Cichlid Radiations

"Here, we demonstrate that multiple large animal adaptive radiations arose from a hybrid swarm." — Joana Meier and colleagues, Nature Communications, 2017¹

"For all the diversity of species, I found the cichlids to be an unmistakably natural group, a created kind." — Arthur Jones, 2001²

Summary

Lake Victoria in East Africa holds about 500 species of haplochromine cichlid fish found nowhere else.  The lake dried out near the end of the last ice age, so the species are often said to have formed in about 15,000 years.  Creationists cite the flock as proof that many species can come from few founders quickly, which a post-Flood model needs.  The case is real, and it is weaker than either side usually says.  This article scores it with the series taxonomy set out in Measuring Biological Information.

The main findings:

1. "500 species in 15,000 years" is two estimates multiplied. The count ranges from "more than 300" in 1996 to 500 to 1,000 later. The clock is a radiocarbon date for the lake refilling, "before 12,400 carbon-14 years before the present," which is about 14,600 calendar years.^{3 4} Neither number was measured on the fish. The "extremely rapid" verdict is an inference from that date. Johnson's team ruled out refuges in the basin and concluded that "most of these species have evolved during the past 12,400 years."³ Without a refuge, they wrote, the rate "is the fastest ever recorded for such a large number of vertebrate species."³
2. The basin was dry, but it was not empty for long. Cores taken in 2018 confirm that the deep lake floor was exposed until about 16,400 years ago.⁵ From about 16,700 years ago the basin held wetlands with permanent shallow ponds, and a deep lake existed only by about 13,500 years ago.⁵ Fish could have survived in those ponds. So the lake is young, but the fish lineages need not be.
3. The variation is old and was mixed by hybridisation. Genome studies infer that the whole Lake Victoria region flock descends from a hybrid swarm of two river lineages that had been apart for about 1.6 to 5.8 million years on a calibrated clock.¹ The best-studied adaptive gene, a red-light opsin, has two allele classes, one from each parent.¹ All of this is read from genomes. None of it was watched.
4. Nobody watched a cichlid species form. What was watched is the reverse: species losing their colour barriers in muddy water, and about 200 species vanishing from catches after Nile perch were introduced.^{6 7} Small crater lakes show populations caught early in divergence.
5. The species barely differ genetically. Victoria species show "almost no genetic differentiation" at microsatellites, and Lake Malawi species differ by only 0.1 to 0.25% of their DNA.^{8 9} They interbreed without loss of fertility and are held apart mainly by mate choice.
6. Scored in the series taxonomy, the radiation is mostly sorting. The hybridisation that fed it is a transfer. The sorting of old alleles into species involves no new mutation at all. The best candidates for new adaptive mutations are in Lake Malawi, not Victoria: insertions of mobile DNA that retune opsin expression, and possible new protein changes in two vision genes. Both are inferred.
7. The clock still cuts against a young earth. A post-Flood window of about 4,400 years is 3.3 times shorter than 14,600 years. It needs one doubling of species number every 490 years, against about 1,630 years on the short clock and 1,900 to 3,200 years for the fastest comparison lake in the standard compilation. Section 6 shows the arithmetic. The variation itself is dated by clocks to millions of years, and the one measured cichlid mutation rate cannot shrink that.
8. Some mainstream work points the other way. Whole-genome modelling on that measured rate puts the colonisation of Nicaraguan crater lakes at 4,700 to 800 years ago, with new species formed after it.¹⁰ That is inside a post-Flood window, though it is an inference, not an observation.

Verdict.  Cichlids are the best real-world evidence that many species can be sorted from a mixed founding stock in thousands of years with no new genes needed, and that result belongs in the creation model of rapid speciation.  The same data date the stored variation to millions of years, and no young-earth author has published a model of where it came from or a test that could tell created variation from accumulated mutations.

