bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
1 Mapping Vegetation Species Succession in a Mountainous Grassland ecosystem
2 using Landsat and Sentinel-2 data
3 Adagbasa G. E1¶*, Mukwada G1,2,3
4
5 1Department of Geography, University of the Free State, South Africa
6 2Afromontane Research Unit, University of the Free State, South Africa
7 3Department of Geography & W.A. Franke College of Forestry & Conservation of the University of
8 Montana, USA
9
10
11 Corresponding Author:
12 E-mail address: [email protected] (EG)
13
14
15
16
17
18
19
20 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
21 Abstract
22 Vegetation species succession and composition are significant factors determining the rate of
23 ecosystem biodiversity recovery after being disturbed and subsequently vital for sustainable and
24 effective natural resource management and biodiversity. The succession and composition of
25 grasslands ecosystems worldwide have significantly been affected by the accelerated changes in
26 the environment due to natural and anthropogenic activities. Therefore, understanding spatial data
27 on the succession of grassland vegetation species and communities through mapping and
28 monitoring is essential to gain knowledge on the ecosystem and other ecosystem services. This
29 study used a random forest machine learning classifier on the Google Earth Engine platform to
30 classify grass vegetation species with Landsat 7 ETM+ and ASTER multispectral imager (MI)
31 data resampled with the current Sentinel-2 MSI data to map and estimate the changes in vegetation
32 species succession. The results indicate that ASTER IM has the least accuracy of 72%, Landsat 7
33 ETM+ 84%, and Sentinel-2 had the highest of 87%. The result also show that other species had
34 replaced four dominant grass species totaling an area of about 49 km2 throughout the study.
35 Keywords: Grassland, Succession, Mountains, Remote sensing
36
37 1. Introduction
38 Vegetation succession has many economically significant attributes, including high overall
39 biomass and productivity, a wider variety of species, and minimal nutrients or energy from the
40 ecosystem (1). Nevertheless, an ecosystem with naturally occurring succession stages will be more
41 resilient to natural and anthropogenic disturbances, suppose these disturbances increase in severity,
42 frequency, and magnitude because of human activities and weather conditions. In that case, the bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
43 pressure on plant communities increases, thereby causing an accelerated succession and creating
44 a new vegetation community and allowing the succession of non-native species (2).
45 Luken (1) describes vegetation succession as a change in vegetation composition over 500 years
46 without been disturbed to achieve a stable species composition called a climax. Climate change
47 and other disturbances within a short period may result in fluctuations of species composition,
48 promote non-native species and delay the natural vegetational succession from reaching its climax.
49 The succession of non-native species could impact biodiversity and the natural ecosystem. Non-
50 native invasive species quickly inhabit disturbed spaces and delay native species from achieving
51 seral or climax states. In some cases, the succession is entirely taken over and held for an extended
52 period at an intermediate state, affecting biodiversity (3-5). Invasive vegetation species threaten
53 native vegetation species and water resources because they grow faster, consume more water, and
54 spread more than the native species (6, 7). The encroachment of these vegetation species tends to
55 alter the balance of ecosystems, thereby accelerating succession. Vegetation species succession
56 and composition are significant factors determining the rate of ecosystem biodiversity recovery
57 after being disturbed (8) and subsequently vital for sustainable and effective natural resource
58 management and biodiversity. For example, in South Africa, changes in vegetation succession
59 resulting from disturbances have led to significant losses in biodiversity (9).
60 Worldwide, the succession and composition of grasslands ecosystems have been significantly
61 affected by the accelerated changes in the environment due to natural and anthropogenic activities
62 (10, 11). It has resulted in shortages in grasslands taxonomy and efficient functioning of ecosystem
63 services (12). Grasslands' changing diversity and composition impact ecosystem services like
64 precipitation and temperature controls, freshwater supply, erosion control, and soil formation (13-
65 15). They can likely result in biodiversity loss (16). About one-third of South African land surface bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
66 is covered by the grassland biome (17). It has just less than 3% located in protected areas, and 40-
67 60% have been altered with little chance of been salvaged and returned. It makes the grassland one
68 of the most vulnerable biomes in South Africa (18). Therefore, understanding spatial data on the
69 succession of grassland vegetation species and communities through mapping and monitoring is
70 essential to gain knowledge on the ecosystem and other ecosystem services (9, 19, 20).
