Environmental changes in biodiversity hotspot ecosystems of
South Ecuador: RESPonse and feedback effECTs (FOR2730)

Abstract:

This study addresses transpiration in a tropical evergreen mountain forest in the Ecuadorian Andes from the leaf to the stand level, with emphasis on nocturnal plant-water relations. The stand level: Evapotranspiration (ET) measured over 12 months with the Eddy-Covariance (ECov) technique proved as the major share (79%) of water received from precipitation. Irre- spective of the humid climate, the vegetation transpired day and night. On average, 15.3% of the total daily ET were due to nocturnal transpiration. Short spells of drought increased daily ET, mainly by enhanced nighttime transpiration. Following leaf transpiration rather than air temperature and atmospheric water vapor deficit, ET showed its maximum already in the morning hours. The tree level: Due to the humid climate, the total water consumption of trees was generally low. Nevertheless, xylem sap flux measurements separated the investigated tree species into a group showing relatively high and another one with low sap flux rates. The leaf level: Transpiration rates of Tapirira guianensis, a member of the high- flux-rate group, were more than twice those of Ocotea aciphylla, a representative of the group showing low sap flux rates. Representatives of the Tapirira group operated at a rela- tively high leaf water potential but with a considerable diurnal amplitude, while the leaves of the Ocotea group showed low water potential and small diurnal fluctuations. Overall, the Tapirira group performed anisohydrically and the Ocotea group isohydrically. Grouping of the tree species by their water relations complied with the extents of the diurnal stem circum- ference fluctuations. Nighttime transpiration and hydrological type: In contrast to the isohy- drically performing trees of the Ocotea group, the anisohydric trees showed considerable water vapour pressure deficit (VPD)-dependent nocturnal transpiration. Therefore, we con- clude that nighttime ET at the forest level is mainly sourced by the tree species with aniso- hydric performance.

Abstract:

Few land-allocation models consider the impact of off-farm income on tropical deforestation. We provide a concept to integrate off-farm income in a mechanistic multiple-objective land-allocation model, while distinguishing between farms with and without re-allocation of on-farm labor to obtain off-farm income. On farms with re-allocation of labor we found that off-farm income reduced farmers’ financial dependency on deforestation- related agricultural income leading to less tropical deforestation. The influence of off-farm income covered two aspects: availability of additional income and re-allocation of on-farm labor to off-farm activities. The labor effect tended to reduce deforestation slightly more than the income effect. On farms without re-allocation of on-farm labor we showed how farmers can use off-farm income to purchase additional labor to accelerate deforestation. Our study highlights the importance of considering off-farm income in land-use models to better understand, model and possibly curb tropical deforestation.

Abstract:

1. How to simultaneously combat biodiversity loss and maintain ecosystem functioning while improving human welfare remains an open question. Optimization approaches have proven helpful in revealing the trade-offs between multiple functions and goals provided by land-cover configurations. The R package optim- Landuse provides tools for easy and systematic applications of the robust multiobjective land-cover composition optimization approach of Knoke et al. (2016). 2. The package includes tools to determine the land-cover composition that best balances the multiple functions a landscape can provide, and tools for understanding and visualizing the reasoning behind these compromises. A tutorial based on a published dataset guides users through the application and highlights possible use-cases. 3. Illustrating the consequences of alternative ecosystem functions on the theoretically optimal landscape composition provides easily interpretable information for landscape modelling and decision-making. 4. The package opens the approach of Knoke et al. (2016) to the community of landscape modellers and planners and provides opportunities for straightforward systematic or batch applications.

Abstract:

Nature Geoscience, doi:10.1038/s41561-022-00911-8

Abstract:

We introduce the FunAndes database, a compilation of functional trait data for the Andean flora spanning six countries. FunAndes contains data on 24 traits across 2,694 taxa, for a total of 105,466 entries. The database features plant-morphological attributes including growth form, and leaf, stem, and wood traits measured at the species or individual level, together with geographic metadata (i.e., coordinates and elevation). FunAndes follows the field names, trait descriptions and units of measurement of the TRY database. It is currently available in open access in the FIGSHARE data repository, and will be part of TRY’s next release. Open access trait data from Andean plants will contribute to ecological research in the region, the most species rich terrestrial biodiversity hotspot.

Abstract:

With increasing elevation, trees in tropical montane forests have to invest larger frac-tions of their resources into their fine roots in order to compensate for increasingly unfavorable soil conditions. It is unclear how elevation and related edaphic changes influence the variability in tree fine root traits and belowground functional diversity. We measured six fine root traits related to resource acquisition on absorptive fine roots of 288 trees from 145 species along an elevational gradient from 1000 m to 3000 m a.s.l. in tropical montane forests of the Ecuadorian Andes. We analyzed trait relation-ships with elevation and soil nutrient availability, and tested whether root functional diversity varied along these gradients. Fine roots at higher elevations and at more nutrient-poor sites were thicker, had higher tissue densities, and lower specific root length and nutrient concentrations than at lower elevations. These trends were diluted by the coexistence of tree species with a broad range of different root traits within communities particularly towards lower elevations, where root functional diversity was significantly higher. We conclude that nutrient limitation and potentially further adverse conditions at higher elevations are strong environmental filters that lead to trait convergence towards a conservative resource use strategy, whereas different trait syndromes are equally successful at lower elevations.

