Table 1 Enabling conditions variables analyzed and their relationships with IWS

From: Global state and potential scope of investments in watershed services for large cities

Count

Predictor Variable (using available data to represent enabling conditions)

Global Cities

Non-USA Cities

Enabling Condition (based on theory)

Enabling Condition Category

1

Average Annual Growth

~

+

Economic growth

Economic

2

Average Distance

 

Resource location and arrangement

Biophysical

    

Proximity of actors to each other

Sociocultural

3

Average Elevation

Resource location and arrangement

Biophysical

4

Average Governance Indicators

 

~

Strong existing institutions

Governance

5

City Population

Large/small number of actors

Sociocultural

6

Conservation Spending

Strong capacity among actors

Governance

7

Enforcing Contracts Indicator

~

~

Manageable transaction costs

Economic

    

Pre-existing market-based culture

Sociocultural

8

IUCN Organizations Per Million People

 

+

Presence/absence of Intermediaries

Governance

    

Strong capacity among actors

Governance

    

Influential champion

Governance

9

National GDP per capita

 

+

Economic growth

Economic

10

Percent Agriculture Cover

+

+

Clear threat or risk to ES provision

Biophysical

11

Percent Forest Cover

 

Clear threat or risk to ES provision

Biophysical

12

Percent Protected Area

Secure land tenure and property type

Governance

13

Registering Property

+

+

Secure land tenure and property type

Governance

    

Pre-existing market-based culture

Sociocultural

14

Total Diversion Volume

  

Small resource area

Biophysical

15

Watershed Area

+

 

Small resource area

Biophysical

16

Watershed Population Density

  

Large/small number of actors

Sociocultural

17

Weighted Drought Vulnerability Index

 

Resource location and arrangement

Biophysical

  1. Variables are based on the biophysical, economic, governance and social-cultural enabling conditions and groups identified by Huber Stearns et al.21 (Fig. 1). Not all variables identified by Huber-Stearns et al. were included in the analysis, because of unavailable or limited data (Supplementary Data 1). For each random forest model, important enabling conditions are provided a value (+/-/~) and unimportant conditions are left blank. Signs indicate the direction of the relationship between each condition and presence of IWS (Supplementary Fig. 1). Some important conditions have relationships that are neither positive nor negative overall, but vary in direction dependent on the underlying data gradients. These relationships are signified by ~ Some predictor variables were representative of multiple enabling conditions. In these cases, all potential representation are included in the Enabling Condition column