Impact of Tree Litter Identity, Litter Diversity and Habitat Quality on Litter Decomposition Rates in Tropical Moist Evergreen Forest.

Background: Attempts to restore degraded highlands by tree planting are common in East Africa. However, up till now, little attention has been given to effects of tree species choice on litter decomposition and nutrient recycling. Method: In this study, three indigenous and two exotic tree species were selected for a litter decomposition study. The objective was to identify optimal tree species combinations and tree diversity levels for the restoration of degraded land via enhanced litter turnover. Litterbags were installed in June 2019 into potential restoration sites (disturbed natural forest and forest plantation) and compared to intact natural forest. The tested tree leaf litters included ve monospecic litters, ten mixtures of three species and one mixture of ve species. Standard green and rooibos tea were used for comparison. A total of 1033 litters were retrieved for weight loss analysis after one, three, six, and twelve months of incubation. Results: The nding indicates a signicant effect of both litter quality and litter diversity on litter decomposition. The nitrogen-xing native tree Millettia ferruginea showed a comparable decomposition rate as the fast decomposing green tea. The exotic conifer Cupressus lusitanica and the native recalcitrant Syzygium guineense have even a lower decomposition rate than the slowly decomposing rooibos tea. A signicant correlation was observed between litter mass loss and initial leaf litter chemical composition. Moreover, we found positive non-additive effects for litter mixtures including nutrient-rich and negative non-additive effects for litter mixtures including poor leaf litters respectively. Conclusion: These ndings suggest that both litter quality and litter diversity play an important role in decomposition processes and therefore in the restoration of degraded tropical moist evergreen forest. with


| Introduction
Although restoring degraded tropical forest land has received some attention in the past (Lamb, 1998;Holl et al., 2000;Lamb et al., 2005), much additional restoration work is needed to achieve the global protection area target (Mappin et al., 2019). So, efforts are still underway to restore tropical forests on human-disturbed lands across the globe (Fagan et al., 2020). In East Africa, Ethiopia has taken an ambitious commitment to restore 22 million ha of degraded land (Kassa, 2018). However, the success of secondary forest development will largely depend on soil fertility (Uriarte et al., 2015).
Tropical forest soils often function with a limited budget of essential nutrients, with the sustainable availability of these scarce resources largely depending on the e ciency of the nutrient recycling process. Undisturbed tropical forest soils can show nutrient recycling rates up to 2 to 5 times faster than grassland soils (Ochoa-Hueso et al., 2019).
In tropical forests, plant leaf litter is the major source of nutrients (Xue et al., 2019), supplying more than 70% of the tropical forest nutrient demand (Chapin et al., 2011). This shows both the e ciency and degree of dependency of tropical forest ecosystems on litter decomposition in order to sustain their exceptional biodiversity (Vitousek, 1984;Powers et al., 2009;Krishna and Mohan, 2017). In that regard, litter decomposition is a crucial process to rebuild soil organic matter on degraded tropical forest land and to ensure sustainable availability of nutrients for the recovering secondary forest (Cleveland et al., 2006;Tang et al., 2010). Despite its importance, there is still a limited knowledge on the factors affecting the litter decomposition process in tropical forest ecosystems (Giweta, 2020). , which may not be su cient to understand the in situ decomposition patterns of these forests (Hättenschwiler et al., 2011;Jacob et al., 2010). Indeed, high diversity of litter types has been shown to enhance litter decomposition rate via positive diversity effects, e.g. by promoting fungal diversity (Mori et al., 2020;Song et al., 2010). Furthermore, overall litter quality and diversity, climatic variation, decomposer community (Desie et Mayer, 2008) may have a strong effect on litter decomposition rates. Thus, it is crucial to understand which tree species, and in which composition, can stimulate the nutrient cycling in the Ethiopia Afromontane forest ecosystem, where there is currently active replacement going on of native tree species with exotics for economic purposes. Restoring the forest with species having contrasting leaf litter combinations might help to enhance the decomposition rate of the recently introduced exotic trees through induced functional traits that change antagonistic mechanisms into synergistic. In other words, competitors could become a facilitator of resource acquisition that increases leaf litter quality (Helsen et al., 2018(Helsen et al., , 2020. Finally, studies that have compared both native and exotic tree leaf litter mixtures with the standardized tea litters (Djukic et al., 2021) in both natural and strongly disturbed forest ecosystems are rare (Cizungu et al., 2014;Peh et al., 2012) In this study, we compared litter decomposition rates of native tree litter (Croton macrostachyus, Millettia ferruginea, Syzygium guineense), exotic tree litter (Eucalyptus globulus, Cupressus lusitanica) and the standard litter types of the 'tea bag index' (Keuskamp et al., 2013) during a one-year incubation period in the Afromontane moist evergreen forests of the Gerese district of SW-Ethiopia. The objectives were fourfold: (1) to measure the impact of species identity on litter decomposition rate, (2) to evaluate the effect of litter diversity on litter decomposition rate, (3) to examine the impact of local habitat quality on litter decomposition rate, and (4) to compare the litter decomposition rates of the indigenous and exotic trees with the standard green and rooibos tea. We hypothesize that (a) litter identity, litter diversity, and habitat quality play a signi cant role in litter decomposition rate and that (b) litter decomposition can be enhanced by restoring the degraded land with native rather than with exotic tree species.