Part 1: The claim and its clock

What the 1996 cores showed

In 1996 Thomas Johnson of the University of Minnesota Duluth and eight colleagues reported seismic surveys and sediment cores from Lake Victoria.  Their abstract is short:

"Lake Victoria is the largest lake in Africa and harbors more than 300 endemic species of haplochromine cichlid fish.  Seismic reflection profiles and piston cores show that the lake not only was at a low stand but dried up completely during the Late Pleistocene, before 12,400 carbon-14 years before the present.  These results imply that the rate of speciation of cichlid fish in this tropical lake has been extremely rapid."³

The full paper gives the details.  The team surveyed the lake in March and April 1995 and took piston cores at seven sites.³  "Five of the seven cores that we recovered from the lake penetrated the entire Holocene sequence and sampled a soil horizon."  The horizon shows vertically oriented rootlets and "prismatic cracking characteristic of vertisols."³  A vertisol is a cracking clay soil that forms on dry land.  The deepest core, V95-2P, came from about 67 m of water, and a 1996 core from the deepest point in the lake, 68 m, "again recovered vertisol at the base of unit A."³  The soil itself was not dated.  The date comes from "plant macrofossils taken from just above the soil horizon at site V95-2P" and is 12,400 ± 70 carbon-14 years, an age that "marks the onset of flooding of the lake basin at this site."³  Two pollen dates in the same core, 11,140 and 12,910 years, were set aside as too old, from reworked pollen.³  So 12,400 is a minimum age for the dry phase, not a measurement of how long it lasted.  The seismic profiles carried the result across the basin:  "the hard reflector is present everywhere beneath the Holocene diatom muds."³  The authors also limited their own claim:  "If this site was dry, then the entire lake basin was dry, at least on a seasonal basis."³  Ole Seehausen, summarising the same work in 2002, put the dry interval at "13 200 to 12 400 14C-years BP."⁴

The date is a radiocarbon date, counted in carbon-14 years.  Seehausen converted it to 14,600 calendar years, and that is where the familiar "15,000 years" comes from.⁴  It dates the return of water to the lake floor, not the fish.

The dispute

Geoffrey Fryer, a freshwater biologist, rejected the dry lake in 2001.  His argument was biological.  If the lake dried out, everything in it died, yet the lake holds endemic catfish, other fish and invertebrates that could not have waited in rivers:

"the account completely overlooks the fact that any such desiccation must have eliminated not only the haplochromine cichlids but the entire biota of the lake."¹¹

He pointed to a deep-water catfish, Xenoclarias, and wrote that "Its presence indicates beyond reasonable doubt that Lake Victoria was not dry some 12 400 years ago."¹¹  He also cited a shallow core from 1969 that showed "an unbroken accumulation of sediment to almost 14 000 years before present (BP), by which time, according to Johnson et al. (1996), the lake had been dry for well over 1000 years."¹¹  Johnson's paper itself gives no length for the dry phase.  It reports that the lake fell "at ~17,300 ¹⁴C years B.P." and began to refill by 12,400.³  Fryer's conclusion was sharp:  if a swamp survived, that "destroys the argument for express speciation."¹¹

Seehausen answered the next year.  He computed speciation rates for all 41 fish lineages in the lake on a 14,600-year history and found that only one needed rates faster than other young lakes:

"Except for one cichlid lineage, accepting a 14 600 year history does not require any speciation rates that fall outside the range observed in fishes in other young lakes around the world."⁴

He added a caution that matters for the creation model:  "since it is unknown how many founding species it has, it is not certain that its speciation rates are really outside the range observed in fishes in other young lakes."⁴

In 2003 Erik Verheyen, Axel Meyer and colleagues attacked the date from the genetic side.  They counted mitochondrial mutations in about 300 fish and dated the flock's population expansion to "about 98,000 to 132,700 years ago."¹²  They concluded that the presence of older haplotypes "strongly argues against the view that LV dried out completely."¹²  The geologists replied that this reading was "incompatible with the geological context."¹³  In 2008 Curt Stager and Johnson restated the case for desiccation and allowed more than one low stand.  As quoted in a 2023 paper, they wrote that the lake "began to re-fill ca. 15,000 years ago […] then fell again to virtually complete desiccation some time between 15,900 and 14,200 years ago."⁵  Their paper itself was not obtained.

A 2009 study of nuclear markers took a middle position.  It found a 30- to 50-fold population crash about 18,000 to 15,000 years ago, and a common ancestor about 4.5 million years ago, and titled the result "bottlenecked but did not extirpate."⁸

The genetic dates in this dispute are not measurements.  Verheyen's 98,000 to 132,700 years converts 4.2 mutations with a rate calibrated on lake-level events.  Meier's 2017 paper warns that its own mitochondrial dates under a million years "are most likely overestimates because of the increase of molecular rates towards the recent."¹  The geology is the firmer of the two clocks.

