Evidence that Biology is Missing a Causal Layer

The size of the genome often does not correspond to the complexity of the organism¹² (for ex: the axolotl salamander genome is 32 Gb in length³ while the human genome is only 2.9 Gb⁴, and one amoeba species has 100x more DNA than humans⁵).

Rates of genetic change can be high over long periods even when the phenotype doesn't evolve (for ex: in the tuatara⁶⁷⁸, in African freshwater butterfly fish⁹, in horseshoe crabs¹⁰¹¹, in Stygocapitella annelid worms¹², in reef-building coral¹³).

Conversely, morphology can change rapidly despite gene flow between lineages (for ex: in Lake Malawi cichlids¹⁴, in Darwin's finches¹⁵, in threespine sticklebacks¹⁶¹⁷).

Species can remain distinct by means of prezygotic (behavioral) barriers even when sympatric and sexually compatible (for ex: in Heliconius butterflies¹⁸, in wildcats and domestic cats¹⁹²⁰, in wolves and coyotes²¹, in European oaks²², in annual sunflowers²³, in mallard and black ducks²⁴²⁵, in fire-bellied toads²⁶²⁷).

Closely related species can sometimes exhibit phenotype to genotype mismatches, and intermediate morphologies among sympatric species do not always reflect genetic hybridization (for ex: a Belgian study of willows showed that 41% of the individuals had a genotype and morphology both corresponding to S. alba but that 23% had a genotype typical of S. alba, but with a S. fragilis morphology. Similarly, 28% of the individuals had a genotype and morphology corresponding to S. fragilis but 8% had a genotype typical of S. fragilis and a morphology typical of S. alba²⁸²⁹; likewise in white oaks morphology does not always accurately predict genotype³⁰, intermediate types between species are often morphological outliers rather than genetic hybrids³¹³², white oaks have been termed the "worst case scenario" for the biological species concept in part because of these discrepancies).

Organisms can be dependent on two separate genomes, mothers giving birth to two separate species (for ex: in Messor ibericus queens produce not only their own species' offspring, but also produce M. structor males, mate with them, and then use their sperm to create hybrid workers for their colonies³³³⁴³⁵).

Many coenocytic (syncytial) organisms can contain multiple nuclei of different genotypes within a shared cytoplasm (i.e., heterokaryosis / dikaryosis / multinucleate genetic chimerism) (for ex: in ascomycete fungi³⁶³⁷, in oomycete fungi³⁸, in rhodophyta red algae³⁹).

A large proportion of genes are phenotypically silent. In some organisms, the majority of gene knockouts have no phenotypic effect according to known screenings (for ex: in yeast⁴⁰⁴¹, in E. coli⁴²⁴³, in Bacillus subtilis⁴⁴, in Arabidopsis plants⁴⁵).

The number of functionally distinct proteins that could be encoded by the genome is far greater than the number of genes due to alternative splicing, use of alternative transcription start sites, alternative poly-adenylation, RNA editing, and post-translational modification. Genetic sequences do not lead to specific protein structures in metazoans⁴⁶ (for ex: humans have 3–5 more kinds of proteins than genes⁴⁷, while in an extreme case just one Drosophila gene encodes 38,016 distinct mRNA isoforms⁴⁸).

The link between DNA and function is further weakened by the fact that many proteins function despite disordered structure (for ex: disordered proteins may be 35–55% of protein totals in humans⁴⁹, nearly 60% of human proteins are estimated to have at least one intrinsically disordered region⁵⁰).

In some species, the egg cell apart from the nucleus contributes significantly to the traits of offspring (for ex: in cross-genus nuclear transfer experiments — common carp nuclei into enucleated goldfish eggs — the resulting fish had a carp nuclear genome, but vertebral counts were consistent with the egg (goldfish) cytoplasm⁵¹⁵²⁵³).

Cell shape and mechanical constraints can act upstream of transcription; forcing a cell into a particular geometry can push it into a matching gene-expression state, even if its prior expression pattern would ordinarily be taken as "determinative" of shape (for ex: in NIH 3T3 fibroblast cells⁵⁴, in mesenchymal stem cells⁵⁵, in human and bovine capillary endothelial cells⁵⁶).