| Study area
The study area is located in the SW of Ethiopia, close to the city of Arba Minch and the town of Gerese. It is geographically located in the range of 37.30 -37.33º E and 5.92º -5.95º N at an elevation between 2100 and 2525m above sea level. The area is characterized by year-round rainfall, estimated at 1700mm per year by Friis et al. (2010).
Heavy rainfall starts in March and extends up to May. The average monthly minimum and maximum temperature during the cold and warm months are 11°C and 24°C, respectively (EMA, 2018).
The forest in our study area is categorized as Afromontane moist evergreen forest (Friis et al., 2010) and covers an area of 363ha. The forest consists of three different forest habitat types, intact natural forest, disturbed natural forest, and plantation forest. The natural forest is characterized by a closed canopy dominated by old-growth trees and an abundance of smaller sized understory plants. The dominant tree species include Syzygium guineense, Macaranga capensis, Ocotea kenyensis and Ilex mitis. In the disturbed forest, native trees are scattered and mixed with planted exotic Eucalyptus globulus, Cupressus lusitanica and Grevillea robusta trees. This forest is characterized by large canopy gaps and the forest oor is partly covered by pioneer weeds and grasses. The local foresters use the gaps for exotic tree plantation as a restoration measure. As a result, the species composition in the degraded forest changed from natural to semi-natural. The plantation forest resulted from an attempt to protect the natural forest from human disturbance by an exotic tree plantation buffer (Eshetu, 2014;Eshetu and Högberg, 2000). This mature plantation is 46 years old and is mainly composed of Eucalyptus globulus. As it also serves as a source of wood for the local community, this buffer zone is not uniform and some parts are highly disturbed.

| Litter collection and decomposition assesment
Undamaged senescent leaves were collected between February and April 2019 from three indigenous (Croton macrostachyus, Millettia ferruginea and Syzygium guineense) and two exotic (Cupressus lusitanica and Eucalyptus globulus) tree species. The senescent leaves were picked by hand from accessible branches of multiple mother trees of each species that were randomly distributed across the three forest types. Both ecological and socio-economical criteria were used for the species selection. S. guineense is a dominant keystone species in the Gerese highland forest. Both C. macrostachyus and M. ferruginea have been used by local people as a source of medicine, fodder, sh poisoning, timber, and soil fertility. The other two exotic trees are both present in the study area, where they were mainly planted for their economic value. Standard Lipton green (EAN: 87 22700 05552 5) and rooibos teas (EAN: 87 22700 18843 8) were used for decomposition rate comparison with the selected indigenous and exotic tree leaf litters, following the standard 'tea bag index' of Keuskamp et al. (2013).
All litter samples were transported to the Arba Minch University botany lab (hereafter AMU) in tagged plastic bags and were dried at 70°C for a minimum of 48 hours. The dried litter was subsequently ground and sieved with a 2 mm and 500 µm mesh size sieve. The 2mm sieve was put on top of the 500µm sieve. Litter that passed the 2mm sieve but was trapped by the 500 µm sieve was used. For each individual leaf litter (3 indigenous and 2 exotic tree species), 2g ground leaf litter was transferred into empty nylon mesh tea bags and sealed with a plastic sealer. In addition, 2g mixture litters of two diversity levels were prepared from the 5 selected species (all ten possible 3-species combinations and the single 5-species combination). In the litter mixtures, all involved species contributed with an equal amount of litter. Together with the Lipton green and rooibos teas, this resulted in a total of 1080 litterbags from the 18 different litter types. Before installation, all litterbags were labeled with aluminum metal sheets indicating litter types, incubation period, and forest type. In each of the three forest types we subsequently selected ve random 3m × 3m plots where we installed four litterbags per litter type on June 12 2019, at 30cm intervals. Initial P, K, Ca, Mg, Na, Fe, Cu, Zn and Mn content of the grounded litter of each study species were quanti ed using Inductively Coupled Plasma Atomic Emission Spectroscopy. Laboratory procedure details are available in Appendix1.
One litterbag of each litter type was then retrieved after respectively one, three, six, and twelve months. Of the 1080 litterbags, only 1033 were successfully recovered. The other 47 litterbags were damaged by earthworms, termites or root penetration. All retrieved litterbags were transported to AMU for further processing. Adhering soil was manually removed from the litterbags with a test tube brush. The cleaned litterbags were dried for 48 hours at 70°C. After drying, the litterbags were opened and their content was transferred to an aluminum foil cap for weight loss analysis with a 0.01g precision electronic balance.

