Appendix 2 Assessing if host plant and nectar plant species compositions determine butterfly species compositions

As butterflies generally have a narrow host plant diet breadth, host plant species composition may be important to explain observed species compositions instead of host plant species density per se. We used co-correspondence analysis (CoCA) (ter Braak and Schaffers 2004) to test if tree host plant species composition can predict butterfly species compositions. The butterfly species compositions were only log-transformed before CoCA to improve normality (Legendre and Legendre 2012). Since butterflies can be oligophagous, sites with different plant species can have the same butterfly species compositions and CoCA may fail to reveal this relationship. Hence, before we conducted CoCA, each host plant species was converted to butterfly species that use the plants according to the literature (Bascombe et al. 1999, Young et al. 2011, Young et al. 2007a, Young et al. 2007b, Young et al. 2008). We used cross-validatory fits to test if the host plant species compositions can predict butterfly species compositions better than random (ter Braak and Schaffers 2004). We did the same for shrub host plant and shrub flower species composition. Because the nectar diets of Hong Kong butterflies were not well-studied, we used the flower species compositions instead of butterfly species that use the flowers.The effect of flowering tree was not tested since not all sites contained flowering tree and those sites must be dropped from the analysis. CoCA was conducted using R package “cocorresp” (Simpson 2014). We found that all host plant and nectar plant species compositions do not explain butterfly species compositions better than random, as cross-validatory fits give a negative value for all axes or permutation test indicates no axes are significant(ter Braak and Schaffers 2004).

References

Bascombe MJ, Johnston G, Bascome FS (1999) The Butterflies of Hong Kong. Academic Press, Cambridge.

Legendre P, Legendre LF (2012) Numerical ecology. Elsevier

Simpson GL (2014) cocorresp: Co-correspondence Analysis Methods.

ter Braak CJF, Schaffers AP (2004) Co-correspondence analysis : A new ordination method to relate two community compositions. Ecology 85:834-846

Young JJ, Law YS, Choi KP et al (2011) A Photographic Monograph on Hong Kong Butterflies Volume 4. Hong Kong Lepidopterists' Society Limited, Hong Kong

Young JJ, Yiu V, Yau SM (2007) A Photographic Monograph on Hong Kong Butterflies Volume 1. Hong Kong Lepidopterists' Society Limited, Hong Kong

Young JJ, Yiu V, Yau SM, Choi KP, Au WK, Sha WCS (2007) A Photographic Monograph on Hong Kong Butterflies Volume 2. Hong Kong Lepidopterists' Society Limited, Hong Kong.

Young JJ, Yiu V, Yau SM et al (2008) A Photographic Monograph on Hong Kong Butterflies Volume 3. Hong Kong Lepidopterists' Society Limited, Hong Kong