Implanting tissue/primordia in an unusual location can result in the grafted material (or the graft–host system) producing a different anatomical structure than it would in its native context (for ex: fragments of a wing imaginal disc transplanted and cultured in vivo can later produce non-wing structures in Drosophila⁵⁷, newt iris epithelial cells implanted into a limb blastema can undergo lens formation⁵⁸).

Though for some monogenic diseases genetics are highly determinative, in general the practice of utilizing the DNA of an individual to predict disease has been judged to add little or no useful information⁵⁹ (for ex: in a hospital-cohort evaluation computing polygenic scores across 457 phenotypes, only 49 phenotypes had AUC > 0.6⁶⁰; a recent clinical appraisal concluded that PRS-alone diagnostic/prognostic performance is "consistently low," and that adding PRS to clinical scores typically yields at best moderate improvement⁶¹).

There is no such thing as selection "pressure." Removing lineages through selection does not directly cause adaptive responses in the lineages that remain, though it may increase the availability of resources. Karl Popper offered a thought experiment demonstrating this: imagine an expanding world of infinite resources where there was no extinction; all lineages that ever occurred were still living. There would be no removal of variants, although there would still be differential rates of reproduction. In this world without a filtering process, evolution would still proceed, including the invention of all the adaptations that are often explained by reference to natural selection⁶².

In sexual eukaryotes, all specific DNA sequences are recombined during the crossing-over phase of meiosis with homologous (but distinct) pairs, thus in these taxa (including humans) no specific gene sequence comprises a unit of Darwinian selection or can rightly be called "selfish"⁶³. Protein domains behave as modular evolutionary units, and domain gain, loss, and rearrangement are powerful mechanisms for generating new functions⁶⁴⁶⁵. In addition, while plasmids and viruses may encode some replicative function, 100% of chromosomal genes depend on the products of other genes for their replication⁶⁶. Thus, there is no particular level at which DNA could comprise self-replicating entities. This argument does not by itself prove that life isn't based on replication (an idea that is disproved elsewhere in this paper); it only shows that it is not based on self-replicators, which do not occur in nature. This distinction is important because for example, artificial life programs based on self-replicators are offered up as arguments for demonstrating the origin of phenomena in organic life, and we argue that this claim is specious.

Genetic assimilation. Selection can drive canalization of an environmentally induced phenotype, where a trait first produced by stress becomes stabilized and expressed under ordinary conditions (for ex: in fruit flies⁶⁷⁶⁸, in spadefoot toads⁶⁹, in nematodes⁷⁰, in plants⁷¹). Genetic assimilation has been traditionally explained in a non-Lamarckian fashion, attributing the entire source of the variation under selection to genetics plus randomness. We dispute this interpretation of the original source of variation⁷², but nevertheless, the concept shows that causation responsible for a given trait can be shifted from environmental determinants of plasticity to the genome. Nonheritable, induced traits can become heritable over time by means of gene mutations and selection. Thus, instances of the genetic encoding of traits are not proof that the traits were randomly acquired.

Adaptive plasticity can be induced that, if found in the fossil record, might be interpreted as gene-based evolutionary novelty (for ex: a goat born without its front legs developed a pelvis typical of bipeds⁷³⁷⁴, exercise can induce increased bone mass in mice⁷⁵⁷⁶, diet regime can induce substantial craniofacial differences in spadefoot toad tadpoles⁷⁷⁷⁸, and differences in adult horn beetle horn size/morphology⁷⁹, predators can induce "helmets" and spines in Daphnia cucullata⁸⁰, shell thickening/shape change in marine snails⁸¹⁸², and deeper-bodied forms in crucian carp⁸³; while temperature can induce musculoskeletal changes in zebrafish⁸⁴).

DNA does not self-replicate with fidelity, but rather depends on cellular machinery to increase error correction approximately 10,000-fold through a combination of proofreading and mismatch repair (for ex: in humans⁸⁵, in E. coli⁸⁶).

At least some biological cells can control and target gene point mutations adaptively⁸⁷⁸⁸ (for ex: activated B lymphocytes intentionally deploy a programmed, locus-targeted mutagenesis system to diversify antibodies during somatic hypermutation⁸⁹⁹⁰).

The properties of life do not travel with DNA but rather with cells (for ex: viruses can't replicate without hosts⁹¹⁹², but enucleated red blood cells can live for ~120 days with no genetic material^93949596).