| Environmental data collection
In each plot, soil samples were taken from three points at 30 cm depth from the surface, which were merged into a composite sample for further analysis at AMU. The soil samples were air-dried at room temperature for a week and ground with mortar and pestle to pass through a 2.5 mm sieve. Ten grams of the sieved soil sample was used to prepare a 1:2.5 soil/water suspension to measure soil pH and electric conductivity using an HACH HQ40d digital multi-meter. The remaining soil macro-(i.e., P, K, Ca, Na, Mg) and micronutrient (i.e., Cu, Fe, Mn, Zn) analyses were carried out at KU Leuven, Belgium using an ICP-OES (Appendix 1). Other environmental parameters like soil temperature and moisture were measured once per month using a digital soil thermometer and TDR 150 soil moisture meter, respectively. Since soil moisture content was uniform across the three forest types during the study period, it was not used in the nal analyses. Location, elevation and aspect were collected for each plot using a Garmin handheld GPS 64S.

| Data analysis
We rst performed a principal component analysis (PCA) on all measured soil variables for all plots, to visualize potential differences in the soil chemical composition of the three forest types. Moreover, we estimated 'litter quality' of each litter mixture, prior to the decomposition experiment, as the percentage of evergreen leaves in the mixture (Table 1), since evergreen species are theoretically expected to have lower litter quality than deciduous species (Silva et al., 2015). Thus 'litter quality' acts as a predictor variable with ve levels (0%, 1/3 = 33%, 2/5 = 40%, 2/3 = 67%, 100%). For example, a litter mixture that contains litter from one evergreen and two deciduous tree species received a 'litter quality' of 33%. Before statistical analysis, the remaining litter mass was log-transformed for all litter types (Appendix 2), to satisfy model assumptions. We constructed an initial linear mixed model (LMM) for the log-transformed remaining litter mass of only the monospeci c litter types (response) as a function of forest type (plantation, degraded and natural), litter type (7 monospeci c litter types), observational period (after 1, 3, 6 and 12 months), and the interactions between litter type and forest type, and between litter type and observational period. Tukey multiple mean comparisons were also carried out among the seven litter types and the three forest types. A second LMM modelled the log-transformed remaining litter mass of all 17 litter types as a function of forest type, observation period, litter quality and litter diversity, including all rst-order interaction terms.
Since four litter mass measurements were taken from the same plot during the one-year incubation period, plot ID was included as a random factor in both LMMs. Where Mixture represents the mean mass loss of each mixture litter and Single the mean mass loss of unmixed, monospeci c litters.
One sample Student t-tests to mean zero were carried out to reveal the signi cance of potential non-additive effects of each monospeci c litter, where zero represents the additive effect in which both synergistic and antagonistic nonadditive effect cancel each other out. 3 | Results

Description of forest and litter types
The three forest types differed in their soil chemical composition (Fig. 2). Especially the natural forest showed larger variation and higher nutrient concentrations compared to both plantation and disturbed forest (Fig. 2).

| Monospeci c litter decomposition
Decomposition of the 402 retrieved monospeci c litter bags varied considerably, both within and across the four observational periods and 7 litter types. The litter mass remaining during the rst observational period ranged from 84.2 to 49.4% (Fig. 3A, Table S1). Litter type (species) had a signi cant effect on the remaining mass over time ( Table  2). Among the selected ve tree species, M. ferruginea demonstrates the fastest mass loss (Fig. 3A, Table S1). Compared to the standard green and rooibos teas, only M. ferruginea had a fast decomposition comparable to green tea (Fig. 3A, Table S1). Both C. lusitanica and S. guineese had slower decomposition than the slowly decomposing rooibos tea, whereas the decomposition rates of C. maycrostachyus and E. globulus were intermediate between the above two groups (Fig. 3A). The linear mixed model further showed that monospeci c litter decomposition was not different across the three forest types (Table 2).

