When rare species are not important: linking plot-based vegetation classifications and landscape-scale mapping in Australian savanna vegetation

Abstract

Plant communities in extensive landscapes are often mapped remotely using detectable patterns based on vegetation structure and canopy species with a high relative cover. A plot-based classification which includes species with low relative canopy cover and ignores vegetation structure, may result in plant communities not easily reconcilable with the landscape patterns represented in mapping. In our study, we investigate the effects on classification outcomes if we (1) remove rare species based on canopy cover, and (2) incorporate vegetation structure by weighting species' cover by different measures of vegetation height. Using a dataset of 101 plots of savanna vegetation in north-eastern Australia we investigated first, the effect of removing rare species using four cover thresholds (1, 5, 8 and 10% contribution to total cover) and second, weighting species by four height measures including actual height as well as continuous and categorical transformations. Using agglomerative hierarchical clustering we produced a classification for each dataset and compared them for differences in: patterns of plot similarity, clustering, species richness and evenness, and characteristic species. We estimated the ability of each classification to predict species cover using generalised linear models. We found removing rare species at any cover threshold produced characteristic species appearing to correspond to landscape scale changes and better predicted species cover in grasslands and shrublands. However, in woodlands it made no difference. Using actual height of vegetation layer maintained vegetation structure, emphasised canopy and then sub-canopy species in clustering, and predicted species cover best of the height-measures tested. Thus, removing rare species and weighting species by height are useful techniques for identifying plant communities from plot-based classifications which are conceptually consistent with those in landscape scale mapping. This increases the confidence of end-users in both the classifications and the maps, thus enhancing their use in land management decisions.Abbreviations: ALL-dataset consisting of the full species pool; C>1-dataset with species contributing >1% to TFC included; C>5 -dataset with species contributing >5%; C>8 -dataset with species contributing >8%; C>10 -dataset with species contributing >10% to TFC included; IS -Indicator Species; ISA -Indicator Species Analysis; TFC -Total Foliage Cover.

The intertidal plant communities in north‐eastern Australia, their carbon stores and vulnerability to extreme climate events

“…Plant communities were identified following the standardized quantitative classification methods adopted by the Queensland government and outlined by Addicott et al (2018) and Neldner et al (2019). This was a four‐step process involving: (i) excluding species with <1% contribution to total foliage cover in any plot (Addicott, Newton, et al, 2018), resulting in 42 species in the analysis; (ii) recognizing differences in structural formation by multiplying %PFC of each species by vegetation layer height to form species importance values for each plot (Addicott, Laurance, Lyons, Butler, & Neldner, 2018); (iii) forming clusters with these species importance values from a similarity matrix (Bray–Curtis dissimilarity coefficient on square‐root transformed data) then agglomerative hierarchical clustering with unweighted pair group mean average; and (iv) determining the cluster division level to form plant communities using a combination of three evaluators: (i) SIMPROF (Clarke, Somerfield, & Gorley, 2008); (ii) Indicator Species Analysis (Dufrêne & Legendre, 1997) in the ‘labdsv’ R package (Roberts, 2013); and (iii) the ability of the classification to predict foliage cover distribution. This last method used generalized linear models in a multivariate framework (M. Lyons, 2017; Lyons, Keith, Warton, Somerville, & Kingsford, 2016).…”

“…Plant communities were identified following the standardized quantitative classification methods adopted by the Queensland government and outlined by and Neldner et al (2019). This was a four-step process involving: (i) excluding species with <1% contribution to total foliage cover in any plot (Addicott, Newton, et al, 2018), resulting in 42 species in the analysis; (ii) recognizing differences in structural formation by multiplying %PFC of each species by vegetation layer height to form species importance values for each plot (Addicott, Laurance, Lyons, Butler, & Neldner, 2018);…”

Supervised versus un‐supervised classification: A quantitative comparison of plant communities in savanna vegetation

“…Queensland Department of Science, Information Technology and Innovation, unpublished data). Addicott, Laurance, Lyons, Butler, and Neldner () quantified these criteria by removing sparse species based on their percent contribution to the total foliage cover at a site (identifying dominant and sub‐dominant species) and multiplying %PFC by the height of the vegetation layer (identifying the predominant layer and consistent sub‐canopy or shrub layers). Addicott, Newton, et al.…”

Aligning quantitative vegetation classification and landscape scale mapping: updating the classification approach of the Regional Ecosystem classification system used in Queensland

“…The second tier is broad geological or geomorphological groups (labelled land zones). The third tier are plant communities recognised at the association level (labelled vegetation communities) (Reprinted with permission from Addicott et al 2018b) (Neldner et al 2020, p. 116) noting that a RE may contain more than one vegetation community, but a vegetation community cannot occur in more than one RE. The classification approach used to identify vegetation communities specifies that communities be identified at the plant association level using plot-based records and characteristics of the pre-dominant layer (defined as that layer contributing most to the above-ground biomass; .…”

“…Removing species from a classification exercise based on contribution to percentage cover rather than frequency is therefore necessary within the RE system. Testing the levels of contribution to total foliage cover that represented dominance in grassland, shrubland and woodland formations in north-eastern Queensland, Addicott et al (2018b) found the levels of contribution to percentage total foliage cover that influenced classification outcomes relevant to broad-scale map communities differed between the three formations. In grasslands a contribution of 8% to total foliage cover in any plot was the optimal threshold for removing species, in shrublands a contribution of 1% was optimal and in woodlands a contribution of 10% was optimal.…”