2. Landscape Function Analysis and Ecological Results
2.4. Discussion
2.4.3 Patch Indices and other Variables at Patch Scale
The overall trend in ecological measurements for patch types tended to follow the degree of 1378
biomass found in a patch (Tables 2.5 and 2.7), although at the high biomass end of the 1379
continuum, differences between grass, tall grass, Schoenoplectus or S. plumosum patches 1380
tended to be small or reflect unique conditions. Therefore, Schoenoplectus patches at Vaal 1381
River, which had significantly higher above-ground biomass together with S. plumosum and 1382
tall grass patches, when compared with bare grass patches, had the highest mean EC value 1383
(Table 2.5). This patch type is defined by the sedge Schoenoplectus corymbosus (Roth ex 1384
Roem. & Schult.) J. Raynal, which is always associated with wetlands (Marnewecke and 1385
64 Kotze, 1999), was found on a single transect (vr8, Table 2.4) located close to a tailings 1386
storage facility. This transect recorded a significantly higher mean EC than any other 1387
transect at Vaal River (Table 2.4) with the exception of the transect next to it (vr9). Electrical 1388
conductivity is a measure of the presence of soluble salts in a soil solution (Benton Jones Jr., 1389
2001) and therefore may be an indicator of contaminated soil. Transect vr8, with the 1390
highest mean EC of 881.5 ± 178.5 μS/cm, could be regarded as very slightly to moderately 1391
saline (Benton Jones Jr., 2001) which is considerably lower than values quoted in the Vaal 1392
River Environmental Management Plan (VR EMP) of 1520 to 28782 μS/cm for contaminated 1393
areas (Ellis et al., 2009), some of which were located fairly close (a few 100’s of metres) to 1394
transect 8 in the vicinity of the same tailings storage facility (TSF). 1395
Seriphium plumosum patches, which also had a high mean EC (Table 2.5), were found on 1396
four contiguous transects (vr 8, 9, 10 and 11, Table 2.4) at Vaal River, two of which had the 1397
highest EC values overall. In total, three transects (vr 3, 8 and 9) had high EC values and 1398
these were also transects that had visible white evaporite crusts on the soil surface. All 1399
three transects were situated on lower slopes and flats and presumably are in close contact 1400
with the water table. Evaluation of soil profiles near vr8 and 9, as documented in the Vaal 1401
River EMP, recorded seepage at 0.6 mbgl (Ellis et al., 2009). Seepage from groundwater and 1402
capillarity together with high evaporation rates at the soil surface and low rainfall resulting 1403
in minimal leaching are believed to cause evaporite crusts or efflorescences (Dultz and 1404
Kühn, 2005). As no chemical analysis was carried out on these evaporite crusts, it is unclear 1405
whether they are the weathered soluble products of dolomite (Ca/MgCO3) which outcrops
1406
throughout the VR study site, gypsum (CaSO4) or other sulphate salts, or a product of
1407
contamination from tailings storage facilities and mining activities or a combination of all 1408
three (Meza-Figueroa et al., 2009, Carmona et al., 2008, Dultz and Kühn, 2005, Joeckel et al., 1409
2005, Singer et al., 1999, Keller et al., 1986). Values of soil EC for all sites at West Wits were 1410
very low (Table 2.6), lower than any values for Vaal River recorded in this study, suggesting 1411
that surface soils contaminated through groundwater or atmospheric pathways (Singer et 1412
al., 1999) was not an issue at the sites studied at West Wits. 1413
Seriphium plumosum patches had the highest values for the LFA indices of soil stability, 1414
infiltration and nutrient cycling at both Vaal River and West Wits (Table 2.5 and 2.7, 1415
respectively), and the highest above-ground biomass, SOM and soil pH at West Wits. This 1416
65 suggests that S. plumosum plays a highly functional role in these grasslands, which according 1417
to literature, is contradictory. S. plumosum has been considered an indicator of severe 1418
overgrazing and rangeland degradation (Snyman, 2009b, Roux, 1969a) or poor management 1419
(Jordaan, 2009) as illustrated in one of its common names, bankrupt bush, and its 1420
encroachment into Highveld grasslands has been well documented although the 1421
mechanism/s behind this encroachment remain unknown (Snyman, 2009a). The shrub is a 1422
prolific producer of seeds, some of which seem to have dormancy mechanisms, and even 1423
after three years may have high viability (Snyman, 2009a). In contradiction to the 1424
conclusions that it invades overgrazed areas and may be a pioneer species, its seedlings only 1425
germinate in low-light conditions such as those formed close to unpalatable grasses such as 1426
Cymbopogon or Elionurus species (Snyman, 2010, Lecatsas, 1962, Cohen, 1937). Snyman 1427
(2010) provided evidence that S. plumosum has allelopathic effects on its own seedlings as 1428
well as interspecific competitor seedlings through inhibition of germination and seedling 1429
