Extinction risk, small populations and genetic consequences

Extinction risk has a boundary, horizon and conditions

Global species extinction means that no living member of the species remains. Local population loss, often called extirpation, means disappearance from a specified area while the species can remain elsewhere. Loss of an allele is another event: a genetic variant disappears from the included copy pool. A declining local count cannot automatically be renamed a global extinction or an allele-frequency measurement.

Risk concerns a possible event under defined conditions and over a stated horizon. Two populations with the same abundance can differ in reproductive contribution, isolation, trends, habitat and exposure to shared threats. Natural rarity and rapid decline are also distinct: a historically restricted population may face different conditions from a once-abundant population undergoing sustained losses. Neither description removes the need to assess its actual pressures.

An extinction-risk assessment summarises evidence such as population trend, distribution and threats. IUCN’s threatened categories include Vulnerable, Endangered and Critically Endangered. They are not statements that extinction is certain or a complete ordering of every conservation decision. Data Deficient means that available evidence does not resolve an appropriate risk category; it does not establish that the species is safe. No current species assignment is made here.

A mean is not a persistence guarantee

Demographic stochasticity is variation arising from individual survival and reproductive events in a finite population. In a small account, a few different individual outcomes can change the whole total substantially. The independent-survival example makes that idea visible without fitting a real population model. Every member has the same survival probability 0.5, but a zero-survivor outcome is possible.

For two starting members there are four equally likely fate assignments: neither, first only, second only or both survive. Their survivor counts 0, 1, 1, 2 have mean 1. For four starting members there are sixteen assignments and mean 2; the all-loss assignment has probability 1/16 instead of 1/4. These are one-interval local accounts with no reproduction, arrival or departure, not multi-year population viability estimates.

Environmental variation and major disturbances can affect many members together, so independent events are often an incomplete model. Larger populations can remain vulnerable to shared habitat losses or sustained negative demographic contributions. Persistence estimates therefore depend on events, dependence, time horizon and uncertainty as well as size. A universal “safe number” would hide these conditions and is not supplied by this exercise.

Copy sampling, inbreeding and adaptive potential differ

Random genetic drift changes allele frequencies through finite contributions from one generation to the next. In the neutral four-copy example, each next-generation copy is an independent replacement draw from an equal A/a pool. Some outcomes lose A and some lose a; they cancel in the mean A frequency. Variation within the resulting pools nonetheless falls on average. A neutral loss does not show that the lost allele was harmful.

Inbreeding concerns reproduction among related individuals and increased probability of alleles being identical by descent. It can increase homozygosity and expose effects of recessive deleterious variants; it does not create those variants simply by pairing relatives. Its consequences depend on genetic history and context. Observed heterozygosity, allele richness, relatedness and fitness are therefore connected questions rather than interchangeable measurements.

The genetically effective population size compares a real population’s drift or inbreeding behaviour with a specified idealised model. Unequal contributions and other departures from ideal assumptions can make it differ from a census count. Gene flow can add variation and has benefited selected inbred populations, but adaptation, ecological compatibility and possible adverse consequences need assessment. The possibility of genetic rescue is not a guarantee of demographic recovery, and these concepts supply no release or breeding instructions.

Worked example

Check your understanding

1. Why does an expected surviving count of one not establish that the two-member population will retain a member?

   Show reasoning

   The expectation averages four different outcomes with survivor counts 0, 1, 1, 2. The zero outcome has probability 0.25 under the stipulated independence conditions. An average can be positive even when complete loss is possible; a shared-event model would require a different account.

2. How can mean A frequency stay 0.5 while genetic variation falls in the neutral model?

   Show reasoning

   A-rich and a-rich outcomes balance in the mean frequency, but each shifted pool has less balanced variation, and single-allele endpoints have none. Averaging their different-copy probabilities gives 0.375 rather than 0.5. Mean allele frequency does not measure retained diversity within each outcome.

3. Does a small population with low variation necessarily require the same conservation action as every other small population?

   Show reasoning

   No. Threats, demographic trends, genetic history, habitat, potential gene-flow consequences and the decision’s ecological and social conditions differ. Census size and one genetic measure do not supply a universal safe threshold, a guaranteed fate or an appropriate practical action.

What to carry forward

Separate the event, boundary and horizon: population loss, allele loss, expected abundance and extinction-risk assessment are different accounts.