71 Remote sensing provides an efficient approach for mapping grassland vegetation species by
72 reducing rigorous fieldwork necessitated by standard mapping methods. It does this effectively by
73 offering a wide range of recent data on vegetation species distribution from hyperspectral and
74 multispectral imagery (21, 22). Extensive studies have been undertaken in monitoring spatio-
75 temporal changes in vegetation species composition and diversity using remote sensing data (23-
76 26). However, these studies focus briefly on a short period, usually between one to five years,
77 because, before now, only high-resolution hyperspectral images could give accurate vegetation
78 species discrimination at individual levels (9, 19, 27-30). Nevertheless, recent studies have shown
79 that free low-resolution satellite images like Sentinel-2 MSI and Landsat 8 OLI can be used to
80 accurately map and monitor grassland vegetation species (31-33). Vegetation species succession
81 and diversity monitoring can be done over an extended period using these low-resolution imageries
82 in combination with machine learning. Therefore, this study used Landsat 7 ETM+ and ASTER
83 MI data fused with the current Sentinel-2 MSI data to map and estimate the changes in vegetation
84 species succession.
85 2. Study Area
86 The study was conducted at the Golden Gate Highlands National Park, located in Free State, South
87 Africa, near the Lesotho border (Figure 1). The Park covers 340 km2 and is located at the foothills
88 of the Maloti Mountains of the Eastern Free State. The highest peak in the park is 2,829 m (9,281 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
89 ft) above sea level. The Park is positioned in the Eastern Highveld region of South Africa and
90 experiences a dry, sunny climate from June to August with showers, hail, and thunderstorms
91 between October and April and snow in winter. The Park has a relatively high annual rainfall of
92 800 mm (31 in). The park contains over 60 grass species (34).
93
94
95
96
97
98
99
100 101 102 Figure 1: The Study Area 103 104 105
106 3. Materials and Methods
107 The study used satellite images from different sensors to cover the period of study. The sensors
108 used include the European Union/ESA/Copernicus Sentinel-2 MSI, the United States Geological
109 Survey (USGS) Landsat series ETM+, and the Advanced Spaceborne Thermal Emission and
110 Reflection Radiometer (ASTER) multispectral imager satellite images with UTM Projection
111 Zone35S, and Datum WGS84 available on the Google earth engine (GEE). The imageries were
112 selected for the rainy months from November to April. The year 2001 was chosen for the Landsat
113 +ETM, 2011 for the ASTER MI, and 2021 for Sentinel-2. Landsat ETM+ and OLI were selected bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
114 for this study because they have a 15m panchromatic band used to pansharpening the images from
115 30 m to 15 m. ASTER IM was selected for 2011 because Landsat ETM+ had the scan line error
116 for 2011. Also, the ASTER IM has 15 m, resolution bands. The atmospheric and geometric
117 correction was done on the images, and the 10 m bands of Sentinel-2 were then used to resample
118 the pan sharped 15m Landsat images and ASTER IM to 10 m using bicubic interpolation. All the
119 images from the different sensors were calibrated to top-of-atmosphere (TOA) reflectance. A
120 spectral library of the 12 dominant grass vegetation species in the study area was acquired from a
121 previous study (31) that used deep learning and machine learning models to discriminating grass
122 species at the individual level. Their study recommended Sentinel-2 MSI bands 6, 7 (red edge),
123 bands 8 and 8A, band 11, and band 12 to produce optimum classification accuracy. Therefore, the
124 spectral resolution of these bands was used to match and select the bands in the Landsat ETM+
125 and ASTER MI as presented in table 1. The spectral library was used to generate 100 random
126 sample locations for training and cross-validation. The locations were randomly split into a
127 training set (70%) to train the classifiers (31, 35) and a test set (30%) for testing purposes (31, 36).
128 The GEE code editor random forest machine learning classifier with ten trees was used to process
129 the image collections to classify the images into induvial species classes (31, 37). Ancillary data
130 such as Normalized Difference Vegetation Index (NDVI) were mapped into the image collection
131 on GEE before classification was done to improve classification accuracy. The difference in
132 illuminating effect by high mountains was accounted for by mapping a 15 meters resolution
133 ASTER Global Digital Elevation Model (GDEM) version 2 scale down to 10 m into the image
134 collection (31, 36, 38, 39).