development. This may explain the bare ground that tended to surround S. plumosum 1430
shrubs on transect ww3 at West Wits (Figure 2.14b). However this was the only site at both 1431
Vaal River and West Wits where there was evidence that S. plumosum may be exerting a 1432
negative effect on surrounding vegetation and no investigation has yet been conducted on 1433
allelopathic interactions between S. plumosum and adult plants of the same or different 1434
species. Furthermore, signs of remnant burnt stubble in these bare patches surrounding S. 1435
plumosum patches in transect ww3 support Snyman’s (2011) conclusion that increased 1436
temperatures due to the burning of the S. plumosum canopy cause increased local mortality 1437
of perennial grass species (Figure 2.14a and b) in the vicinity of these shrubs. 1438
According to Snyman (2012, 2011, 2010), S. plumosum is not found on soils that get water- 1439
logged, preferring well-drained soils. Yet forty-three S. plumosum patches were recorded on 1440
the 180 metres of transect vr8 (Table 2.3). This transect also had the only recorded patches 1441
at any site in this study of Schoenoplectus species, a sedge that definitely prefers at least 1442
seasonally waterlogged or wetland conditions (Marnewecke and Kotze, 1999). Possibly 1443
these S. plumosum patches are a remnant of drier conditions prior to the establishment of 1444
the TSF near the foot of transect vr8. Aerial surveys from 1980 show no TSF but it was 1445
present in 1992. Hattingh (1953) estimated that S. plumosum had an average lifespan of 1446
66 a.)
1447
b.) 1448
Figure 2.14 (a) Seriphium plumosum burning as a fire swept through transect vr15
1449
showing the increased intensity of the fire as it burns the S. plumosum 1450
canopy compared to the grass canopy. Notice the grass right up to the S. 1451
Plumosum suggesting no allelopathic response. (b) Remnant burnt grass 1452
stubble and bare patch in the vicinity of an S. Plumosum shrub on ww3 1453
transect. The bare patch and dead grass stubble possibly indicative of the 1454
increased mortality of grasses close to these shrubs through increased fire 1455
intensity when the shrub burns (Snyman, 2011). 1456
67 about 15 years which is significantly shorter than the time that the TSF has been present 1457
(> 20 but < 34 years). However, it is unknown whether or when surface/groundwater 1458
conditions might have switched from being possibly well-drained to hydromorphic with the 1459
establishment of the TSF. Seriphium plumosum shrubs have been associated with low soil 1460
nutrient conditions (SOM and % P) and Snyman (2009b) observed that they preferentially 1461
invaded abandoned marginal cultivated lands over more fertile sites. Nevertheless, in this 1462
study there was no significant difference in SOM between patch types at Vaal River, while at 1463
West Wits, SOM values were highest for S. plumosum patches and these differences were 1464
significant for bare patches but not for other vegetated patch types. Observations while 1465
sampling these patches found there was often a well developed fine litter layer of 1466
predominantly seeds shed from the shrub. Phosphorus was not examined in this study but 1467
no significant differences in soil nitrogen, either organic or inorganic, were found between 1468
patch types at West Wits. 1469
Of the seven patch types described from Vaal River, four were some form of graminoid- 1470
dominated patch, namely bare grass, sparse grass, grass patch and tall grass, which 1471
essentially describes an above-ground biomass gradient (Table 2.5). At West Wits, two of 1472
the four described patch types were graminoid-dominated patch types and were similarly 1473
named as two of the Vaal River patch types, i.e. sparse grass and grass patch, again based 1474
on differences in above-ground cover reflecting a gradient in above-ground biomass (Table 1475
2.1 and 2.7). One factor influencing or influenced by this continuum is soil moisture. 1476
However, transect averages for soil moisture tended to follow the date of sampling (not 1477
shown) with transects like vr10 and vr11 with the highest average soil moisture being 1478
sampled in mid winter whereas transects vr13 and vr14, with the lowest mean soil moisture 1479
sampled last towards the end of the dry winter (Table 2.4). But even with the influence of 1480
sampling date and winter dry-season effects, above-ground biomass (Figure 2.8) and root 1481
biomass (Figure 2.10a) were weakly correlated with soil moisture. Patch type showed 1482
significant differences based on soil moisture (Table 2.5 and 2.7) and generally followed the 1483
trend of increasing vegetation biomass although there were exceptions. For instance at Vaal 1484
River, grass patches had significantly higher above-ground biomass when compared with 1485
bare grass and biological soil crust patches, the latter having no measureable above-ground 1486
