Tree rings are a valuable tool of retrospective bio-indication, providing information on growth rates of trees, past climate conditions, dynamics and carbon sequestration rates of natural forest stands and much more [1
]. Tree ring science in the tropics still often struggles with the old, oft-repeated and wrong assumption that tropical climates are uniform, which led to the likewise wrong assumption that tropical trees would not form annual tree rings [2
]. In fact, there is overwhelming evidence showing distinct climate seasonality in respect to rainfall patterns in most part of the tropics [4
]. These seasonal changes induce a cambial dormancy and consequently annual tree rings in the wood [5
Globally, the first ever intensive studies on the periodicity of tropical tree growth and the existence of annual tree rings in the tropics were carried out on Java Island, Indonesia, already at the beginning of the 20th century [6
]. To date, a large number of studies on tropical tree ring formation have been carried out in many regions of the world and on a huge number of tree species [10
]. Recent tree ring studies on Java concentrated on the potential of teak chronologies for climate reconstruction [11
]. These studies were based on the fact that tree rings are a consequence of seasonally varying precipitation patterns with annually occurring dry seasons. Another triggering factor for annual growth rhythmicity in trees is the annual flood pulse of large river systems such as the Amazon and its tributaries [15
Peat swamp forests in SE-Asia are gaining public interest due to their fast deforestation, degradation and conversion into oil palm plantations and other agricultural systems, and the accompanying, often devastating, fires and massive greenhouse gas emissions (e.g., [17
]). In contrast to their importance for global carbon storage [19
], knowledge on the ecology, vegetation and growing conditions of swamp forests is rather poor. In particular, nothing is known about growth rates of tree species or carbon sequestration rates of intact forest stands in peat swamps. Tree ring research can be used for fast increment estimations to close this gap. However, as a precondition, basic research is needed to establish whether growth rings exist in trees of this unique forest type and if yes, whether they are annual or represent other environmental seasonalities. Finally, a possible climatic trigger for the annual growth ring formation needs to be indentified.
In hydrological terms, the peat swamp ecosystem seems to occupy an intermediate position between floodplains (e.g., in the Amazon [21
]) and non-flooded forests. The apparent surplus of water in the ecosystem throughout the year makes seasonal drought stress (that could induce rings) at the first view unlikely. The level of inundation in the rainy season is much lower than in floodplains, and seasonal flood stress as a growth ring triggering factor seems to be unlikely as well. In order to understand growth ring formation in Borneo’s peatlands, one has to understand the more general dynamics of growth ring formation in non-flooded tropical rainforests of the region.
The current paper represents the first ever study on tree rings in a tropical peat swamp forest. We demonstrate annual ring formation in the specific hydrological situation of a peat swamp forest in Borneo. We collected additional samples in an “indigenous” rubber tree plantation in a non-flooded site on mineral soil, for comparison.
The idea of a year-round uniform climate, an absence of seasons and consequently, absence of annual tree rings in the tropics belongs to the most persistent but wrong paradigms in ecology and forest science since the beginning of the 20th century. In reality, seasonal rainfall patterns with at least one distinct dry period are found over most parts of the tropics [5
]. In these regions, the existence of annual rings in trees has been confirmed for many species and ecosystems around the globe [10
]. In some parts of the world, e.g., in East Africa, a pronounced bimodal climate type prevails with two rainy and two dry seasons. Consequently, trees form two growth zones per year [34
Another triggering factor for an annual growth rhythm is seasonally occurring inundation in the floodplains of large river systems such as the Amazon, the Orinoco and their tributaries [5
]. In the floodplains, the inundation causes a lack of oxygen in the root system, followed by reduced root metabolism and water transport, leaf shedding and cambial dormancy, which are the basic processes for the formation of visible growth zones in the wood [37
Despite the numerous reports on tree rings in the tropics, the situation seems to be more complicated in Southeast Asia and Indian forests. While the pioneer of tropical dendrochronology, Coster [8
], clearly demonstrated the existence of annual tree rings for many species and different climate zones on Java, recent tree ring studies in Southeast Asia seem to be restricted mainly to teak [38
]. In studies of Malaysian forests, a lack of rings or the prevailing of indistinct rings are mentioned [39
], and the same is reported for India (e.g., [42
]), based on the interpretation of wood anatomical slides alone. In a recent study of Congo timber using wood anatomical properties, 40% of the evergreen species from moist forests were classified into the group of trees with distinct rings [26
]. For the Indonesian islands, besides Java, tree ring studies are completely lacking.