What the newer cores show

In 2018 a team from Bern took four new cores along a line from 13 to 63 metres of water depth, with pollen analysis and new radiocarbon dates.⁵  They confirm the dry phase:  at the deepest site the lake floor was exposed "from at least ~20.2 ka (bottom of LV4), until prior to ~16.4 ka."⁵  Here "ka" means thousands of calendar years ago.  What followed was not an empty basin waiting for rain:

"At our coring site in the central basin, local Typha wetlands existed >16.7 ka, alternating with periods of desiccation.  Moisture increased slightly between ca. 16.7 – 14.5 ka and wetlands with permanent, shallow ponds established simultaneously in the center and the marginal, more elevated parts of the flat lake basin."⁵

Typha is cattail.  The wetlands appear at sites from 63 down to 22.6 metres at once, which "suggests that shallow water bodies … persisted simultaneously in large parts of the basin."⁵  The lake then rose fast:  "A 40–50 m deep lake was established by ~13.6 ka."⁵

The 2023 genomic study of the flock by Meier and colleagues fits this picture.  It says the largest radiation took place "in a lake less than 16,000 years old," and that it began when "several swamp-dwelling refugial populations, each of which were of older hybrid descent, met in the newly forming lake."¹⁴  The core study itself does not mention fish.  Linking its ponds to Meier's swamp refuges is this article's inference.

What this does to "500 species in 15,000 years"

Each newer result changes the headline differently.  The 1996 cores make the lake young.  The 2023 cores keep the lake young but give the fish somewhere to wait, which answers Fryer's main objection without his conclusion.  Once fish can survive in the basin, the age of the lake is no longer the age of the fish lineages, and "500 species in 15,000 years" becomes 500 species' worth of ecological sorting in a lake about 14,000 to 16,000 years old, from lineages that may be older.  Meier's 2023 abstract still claims that "all of its ecological guilds evolved in situ,"¹⁴ so the short clock survives as a clock for the ecological radiation.  The rate claim is only as good as that date and that count, and the article's arithmetic in Part 6 uses both with their ranges.

Part 2: Where the variation came from

A hybrid swarm at the root

In 2017 Joana Meier, Ole Seehausen and colleagues sequenced cichlids from every major African river system and from all the lake radiations in the Lake Victoria region.¹  What they measured was patterns of shared DNA.  Every lake radiation shares extra alleles with two Upper Nile species, and a statistical ratio puts the Upper Nile share at about 20% in every lake.¹  The shared stretches of genome average about 3,000 bases, which they read as mixing "many thousands of generations ago in the common ancestor of all superflock radiations."¹

From these measurements they inferred a history.  A Congolese lineage entered the Lake Victoria region and met an Upper Nile lineage from which it had been separated for a long time.  They date that separation "to ~1.6–5.8 million years ago," using a mitochondrial tree dated with four different sets of calibration points.¹  The two lineages hybridised, and the mixed population became the ancestor of about 700 species across several lakes.

The red-light opsin gene LWS is the showcase.  It has two allele classes in the flock, one tuned for clear shallow water and one for murky deep water, differing partly by a single amino acid that shifts peak sensitivity "by 7 nm."¹  Meier's team found that the two classes "are each shared exclusively with just one of these parental lineages."¹  An earlier study had already concluded that these alleles "originated prior to establishment of the species" and must have come "either from a single polymorphic ancestral population or through hybridization between species."¹⁵

How much of the difference is old

Meier's conclusion is that "speciation has commonly been associated with sorting of alleles brought together in the radiation ancestor by the admixture event."¹  That is a claim about enrichment, and its own figure limits it.  Of 12,890 variable sites in six Victoria species, 340 are strongly differentiated.  The text says that at "30% of sites" of that kind one allele came in through the Upper Nile hybridisation.¹  The largest category in the figure holds 10,799 sites where only one of the two alleles was found in the sampled parents, labelled "novel LV allele or unsampled in parental lineages."  At 1.4% outliers, that category supplies about 150 of the 340.¹  With only a few parental fish sampled, "unsampled" and "new" cannot be told apart.  The paper shows that old parental differences are over-represented where species differ.  It does not show that every difference is old.

