Table 3.

Results and cline parameters for best fitting model from hzar analyses for each marker type and year

Marker and year	AICc	Centre estimates (km)		Width estimates (km)		Best model (scaling, tails)
		ML	2LL low–high	ML	2LL low–high
Microsatellites
2000	24.42	1609a	1,531–1,675	619abc	411–947	Fixed scaling, neither tail
2002	22.78	1544abc	1,436–1,551	43d	21–401	No scaling, mirror tails
2007	6.53	1356cde	1,252–1,448	694abcd	393–955	Fixed scaling, neither tail
2015	7.92	1244e	1,178–1,310	708bc	562–865	Fixed scaling, neither tail
COI
2000	18.61	1522ab	1,467–1,574	481cd	370–638	Fixed scaling, neither tail
2002	56.38	1491bc	1,438–1,523	395cd	91–606	Fixed scaling, mirror tails
2007	91.43	1404bcd	1,333–1,482	1100a	868–1,387	Fixed scaling, mirror tails
2015	32.04	1230e	1,168–1,298	849ab	705–1,037	Fixed scaling, neither tail
Prediction (2014)	10.38	936f	766–1,047	828abcd	400–1,294	No scaling, right tail only
SNP
Q‐value	53.77	1389d	1,355–1,418	925ab	830–1,052	No scaling, mirror tails
Allele frequency*	—	1374cde	1,272–1,463	1006ab	701–1,442	—

Maximum‐likelihood (ML) cline centre and width are provided with their two log‐likelihood (2LL) low and high support limits. Significant differences (based on 2LL high and low) are denoted by different letters, where overlapping cline centre and width estimates represent clines that are coincident and concordant, respectively. For single nucleotide polymorphisms (SNPs), mean cline parameters are provided based on clines for all 96 SNPs. Models for SNPs and microsatellites were weighted by effective number of alleles (see Section 2).

*Mean of all SNP allele frequency models.