Groundwater decline and tree change in floodplain landscapes: Identifying non-linear threshold responses in canopy condition

Jarrod KATH, K. Reardon-Smith, Andrew Le Brocque, Fiona DYER, E. Dafny, L. Fritz, M Batterham

Research output: Contribution to journalArticlepeer-review

43 Citations (Scopus)
115 Downloads (Pure)


Groundwater decline is widespread, yet its implications for natural systems are poorly understood. Previous research has revealed links between groundwater depth and tree condition; however, critical thresholds which might indicate ecological ‘tipping points’ associated with rapid and potentially irreversible change have been difficult to quantify. This study collated data for two dominant floodplain species, Eucalyptus camaldulensis (river red gum) and E. populnea (poplar box) from 118 sites in eastern Australia where significant groundwater decline has occurred. Boosted regression trees, quantile regression and Threshold Indicator Taxa Analysis were used to investigate the relationship between tree condition and groundwater depth. Distinct non-linear responses were found, with groundwater depth thresholds identified in the range from 12.1 m to 22.6 m for E. camaldulensis and 12.6 m to 26.6 m for E. populnea beyond which canopy condition declined abruptly. Non-linear threshold responses in canopy condition in these species may be linked to rooting depth, with chronic groundwater decline decoupling trees from deep soil moisture resources. The quantification of groundwater depth thresholds is likely to be critical for management aimed at conserving groundwater dependent biodiversity. Identifying thresholds will be important in regions where water extraction and drying climates may contribute to further groundwater decline.
Original languageEnglish
Pages (from-to)148-160
Number of pages13
JournalGlobal Ecology and Conservation
Publication statusPublished - 1 Dec 2014


Dive into the research topics of 'Groundwater decline and tree change in floodplain landscapes: Identifying non-linear threshold responses in canopy condition'. Together they form a unique fingerprint.

Cite this