The five-genome study of 2014 put a floor on the old share.  It found that "14–15% of all Victoria SNPs are also variable among the other cichlid genomes," and that the divergent variants between species differing in colour are rich in these ancient ones.¹⁶  It also found that "Fixation of alternative alleles between species happens but is restricted to a minority of the many divergent loci, consistent with models of polygenic adaptation from standing genetic variation."¹⁶  A 2020 study of 100 Victoria genomes found "hundreds of ancient haplotypes" carrying insertion and deletion variants, "shared with ecologically similar species from other older radiations elsewhere in Africa."¹⁷  A 2023 review traced some of them further back, to alleles "that may have originated in the ancestral lineages of Lake Tanganyika cichlids."¹⁸

Lake Malawi repeats the pattern

Lake Malawi's flock of more than 500 species shows the same signature independently.  A 2020 study found hybridisation between "two lineages that split 3–4 Ma," with "tens of thousands of polymorphic variants apparently derived from the hybridization" over-represented among the loci that separate ecological groups, opsins included.¹⁹  Two unrelated lake flocks both starting from hybrid swarms makes the pattern a regularity, not a single lucky case.

How old the old variation is

Only the last row uses a mutation rate measured by sequencing parents and offspring:  3.5 × 10⁻⁹ per base per generation, "approximately three to four times lower than in human."⁹  The other rows rest on fossil and geological calibrations, or on an assumed rate.  In sticklebacks, measuring the rate made the dates older, not younger;²⁰ in cichlids the one measured rate gives hundreds of thousands of years for splits inside a single lake.

Part 3: What was actually watched

This series admits a case only for what was watched happening.  By that rule the Lake Victoria radiation itself is not an observation.  It is a rate inferred from a species count and a geological date, and its history is read from genomes.  To make it observed, someone would have had to follow one population through generations, genotyping it, until two groups stopped interbreeding.  Nobody has done that for any cichlid.  Speciation that has been watched, mainly in plants, is covered in Observed Speciation.

Three things in cichlid lakes have been watched.

Speciation running backwards

In 1997 Seehausen, Jacques van Alphen and Frans Witte reported that Lake Victoria species "can interbreed without loss of fertility but are sexually isolated by mate choice," and that mate choice depends on colour.  Where the water had turned murky from pollution they found "Dull fish coloration, few color morphs, and low species diversity."  Turbidity, they concluded, "blocks the mechanism of reproductive isolation."⁶  The 2008 sensory-drive study says it provides "a mechanistic explanation for the collapse of cichlid fish species diversity during the anthropogenic eutrophication of Lake Victoria."²¹

This is the reverse of speciation:  species merging when the signal that kept them apart is lost.  It was documented by comparing clear and murky sites, and only the abstract of the 1997 paper was available here, so no time span for the merging is given.  The Institute for Creation Research drew the right lesson from it in 2008:  "the cichlids can rapidly favor certain colors, but then rapidly change back to having no particular preference if the environment is no longer color selective."²²

The Nile perch

Nile perch, a large predator introduced to Lake Victoria, multiplied explosively in the 1980s.  A 1993 study by Tijs Goldschmidt and colleagues reported that the perch boom "caused the destruction of approximately 65% of the endemic haplochromine cichlids," and called "the eradication of approximately 200 vertebrate species in less than a decade" possibly the century's largest vertebrate extinction.⁷  The decline was recorded as it happened, so this was watched;  the survey data are in a 1992 paper by Witte and colleagues that was not obtained.  It is loss, not origin, and "destruction" in 1993 meant disappearance from catches.

