Table 1

Summary of statistical method and result for each comparison. (Habitats are abbreviated according to their relative elevation; high refers to cloud forest (2100 m), mid- to lower montane rainforest (1000 m) and low to lowland rainforest (400 m).)

speciestreatments comparedcomparisonapproach (one-sided unless stated)no. observations per habitat high, mid, low/closed, opentest statisticmodeldirection

test statisticp-value
survival results
A. baezahigh, mid and lownest extinctionlogrank (Mantel-Cox) testH: 14, M: 18, L: 15Χ241.37<0.0001as predicted: high>mid>low
A. baezahigh and midproportion of spiders remainingHLM (mixed)H: 12, M: 9F11.740.007as predicted: high>mid
dispersal results
A. baezahigh and midnumber of propagulesHLM (mixed)H: 18, M: 17F7.960.009as predicted: high>mid
A. baezahigh and midpropagule survivallogrank (Mantel–Cox) testH: 47, M: 18Χ22.880.045as predicted: high>mid
rain exclosure
A. baezalow, exposed versus coveredproportion of spiders remainingHLM (GLIMMIX)C: 22, O: 22F3.680.032as predicted: covered>exposed
A. baezalow, exposed versus coverednumber of web strands on plantHLM (mixed)C: 17, O: 19F5.580.05as predicted: covered>exposed
A. eximiuslow, exposed versus coveredproportion of spiders remainingWilcoxon testC: 11, O: 10Χ20.240.31as predicted: covered>exposed
A. eximiuslow, exposed versus coverednest volumet-testC: 11, O: 11T1.850.04as predicted: covered>exposed
predation
A. baezahigh, mid and lowpresence/absence of antstwo-sided nested modelH: 18, M: 17, L: 50Χ2260.038as predicted: high<mid<low
A. baezahigh, mid and lowpresence/absence of spiderstwo-sided nested modelH: 18, M: 17, L: 50Χ242.90.0002opposite of predicted: high>mid>low