Table 2. General linear model (type 3) showing the effects of hemisphere, habitat type^a, and temperature on ant species richness^b (Akaike weight: 0.656241^c).

Effect	Num df	SSQ	MSQ	Den df	F	P
Intercept	1	1.041	1.041	319	2.06	0.0403
Hemisphere	1	13.715	13.715	319	34.52	<0.0001
Habitat type	7	30.890	4.413	319	11.11	<0.0001
Temperature	1	16.409	16.409	319	41.30	<0.0001
Temperature2	1	8.316	8.316	319	20.93	<0.0001
Residual	319	126.758	0.397

^aHabitat types: arid shrubland, other shrubland (including a variety of shrublands, scrublands and thickets and other bush dominated habitat types), desert, forest, grassland, savanna, wetland, woodland), and temperature on ant species richness.

^bSpecies richness was ln(x+1)-transformed for analysis.

^cA global dataset on ant species richness derived from the literature that excludes sites from the FB and SKB is used. The model shown is the best model based on Akaike weights from a set of all possible models testing the above factors and also the effects of linear and squared terms of latitude (absolute value rounded to the nearest degree) and NDVI (productive energy).