3.5 The Macroecology of Plant Invasions in South Africa
3.5.1 Plant Invasions as a Biogeographical Assay
Previous research showed that the distribution of naturalised alien plants in
South Africa can be viewed as a “biogeographical assay” (Rouget et al. 2015; see
also Richardson et al. 2004, 2005). Patterns of distribution, co-occurrence and
turnover of well-established alien species at the scale of QDGCs show that “invasive
alien [plant] species assemblages” (sensu Rouget et al. 2015) closely match the
traditional biomes of South Africa (see van Wilgen et al. 2020a, Chap. 1 Fig. 1.1),
which are defined on the basis of native plant biogeography and environmental
conditions (Rutherford 1997). We used the latest SAPIA data (see above) to determine an optimum number of “alien plant species assemblage zones” in South Africa,
i.e. regions characterised by similar alien plant species composition. Species compositions in QDGCs were compared in a pairwise fashion using the Simpson Dissimilarity Index. Non-metric dimensional scaling (nMDS) was then applied to plot each
QGDC in three-dimensions (red-green-blue) so that QDGCs with similar colours
have similar species composition (see methodological details in Supplementary
Appendix 3.3). A K-means clustering algorithm was then used to identify distinct
zones based on consensus over 30 different criteria. Results of the clustering analysis
revealed that four zones provide a good summary of current alien plant distribution
data at the scale of QDGCs (Fig. 3.4). This contrasts with the six clusters defined by
Rouget et al. (2015), based on the number of commonly defined native biomes. Two
of the zones defined in Fig. 3.4. (“fynbos-specific invaders” and “grassland-specific
invaders”) are very similar to clusters defined by Rouget et al. (2015)—these equate
closely with the Fynbos and Grassland Biomes of South Africa, respectively. The
“moist subtropical invaders” and “semi-arid invaders” zones correspond with the
mesic parts of the Savanna Biome, and the interface between the Nama Karoo and
arid parts of the Savanna Biome, respectively. Large parts of the Nama Karoo and
Succulent Karoo Biomes (a complex mixture of clusters 1, 2 and 3 in Rouget et al.
2015) were not characterised by any cluster in our analysis, as these cells contained
fewer records compared to the rest of the country. This low number of records led to
biases in the comparisons of QDGCs and prevented the nMDS algorithm from
generating sensible results. We believe that the clustering resulting from our analysis
provides an ecologically meaningful basis for discussing broad-scale patterns of plant
invasions in South Africa. Note that species composition is not perfectly homogeneous within each zone, and that species composition varies gradually in space, even
within zones (Supplementary Fig. S3.1). For example, visual inspection of the
differences in species composition suggests that the northeastern and southwestern
parts of the “grassland-specific invaders” zones are slightly different from each other,
as are the northern and southern parts of the “moist subtropical invaders” zone.
Finally, we identified “signature taxa” - those that typify each alien plant species
assemblage based on the proportion of QDGCs occupied by the taxa that fall within
the assemblage. We also identified widespread naturalised taxa that have large parts
84
D. M. Richardson et al.
3.5.1 Plant Invasions as a Biogeographical Assay
Previous research showed that the distribution of naturalised alien plants in
South Africa can be viewed as a “biogeographical assay” (Rouget et al. 2015; see
also Richardson et al. 2004, 2005). Patterns of distribution, co-occurrence and
turnover of well-established alien species at the scale of QDGCs show that “invasive
alien [plant] species assemblages” (sensu Rouget et al. 2015) closely match the
traditional biomes of South Africa (see van Wilgen et al. 2020a, Chap. 1 Fig. 1.1),
which are defined on the basis of native plant biogeography and environmental
conditions (Rutherford 1997). We used the latest SAPIA data (see above) to determine an optimum number of “alien plant species assemblage zones” in South Africa,
i.e. regions characterised by similar alien plant species composition. Species compositions in QDGCs were compared in a pairwise fashion using the Simpson Dissimilarity Index. Non-metric dimensional scaling (nMDS) was then applied to plot each
QGDC in three-dimensions (red-green-blue) so that QDGCs with similar colours
have similar species composition (see methodological details in Supplementary
Appendix 3.3). A K-means clustering algorithm was then used to identify distinct
zones based on consensus over 30 different criteria. Results of the clustering analysis
revealed that four zones provide a good summary of current alien plant distribution
data at the scale of QDGCs (Fig. 3.4). This contrasts with the six clusters defined by
Rouget et al. (2015), based on the number of commonly defined native biomes. Two
of the zones defined in Fig. 3.4. (“fynbos-specific invaders” and “grassland-specific
invaders”) are very similar to clusters defined by Rouget et al. (2015)—these equate
closely with the Fynbos and Grassland Biomes of South Africa, respectively. The
“moist subtropical invaders” and “semi-arid invaders” zones correspond with the
mesic parts of the Savanna Biome, and the interface between the Nama Karoo and
arid parts of the Savanna Biome, respectively. Large parts of the Nama Karoo and
Succulent Karoo Biomes (a complex mixture of clusters 1, 2 and 3 in Rouget et al.
2015) were not characterised by any cluster in our analysis, as these cells contained
fewer records compared to the rest of the country. This low number of records led to
biases in the comparisons of QDGCs and prevented the nMDS algorithm from
generating sensible results. We believe that the clustering resulting from our analysis
provides an ecologically meaningful basis for discussing broad-scale patterns of plant
invasions in South Africa. Note that species composition is not perfectly homogeneous within each zone, and that species composition varies gradually in space, even
within zones (Supplementary Fig. S3.1). For example, visual inspection of the
differences in species composition suggests that the northeastern and southwestern
parts of the “grassland-specific invaders” zones are slightly different from each other,
as are the northern and southern parts of the “moist subtropical invaders” zone.
Finally, we identified “signature taxa” - those that typify each alien plant species
assemblage based on the proportion of QDGCs occupied by the taxa that fall within
the assemblage. We also identified widespread naturalised taxa that have large parts
84
D. M. Richardson et al.