The monomodal rainfall pattern that characterizes Kalimantan typically generally induces seasonal tree growth and, consequently, annual tree rings. Hence, it would have been surprising if annual tree ring formation were confined to peat swamp forest alone. Considering that tree rings had never before been demonstrated for Kalimantan, we carried out a simple test by including a tree from a non-flooded rainforest in the study. With this, indeed, we could demonstrate the existence of annual rings in Hevea brasiliensis
trees from a plantation on mineral soil in West Kalimantan, as was shown earlier by Ogata et al. [43
] for young rubber trees in peninsular Malaysia. This aligns with findings from other tropical regions with similar or even wetter climate conditions, e.g., in a Cameroonian [44
] and in a Costa Rican lowland rainforest. At that latter site, the rainfall is higher and the dry period is shorter as in Kalimantan; nevertheless, some tree species shed their leaves in a period of a few weeks with little or no rainfall [45
] and form annual rings [46
In the peat swamp forests, however, the hydrological situation is comparatively complicated. The existence of pneumatophores and knee roots at our research site (and elsewhere) indicates high groundwater table and moderate flooding in certain periods of the year. Pneumatophores do not occur in ecosystems with high rising inundations as in the Amazon floodplains. The peat soils in the swamp forests of Borneo have a high water holding capacity which may dampen the effects of the dry periods on growth periodicity. However, during the dry period, the ground water table is lowering down considerably [20
]. This could be the explanation for a peak of leaf shedding in Shorea uliginosa
during the driest months in a peat forest on the west coast of Malaysia [48
The tree ring structures of different species from our research in the swamp forest suggest several conclusions:
A periodically occurring stress factor induces cambial dormancy and consequently tree rings in the wood.
Specifically, we found a varying distinctiveness of the growth zone boundaries, which is also observed for other sets of species from regions with a much more pronounced dry season [9
, we found distinct growth zones (Figure 4
b) based on density variations which obviously are not annual, but point to a stress factor with a non-annual periodicity.
, we demonstrated the existence of annual rings based on two independent methods. Successful radiocarbon based dating is supported by the similarity of time series from tree rings and precipitation (e.g., [49
]). The anatomical feature of growth ring boundaries is the parenchyma bands as limits of growth zones overlaying less frequent patterns of density variations.
In a next step, we measured ring width curves from all species with marginal parenchyma bands as a growth zone delimiter. The successful cross dating of the tree ring sequences from Magnolia and Calophyllum with those from Horsfieldia indicates the annual nature of the growth rings in these species as well. The time series from Knema intermedia is too short for significant statistical comparisons but visually shows a good congruence with Horsfieldia and Calophyllum samples and the dry period rainfall curve. These findings indicate that all species with marginal parenchyma bands form annual rings.
The discrepancy between the findings from Diospyros and those of other species needs an explanation. The anatomical structure of tree rings is genetically fixed—this determines the pattern of the different cell types and the structure at the tree ring boundaries. The presence or absence of a triggering factor, however, may determine whether tree rings are formed or not.
While Worbes et al. [50
] explained the indistinctiveness of growth zones in stem succulent tree species from Costa Rica with a very specific leaf-fall behavior and physiological adaptations, we hypothesize here that the reasons for the observed differences between the two tested species from the peat forest lie in the wood formation itself. The ring boundary in Knema
is formed by a single row of parenchyma cells. This marginal parenchyma is a widespread feature in many tropical families [49
]. The time for its formation at the end of a growing period might be relatively short [51
] and the inducing factor probably weak. The formation of dense wood at the end of a growth zone (late wood), however, is a process over a longer period where several cell rows must be involved for the ring to become visible.
From conifer species in the tropics, we know intra-annual density variations in the wood reflecting either rainfall events in the dry season [5
] or droughts in the rainy season [52
]. This indicates the process of late wood formation as a sensitive process, which, however, is not distinct when the stress is low, e.g., in very wet dry seasons.
The detailed analysis of the rainfall patterns indicates a pronounced and extended dry season of up to four months for our Kalimantan site in some years while in other years there is—by our definition of drought as months with less than 80 mm precipitation—no drought stress at all. However, in all years, a more or less pronounced decrease in precipitation happens in the dry season. The mean dry season month has only 50% of the rainfall of an average rainy season month. The number of distinct dry seasons since 1960 obviously coincides with the number of visible density variations in Diospyros. The more sensitive reaction of parenchyma band formation, however, is triggered by a moderate decrease in precipitation during the dry season months.
The long-term pattern of the growth curves shows a constant and very low growth level in Magnolia
. We assume that the sampled tree grew in the understorey under shadow during its entire life. For Horsfieldia
, we observed in the same time period a sudden release in three out of four randomly sampled discs. This is the consequence of an external influence, very likely the reduction of competition [53
]. Locals hinted that the studied forest was partially logged years before with a focus on the most valuable timber species (creaming).
Considering the results of our tree ring analysis and the radiocarbon dates, we measured very low growth rates of the tested species. Compared with tree ring based results from other tropical forests, Horsfieldia
attains less than 30–50% of the growth rates of tree species with similar wood density [54
]. The same is true for Magnolia
compared with understorey tree species of a Cameroonian lowland forest [55
]. This coincides with findings on general slow plant growth as a result of the low nutrient status of the peat soils and the complicated hydrological situation [56