135
136 Table 1: Selected bands for classification bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
137 Name Pixel Size Wavelength Description Sentinel-2 MSI 2021 138 B6 20 meters 0.74 µm Red Edge 2 B7 20 meters 0.78 µm Red Edge 3 139 B8 10 meters 0.84 µm NIR B8A 20 meters 0.87 µm Red Edge 4 B11 20 meters 1.614 µm SWIR 1 140 B12 20 meters 2.202 µm SWIR 2 ASTER MI 2011 B3N 15 meters 0.780-0.860µm VNIR_Band3N (near infrared, nadir pointing) 141 B04 30 meters 1.600-1.700µm SWIR_Band4 (short-wave infrared) B05 30 meters 2.145-2.185µm SWIR_Band5 (short-wave infrared) B06 30 meters 2.185-2.225µm SWIR_Band6 (short-wave infrared) 142 Landsat ETM+ 2001 B4 30 meters 0.77 - 0.90 µm Near infrared B5 30 meters 1.55 - 1.75 µm Shortwave infrared 1 143 B7 30 meters 2.08 - 2.35 µm Shortwave infrared 2 B8 15 meters 0.52 - 0.90 µm Panchromatic
144 4. Results and Discussion
145 4.1. Sensor performance and spectral reflectance
146 Table 2 shows the accuracy of each sensor. ASTER IM has the least accuracy of 72%, Landsat 7
147 +ETM 84%, and Sentinel-2 had the highest of 87%. The ASTER IM had the lowest level of
148 accuracy, possibly because (40) stated that each scene does not have all 14 bands. Therefore, some
149 scenes may have fewer bands than others. Hence, only bands 1 to 3 were available for that period.
150 However, these bands' spectral range can be compared to bands 1-5 of Landsat +ETM and OLI.
151 Another possible reason is that the three bands available didn't adequately separate the grass
152 species from each other, as shown in the spectral reflectance curve in Figure 2. Nevertheless, if
153 all the bands were available, ASTER IM should discriminate the grass species effectively to attain
154 a higher accuracy using machine learning classifiers. The accuracy of the ASTER image agrees
155 with a study done by (41). Their study used ASTER NDVI and EVI to discriminate rice and citrus
156 fields with 75% and 65% accuracy, respectively. They also used Landsat 5 TM NDVI and EVI, bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
157 which had a lower accuracy of 60% and 65% than the accuracy reached in this study with Landsat
158 7 ETM+. Landsat 7 ETM+ was able to get a greater accuracy because the bands were pan-
159 sharpened with the 15m panchromatic bands, unavailable on the Landsat 5 T.M., and resampled
160 from 30 m to10 m using the Sentinel-2 10 m bands. Also, the R.F. machine learning classifier,
161 which many studies have proved to improve classification (31, 32, 42, 43), contributed to the
162 higher accuracy in this study than the density slicing classification used in their research.
163 Table 2: Accuracy of different sensors S/N Sensors Accuracy % 1 Sentinel-2 MSI (2021) 87% 2 Landsat7 +ETM (2001) 84% 3 ASTER MI (2011) 72%
(a) (b) Reflectance
(c)
164 Wavelength (micrometers)
165 Figure 2: Species spectral reflectance curves of twelve grass species extracted from Landsat 7 166 ETM+ (a), ASTER MI (b), and Sentinel-2 (c) bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
167 Figure 2 shows the spectral reflectance of the twelve grass species extracted from all the sensors.
168 In the Landsat 7 ETM+, the species were discriminated in wavelengths of 0.52 - 0.77 µm and 1.55
169 – 2.08 µm, representing bands at the start of the wavelength for the panchromatic near-infrared,
170 shortwave infrared 1, and shortwave infrared 2. The ASTER MI has its best spectral separation
171 wavelength of 0.780-0.860µm (VNIR near-infrared, nadir pointing band). At the same time, the
172 Sentinel-2 separated it best in the bands 6, 7 (red edge), bands 8 and 8A, band 11, and band 12 as
173 recommended by the study by (31, 44-46), hence the difference in classification accuracy.
174 4.2. Grass Species changes and succession.
175 Figure 3 shows the map of twelve dominant vegetation species discrimination for 2001, 2011, and
176 2021. However, the classified map for 2010 was not analyzed further because of the low level of
177 accuracy. The difference of 12% from 2001 and 15% to 2021 might misrepresent the changes that
178 occurred with the classified maps of 2001 and 2021.
179 (b) (a)
180
181
182
(c) 183
184
185
186 187 Figure 3: The map of grass species derived from the Landsat 7 ETM+ (a), ASTER MI (b), 188 and Sentinel-2 MSI (c). 189 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
190 Figure 4 shows the vegetation changes from 2001 to 2021. It shows that four grass species had the
191 most significant transformation into other species in area coverage over twenty years. S.
192 centrifugus had an enormous shift of 22.6 km2, E. curvula had a change of 11.42 km2, S. Conrathii
193 had a change of 9.7 km2, and P.australis changed 5.14 km2. The other seven grass species had
194 gained and losses over the other four species.
