The mafic-intermediate source lithologies qualify the red clays and brown loams of BCI as basisols (Young, 1976) or mafisols (Baillie, 1996). These are moderately frequent in plate tectonic subduction zones or on oceanic mantle plumes. They differ significantly from the much more extensive dystrophic tropical forest soils on siliceous and felsic lithologies of the shields and sedimentary basins of continental plate interiors.
The studies at La Selva in Costa Rica and across Hawaii are particularly appropriate for comparison with BCI, as their study sites are also on mafisols and because they relate pedological development to forest features and processes. The Luquillo CTFS plot at Luquillo in Puerto Rico is also on volcanic parent material of intermediate lithology (Brown et al., 1983; Roberts et al., 1942), but its pre-montane climate (Thompson et al., 2004) complicates pedogenic comparisons.
The non-alluvial soils at La Selva are developed on andesitic-basaltic lava flows, the most of recent of which occurred in the Pleistocene and is therefore younger than any of the geological formations on BCI. Nonetheless, they weather similarly and the main mature soils are deep, weathered, leached, reddish clays. Morphologically they are similar to the red clays of BCI (Sollins et al., 1994), but have less compact subsoils (E Veldkamp, personal communication, 2006; Sollins et al., 1994). The La Selva climate is wetter and less seasonal than that of BCI, and the soils are more intensively leached. The chemical ranges of the red clays on BCI and La Selva lava flow residual soils overlap, but the La Selva soils are generally more acid and base–depleted. They contain substantial quantities of non-melanising organic matter that extend downwards to well below 1 m (Veldkamp et al., 2003). Another feature of La Selva not so far noted on BCI is the localised nutrient enrichment of streams by seepage of geothermal waters (Pringle, 1991). The significance of this to soils and terrestrial nutrient cycles is not clear.
Soils on intermediate-mafic ash and lavas elsewhere in lowland Costa Rica have high contents of volcanic glass inherited from their parent materials (Nieuwenhuyse et al. 2005). The clay minerals develop through a sequence of volcanic glass – allophane – halloysite, and the main soils are classified as Andosols (WRB)/Andisols (ST). These are characterised by high contents of poorly crystalline allophanic clay minerals, very high porosity, and high capacities to bond with organic matter and to sorb phosphate and other anions. The alluvial soils at La Selva still have some andic characteristics (Johnson & Todd, 1983) but the residual soils on the lava flows are classified as mature Ultisols or Oxisols. Their low-moderate bulk densities, substantial and stable organic matter at depth (Mikutta et al., 2005), and residual rock-derived nutrient fertility (Bern et al., 2005; Porder et al., 2006) suggest that they have reached their present state along the allophane-halloysite weathering trajectory that prevails in volcanic soils elsewhere in Costa Rica (Sollins et al., 1994).
The main mature soils on the basaltic and related lithologies in the humid forested zones of Hawaii are also red clays that have weathered along the glass-allophane–halloysite-gibbsite trajectory. The intermediate allophanic stage is surprisingly prolonged, with the poorly crystalline minerals disappearing only in very mature soils on geomorphically stable sites on the older volcanoes (Chadwick, 2006; Chorover et al., 2004; Cline, 1955; Foote et al., 1972; Hedin et al., 2003; Vitousek, 2004; Wada et al., 1972).
Allophane and halloysite occur as intermediate stages in the weathering of basalt to kaolinite and sesquioxides in many other humid tropical areas (Auxtero et al., 1996; Curi & Franzmeier, 1981; Naidu et al., 1987; Schirrmeister & Störr, 1994; Siefferman & Milliot, 1969).
The BCI pattern of the weathering of intermediate-mafic volcanics from primary minerals to kaolinite + sesquioxides via smectites, rather than allophanes, therefore appears to be uncommon in tropical climates moist enough to support forests. Smectites are common weathering products in tropical basaltic soils, but usually in dry areas with savannah vegetation (Young, 1976; Chadwick, 2006). Smectites do occur in tropical forest soils, but are usually only significant in immature dark cracking clays on calcareous parent materials (Wright et al., 1959: Baillie, 1996). Factors that may possibly favour smectitic over allophanic weathering on BCI include: low ash: lava ratios; low contents of volcanic glass; significant contents of phyllosilicates; low Al:Mg ratios; and a moderate dry season.
Not only is the occurrence of smectites on BCI unusual, so also is their persistence in the pale swelling clays. Thermodynamic and electrochemical considerations suggest that these minerals should be unstable in acid leaching environments, and should rapidly desilicate to 1:l kanditic structures (Karathanasis & Hajek, 1983). On BCI this appears to happen in the disappearance of smectites as the brown fine loams mature to predominantly kanditic red clays. Less usual alternatives for smectites in Al-rich environments is transformation to chloritic 2:1:1 structures by the precipitation of Al as gibbsite or Mg as brucite interlayers (Jannet et al., 1996), or the formation of zeolites or similar intermediates (Bhattacharya et al., 1999). However, we have no evidence that these processes occur on BCI (R. Grim, unpublished data, 2007).