The study of the origin of life has not yet offered a plausible model of abiogenesis. High accuracy self-replication appears to require catalysts and processes that are too complex to have arisen by chance^979899100.

Directed evolution of novel organisms works much better than rational design gene-by-gene^101102103104.

Evolutionary transitions in individuality transform the self to which natural selection applies. In order to fit this into a selfish-gene framework, mechanisms have to be identified that explain assortment, conflict mediation, inheritance at the new level, and enforcement/policing¹⁰⁵.

Darwin's theory in On The Origin of Species, widely understood as the intellectual foundation of modern gene-centered evolutionary thought, was in fact a Lamarckian theory, relying on use and disuse as a source of heritable variation¹⁰⁶¹⁰⁷. Darwin spent the latter years of his life investigating his theory of somatic inheritance through "gemmules," a physiological system he conjectured that was analogous to what has now been identified as exosomal vesicles¹⁰⁸.

Stress can activate regulated "variation-generator" pathways like transposition^109110111112. If variation is induced by stress, it cannot be said to be random with respect to function, even if it may result in unpredictable new genetic sequences. DNA transposons and retrotransposons (e.g., LINEs, SINEs, LTR elements) are major structural components of many eukaryotic genomes¹¹³ (for ex: two-thirds of the human genome¹¹⁴, and around 85% of the maize genome¹¹⁵ and other large-genome angiosperms¹¹⁶).

Stress responses can upregulate or reprogram homologous recombination or recombinational DNA repair in ways that increase genetic variation¹¹⁷ (for ex: in E. coli¹¹⁸, and in fission yeast¹¹⁹).

Stress can induce whole chromosome rearrangements including double-strand breaks (DSBs), translocations, deletions, and complex reassemblies such as chromothripsis^120121122. Chromosomal rearrangements are thought to be involved in the reproductive isolation leading to speciation in sexual eukaryotic lineages, but such a change requires at least one compatible partner to succeed^123124125.

Prokaryotes exhibit abilities to both read from and write to the genome¹²⁶ (for ex: CRISPR spacer acquisition, which integrates fragments of foreign DNA into a host as a heritable genomic memory¹²⁷¹²⁸).

A species may exhibit two different phenotypes maintained by means of heritable epimutations¹²⁹ (for ex: in Linaria vulgaris toadflax¹³⁰, in tomato¹³¹¹³²).

Acquired traits can be inherited through transgenerational epigenetic inheritance¹³³ (for ex: in C. elegans^134135136 including an example of long-term silencing by RNAi of at least 80 generations¹³⁷, in mice via paramutation¹³⁸¹³⁹, in Arabidopsis¹⁴⁰, in maize¹⁴¹, in wheat¹⁴², in yeast via protein-based inheritance¹⁴³, via histone modifications¹⁴⁴ and as phenotype switching¹⁴⁵, in water fleas¹⁴⁶, in Drosophila via paramutation for at least 50 generations¹⁴⁷, in moths¹⁴⁸, in foxes¹⁴⁹, in Ascobolus via methylation induced premeiotically¹⁵⁰, in Coprinus¹⁵¹, in Podospora fungus via prions¹⁵², in Oxytricha ciliates via an RNA-mediated genome rearrangement pathway¹⁵³). In their famous 2009 review, Jablonka and Raz write that epigenetic inheritance "may be ubiquitous" and that "epigenetic variants in every locus in the eukaryotic genome can be inherited, but in what manner, for how long, and under what conditions has yet to be qualified"¹⁵⁴.

Organisms can also develop habits or behavioral patterns during their lifetimes which are then transmitted through the germline to offspring that never experienced the original stimulus (for ex: pathogen avoidance in C. elegans¹⁵⁵¹⁵⁶, olfactory fear conditioning in mice¹⁵⁷¹⁵⁸, feeding behavior after ancestral high-sugar diet in Drosophila¹⁵⁹, neurobehavioral alteration from ancestral methylmercury exposure in zebrafish¹⁶⁰, acquired resistance to downy mildew in Arabidopsis plants¹⁶¹).

