements covered short distances and only rarely did people venture out of the short range in which daily movement took place". This is supported by parish-records that show during the Middle Ages in England most men chose a spouse within a mile of his residence; a minority from nearby parishes, but almost none from distant parishes. Since isolation-by-distance restricts gene flow over large distances: "individuals tend to mate with those who live nearby and geographically close populations tend to exchange more genes [gene flow] than geographically distant" (Bolnick, 2008: 72)

A popular argument against racialism is human inter-regional genetic variation between continents is small (between 0.03 and 0.15 Fst using many different neutral genetic markers), relative to intra-regional variation (Relethford, 1994; Lewontin, 1972, 1974: 156-157, 1987, 2006; Brown & Armelagos, 2001; Li et al. 2008; Rosenberg, 2011) thus invalidating social constuctivist races as phylogenetic units. In a recent study on post-crania, the inter-regional variation is as low as 0.056 Fst (Betti et al. 2013). As a comparison – chimpanzees have a lot more infraspecific population structure (Templeton, 2013); the human species is monotypic, not polytypic with subspecies. In 2003, a geneticist published a paper named "Lewontin's Fallacy" and argued it is possible to identify an individual's biogeographical ancestry regardless of whether the proportion of genetic variation is low (Edwards, 2003). However, this argument is a red-herring logical fallacy itself: "...geographical correlations are far weaker hypotheses than genetically discrete races, and they obviously exist" (Marks, 2010: 270-271).

The fact human populations smoothly grade into each other in neutral genetic markers, whose isolation-by-dstance pattern of genetic variation results from stochastic processes (genetic drift, gene flow) rather than selection – first led anthropologists from the 1960s [8] to deny the existence of races (Livingstone, 1962 [1964]; Hiernaux, 1964, 1965a; Brace, 1964a, 1964b; Fried, 1968) that was further supported by Richard Lewontin's (1972) study that showed low inter-regional quantitative variation (0.063 Fst) in classical polymorphisms (blood groups). The faction of scientists who denied races for pheno-clines, the so-called clinalists, became the majority by the 1980s, e.g. "the view that races do not exist was expressed in 10 textbooks [published 1975-79] and became the modal position, with only 5 texts arguing races are real" (Littlefield, et al. 1982). There is now scientific consensus human races don't exist (Edgar & Hunley, 2009; Wagner et al. 2016) e.g. the Human Genome Project notes: "DNA studies do not indicate that classifiable subspecies (races) exist in modern humans". This doesn't mean to say there isn't some (albeit weak) human population structure, but it's not racial and is best analysed by local populations.

Local populations as small biopackages

Although C. Loring Brace (1964a, 1964b) went as far as discarding local populations as operational units of analysis, it is seemingly impossible to map clines (individual character gradients across space) without them: "the clinal approach has its strength, but only when used in conjunction with the actual basis for trait distribution, the population" (Molnar, 1983: 146) and Brace et al. (1980) later himself used local populations to map tooth-size pheno-clines. For this reason, anthropologist William W. Howells amended Livingstone's (1962) aphorism "there are no races, there are only clines" to "there are no races, there are only populations" (Ousley et al. 2009) having taken 57 measurements from 2500 skulls across the globe. Howell's monumental study found neutral craniometric variation doesn't cluster into groups of continental/regional populations because averages for craniometric variables are spatially gradual, not abrupt, but local populations are more distinguishable, i.e. "individuals assign themselves to specific populations better than to 'races' or regional samples" (Howells, 1995: 103-104) such as Moriori, Ainu and other local ethnies or tribes.

Forensic anthropologists who study osteology are able to determine the biogeographical ancestry of human remains at "relatively high-allocation accuracies (often more than 80%, but rarely more than 90%)" (Albanese & Sanders, 2006: 309). This is lower than 100%, but is nevertheless quite remarkable. However, these relatively high accuracy rates are restricted to local populations. As an example Giles & Elliot (1962) took 8 measurements from skulls at the archaeological site Indian Knoll, Kentucky to construct a reference sample to identify an "American Indian" race based on discriminant function analysis (based on multivariate distances to the reference sample centroid). When tested on crania from Inuit/Eskimo people (Labrador, Canada): only 64.4% were correctly identified (Birkby, 1966), while Fisher & Gill (1990: 59) accurately classified as low as 25.9% (m) and 37.5% (f) skulls (= 31.7%) from north-west US Plains Indian tribes (Goodman, 1997).

In a separate test on skulls from the same site of the reference sample (Indian Knoll): 92% of crania were correctly identified (Birkby, 1966) meaning while it is possible at relatively high (but not total) accuracy to determine biogeographical ancestry of human remains from local populations that form reference samples: "the method performed very poorly when it was applied to samples outside [of] the reference sample used to develop the original method" (Albanese & Sanders, 2006: 285). In other words Indian Knoll is an operational population of use to a physical anthropologist, but an "American Indian" isn't. Similarly, non-metric analyses that score average trait frequencies (rather than mean measurements) reveal "the clinal nature of discrete cranial trait variation across regions" (Hanihara et al. 2003) or as Brace (2000b) puts it: "each region grades without break into... next door". Forensic scientists recognise metric/non-metric techniques (for ancestry assessment) that apply well to local populations (e.g. Indian Knoll) are much less effective when applied to broad groups of populations (Brues, 1992; Smay & Armelagos, 2000; Ta'ala, 2017: 10-11).

