Online Resources
Online Resource 1
Variable importance in MARS species distribution models is represented. Some variables had non-normal distributions and were therefore transformed prior to analysis. When the lowest non-zero value for variables were small in magnitude (e.g., agriculture), this value was added as a constant in the transformation. For native and non-native species, we report the δ deviance for each variable averaged over all species and the average rank of each variable.
Native / Non-nativeCategory / Variable / Description / Transformation / δ deviance / rank / δ deviance / rank
Land-use / Agriculture / Proportion of land that is agriculture / log + 0.000004 / 4.1 / 7 / 4.1 / 7
Land-use / Roads / Road density (m/km2) / log + 0.83 / NS / - / NS / -
Climate / Avg precip / Average annual precipitation (1970-2000; mm) / none / 27.9 / 3 / 13.0 / 5
Climate / Avg temp / Average annual temperature (1970-2000; °C) / none / 40.5 / 2 / 38.5 / 2
Climate / CV spring precip / Coefficient of variation for spring precipitation (1970-2000; March – April) / sqrt / NS / - / NS / -
Climate / CV winter precip / Coefficient of variation for winter precipitation (1970-2000; November – February) / sqrt / 18.3 / 5 / 20.8 / 4
Hydrology / TWI / Length weighted mean of the topographic wetness index (range 0 to 1, xeric to mesic) / log + 1 / 10.5 / 6 / 6.8 / 6
Topography / Elevation / Average elevation of stream segment calculated by averaging the sum of the elevation at the upstream and downstream ends (m) / sqrt / 19.6 / 4 / 38.6 / 1
Topography / Gradient / Gradient (m/m) / log + 0.01 / NS / - / NS / -
Topography / Shreve / Shreve link value of segment, measure of stream size / log + 1 / 41.0 / 1 / 29.1 / 3
Topography / TPI1 / Canyons, deeply incised streams / none / NS / - / NS / -
Topography / TPI2 / Midslope drainages, shallow valleys / none / NS / - / NS / -
Topography / TPI4 / U-shaped valleys / none / NS / - / NS / -
Topography / TPI5 / Plains / none / NS / - / NS / -
Topography / TPI6 / Open slopes / none / NS / - / NS / -
1
Online Resource 2
Spatial conservation prioritization approaches typically use species occurrence data to identify high priority areas for management activity. The conservation prioritization program Zonation (version 2.0 and 3.1) permits users to add complexity by incorporating positive or negative interactions between species into the planning process. Here we describe how interactions between native and non-native species were incorporated into our analysis to identify restoration and preservation PAs for native fish conservation planning in the Gila River basin of the Southwestern, USA.
For restoration PAs, a positive interaction was assigned for native species with non-native species prioritizing areas of high overlap of the species distributions. For preservation PAs, a negative interaction was assigned de-emphasizing areas of high overlap between the two species types prioritizing areas with exclusively native species. Zonation determines the interaction between species distributions by calculating the connectivity of species distribution layers. We have outlined below the methodology utilized for applying species interactions into our analysis (also see Moilanen and Kujala 2008).
The terms eij and fik represent the local occurrence of non-native species j and native species k in grid cell i, respectively. Let βk be the parameter modeling the spatial scale of potential stream travel distance for native species k. βk is the parameter of a negative exponential function. We specify the interaction intensity of non-native species j at cell i by native species k is Eij, which is the local non-native species occurrence multiplied by the connectivity of the cell to the native species distribution, Sijk, using parameter βk to model the travel distance of the native species (d in is distance between cells i and n). Thus locations with high Eij have the occurrence of the non-native species well within the travel distance to the native species. Essentially, Eq. (1) represents the positive association of the non-native species to the distribution of the native species.
Restoration Priority Areas (Eq. 1)
Eijk=eijmin1.0,Sijkγjmaxi Sijk =eijmin1.0,n-1Nexp-βkdinfnkγjmaxn-1Nexp-βkdinfnk
The interaction component between native and non-native species was then calculated for the preservation PAs (Eq. 2) by specifying that the discounted value of feature j at cell i is Eij, which is the local occurrence of the focal feature eij discounted by connectivity to the native species to be avoided f, using parameter βk to model the distances to which the undesirable influence spreads.
Preservation Priority Areas (Eq. 2)
Eijk=eijmin1.0,Sijkγjmaxi Sijk =eijmin1.0,n-1Nexp-βkdinfnkγjmaxn-1Nexp-βkdinfnk
Online Resource 3
A map of the (a) Gila River basin within the American Southwest, USA, displaying the major rivers, cities, dams, and basin boundaries (b) the top 10% of preservation and restoration PAs overlaid with the species richness within each catchment (>1 species per catchment minimum to capture community richness) from the original field records, and (c) a section of the mainstem Verde River displaying species richness and the top PAs in the basin.
1