Instead, the large quantities of smectites that weather from the Caimito Marine formation persist in the pale swelling clays. Unlike on the other geological formations, these smectites are not transient, and they survive even in deep and apparently mature soils with low base and high Al saturations. This combination is uncommon but not unknown, and Dystruderts occur in southeastern USA (Karathanasis & Hajek 1985; Soil Survey Staff, 1999).
Complexation with organic matter may be involved in this stabilisation of the smectites. It is known that the resistance of organic matter to oxidation is increased by intercalation as interlayers in smectites (Deng et al., 2003; Righi et al., 1995; Wattel-Koekkoek et al., 2003). It is possible that such associations may also stabilise the smectite components of the complexes against hydrolysis and desilication. There may be parallels with the persistence of organically complexed allophanic minerals in the basalt soils of Hawaii. The organic components of such complexes are protected against degradation (Chorover et al., 2004), and the allophanes may be likewise stabilised by the association.
A further speculative hypothesis is that complexation with organic matter not only stabilises some clay minerals, but also affects weathering trajectories. Thus, the various components of the instantaneous ionic-colloidal mix produced in the early stages of mineral hydrolysis are likely to have specific bonding patterns with different organic materials, depending on aliphatic:aromatic and hydrophilic:hydrophobic ratios, the proportions of specific groups, and their stereo-chemical juxtapositions (Chorover & Amistadi, 2001; Righi et al., 1995). Such complexation may influence the crystallisation pathways of the secondary clay minerals and could thus influence the balance between allophanic versus smectitic weathering. Forests may affect pedo-mineralogical trajectories through biochemical variations in their exudates, litter and decomposition products. These putative influences are likely to be mutual and bi-directional, as the biochemical composition and breakdown of litter is affected by the forest’s supply of nutrients, and thence by soil chemistry and mineralogy.
Although mainly pedological, the survey elucidates some aspects of BCI’s edaphic environments. It shows that, within the bounds set by relatively uniform climate and lithology, BCI forests grow in a considerable range of edaphic conditions, with respect to site stability, root aeration, and supplies of water and nutrients.
Aeration and stability are aspects of soil fertility that tend to be neglected in tropical forest ecology, but may be important edaphic differentiae on BCI (Kursar, 1989). The brown and dark loams and red clays are all well drained and apparently well aerated. The limited areas of swamp soils are imperfectly or poorly drained, and root aeration is severely restricted for most of the year. The drainage and aeration status of the pale swelling clays vary substantially between seasons. Once the smectites are thoroughly wetted and expanded in the early wet season, the subsoils are saturated for weeks or months on end, and aeration is severely restricted. When the clays contract in the early dry season, surface cracking facilitates free airflow into the topsoil and upper subsoil. This is likely to reactivate those roots that survive the waterlogging and accelerate the growth of fresh roots, uptake of water, and hence further and deeper drying and cracking.
As to site stability, the red clays occur mainly on gentle and apparently stable slopes. In contrast, the steep slopes, shallow sola, limited weathering and frequent treefalls (Putz, 1983; Putz et al., 1983 & 1985) suggest that profile truncation and site disturbance recur in the brown fine loams within time spans of the order 10 – 104 years
Many of the pale swelling clays occur on gentle slopes and most of their sola appear moderately deep. However, rapid spalling from the faces in our profile pits show that these soils are unable to retain a free face for much more than a single seasonal cycle of shrink-swell. This may have implications for forest site stability. Natural free faces are uncommon under intact forest on the gentle slopes. However, as these soils heave and buckle with each seasonal moisture cycle, root systems may be loosened and trees rendered vulnerable to windthrow. Once a tree is uprooted, the resultant pit has free faces. These can spall and slump, and destabilise adjacent soils. These effects may be sufficiently widespread and severe for site instability and enhanced vulnerability to windthrow to be significant edaphecological factors on these soils (Foster & Brokaw, 1996).
Some aspects of the water supply dimension of soil fertility have been elucidated by hydrogen isotope studies, which indicate that evergreen trees and lianas on BCI continue and even increase their water uptake during the dry season. They tap moisture from successively deeper horizons, down to about 1m, as the dry season progresses and the upper layers dry out (Andrade et al., 2005; Jackson et al., 1995; Meinzer et al., 2004).