195 Phragmites australis Miscanthus junceus 196 Stiburus Conrathii Eragrostis Plane Nees 197 Hermania Depressa Aristida junciformis Themeda triandra 198 Sporobolus centrifugus Mixed 199 Seriphium plumosum Hyparrhenia cf.Tamba Eragrostis curvula 200 -25 -20 -15 -10 -5 0 5 10 15 20 Km2 201 Figure 4: Vegetation transformation between 2001 and 2021 202 203 M. junceus has the highest success rate. It has replaced different species covering a total land area
204 of 17.22 km2. Another species fast replacing other species is the T. triandra species, replacing
205 11.4km2 that formerly contained other species. E. curvula, termed an increaser species by many
206 studies (9, 31), and grows very fast in disturbed environments have been replaced by eight different
207 species. T. triandra accounts for 50% of the total area that other species have replaced the E.
208 curvula. Although the E. curvula being an increaser species, it had replaced other species like the
209 S. centrifugus and area dominated by mixed species in different study locations and gained back
210 almost 90% of the area lost to other species in Figures 5 and 6. Figure 6 shows that the replacement
211 of E. curvula by T. triandra happens all over the study area. Still, it is more concentrated around
212 the North, North-East, South-west, and roads of the study area. bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
213 Phragmites australis 214 Miscanthus junceus Stiburus Conrathii 215 Eragrostis Plane Nees Hermania Depressa Aristida junciformis 216 Themeda triandra Sporobolus centrifugus 217 Mixed Seriphium plumosum Hyparrhenia cf.Tamba 218 Eragrostis curvula -8 -6 -4 -2 0 2 4 6 219 Km2
220 Figure 5: Species contributions to changes in Eragrostis curvula 221
222
223
224
225
226
227
228
229
230 231 232 233 Figure 6: Areas where Themeda triandra succeeded from Eragrostis curvula 234 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
235 S. centrifugus, the highest replacement species, is replaced by all the other species in the study
236 area, especially M. junceus, E. curvula, and T. triandra, accounting for 4.8km2, 4.7km2, and, 3.3
237 km2 respectively in figure 7. The S.centrifugus species is monocotyledon and belongs to the
238 Poaceae family. It is a native species of South Africa and is termed one of the least concerned
239 threatened species in the red list of South African plants (47). The succession appears to be
240 occurring around the South, South-western part of the study area, where there are very high
241 elevations (figure 8)
242 Phragmites australis Miscanthus junceus 243 Stiburus Conrathii Eragrostis Plane Nees Hermania Depressa 244 Aristida junciformis Themeda triandra Sporobolus centrifugus 245 Mixed Seriphium plumosum 246 Hyparrhenia cf.Tamba Eragrostis curvula -6 -5 -4 -3 -2 -1 0 247 Km2 248
249 Figure 7: Species contributions to change in Sporobolus centrifugus bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
250
251
252
253
254
255
256
257
258 259 260 261 262 Figure 8: Areas where M. junceus succeeded from S. centrifugus
263 S. Conrathii is a native species in South Africa but not endemic to the country. It is also not seen
264 as threatened plant species (47). This species has about 6.26 km2 replaced by M. junceus, majorly
265 in the southwestern (figure 10) part of the study area but gains 1.06km2 by replacing S. centrifugus
266 and 0.34km2 of H. depressa (Figure 9). bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
267 Phragmites australis Miscanthus junceus 268 Stiburus Conrathii Eragrostis Plane Nees 269 Hermania Depressa Aristida junciformis Themeda triandra 270 Sporobolus centrifugus Mixed Seriphium plumosum 271 Hyparrhenia cf.Tamba Eragrostis curvula 272 -7 -6 -5 -4 -3 -2 -1 0 1 2
273 Figure 9: Species contributions to changes in Stiburus Conrathii
274
275
276
277
278
279
280
281
282
283
284
285 Figure 10: Areas where M. junceus succeeded from S. Conrathii
286 P. australis is a decreaser species quickly affected by overgrazing and has a slow recovery rate
287 after a disturbance (48). It is a tall grass found across South Africa, especially around river beds bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
288 and wet environments, and is not at risk of extinction. (47, 49). Nevertheless, figure 11 shows that
289 it is being replaced mainly by T. triandra (2.55 km2), E. plane Nees (2.3 km2), and M. junceus (1.4
290 km2). It is also gaining back by replacing S. centrifugus and E. curvula. It is found across the study
291 area around the river channels and is replaced mainly in the southern and northern parts of the
292 study area (Figure 12).