Crater lakes caught early

Small crater lakes are the nearest thing to a controlled setting.  Lake Massoko in Tanzania is 700 metres across and formed about 50,000 years ago.²³  Its Astatotilapia fish have split into a shallow and a deep form that differ in colour, body shape and diet, and females prefer males of their own form.  Their genomes differ very little:  "there are no fixed differences between Massoko benthic and littoral ecomorphs.  Genome-wide divergence FST is 0.038."²³  The deep form carries a rhodopsin allele that differs by four amino acids and absorbs bluer light.²³  The authors call them "a pair of incipient species."²³  The fish were sampled once.  Their history, divergence "within the last 500-1,000 years," is modelled from genomes.²³

Lake Apoyo in Nicaragua holds Midas cichlids.  In 2006 Marta Barluenga and colleagues argued that the Arrow cichlid formed inside this closed crater lake "within less than ~10,000 yr," in a lake about 23,000 years old, after a single colonisation.²⁴  In 2020 a study of 453 genomes from the same research group revised that picture:  speciation in Apoyo "was preceded by admixture from a secondary wave of colonization."¹⁰  The same study, using a mutation rate measured in Malawi cichlids, dates the whole 13-species Midas complex to about 16,700 years and the colonisation of the crater lakes to "between 4,700 and 800 years ago."^{10 10}  Apoyo now holds six described species.¹⁰  These are model dates, not observations, but they are the youngest species ages in the cichlid literature found for this article.

Part 4: What separates the species

Colour, vision and sensory drive

Victoria cichlid species are held apart mainly by female choice of male colour.  In the best-studied pairs, one species lives deeper, where light is redder, carries the red-shifted LWS allele class and has red males;  its sister lives shallower with the other allele class and blue males.¹⁵  Seehausen and colleagues reported that "speciation occurs on all but the steepest gradients" of light along island shores.²¹  The difference in the pigment is small.  Measured in the laboratory, the red-light pigment of the deep-water allele is shifted "by 7 nm" with one of the two chromophores fish use, and not at all with the other.^{15 15}  Both allele classes are older than the species, as Part 2 showed.

Jaws and teeth also differ, sometimes dramatically, between algae scrapers, snail crushers, fish eaters and scale eaters.  Meier's 2017 paper lists tooth shape with colour and opsin alleles as traits that diverged again and again within the radiations.¹  A 2020 genome study found that "the origins of ecological guilds are concentrated early in the radiation."¹⁷  Arthur Jones, who did his doctorate on cichlid development, described the same pattern in 2001 as "the endless permutation of a relatively small number of character states: 4 colors, ten or so basic pigment patterns and so on."²

How little genetic difference there is

In Lake Malawi, "the sequence divergence within a single diploid individual is sometimes higher than the divergence between two distinct species."⁹  At microsatellites, "even a reasonable number of microsatellites is insufficient to distinguish among many LVRS species within a lake."⁸  Fertility is not the barrier either.  Meier notes that "allopatric cichlid species, divergent by even millions of years, readily produce fertile offspring in the lab."¹

What a species means here

In this flock a species is a population whose females choose their own males, usually by colour, and which keeps its own ecology.  It is not a population that cannot interbreed.  Many are separated at a handful of regions of the genome while sharing the rest.  Nathaniel Jeanson's comment on species in general applies with force:  "The term “species” is a moving target."²⁵

This bears directly on the count of 500.  If a large share of the species are colour forms separated by mate choice at a few loci, then "500 species" measures how many times isolation by behaviour arose, not how much new genetic information appeared.  It does not lower the rate of isolation itself, and that rate is what a post-Flood model needs.

Part 5: Scoring the radiation

Three different processes

The series scores each change with the four classes of Measuring Biological Information:  transfer, loss, gain and modification, asked in that order, with every case marked observed or inferred.  The cichlid record contains three different processes, and only one of them is a mutation.

1. Hybridisation of the parental lineages. DNA of Upper Nile origin entered the Congolese lineage. That is introgression, which the taxonomy scores as transfer, worth zero new nucleotides to the receiving lineage. Inferred.
2. Sorting of old alleles into species. Changing the frequency of alleles already present is not a mutation and is not scored. The sticklebacks article treats the rise of the old Eda allele in lakes the same way.²⁶ Most of what separates Victoria species falls here. Inferred from genomes, and its reverse was watched in turbid water.
3. New mutations during the radiation. These are the only changes that could add new sequence. The next section looks for them.

Looking hard for a new adaptive mutation

Four candidates came up.