biomass. Yet soil moisture was not significantly different between the three patch types 1487
68 mentioned, although it was significantly lower than that for Schoenoplectus and S. 1488
plumosum patch types which had the highest mean above-ground biomasses (Table 2.5). As 1489
these grass patches belong to transect vr9 and this was sampled towards the end of winter, 1490
it is likely that this anomaly in low soil moisture is a result of the length of time the soil has 1491
had to dry out over the winter dry period prior to sampling. However, the dominant trend of 1492
there being a direct correlation between soil moisture and biomass, and therefore patch 1493
type, is one that is well described in the literature (Li et al., 2013, D'odorico et al., 2007, 1494
Eldridge et al., 2002, Breshears and Barnes, 1999, Greene, 1992). Bhark and Small (2003) 1495
showed that shrub and grassland canopies had higher infiltration and soil moisture than did 1496
the inter-canopy spaces and that this effect was stronger for shrubs in their study area than 1497
for grass canopies. Pockman and Small (2010) however, found grass patches in the 1498
Chihuahuan Desert had higher soil moisture after a rain event than did the shrub Larrea 1499
tridentate with bare patches the lowest. Thus the relationship between vegetation cover or 1500
above-ground biomass and soil moisture is complex. 1501
The primary determinant of soil moisture is climate (Legates et al., 2011, Noy-Meir, 1981). 1502
In the United States, Anderson (2006) describes how the central grasslands are separated 1503
into three categories which form a continuum from west to east, namely short-grass (0.3 – 1504
0.5 m tall), mixed grass (0.8 – 1.2 m) and tall grass (1.8 – 2.4 m). This continuum essentially 1505
reflects an annual precipitation gradient with the arid central continental short-grass prairie 1506
in the west receiving 260 – 375 mm annual precipitation, which increases as one moves 1507
eastward with the wetter east receiving 625 – 1200 mm (Anderson, 2006), and mixed prairie 1508
situated between the two and in the middle of the moisture gradient. It is likely that 1509
differences in precipitation underlie differences in biomass and canopy density between the 1510
two research sites in this study. At Vaal River 50% of all patches described were sparse grass 1511
and this patch type constituted 67% of the graminoid-based patches. At West Wits, 52% of 1512
patches were graminoid in nature, split slightly in favour of grass patches (56%) over sparse 1513
grass (44%) with no tall grass patches on the transect, but such tall grass patches were 1514
observed in off-transect areas of West Wits. Although the composition of the sparse and 1515
grass patches was not sampled the only obvious difference was in density of cover as 1516
reflected in the above-ground biomass. The fact that this was only significantly higher at 1517
West Wits is probably at least partially a function of the higher annual precipitation at West 1518
69 Wits (Anderson, 2006). Height in these two patch types was not measured but never got 1519
higher than knee height (0.6 m). However, tall grass patches consisting of Hyparrhenia hirta 1520
tussocks that can reach a height of 1.5 m (Van Wyk and Van Oudtshoorn, 1999), made up 1521
5% of sampled patches and was limited to particular transects at Vaal River. Although it was 1522
present at West Wits, tall grass patches were not recorded from any transects sampled 1523
there. However its presence is not related to a moisture gradient as H. hirta is regarded as a 1524
drought resistant species generally associated with climax or sub-climax grassland 1525
communities and secondary succession (Van Wyk and Van Oudtshoorn, 1999, Roux, 1969b). 1526
Soil organic matter (SOM) was significantly different between patch types at West Wits 1527
where bare patches had the lowest mean values and S. plumosum patches had the highest 1528
values (Table 2.7). However at Vaal River there were no significant differences in SOM 1529
between patch types (Table 2.5). Furthermore, all patches at Vaal River had lower mean 1530
values for SOM than did any patch type at West Wits although these differences were not 1531
tested statistically. The reasons for these differences are unknown but the formation and 1532
breakdown of SOM is a complex process (Jenkinson et al., 1990, Jenkinson and Rayner, 1533
1977) and a number of factors are known to affect SOM. Temperature (Kätterer et al., 1998, 1534
Lloyd and Taylor, 1994) and soil moisture (Yang et al., 2007, Lee et al., 2004) have been 1535
shown to influence SOM where increases in either variable lead to increased soil microbial 1536
activity and rates of SOM turnover which may result in decreased levels of SOM in soils. 1537
However, increases in both these factors can also result in increased inputs of carbon to 1538
soils through increased net primary productivity. Jenkinson and Rayner (1977) showed that 1539