293 Phragmites australis 294 Miscanthus junceus Stiburus Conrathii 295 Eragrostis Plane Nees Hermania Depressa 296 Aristida junciformis Themeda triandra 297 Sporobolus centrifugus 298 Mixed Seriphium plumosum 299 Hyparrhenia cf.Tamba Eragrostis curvula 300 -3 -2 -1 0 1 2 3
301 Figure 11: Species contributions to change in Phragmites australis
302 The replacement of species may be happening because several factors or disturbances like that
303 could either be natural or anthropogenic. Each species may have a unique or several reasons for
304 replacing others or has been replaced. Some of these species are used for human activities like
305 thatching and medicinal purposes, and some are palatable for grazing. Climate change and fires
306 are common factors that can also affect these successions (50, 51). The study area is a region
307 constantly affected by wildfires, and the fire severity and magnitude have been mapped by (31).
308 Their research showed some parts of the study area had constantly been burnt with high fire
309 severity over 20 years. These parts of the study area may be experiencing changes in species
310 composition, leaving only the fire-tolerant species or invasive species like S. plumosum, which is bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
311 a known species that promote the spread of wildfires (52-54). In fires recently disturbed areas in
312 the park, S. plumosum can sprout and lie dormant when encountering higher temperatures and low-
313 moisture conditions for the remainder of winter while awaiting the emergence of spring (55).
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328 Figure 12: Areas where T. triandra succeeded from P. australis 329 330 Several studies have found that climatic variables like temperature changes significantly impact
331 the distribution of ecological characteristics and environmental dynamics for many types of
332 vegetation species, including alien or native hosts (55-57). Temperature plays a significant role in
333 the distribution of species, with substantial effects on fire risk. The region's climate often has
334 extended periods of pronounced temperature and low precipitation, resulting in large, devastating bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
335 fires that devastate populations of plant life (57). In their research in the study area, Adepoju,
336 Adelabu (55) noted the possibility that the distribution of some types of grasses could be better
337 enhanced under conditions with higher daytime temperatures, low to moderate levels of rain at
338 lower elevations. They also noted that climate is an essential factor in determining grass species
339 distribution in the mountainous grasslands of South Africa, where periods of warm and cold
340 weather considerably fluctuate.
341
342 Areas that are subject to overgrazing and human activities are more likely to experience the
343 spreading of the species that can lead to gaining of new species and loss others. Factors like
344 distance from settlement, land near grasslands and agricultural, and distance from roads impact
345 how disturbances affect different vegetation species (55, 58, 59). The study area has been a national
346 park with a history of incorporating farms lands to expand the conservation area (60). In February
347 1991, the Qwaqwa National Park, which initially comprised multiple crop farmlands, agricultural
348 activities like domestic animal grazing, was also integrated into the study area. To date, some of
349 these farmers are still located in the park, and their livestock is still grazing the vegetation (60, 61).
350 However, some farmers have stopped planting on them. The disturbed land has been left to recover
351 by the park managers as a conservation strategy. Nevertheless, not all these previous farmlands
352 have recovered fully because the soils have been over-exploited. The vegetation species in these
353 locations struggle to survive with the effects of disturbances like frequent wildfires and
354 overgrazing from agricultural animals from within the park or the surrounding communal areas or
355 from the herbivorous animals within the park.
356
357 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
358 5. Conclusion
359 This study has shown that the Landsat 7 ETM+ can be used for vegetation species discrimination
360 if the panchromatic band is used to pan sharpening the 30 m bands to 15 m and then resampled
361 with the 10m bands Sentinel-2 MSI. It will allow for research in monitoring vegetation species
362 changes over a long period. Although the ASTER MI wasn't used to analyze the vegetation species
363 changes, it also has a prospect of being used if all the recommended bands are available. The study
364 explored the differences in vegetation species that have occurred over 20 years but didn't explore
365 the precise reasons why others were replacing some vegetation species. The causes and factors
366 influencing the shift in vegetation species in some park locations can be done in a further study. It
367 will help the park managers appropriately manage the park and prevent key vegetation species
368 from total annihilation by other more aggressive vegetation species, preserving the animal
369 population in the park and keeping the ecosystem healthy.
370
371 6. Acknowledgments
372 The authors would like to acknowledge the support of the South African National Park
373 (SANPARKS) South Africa throughout data collection. The authors would also like to
374 acknowledge the contribution of Prof. S.A Adelabu without whom this research would not have
375 been possible.