• LWS position 177. The substitution that shifts the red pigment is a real, function-changing change of one amino acid. It separates the two allele classes that the two parental lineages brought in, so it arose before the lake flock existed.¹ It cannot be credited to the radiation.
• Massoko rhodopsin. The deep-water allele differs by four amino acids and is blue-shifted.²³ The paper does not say whether the allele arose in the lake.
• Mobile-DNA insertions at opsin genes in Lake Malawi. This is the strongest case. Karen Carleton and colleagues traced four variants that change opsin expression to "indels in the promoters of either key transcription factors or an opsin gene," found that they "are the result of the movement of transposable elements," and that "they are endemic to Lake Malawi, suggesting that these TEs are recently active and are segregating within the Malawi cichlid lineage."²⁷ Here new mutations arose inside a lake radiation and retune colour vision. Only the abstract was available here.
• Two vision genes in deep-water Malawi fish. At peripherin-2 and an arrestin-C gene, protein-changing variants are shared by two deep-water groups while nearby silent variants are not, which "suggests that for these two genes there may have been independent selection on the same de novo mutations."⁹ The authors hedge it, and the effect on the proteins was not measured.

The five-genome study of 2014 also reports "40 cases of de novo miRNA emergence" and extra gene duplicates across East African lineages.¹⁶  Those differences accumulated over millions of years between lineages, not within a lake flock, and belong with Orphan Genes and Gene Duplication and Divergence.

No gain of function in the sense of a new coded element was found.  The transposon insertions come closest, because they put new sequence into control regions, and a new control relationship would score as a gain.  The abstract describes a change in how much of an existing opsin is made, which is a modification.

Why there is no ledger

Organism articles in this series multiply population size by generations and mutation rate to price what was searched.  That cannot be done honestly here.  No census of Lake Victoria haplochromines before the Nile perch collapse was found, generation times in the sources range from one to three years, and the radiation's output is mostly sorting, not new sequence.  The unique-functional-sequence cell is left empty.

Part 6: What each side makes of it

The creationist reading

Arthur Jones stated the creationist model for cichlids earliest and most exactly.  He called the family "a created kind" and wrote that its variation "can occur quite rapidly (since it involves only what was already there)," adding that some cichlid radiations "were indeed dateable (by evolutionists) to within timespans of no more than a few thousand years."²

Mike Wild, writing for Answers in Genesis in 2019, put the model in its current form:  "The problem for evolutionists is that the potential variety already exists within the original pair of species," and "The original cichlids already had the necessary programming to support marvelous variations to fill many niches."²⁸  He cited hybridisation approvingly as the mechanism.  Jeffrey Tomkins of the Institute for Creation Research wrote in 2026 that the standing variation "is pre-engineered and originally placed in creatures at creation by the Creator."²⁹  Jeanson's Replacing Darwin gives the general mechanism:  "Under the creationist model of preexisting genetic differences, massive genetic differences arise in a single generation."²⁵  He works his rate arithmetic on "a theoretical timescale of 12,000 years," which is generous to his opponents.²⁵  His book does not mention cichlids.  No treatment of cichlids by Todd Wood or Jean Lightner was found.

Some creationist claims do not survive a check against their sources.

• Wild says Lake Malawi "may be only 40,000 years old" and that researchers "concluded that the DNA must change more quickly in cichlids" with "a faster mutation rate."²⁸ He cites no source for the lake's age, and the Malawi genome study measured a rate "approximately three to four times lower than in human" per generation.⁹
• Answers in Genesis wrote in 2008 that sensory drive turned an ancestor "that likely could see both types of light" into species "whose senses are limited."³⁰ The two allele classes differ by a few nanometres of peak sensitivity, and heterozygous fish carry both. This is a shift in allele frequency, not a lost sense.
• Tomkins wrote in 2020 that the Malawi transposon mutations were "activated by hybridization" and "brought about by an engineered system of genetic targeting."³¹ The paper he summarised says only that the loci may be "primed for TE insertion" because an independent insertion arose at a similar place in a fish outside the Malawi flock, and that "Increased TE mobility may be associated with interspecific hybridization."²⁷ "May be associated" became "was found to be activated."