In mammals, there is strong evidence that the reproductive tract uses specialized extracellular vesicles (EVs) to deliver developmentally important regulatory RNAs to gametes¹⁶² (for ex: in mice¹⁶³¹⁶⁴, in bovids¹⁶⁵¹⁶⁶, in humans¹⁶⁷¹⁶⁸).

Some organisms have shown transgenerational immune priming, where parental pathogen exposure can enhance offspring resistance without genetic change (for ex: antiviral immunity was inherited in Drosophila and Aedes aegypti¹⁶⁹, paternal and maternal priming was inherited in red flour beetle¹⁷⁰, resistance to Pasteuria ramosa was inherited in Daphnia¹⁷¹, and maternal immune challenge primed worker offspring in bumblebees¹⁷²).

Heritable traits can sometimes be transmitted by means of blood transfusion (for ex: in pipefish, paternal exposure to heat-killed bacteria altered the immune phenotype of offspring through F2, transmitted through paternal blood-borne factors that reach developing embryos¹⁷³; in butterflies, larvae injected with haemolymph from odour-fed donors stopped avoiding the novel odour, and their naive offspring preferred the odour more compared to the offspring of larvae injected with control haemolymph¹⁷⁴).

In certain cases, the phenotype of a female's previous mate can influence her future offspring sired by another male, even without epigenetic effects (for ex: in Telostylinus angusticollis flies, adult body size of offspring was influenced positively by the condition of females' initial mate. Offspring body size was also positively related to maternal body size, but not related to the body size of the second male, which was the putative sire¹⁷⁵).

There are some taxa in complex multicellular lineages (metazoans, plants, fungi) with highly differentiated body plans that do not segregate germline cells from information flow early in ontogeny — so early germline sequestration (a Weissmann barrier) is not essential for complexity¹⁷⁶¹⁷⁷ (for ex: in cnidarians¹⁷⁸, in flatworms¹⁷⁹, in annelids¹⁸⁰, in echinoderms¹⁸¹, in plants¹⁸², in multicellular fungi¹⁸³). Germline cell fates are specified in some lineages by inherited factors present in the maternal oocyte (fruit fly), and in others by cell-cell communication during development (mouse), suggesting that degree of information flow from soma to germline is a gradient under evolutionary selection and not a firm requirement of a complex ontogeny.

Some organisms with early segregation of gametes can nevertheless regenerate gonads from somatic cells if these are removed (for ex: in ascidians¹⁸⁴¹⁸⁵). This indicates that the point in development when gamete segregation occurs is not necessarily the latest time when it could occur given different developmental conditions.

Unicellular life has no germline segregation, therefore all inheritance in these taxa is Lamarckian by definition. Unicellular organisms are the overwhelming majority of individuals on earth by number¹⁸⁶.

In metazoans that reproduce by fission, heritable novel target morphologies can be induced by bioelectricity (for ex: two-headedness can be induced and stably inherited in planaria across rounds of regeneration by transient inhibition of gap junctional communication¹⁸⁷).

Unrelated disparate lineages can fuse in symbiogenesis¹⁸⁸ (for ex: mitochondria originate in integration of an alphaproteobacterial endosymbiont into the lineage leading to all modern eukaryotes¹⁸⁹, chloroplasts originate from incorporation of a cyanobacterium into the common ancestor of glaucophytes, red algae, and green algae/plants¹⁹⁰¹⁹¹, a nitrogen-fixing organelle originates from a cyanobacterium-derived endosymbiont inside a marine haptophyte alga¹⁹²¹⁹³).

Genetically distinct organisms can fuse, becoming mosaics before reproduction (for ex: in colonial ascidians Botryllus schlosseri neighboring colonies can undergo vascular fusion producing a single colony-level organism composed of multiple genotypes¹⁹⁴, in social amoebae Dictyostelium discoideum starvation triggers aggregation to form chimeric slugs with mixed genotype fruiting bodies prior to spore formation¹⁹⁵¹⁹⁶).

Prokaryotes often trade their DNA across distantly related taxa in horizontal gene transfer (HGT). HGT rates are comparable to the rate of point mutations and much greater than the duplication rate¹⁹⁷¹⁹⁸. Thus, in bacteria and archaea, the dominant feature of genome evolution is not gene duplication but rather evolution by extensive gene loss and gene gain via HGT¹⁹⁹.