376
377
378
379
380 bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
381 7. References
382 1. Luken JO. Vegetational succession, climax. Environmental Geology. Dordrecht: Springer 383 Netherlands; 1999. p. 652-4. 384 2. Ma W, Wang X, Zhou N, Jiao L. Relative importance of climate factors and human activities in 385 impacting vegetation dynamics during 2000–2015 in the Otindag Sandy Land, northern China. Journal of 386 Arid Land. 2017;9(4):558-67. 387 3. David AS, Zarnetske PL, Hacker SD, Ruggiero P, Biel RG, Seabloom EWJPo. Invasive 388 congeners differ in successional impacts across space and time. 2015;10(2):e0117283. 389 4. Vera FWM. Grazing ecology and forest history: Cabi; 2000. 390 5. Dyderski MK, Jagodziński AMJF. Impact of invasive tree species on natural regeneration species 391 composition, diversity, and density. 2020;11(4):456. 392 6. Rebelo A, Cowling R, Richardson D, Pierce S. Vegetation of Southern Africa. Cambridge 393 University Press Cambridge; 1997. 394 7. Curhes S, Leigh C, Walton C. Weed risk assessment: African lovegrass Eragrostis curvula. 2009. 395 8. Viana-Soto A, Aguado I, Martínez S. Assessment of Post-Fire Vegetation Recovery Using Fire 396 Severity and Geographical Data in the Mediterranean Region (Spain). Environments. 2017;4(4):90. 397 9. Mansour K, Mutanga O, Everson T, Adam E. Discriminating indicator grass species for 398 rangeland degradation assessment using hyperspectral data resampled to AISA Eagle resolution. ISPRS 399 Journal of Photogrammetry and Remote Sensing. 2012;70:56-65. 400 10. Smart SM, Thompson K, Marrs RH, Le Duc MG, Maskell LC, Firbank LGJPotRSBBS. Biotic 401 homogenization and changes in species diversity across human-modified ecosystems. 402 2006;273(1601):2659-65. 403 11. Dallimer M, Tinch D, Acs S, Hanley N, Southall HR, Gaston KJ, et al. 100 years of change: 404 examining agricultural trends, habitat change and stakeholder perceptions through the 20th century. 405 2009;46(2):334-43. 406 12. Wesche K, Krause B, Culmsee H, Leuschner CJBC. Fifty years of change in Central European 407 grassland vegetation: Large losses in species richness and animal-pollinated plants. 2012;150(1):76-85. 408 13. Hoeffner K, Beylich A, Chabbi A, Cluzeau D, Dascalu D, Graefe U, et al. Legacy effects of 409 temporary grassland in annual crop rotation on soil ecosystem services. Science of The Total 410 Environment. 2021;780:146140. 411 14. Zhao Y, Liu Z, Wu JJLE. Grassland ecosystem services: a systematic review of research 412 advances and future directions. 2020:1-22. 413 15. Muller M, Siebert S, Ntloko B, Siebert FJB-AB, Conservation. A floristic assessment of 414 grassland diversity loss in South Africa. 2021;51(1):1-9. 415 16. Everson CS, Everson TJAJoR, Science F. The long-term effects of fire regime on primary 416 production of montane grasslands in South Africa. 2016;33(1):33-41. 417 17. Carbutt C, Tau M, Stephens A, Escott BJG. The conservation status of temperate grasslands in 418 southern Africa. 2011;11(1):17-23. 419 18. Little IT, Hockey PA, Jansen RJB-AB, Conservation. Impacts of fire and grazing management on 420 South Africa's moist highland grasslands: A case study of the Steenkampsberg Plateau, Mpumalanga, 421 South Africa. 2015;45(1):1-15. 422 19. Cho MA, Malahlela O, Ramoelo A. Assessing the utility WorldView-2 imagery for tree species 423 mapping in South African subtropical humid forest and the conservation implications: Dukuduku forest 424 patch as case study. International Journal of Applied Earth Observation and Geoinformation. 425 2015;38:349-57. 426 20. Ramoelo A, Cho MA, Mathieu R, Madonsela S, Van De Kerchove R, Kaszta Z, et al. Monitoring 427 grass nutrients and biomass as indicators of rangeland quality and quantity using random forest modelling 428 and WorldView-2 data. International journal of applied earth observation and geoinformation. 429 2015;43:43-54. bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