The mainstream reply at full strength

The fuel is old, and old variation had to come from somewhere.  Sorting explains the speed only by moving the work earlier.  The parental lineages differ at thousands of sites, and every clock used dates their separation to millions of years:  1.6 to 5.8 million for Victoria's parents, 3 to 4 million for Malawi's.^{1 19}  Elmer and colleagues reject the idea that cichlids simply mutate faster, which "would require an increase in mutation rate by 2 orders of magnitude and such an anomaly is not found in any other cichlids."⁸  The one pedigree rate measured in cichlids, applied to Malawi's genomes, gives lineage splits of 390,000 to 460,000 years.⁹

Even the short clock is too long.  A young-earth model allows about 4,400 years since the Flood, and less for Lake Victoria, whose desiccation must fall after it.  The short clock of 14,600 years is 3.3 times longer.

What a young-earth model has to claim

A young-earth account of the flock has to make two claims, and neither has been published as a worked model.

1. The variation was created, or it arose far faster before or just after the Flood. Tomkins asserts creation. Jeanson's created-heterozygosity model is the general framework. Nobody has shown how a created cichlid kind would carry the specific two-lineage pattern Meier found, with blocks of Upper Nile ancestry about 3,000 bases long in every lake. A test exists. Mutations leave a signature: a characteristic mix of base changes and a pile-up at CpG sites. If the differences between the parental lineages have the same mix as new mutations counted in parents and offspring, they look accumulated. If not, created variation gains support. Nobody has run it for cichlids.
2. Lake Victoria dried and refilled after the Flood. That requires compressing the radiocarbon timescale so that 12,400 carbon-14 years fit into a few thousand calendar years. The site's radiocarbon article deals with that problem in general. No creationist paper applying it to Lake Victoria was found.

The rates on each timeline

The standard measure is Seehausen's net speciation interval, the average time for the number of species to double:  SI = t × ln 2 ÷ ln(N ÷ N₀), where t is the time available, N the species now and N₀ the founders.⁴  The arithmetic is in the note.³²

Seehausen's own table gives Lake Victoria's haplochromines 1,500 to 22,000 years per doubling across the 14,600- and 200,000-year clocks.⁴  For comparison, his table of other young lakes gives pupfish, three species in 3,000 to 5,000 years, a doubling time of 1,900 to 3,200 years;  and Great Lakes ciscoes 6,000 years.⁴  If Apoyo's six species all arose after a colonisation no more than 4,700 years ago, their doubling time is at most about 1,820 years.³²  Every one of these is also inferred.

On the short clock Lake Victoria is already the fastest large radiation known, with a doubling time close to the fastest small cases.  A post-Flood Lake Victoria needs doubling about every 490 years, 3.3 times faster than its own record and four to six times faster than the pupfish, the fastest comparison lake.  At 1.5 to 3 years a generation, 490 years is only about 160 to 330 generations per doubling.  Two founders instead of one helps little, as the table shows.  Many founders would help more, which is why the hybrid swarm and Meier's "several swamp-dwelling refugial populations" matter to the creation model.

How far the mutation route falls short

No creationist proposes that the differences between species arose by mutation since the Flood.  The following number is for scale only.  Malawi species pairs differ at about 2 bases in 1,000.⁹  Making that by mutation in 4,400 years would need a rate of 3.4 to 6.8 × 10⁻⁷ per base per generation, 97 to 195 times the measured rate.³³  Most of that divergence is shared old variation, which is exactly why the creation model depends on created variation and cannot rely on mutation.

Objections and responses

"Five hundred species in 15,000 years proves the creation model."  It shows that species defined by mate choice can be sorted from a mixed stock in thousands of years.  It does not show where the stock came from, and the short clock is still more than three times the post-Flood window.  The case supports one part of the creation model, fast sorting, and leaves the other part, the origin of the variation, untouched.

"The lake never dried out, so the flock is old and the creation argument fails."  The deep basin was dry;  cores from 1996 and 2023 agree on that.^{5 5}  What the 2023 cores add is ponds and wetlands after about 16,700 years ago, in which fish could survive.  That makes the fish lineages potentially older than the lake, but it does not move the ecological radiation, which Meier's 2023 study places inside a lake "less than 16,000 years old."¹⁴

"Hybridisation proves the variation was created in one pair."  Hybridisation shows only that two lineages carrying different variants met and mixed.  The creation model can read those lineages as descendants of one created kind, but it then has to explain why their differences fall into long blocks of ancestry and why each lineage carries its own set, which is what separate histories produce.  That explanation has not been published.

