Treatments
We used a fully factorial experiment to investigate the interactive effects of corridor quality with drought severity and the number of drought patches across landscape on the persistence of metapopulations. In the good corridor quality treatment, corridors had 2ml of water added each week to keep them moist (Li et al. 2021), whilst in the poor corridor quality treatment no water was added to the corridors. For each corridor quality treatment (i.e., good or poor), we manipulated drought in two different ways: changing the rate at which the moisture content of patches declined (i.e., drought severity), and the number of patches simultaneously experiencing drought (i.e., number of drought patches). Severe drought was created by reducing the 2ml of water added to each patch by 0.4ml water per week, whilst weak drought was simulated by reducing the 2ml by 0.2ml water per week (Fig. 1b). In addition, a number of patches were also randomly selected to receive no water at all in any given week, and the number impact in this was increased through time (the number of drought patches treatment). The number of patches which did not receive any water was generated by drawing a number (from 0 to 4) from a binomial distribution with the probability increasing from 0 to 1 at either 0.2 per week (i.e., fast increase) or 0.1 per week (i.e., slow increase) (Fig. 1b). This was done for each replicate landscape, and the location of patches impacted by this was randomized for each landscape. We also implemented a constant (control) treatment for each corridor quality group (i.e., good or poor) where 2ml water were constantly added to all the patches each week. This gave us in total two levels of corridor quality (good vs. poor), two levels of drought severity (severe vs. weak), and two levels of increase in number of drought patches (fast vs. slow), and one control group (no change in water through time), with each treatment combination replicated 6 times for a total of 60 replicates.