Skip to main content
. 2025 Aug 22;87(9):134. doi: 10.1007/s11538-025-01509-y

Table 1.

Definitions of imbalance indices that are applicable to arbitrary trees, i.e., those with domain Tn (notice that the s^-shape statistic is applicable to arbitrary trees, but is only an imbalance index on BTn). It is straightforward to see that these imbalance indices are induced by the clade size metaconcept ΦfN and the leaf depth metaconcept ΦfΔ, respectively, when the function f is chosen as specified in the two rightmost columns. The Sackin index and the s^-shape statistic are induced by the first-order metaconcept with id as the identity function. In contrast, the average leaf depth is induced by the second-order metaconcept. Moreover, the total cophenetic index is induced on BTn by the second-order and on Tn by the third-order metaconcept. This is because for binary trees we have |V˚(T)|=n-1, so no further additional value than n is needed. For further details on the total cophenetic index, see Remark 3.14

imbalance index definition CSM ΦfN LDM ΦfΔ
Sackin index Sackin (1972); Shao and Sokal (1990); Blum and François (2005); Fischer (2021) S(T):=xVL(T)δT(x) S(T)=vV˚(T)nT(v) id id
Average leaf depth Kirkpatrick and Slatkin (1993) N¯(T):=1nxVL(T)δT(x) N¯(T)=1n·S(T) fN¯(nv,n)=1n·nv fN¯(δ,n)=1n·δ
s^-shape statistic Blum and François (2006) s^(T):=vV˚(T)lognT(v)-1 fs^(nv)=log(nv-1)
Total cophenetic index Mir et al. (2013) Φ(T):=(x,y)VL(T)2xyφT(x,y) Φ(T)=vV˚(T)\ρnT(v)2 fΦ(nv,n,|V˚(T)|)=nv2-n2|V˚(T)|