Stones have little available moisture storage capacity and diminish overall soil reserves of usable moisture. High clast contents therefore tend to intensify soil droughtiness, and this is likely to be a significant constraint in the BCI brown fine loams, especially Hood. However, the droughtiness of stony (and shallow) soils is alleviated if the underlying saprolite is weathered enough to have some available moisture capacity and soft enough to be rootable. It appears that some understorey evergreens on BCI root as deep as 3 m (Mulkey et al., 1991; Wright et al., 1992). Larger trees on BCI may root even deeper (Nepstad et al., 1994; Pozwa et al., 2002) and into saprolite, (Baillie & Mamit, 1976).
Full characterisation of soil nutrient fertility requires that each nutrient be considered separately with respect to: distribution between mineral and organic components and between horizons; physical accessibility; chemical lability; and interactions and stoichiometric balance. Methods for full nutrient characterisation include: soil sampling from several depths; nutrient fractionation with extractants of increasing intensity; and equilibration with solutions of varying composition. Isotope spectrometry can be particularly valuable in identifying nutrient sources, loci and pathways. Nutrients need to be reported relative to others, as well as in absolute values.
Most BCI soil nutrient data, including our own, are restricted to labile forms. Because differently accessible forms of the same nutrient are often correlated, data for one form may be sufficient to depict general spatial distributions and ecological associations. However, they are inadequate to elucidate nutrient dynamics (Baillie et al., 2006). Also, the interpretation of just one set of extracts can be problematic, as seen in the finding that extractable P levels in the soils of BCI are lower than for forests at La Selva in Costa Rica and Cocha Cashu in Peru (Harms et al., 2004). A priori expectations are that the mafic lithology results in moderate or better P levels in BCI soils, and this accords with moderate P contents in leaf litter (Leigh et al., 1996; Yavitt, 2000). This discrepancy may be a methodological artifact, but may also be ecologically significant and indicate that there is substantial P bonded sufficiently firmly on clay mineral and organic surfaces to resist extraction by mild reagents, yet able to desorb sufficiently to contribute to forest nutrition.
In general, there are now sufficient data to indicate that, compared with many tropical forests, the soils of BCI are well endowed with available forms of Ca and Mg, but less with K. P supply appears to be moderate.
The combination of high Ca and Mg and low K in BCI soils contrasts with the soils of some Asian CTFS plots. Although acid and dystrophic, the soils at Lambir in Sarawak are relatively well endowed with exchangeable K, and Ca appears to be the critically deficient cation (S, Tan, unpublished data). The soils of the perhumid Sinharaja plot in Sri Lanka have extremely low contents of all the main cationic nutrients (I.A.U.N. Gunatilleke, unpublished data). These proportional differences can be depicted in Alvim (1979) stoichiometric roses (Figure 8.1).
Some of the BCI red clays have mid-profile minima for pH, TEB, and BS. The higher values in the topsoils are attributed to nutrient cycling and returns to the surface and topsoil in throughfall and litter. The higher values in the lower subsoil are attributed to inputs from mineral weathering. Slight but significant nutrient augmentation by weathering is widespread in a variety of intensively weathered and highly leached soils under tropical forests (Baillie et al. 2006; Bern et al., 2005; Porder et al., 2006). However, Barthold et al. (2007) show that the weathering augmentation of subsoil exchangeable K is less than litter return enhancement of topsoil K in deep BCI soils.
Sodium is normally regarded as a highly mobile element, the labile forms of which are rapidly depleted by intense leaching. However, exchangeable Na levels are moderate in several BCI profiles. This may be a result of weathering being less advanced than morphology indicates. This is corroborated by the survival of small quantities of the weatherable mineral albite, the most sodic of the plagioclase feldspars, in the subsoils of two profiles on the Bohio formation.
Neutral theory assembly-dispersal and other non-edaphic biotic processes appear to account for much of the spatial pattern of the forests on BCI. However, there are some edaphic influences, even within the homogeneous landscape and soils of the 50 ha LTER plot (Harms et al., 2001; John et al, 2007). Nonetheless, the forests on BCI lack the visually obvious edaphic associations with forest structure, floristics and dynamics seen in some Asian CTFS plots, such as the striking topographically-related congeneric segregations at Sinharaja, Sri Lanka (Gunatilleke et al. 2006) and the litho-pedologically related pattern in the dipterocarp forest at Lambir, Sarawak (Lee et al. 2002; Davies et al.2005; Russo et al., 2006). This may be because the more fertile BCI soils impose less nutrient stress, and allow non-edaphic, processes more latitude. This accords with the decreasing clarity of edaphic associations in Sarawak mixed dipterocarp forests as soil nutrient fertility increases (Ashton & Hall, 1992; Potts et al. 2002).
Our soil map may prompt identification of more edaphic associations in BCI forests than apparent at present. Possibilities include: instability as a determinant of forest structure and dynamics in pale swelling clays in the west and southwest of the island; and direct physical disruption effects of treefalls plus indirect effects of nutrient supplementation by erosion and revitalised weathering in the forests on the shallow brown fine loams.