Aarhus University Seal / Aarhus Universitets segl

No. 151: Structural equation modeling for decision-making within nature management

Lykke, I.M.Ø., Strandberg, M., Nielsen, K.E., Barfod, A. & Damgaard, C. 2015. Strukturel ligningsmodelering som beslutningsgrundlag indenfor naturforvaltningen. Et eksempel fra pleje af klokkelyng på våde heder. Aarhus Universitet, DCE – Nationalt Center for Miljø og Energi, 100 s. - Videnskabelig rapport fra DCE - Nationalt Center for Miljø og Energi nr. 151. http://dce2.au.dk/pub/SR151.pdf


We here suggest the usage of structural equation modeling as a tool to be applied in nature management decision making. This usage of structural equation models is demonstrated on cross-leaved heath (Erica tetralix) in Danish wet heaths. In accordance with the Habitats Directive issued by the European Commission, the nationwide Danish Monitoring Program (NOVANA) has been used to monitor the condition of vulnerable Danish habitats comprised in the Natura 2000 network. Danish wet heaths that constitute a main habitat of Erica tetralix are in decline. We used structural equation modeling on NOVANA data to examine the determinants of the E. tetralix cover. Traditional vegetation models reflect how abiotic and biotic variables influence the cover of plant species. These variables are set in the model to influence the cover either directly or indirectly through an alteration of the growth conditions of competing species. This kind of model has for example been used for comparison of the drivers E. tetralix cover on heathlands and dune heaths, respectively. Our results showed that moisture and soil type were the most important variables determining E. tetralix cover on wet heaths, which is in agreement with the accepted theory. The variables grazing and nitrogen deposition, which affected the cover indirectly by promoting competitive species, had a weaker effect on E. tetralix cover.  Some of the results obtained contradict some of the generally accepted theories. This applies for example to acidification, which had no or limited influence.  This result could be explained by the combined effect of base cation and nitrogen deposition, which may influence competitive species negatively. On dune heaths exposed to disturbance moisture had a promoting effect on competitive species. Although moisture may not have a direct effect on E. tetralix cover, competitive species with a positive influence on E. tetralix may benefit from it. Therefore, species considered to be competing with E. tetralix on wet heaths may not do so on dune heaths. The fit of the models determined by root mean squared of error approximation (RMSEA) was relatively poor, however, with a RMSEA-value exceeding 0.08. This may be due to a small sample size, lack of multivariate normality or model misspecification. More exact measurement of the variables is also needed to gain more detailed insight into their effects on E. tetralix.  We conclude that SEM is suitable to modeling the complex interaction between the various biotic and abiotic variables that influence the cover of individual plant species in plant communities. On a more practical level this implies that before the decision to use grazing as a management tool is taken, various other factors affecting the ecosystem should be taken into account such as seasonality and nutrient balances.