Table 1. Parameters of models implemented.

Model	Parameter combinations	Comment
a	= 3,000, τ = 1	Original EW model
		Strong selection
		Absolute trait effect and fitness effect linearly correlated
b	= 30, τ = 1	Original EW model
		Weak selection
		Absolute trait effect and fitness effect linearly correlated
c	= 3,000, τ = 0	Original EW model
		Strong selection
		Absolute trait effect and fitness effect uncorrelated
d	= 3,000, τ = 0.5	Original EW model
		Strong selection
		Absolute trait effect and fitness effect weakly correlated
e	= 3,000, τ = 2	Original EW model
		Strong selection
		Quadratic relationship between absolute trait effect and fitness effect
f	= 3,000,A = 10,000,B = 10,000	Saturating function
		Strong selection
		Absolute trait effect threshold of 10,000 before noise term (1 + ε)
		Linear relationship between trait and fitness near origin with gradient 1 before noise term δ(1 + ε)
g	= 3,000,A = 4,000,B = 40	Saturating function
		Strong selection
		Absolute trait effect threshold of 4,000 before noise term (1 + ε)
		Linear relationship between trait and fitness near origin with gradient 100 before noise term δ(1 + ε)
h	= 3,000,A = 4,000,B = 4,000	Saturating function
		Strong selection
		Absolute trait effect threshold of 4,000 before noise term (1 + ε)
		Linear relationship between trait and fitness near origin with gradient 1 before noise term δ(1 + ε)