It might be asked what about traditional (early 20th century) arbitrary regional/continental groups of populations that were once used by anthropologists (Coon et al. 1950) such as "Mongoloids", "Caucasoids" and "Negroids". As it happens like "American Indian" these have all been invalidated by studies showing they were originally developed on a "subset of a sample and then tested on the sample from which the subset was derived" (Smay & Armelagos, 2000). The "Mongoloid" is a good example; according to Heard (2008): "traits considered to be characteristic of the (classical) Mongoloid group were not derived from studies encompassing all of the populations that would be classified as Mongoloid". It is now widely known there is a continuous spatial gradient of metric means and average non-metric trait frequencies across East Asia (Howells, 1983: 300) contradicting an outdated view populations in this large region are homogeneous in a suite of characters, including a dental complex: "in East Asia we are dealing more with a gradual [pheno-]cline in dental morphology than a sharp distinction between Sinodonts and Sundadonts" (Bulbeck, 2000).

Granular vs broad classifications of populations

Why is the biogeographical ancestry of humans far better at identifying on a fine-grained level rather regional/continental (Fig. 3)? Variation in neutral genetic markers is spatially gradual and local populations better approximate gradients; broader divisions for example in forensic science are too heterogeneous (Albanese & Sanders, 2006: 285). For the same reason, when mapping pheno-clines, it makes much more sense to plot gradualist changes as small as possible, so that variation is not masked in the analysis: "application of much of population genetics works best when considering variation between local populations and not between aggregates" (Relethford, 2017: 168). The vast majority of geneticists stopped categorising groups of populations decades ago, to focus on local populations, so-called demes, especially that are of use to medical genetics. Lewontin (2006) provides an example of α-thalassemia that has a high frequency in specific local populations in South Asia (e.g. Tharu people in Nepal) but low in others; a "South Asian" race is not useful, although a local ethnic group or tribe is: "there can be considerable genetic heterogeneity within a region, it is most useful to be as specific as possible, about geographic origins, ethnicity or tribal affiliation" (Tishkoff & Kidd, 2004).

In terms of population genetics, local populations reflect actual mating behaviours and are often named demes or breeding populations: groups of spatiotemporal individuals that have a higher tendency to mate with individuals within the same group than with different groups (Hiernaux, 1964: 32; Relethford, 2012: 18). Demes are operational units to study microevolution (genetic change from generation to generation). The fact deme boundaries are somewhat fuzzy isn't a problem since they aren't phylogenetic units and so don't require discontinuity like subspecies do (Templeton, 2013). As summarised by one contemporary scientist: "focus on the population genetics of human demes is what permitted biological anthropologists of the 1970s to avoid race altogether" (Marks, 2010: 269) and demes are the "small biopackages" that replaced the traditional concept of race (Marks, 1995: 274-275).

The population geneticist Luigi L. Cavalli-Sforza and colleagues in 1991 set up the Human Genome Diversity Project that includes 52 local populations across the globe as reference samples. In his book The History and Geography of Human Genes, Cavalli-Sforza et al. (1994: 19) reject race because of human mobility and gene flow; there are no discrete genetic boundaries, and clustering populations into larger divisions is a "futile exercise". Instead, Cavalli-Sforza and colleagues defend demes as the operational unit of analysis to map gradients of allele frequencies across space and identify fine-grained local populations. Aeschylus (talk) 13:21, 6 February 2019 (UTC)

NotPoliticallyCorrect basically argues for races as continental populations, I showed him why these are of no use in population genetics. For example if we look at medicine: "big groups... are too heterogeneous to lump together in a scientific way... if you're doing a DNA study to look for markers for a particular disease, you can't use 'Caucasians' as a group. They're too diverse" (Naggert, 2000). He never had a rebuttal for this. And if we look at ancestry DNA testing, these broad groups also have no use; I provided an example:

Self-identified "White Americans" are by no means homogeneous "due to historical immigration of diverse source populations" from Europe (Price et al. 2008) e.g. Bryc and colleagues found mean Swedish and/or Norwegian ancestry significantly differs in specific US states from under 5% to over 20%, while British (English, Welsh or Scots) ancestry more remarkably by under 25% to over 55%. For this reason, International HapMap Project includes a local population from Utah, who self-identify with a narrower-scope of geographical ancestry (i.e. British Isles), than the whole European continent. Contemporary scientists who study biogeographical ancestry of humans understand to use granular, not broad human population divisions: "ancestry components that geneticists are most commonly including in their analyses are making fine-grained distinctions between people" (Turkenheimer et al. 2017) which confirms Relethford's (2017) main criticism of race: "application of much of population genetics works best when considering variation between local populations and not between aggregates".

Aeschylus (talk) 15:30, 6 February 2019 (UTC)
This is good stuff. :thumbs: — Dysk (contribs) 16:05, 6 February 2019 (UTC)