430 21. Adam E, Mutanga O, Rugege D, Ismail R. Discriminating the papyrus vegetation (Cyperus 431 papyrus L.) and its co-existent species using random forest and hyperspectral data resampled to HYMAP. 432 International Journal of Remote Sensing. 2012;33(2):552-69. 433 22. Schmidt K, Skidmore A. Spectral discrimination of vegetation types in a coastal wetland. Remote 434 sensing of Environment. 2003;85(1):92-108. 435 23. Lu B, He Y. Species classification using Unmanned Aerial Vehicle (UAV)-acquired high spatial 436 resolution imagery in a heterogeneous grassland. ISPRS Journal of Photogrammetry and Remote Sensing. 437 2017;128:73-85. 438 24. Osińska-Skotak K, Radecka A, Piórkowski H, Michalska-Hejduk D, Kopeć D, Tokarska-Guzik 439 B, et al. Mapping Succession in Non-Forest Habitats by Means of Remote Sensing: Is the Data 440 Acquisition Time Critical for Species Discrimination? 2019;11(22):2629. 441 25. Hantson W, Kooistra L, Slim PAJAvs. Mapping invasive woody species in coastal dunes in the N 442 etherlands: a remote sensing approach using LIDAR and high‐resolution aerial photographs. 443 2012;15(4):536-47. 444 26. Möckel T, Dalmayne J, Schmid BC, Prentice HC, Hall K. Airborne Hyperspectral Data Predict 445 Fine-Scale Plant Species Diversity in Grazed Dry Grasslands. 2016;8(2):133. 446 27. Ramoelo A, Skidmore AK, Cho MA, Schlerf M, Mathieu R, Heitkönig IM. Regional estimation 447 of savanna grass nitrogen using the red-edge band of the spaceborne RapidEye sensor. International 448 Journal of Applied Earth Observation and Geoinformation. 2012;19:151-62. 449 28. Mutanga O, Adam E, Adjorlolo C, Abdel-Rahman EM. Evaluating the robustness of models 450 developed from field spectral data in predicting African grass foliar nitrogen concentration using 451 WorldView-2 image as an independent test dataset. International Journal of Applied Earth Observation 452 and Geoinformation. 2015;34:178-87. 453 29. Cushnahan T, Yule I, Pullanagari R, Grafton M. Identifying grass species using hyperspectral 454 sensing. 455 30. Adelabu S, Mutanga O, Adam E, Sebego R. Spectral discrimination of insect defoliation levels in 456 mopane woodland using hyperspectral data. IEEE Journal of Selected Topics in Applied Earth 457 Observations and Remote Sensing. 2014;7(1):177-86. 458 31. Adagbasa EG, Adelabu SA, Okello TW. Application of deep learning with stratified K-fold for 459 vegetation species discrimation in a protected mountainous region using Sentinel-2 image. Geocarto 460 International. 2019:1-21. 461 32. Shoko C, Mutanga O. Examining the strength of the newly-launched Sentinel 2 MSI sensor in 462 detecting and discriminating subtle differences between C3 and C4 grass species. ISPRS journal of 463 photogrammetry and remote sensing. 2017;129:32-40. 464 33. Sothe C, Almeida CMd, Liesenberg V, Schimalski MBJRS. Evaluating Sentinel-2 and Landsat-8 465 data to map sucessional forest stages in a subtropical forest in Southern Brazil. 2017;9(8):838. 466 34. Parks SAN. Annual Report 2006/07. 2007. 467 35. Nkemelu DK, Omeiza D, Lubalo N. Deep Convolutional Neural Network for Plant Seedlings 468 Classification. arXiv preprint arXiv:181108404. 2018. 469 36. Adepoju K, Adelabu S, editors. Improved Landsat-8 OLI and Sentinel-2 MSI Classification in 470 Mountainous Terrain using Machine Learning on Google Earth Engine. Proceedings of the Biennial 471 Conference of the Society of South African Geographers; 2018; University of the Free State. 472 37. Sesnie SE, Finegan B, Gessler PE, Thessler S, Ramos Bendana Z, Smith AM. The multispectral 473 separability of Costa Rican rainforest types with support vector machines and Random Forest decision 474 trees. International Journal of Remote Sensing. 2010;31(11):2885-909. 475 38. Saha AK, Arora MK, Csaplovics E, Gupta RP. Land Cover Classification Using IRS LISS III 476 Image and DEM in a Rugged Terrain: A Case Study in Himalayas. Geocarto International. 477 2005;20(2):33-40. 478 39. Bahadur K.C. K. Improving Landsat and IRS Image Classification: Evaluation of Unsupervised 479 and Supervised Classification through Band Ratios and DEM in a Mountainous Landscape in Nepal. 480 Remote Sensing. 2009;1(4):1257-72. bioRxiv preprint doi: https://doi.org/10.1101/2021.08.18.456865; this version posted August 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license.