reduced plant inputs reduced the long-term SOM values of a soil, and sparse grass, grass 1540
patch and S. plumosum patch above-ground mean biomasses were considerably lower at 1541
Vaal River compared to those for the same patch types respectively at West Wits (Table 2.5 1542
and 2.7). Seriphium plumosum patches at West Wits also had the highest mean above- 1543
ground biomass for any patch type at West Wits which supports the concept of higher 1544
carbon in the form of biomass inputs contributing to higher SOM values. Root biomass was 1545
not quantified at West Wits. At Vaal River, root biomass was found to be only weakly 1546
correlated with SOM and more strongly correlated with above-ground biomass (Figure 1547
2.10a). Furthermore, the relationship between root biomass and patch type was not a direct 1548
relationship between above- and below-ground biomass. This can be seen when comparing 1549
70 above and below-ground biomasses for sparse grass which had higher mean root biomass 1550
and lower above-ground shoot biomass when compared to grass patches at Vaal River. 1551
However, the general trend was that increases in above-ground biomass were mirrored in 1552
increased root biomass (Table 2.5). The effect of roots on SOM is complex as, on the one 1553
hand, they are a source of SOM (Aerts et al., 1992), while Cheng (1990) showed that living 1554
roots enhanced the rate of decomposition of SOM. However, amongst other factors, both 1555
SOM and the presence of above- and below-ground biomass are important influences on 1556
the stability of a soil and its resistance to fluvial based soil erosion (Oades, 1993, Oades, 1557
1984). 1558
The LFA index for soil stability is designed to indicate a site’s susceptibility to erosive forces 1559
and the ability of a soil to reform after disturbance (Tongway and Hindley, 2004). The index 1560
is calculated as a composite of a number of visual or tactile properties of the soil, or features 1561
at the soil surface, and uses a relative scale in which high values indicate stable, resistant 1562
soils. As the index is a visual system it does not include direct measures of SOM and root 1563
presence. Soil organic matter is indirectly estimated through evaluating the degree of 1564
incorporation of a leaf litter into soil and the visual presence of fungi in the litter. Root 1565
biomass is indirectly estimated through the basal area in grasses, and the canopy cover and 1566
density in shrubs and trees (Tongway and Hindley, 2004), on the assumption that the larger 1567
the base of a grass tuft is, the greater the root mass that will extend from it into the soil 1568
volume. The general trend at both West Wits and Vaal River was that values for LFA soil 1569
stability increased from lower values measured for bare soil to highest values for S. 1570
plumosum patches (Table 2.5 and 2.7). This trend in LFA soil stability was mirrored in most 1571
of the environmental factors that promote soil stability and resistance to erosion such as 1572
increased above-ground biomass (Chartier et al., 2013, Dlamini et al., 2011, Okin et al., 1573
2006, Abrahams et al., 1995) which was strongly correlated with LFA soil stability at West 1574
Wits and less so at Vaal River. Similarly, a weaker but still positive correlation between 1575
below-ground biomass and the LFA soil stability index supported the conclusion that this 1576
index was providing a viable relative indicator of soil stability and susceptibility to erosion 1577
(Burylo et al., 2012, De Baets et al., 2007, Cammeraat et al., 2005). 1578
Other environmental factors measured in this study also provide support for the conclusion 1579
that the LFA soil stability index is giving a valid relative indicator of soil stability. Core depth 1580
71 was taken to 10 cm but in a number of occasions stone prevented the attainment of this 1581
depth and there is a statistical correlation between depth of core and patch type. Core 1582
depth was significantly shallower in biological soil crust patches and bare grass patches 1583
(Table 2.5) compared to all other vegetated patches with the exception of tall grass patches. 1584
Also, the amount of stone greater than 2 mm present in the soil in both biological soil crust 1585
patches and bare grass patches was significantly higher than in other patch types with the 1586
exception of grass, and Schoenoplectus patches which were very few and restricted to one 1587
transect. There could be a number of explanations for the differences in soil core depths 1588
and soil stone content which are unrelated to soil stability and erosion. One such 1589
explanation could be that the sub-surface rock is closer to the surface in these patches than 1590
it is in neighbouring or nearby patch types and this leads to conditions that are not 1591
conducive to the establishment of vegetation and thus a bare patch, biological soil crust