SIMPER Dissimilarity Contributions

SIMPER (similarity percentages) decomposes the average Bray-Curtis dissimilarity between two groups into contributions from each variable, ranking the variables that account for most of the difference.

SIMPER는 두 집단 사이 평균 Bray-Curtis 비유사도를 변수별 기여도로 분해하여 차이를 가장 많이 설명하는 변수를 순서대로 보여 주는 기법입니다.

Which species contribute most to the difference in composition between my two groups of sites?

두 지점 집단의 종 구성 차이에 가장 크게 기여하는 종은 무엇인가요?

SIMPER averages, over all pairs of samples from two groups, each species' share of the Bray-Curtis dissimilarity and lists species by their percentage and cumulative contribution. Because contributions mix differences in means with high variability, the results are descriptive and are best read alongside the ratio of average contribution to its standard deviation.

When to use SIMPER Dissimilarity Contributions

• A PERMANOVA or ANOSIM shows groups differ and you want to describe which variables drive the difference
• Data are non-negative abundances suited to Bray-Curtis
• A ranked list of contributing species is needed for a report

When to use something else

• You need formal tests of which species differ: use multivariate GLMs or per-species tests with multiplicity control
• You want an overall test of group difference: use PERMANOVA

Assumptions and requirements

• Non-negative abundances with no empty samples
• Contributions confound mean differences and variance, so they are descriptive

What the analysis reports

• Average Bray-Curtis dissimilarity for each pair of groups
• Average contribution, standard deviation and ratio per species
• Percentage and cumulative contribution, group means, optional permutation p-values
• Bar chart of top contributions

How to report it

SIMPER showed an average Bray-Curtis dissimilarity of [value] between [group A] and [group B]; [species] contributed most ([value]% of the dissimilarity).

Related methods

• ANOSIM
• PERMANOVA
• Indicator Species Analysis

Run SIMPER Dissimilarity Contributions on Quorant