481 40. NASA LD. DAAC: ASTER Level 1 Precision Terrain Corrected Registered At-Sensor Radiance, 482 Version 3, NASA EOSDIS Land Processes DAAC, USGS Earth Resources Observation and Science 483 (EROS) Center, Sioux Falls, South Dakota. 2015. 484 41. Hashim M, Hazini SJJT. Separation of Different Vegetation Types in ASTER and Landsat 485 Satellite Images Using Satellite†derived Vegetation Indices. 2014;71(4). 486 42. Adepoju KA, Adelabu SA. Improving accuracy evaluation of Landsat-8 OLI using image 487 composite and multisource data with Google Earth Engine. Remote Sensing Letters. 2020;11(2):107-16. 488 43. Sharma RC, Hara K, Hirayama H. A Machine Learning and Cross-Validation Approach for the 489 Discrimination of Vegetation Physiognomic Types Using Satellite Based Multispectral and 490 Multitemporal Data. Scientifica. 2017;2017:8. 491 44. Wakulińska M, Marcinkowska-Ochtyra AJRS. Multi-temporal sentinel-2 data in classification of 492 mountain vegetation. 2020;12(17):2696. 493 45. Immitzer M, Neuwirth M, Böck S, Brenner H, Vuolo F, Atzberger CJRS. Optimal input features 494 for tree species classification in Central Europe based on multi-temporal Sentinel-2 data. 495 2019;11(22):2599. 496 46. Puletti N, Chianucci F, Castaldi CJASR. Use of Sentinel-2 for forest classification in 497 Mediterranean environments. 2018;42:32-8. 498 47. Raimondo D, Staden Lv, Foden W, Victor J, Helme N, Turner R, et al. Red list of South African 499 plants 2009: South African National Biodiversity Institute; 2009. 500 48. Volesky JD, Young SL, Jenkins KH. Cattle grazing effects on Phragmites australis in Nebraska. 501 Invasive Plant Science and Management. 2016;9(2):121-7. 502 49. Canavan K, Paterson ID, Lambertini C, Hill MP. Expansive reed populations—alien invasion or 503 disturbed wetlands? AoB PLANTS. 2018;10(2). 504 50. Leistner OA. Seed plants of southern Africa: families and genera: National Botanical Institute; 505 2000. 506 51. Russell GG, Watson L, Koekemoer M, Smook L, Barker N, Anderson H, et al. Grasses of 507 southern Africa1990. 508 52. Snyman H. Seriphium plumosum. Newsletter of the Grassland Society of Southern Africa 2009. 509 53. Dubula B, Tefsamichael SG, Rampedi IT. Assessing the potential of remote sensing to 510 discriminate invasive Seriphium plumosum from grass. South African Journal of Geomatics. 511 2016;5(2):201-13. 512 54. Mashalane MJ, Adjorlolo C. Integrating remote sensing and geostatistics in mapping Seriphium 513 plumosum (bankrupt bush) invasion 2016. 514 55. Adepoju K, Adelabu S, Mokubung CJEm, assessment. Mapping Seriphium plumosum 515 encroachment and interaction with wildfire and environmental factors in a protected mountainous 516 grassland. 2020;192(6):1-15. 517 56. Du Z, Wang Z, Liu Y, Wang H, Xue F, Liu YJIJoID. Ecological niche modeling for predicting 518 the potential risk areas of severe fever with thrombocytopenia syndrome. 2014;26:1-8. 519 57. Liu M, Zhao J, Guo X, Zhang Z, Tan G, Yang JJS. Study on climate and grassland fire in 520 HulunBuir, Inner Mongolia autonomous region, China. 2017;17(3):616. 521 58. Keeley JEJIjowf. Fire intensity, fire severity and burn severity: a brief review and suggested 522 usage. 2009;18(1):116-26. 523 59. Chuvieco E, Riaño D, Danson F, Martin PJJoGRB. Use of a radiative transfer model to simulate 524 the postfire spectral response to burn severity. 2006;111(G4). 525 60. Rademeyer C, van Zyl WJMJoSS. Golden Jubilee for Golden Gate-A Concise History of Golden 526 Gate Highlands National Park, 1963 to 2013. 2014;5(27 P3):1169. 527 61. Slater RJGJ. Between a rock and a hard place: contested livelihoods in Qwaqwa National Park, 528 South Africa. 2002;168(2):116-29. 529