"All the differences are old alleles, so nothing new happened."  Too strong.  Meier's own figure leaves about 150 of 340 strongly differentiated sites with an allele not found in the sampled parents,¹ and Lake Malawi has new mobile-DNA insertions that retune opsin expression.²⁷  New mutations did occur and were used.  None has been shown to create a new coded element.

"Sorting is loss, so the radiation is information loss."  Sorting changes allele frequencies and removes nothing from the flock as a whole.  Each species keeps a subset, which can look like loss at the species level.  The series does not score sorting as a mutation of any kind, and the reversibility seen in turbid water shows that the alleles were not destroyed.

"Radiocarbon dates are unreliable, so the 15,000 years means nothing."  If radiocarbon ages are compressed, they must be compressed for everything, and that claim needs its own model.  Within this article the relative order holds:  dry basin, then wetlands, then a deep lake over a few thousand years.  The same sequence in a young-earth chronology would still leave the flock a few centuries to a couple of thousand years.

"Scientists disagree on the age, so any age is possible."  They disagree between about 15,000 and 200,000 years for the flock and 1.6 to 5.8 million years for its parental lineages.  Every position in that dispute is at least three times longer than a post-Flood window.

"The measured mutation rate is too low to date anything reliably."  It comes from three parent-offspring trios, with a 95% interval of 1.6 to 4.6 × 10⁻⁹.⁹  Even the top of that range leaves Malawi lineage splits hundreds of thousands of years old, and Kautt's use of the same rate gives very young crater-lake ages, so the rate does not simply produce old dates.

Issues and unanswered questions

1. The dry phase is dated from above, and its length is not measured. Johnson 1996 dated plant fragments just above the soil in one core (V95-2P) to 12,400 ± 70 carbon-14 years and traced the soil across the basin by one seismic reflector, tied to five cores. It did not date the dry surface, did not measure how long it lasted, and allowed that the basin may have been dry only "on a seasonal basis." Stager and Johnson 2008 is quoted only through Wienhues 2023.
2. The species count is soft. Counts run from more than 300 to 1,000, and many species are colour forms. No source read here gives a count of genetically distinct populations.
3. Only abstracts were read for Seehausen 1997, Seehausen 2008, McGee 2020, Meier 2023, Carleton 2020, Nakamura 2023 and Goldschmidt 1993. In particular, Carleton's full text would show whether any transposon insertion created a new binding site, which would score as a gain.
4. Nile perch losses. The figure of about 200 species is from a 1993 abstract. Whether some species have since reappeared was not checked.
5. Lake Nabugabo. Eric's research notes quote Ernst Mayr on five endemic species in a lagoon cut off from Lake Victoria by a sand bar dated to "less than 5,000 years," and the same note says "less than 4000 years." Neither Mayr nor Greenwood's original work was read for this article, so Nabugabo is left out of the rate table.
6. The spectrum test that could tell created parental differences from accumulated mutations has not been run by anyone found.
7. Apoyo's own colonisation date sits in a figure and a table that were not readable as text. The article uses only the range across crater lakes.

Verdict

The Lake Victoria cichlids are the best real-world support for the claim a post-Flood model needs:  hundreds of species can be sorted in thousands of years from variation already present, and hybridisation of a few founding populations makes it faster.  The genomic work of the last decade confirms that the flock was built mostly by recombining and sorting old alleles, not by accumulating new mutations, as Arthur Jones said in 2001.  No new coded element was found, and the new mutations that were found, all in Lake Malawi, modify existing opsin and vision genes.

Against the young-earth side, three points stand.  Nothing in the radiation was watched.  The short clock of about 14,600 years, already the fastest large radiation known, is still 3.3 times longer than the post-Flood window, and a post-Flood Lake Victoria needs a doubling of species about every 490 years.  And the variation that did the work is dated by every clock used, including the one measured cichlid mutation rate, to hundreds of thousands to millions of years, while no young-earth author has published a model of its origin or a test that could distinguish created variation from accumulated mutations.  The findings that the creation model of rapid speciation can use are the sorting, the hybrid swarm, the tiny genetic distances and the Nicaraguan crater-lake dates.  It also has to answer the age of the variation.

Sources