| Effects of species richness and litter quality on decomposition
When including the 631 retrieval litter bags containing the species mixtures (11 additional litter types) to the linear mixed model, the overall remaining litter mass still decreased signi cantly with time, but at a varying pace for litter types of different litter quality (signi cant interaction). Decomposition was furthermore faster in the disturbed forest compared to the two other forest types (Table 3). Low quality litters (e.g. S. guineense and C. lusitanica) that were mixed with both the two high quality litter species C. macrostachyus and M. ferruginea showed higher decomposition rates than the same species mixed with only one high quality litter species, except for the litter mixture of M. ferruginea, E. globulus and C. lusitanica (Fig. S1). Lower decomposition rates were revealed for litter mixtures that contained low quality litters (e.g., S. guineense, C. lusitanica and E. globulus) (Fig. S1).
A linear mixed model on decomposition of all litter types based on only the results after 1 year showed a clear positive litter diversity effect on decomposition. As the number of species in the litter mixture increased from one to ve, the remaining litter mass signi cantly decreased (Fig. 4A). While the three forest types showed differences in litter decomposition in the model including all observational periods, decomposition was no longer signi cantly affected by forest type when considering only the last observational period (1 year) ( Table 4 and Fig. 4C).

| Relative individual performance (RIP)
The RIP was different across the ve study species (Fig. 5). A signi cantly negative RIP was observed for M. ferruginea (t 27 = 12.5, p < 0.001), indicating that this species has a higher decomposition rate as monospeci c litter than in mixtures, and thus relatively high litter quality (Fig. 5

| Net diversity effect (NDE)
All litter mixtures except CDE (S. guineense, E. globulus and C. lusitanica) showed a positive net diversity effect (Fig.  6). The highest net diversity effect was observed for those litter mixtures that combine both high and low-quality leaf litter (e.g. BCD and BCE in Fig. 6). The lowest net diversity effect was observed for those litter mixtures that incorporate only poor-quality litter (CDE in Fig. 6). All other leaf litters that were combined intermediate-quality groups demonstrated moderate, but always positive net diversity effects (Fig. 6).

| Soil and leaf litter nutrient concentration
The natural forest was richer in soil macro and micronutrient content compared to the plantation and the disturbed forest. This may be related to the diverse native plant community, which provides nutrient-rich leaf litters and/or is adapted to recycle nutrients more e ciently (Meier et al., 2005). Other studies also showed that leaf litterfall from diverse old-aged natural forest trees are an important source of soil nutrients ( Indeed, in our research, the initial leaf litter nutrient composition strongly varied among the ve selected trees species with native tree species having the highest litter quality. Soil nutrient content largely corresponding with forest type was found to be a signi cant factor in uencing decomposition during the early incubation period. This abiotic factor is probably important in relation to the short-term nutrient leaching processes in early-stage decomposition. Litter species identity was nevertheless the dominant factor affecting litter decomposition across all observational periods. Thus, solubility (leaching) and digestibility (microbial consumption) of leaf litter depend on both environmental (to a larger extent) and initial leaf litter cation concentration

| Conclusion And Recommendation
Three indigenous and two exotic tree leaf litters were incubated for one year in three different forest habitat types to investigate litter decomposition rates in comparison with green and rooibos tea references. The results indicate that both indigenous and exotic tree leaf litters strongly vary in litter quality. This variation plays a signi cant role in both monospeci c litter and litter mixture decomposition rates. The indigenous rich litter tree species M. ferruginea contributes considerably to the decomposition rate of nutrient-poor litter species. Both tree leaf litter types and litter mixture diversity showed a signi cant impact on litter decomposition rate. Habitat variation was expected to be a signi cant factor in litter decomposition rate but in our study, plantation forest and degraded natural forest habitats did not show a difference in litter decomposition rate, which differed signi cantly from the natural forest for all monoculture litters. and in the later decomposition stage, for all mixture litters. Rather, the number of rich litters in a mixture was found to be a good predictor for litter decomposition rate. Besides, the proportion of nutrient-poor litters also had a signi cant negative impact on the litter decomposition rate. In general, native nutrient-rich tree leaf litters play a signi cant role in enhancing soil organic matter and rehabilitating degraded forest land. However, special attention needs to be given to functional diversity. The presence of multiple species that play a similar role (e.g. functionally redundant litter mixture CDE) inhibits nutrient turnover and complementary effect. Thus, functional diversity is strongly recommended while applying the commonly used tree plantation approach to restore degraded forest land. Still, further research is needed to determine the optimal proportion of different functional groups.