Birth-Death Parameters
Display Options
Birth-Death Phylogenetic Trees
The birth-death process is a fundamental model in phylogenetics that generates random evolutionary trees. Species undergo speciation (birth) at rate λ and extinction (death) at rate μ. The resulting tree shape reveals diversification dynamics through time.
Key Concepts
Net diversification rate: r = λ - μ
When r > 0, lineages increase exponentially on average. When r ≤ 0, eventual extinction is certain.
Turnover ratio: ε = μ/λ
High turnover (ε → 1) means many species go extinct, creating asymmetric trees.
Low turnover (ε → 0) is a pure-birth Yule process with balanced, symmetric trees.
Tree balance:
The shape of phylogenetic trees carries information about diversification.
Symmetric trees (balanced) suggest constant-rate birth with little extinction.
Asymmetric trees (unbalanced) suggest heterogeneous rates or high extinction.
Gillespie Algorithm
We simulate the birth-death process using the Gillespie algorithm for continuous-time stochastic processes:
- Start with 1 lineage at time t = 0
- With N active lineages, the total event rate is: λN + μN
- Time to next event: Δt = -ln(U) / (λN + μN), where U ~ Uniform(0,1)
- Event type: birth with probability λN/(λN + μN), otherwise death
- If birth: split a random lineage into two daughter lineages
- If death: remove a random lineage
- Continue until t ≥ T or all lineages extinct
Tree Statistics
Gamma (γ) statistic (Pybus & Harvey 2000):
Tests whether branching times follow a constant-rate birth-death model.
γ < 0 suggests slowdown (early rapid diversification),
γ > 0 suggests speedup (recent rapid diversification).
Under constant rates, γ follows a standard normal distribution.
Lineages-Through-Time (LTT) plot:
The number of lineages N(t) vs time on a log scale.
Under pure birth (Yule), this is linear with slope λ.
Deviations reveal changes in diversification rate.
Model Variants
Yule process (μ = 0): Pure birth with exponential growth. Produces balanced trees with E[N(t)] = e^(λt).
Constant-rate birth-death: Both λ and μ constant. More realistic, allows extinction. Produces more asymmetric trees as μ/λ increases.
Trees are ultrametric: all tips (extant species) align at the present time, with branch lengths proportional to elapsed time.
Applications
- Inferring speciation and extinction rates from molecular phylogenies
- Testing hypotheses about adaptive radiation and key innovations
- Understanding macroevolutionary dynamics and diversity patterns
- Detecting shifts in diversification rates across clades