The phrase "attribute pressure" is correct and usable in written English.
It can be used in contexts where one is discussing the assignment of responsibility or influence related to pressure in various situations, such as in psychology, business, or social dynamics.
Example: "In the study, researchers sought to attribute pressure to external factors that influenced the participants' performance."
Alternatives: "assign pressure" or "ascribe pressure".
First, it implicitly assumes that the qualitative form of an attribute-pressure relationship is stationary.
The goal of this analysis is to distinguish linear (Fig. 1a) from nonlinear (Figs. 1b-d) attribute-pressure relationships.
Thus, for a nonlinear attribute-pressure relationship the potential gains toward achieving EBM goals by reducing a pressure are greatest where the curve is steepest.
Empirical understanding of attribute-pressure relationships based on spatial correlations, time series, and experiments (e.g., [5], [8], [22]), are valuable sources of this information.
There are many nonlinear models besides the piecewise and sigmoidal functions we considered that might be used to determine the general shape of the attribute-pressure relationship [59].
The third step in this analysis is to identify the utility threshold in any attribute-pressure relationship judged to be nonlinear.
The shape of the attribute-pressure relationship and the ease with which an objective threshold point can be defined should drive the choice of which mathematical functions to consider in the model selection analysis.
The fourth step in the utility threshold analysis is to translate the threshold pressure in any nonlinear attribute-pressure relationship to values of easily-measured indicators representative of the status or trend in the ecosystem attribute.
Similar to the attribute-pressure relationships, we compared the relative fits of linear, piecewise, exponential, and parabolic models (see Text S1 for details) to the relationships between the indicators and pressures generated from each Monte Carlo data set using AICc.
To facilitate comparison of attribute-pressure relationships, we re-scaled the attributes (so that small values indicated a stressed condition and large values indicated an unstressed condition) and standardized them to zero mean and unit variance.
We compared the relative fits of a linear, piecewise, and sigmoidal model to the attribute-pressure relationships generated from each Monte Carlo data set using Akaike's information criterion corrected for small samples (AICc; [38]).
