Table 1.
Experimental treatments at Towoomba over the period 1949 to 1981.
Fig 1.
Location of Towoomba Agricultural Development Centre in the southern Springbok Flats of Limpopo Province in South Africa.
This map was generated based on datasets from the following sources: [35, 36].
Fig 2.
(a) Satellite image showing the approximate position of the experimental plots at Towoomba Agricultural Development Centre, South Africa; and (b) a photograph taken in September 2014 of two experimental plots. Black borders in (a) depict the plots shown in (b). Foreground in (b): a plot with no woody encroachment that received the maximum applications of AS (1166 kg ha-1 yr-1) and SP (466 kg ha-1 yr-1). Background in (b): a plot with intense woody encroachment by Vachellia karroo that received minor applications of AS (146 kg ha-1 yr-1) and moderate applications of SP (233 kg ha-1 yr-1). [Depicted is the author Antoni Milewski who confirmed consent for publication].
Fig 3.
Abundance and cumulative height of trees according to experimental treatment at Towoomba in 2011/12.
Box-plots show medians (dark bar), upper and lower quartiles (tops and bottoms of boxes), and ranges (upper and lower whiskers). AS = ammonium sulphate; SP = superphosphate. p values show the results of the Kruskal-Wallis testing for individual effects of AS and SP after Friedman testing for differences across all treatments.
Fig 4.
Concentrations of soil properties according to experimental treatment at Towoomba in 2014.
Box-plots show medians (dark bar), upper and lower quartiles (tops and bottoms of boxes), and ranges (upper and lower whiskers). AS = ammonium sulphate; SP = superphosphate. p values show the results of the Kruskal-Wallis testing for individual effects of AS and SP after Friedman testing for differences across all treatments.
Table 2.
Significant effects for ammonium sulphate (AS) and superphosphate (SP) treatments.
Fig 5.
Principal component analysis of tree abundance in relation to all soil properties.
Left: arrows indicate the strength of the correlation between each element and the first two axes produced by PCA. Right: points show the scores for the samples on axes 1 and 2 of the PCA, and ellipses summarize the variation in soil samples taken from plots containing 3 or fewer woody plant individuals per plot (black) and those containing 4 or more woody plant individuals per plot (grey).
Fig 6.
Tree abundance (categorised as 0–1, 2–4, 5–8 or >8 individual trees per plot) relative to nutrient concentrations, pH (KCl) and acid saturation.
Data points are depicted as grey circles. Means ± standard errors are depicted with black circles and error bars. Asterisks show significant differences in tree abundance according to Kruskal-Wallis rank sum tests (*** p<0.0001; ** p = 0.001–0.009; * p = 0.01–0.05). Different letters designate significant differences between means (p<0.05).
Fig 7.
Tree abundance (individual trees per plot) relative to nutrient concentrations, nutrient ratios, pH (KCl) and acid saturation.
Dashed lines depict boundary lines identified by boundary line analysis (S4 Table). In the bottom row, four categories of tree abundance (i.e. 0–1, 2–4, 5–8 or >8 individual trees per plot) are shown relative to nutrient ratios, with data points depicted as grey circles, and means ± standard errors depicted with black circles and error bars. Asterisks show significant differences in tree abundance according to Kruskal-Wallis rank sum tests (*** p<0.0001; ** p = 0.001–0.009; * p = 0.01–0.05). Different letters designate significant differences between means (p<0.05).
Fig 8.
Theoretical relationships between biotic response and nutrient concentration for various plant species occurring in the same environment.
The response of plants in terms of vigour to a particular nutrient are invariably bell-shaped curves (a) as opposed to distinct steps (b) delineating zones where concentrations of a particular nutrient are deficient, optimal or toxic.