Abstract:

Understanding variation in tree functional traits along topographic gradients and through time provides insights into the processes that will shape community composition and determine ecosystem functioning. In montane environments, complex topography is known to affect forest structure and composition, yet its role in determining trait composition, indices on community climatic tolerances, and responses to changing environmental conditions has not been fully explored. This study investigates how functional trait composition (characterized as community-weighted moments) and community climatic indices vary for the tree community as a whole and for its separate demographic components (i.e., dying, surviving, recruiting trees) over eight years in a topographically complex tropical Andean forest in southern Ecuador. We identified a strong influence of topography on functional composition and on species’ climatic optima, such that communities at lower topographic positions were dominated by acquisitive species adapted to both warmer and wetter conditions compared to communities at upper topographic positions which were dominated by conservative cold adapted species, possibly due to differences in soil conditions and hydrology. Forest functional and climatic composition remained stable through time; and we found limited evidence for trait-based responses to environmental change among demographic groups. Our findings confirm that fine-scale environmental conditions are a critical factor structuring plant communities in tropical forests, and suggest that slow environmental warming and community-based processes may promote short-term community functional stability. This study highlights the need to explore how diverse aspects of community trait composition vary in tropical montane forests, and to further investigate thresholds of forest response to environmental change.

Abstract:

Sustainable intensification of agricultural lands might reconcile the conservation of tropical forest with food production,but in-depth assessments considering uncertainty and extreme values are missing. Uncertainty prohibits mapping probabilities to potential future states or ranking these states in terms of their likelihood. This in turn hampers the assessment of possible decision outcomes. Here, we use simulations to investigate how uncertainty may influence the social acceptability of alternative land-use strategies to halt tropical deforestation (including sustainable intensification), based on indicators representing farmer satisfaction. The results show how extreme values (worst values) for indicators of farmer satisfaction may undermine the adoption of sustainable intensification. We demonstrate that a pure forest conservation strategy leads to lower food production, but outperforms a sustainable intensification strategy that maintains food security. Pure forest conservation performed better, i.e., could secure higher farmer satisfaction, than sustainable intensification across a range of indicator groups. This suggests strong barriers to achieving sustainable intensification. Using agricultural subsidies breaks the dominance of pure forest conservation by enhancing the economic returns of sustainable intensification. We discuss the importance of access to labor and farmers’ preferences for the use of already cleared lands, which achieved the worst values under sustainable intensification and conclude that any assessment of land-use strategies requires careful consideration of uncertainty and extreme values.

Abstract:

Increased nutrient supply can have drastic effects on natural ecosystems, especially in naturally nutrient-poor ones such as most tropical rainforests. Many studies have focused on the reaction of trees to fertilization, but little is known about herbaceous plants. Ferns are a particularly common group in tropical forests, spanning all vegetation types and zones. Here, we assess how seven years of moderate addition of nitrogen (N), phosphorus (P), and N+P along an elevational gradient (1000–3000 m) have impacted richness and composition of fern and lycophyte assemblages in tropical montane rain forests growing on naturally nutrient deficient soils in the Ecuadorian Andes. We found that fertilization does not affect overall species richness, but that there were strong differences in species abundances (~60% of species), both negative and positive, that were apparently related to the systematic affiliations and ecological properties of the affected species. These diverse responses of ferns to fertilization provide insight into the sensitivity and complexity of the relationships of nutrient availability and community composition in tropical forests.

Abstract:

The “hierarchy of factors” hypothesis states that decomposition rates are controlled primarily by climatic, followed by biological and soil variables. Tropical montane forests (TMF) are globally important ecosystems, yet there have been limited efforts to provide a biome-scale characterization of litter decomposition. We designed a common litter decomposition experiment replicated in 23 tropical montane sites across the Americas, Asia, and Africa and combined these results with a previous study of 23 sites in tropical lowland forests (TLF). Specifically, we investigated (1) spatial heterogeneity in decomposition, (2) the relative importance of biological factors that affect leaf and wood decomposition in TMF, and (3) the role of climate in determining leaf litter decomposition rates within and across the TMF and TLF biomes. Litterbags of two mesh sizes containing Laurus nobilis leaves or birchwood popsicle sticks were spatially dispersed and incubated in TMF sites, for 3 and 7 months on the soil surface and at 10–15 cm depth. The within-site replication demonstrated spatial variability in mass loss. Within TMF, litter type was the predominant biological factor influencing decomposition (leaves > wood), with mesh and burial effects playing a minor role. When comparing across TMF and TLF, climate was the predominant control over decomposition,but the Yasso07 global model (based on mean annual temperature and precipitation) only modestly predicted decomposition rate. Differences in controlling factors between biomes suggest that TMF, with their high rates of carbon storage, must be explicitly considered when developing theory and models to elucidate carbon cycling rates in the tropics.

Abstract:

Biodiversity and ecosystem functions are highly threatened by global change. It has been proposed that geodiversity can be used as an easy-to-measure surrogate of biodiversity to guide conservation management. However, so far, there is mixed evidence to what extent geodiversity can predict biodiversity and ecosystem functions at the regional scale relevant for conservation planning. Here, we analyse how geodiversity computed as a compound index is suited to predict the diversity of four taxa and associated ecosystem functions in a tropical mountain hotspot of biodiversity and compare the results with the predictive power of environmental conditions and resources (climate, habitat, soil). We show that combinations of these environmental variables better explain species diversity and ecosystem functions than a geodiversity index and identified climate variables as more important predictors than habitat and soil variables, although the best predictors differ between taxa and functions. We conclude that a compound geodiversity index cannot be used as a single surrogate predictor for species diversity and ecosystem functions in tropical mountain rain forest ecosystems and is thus little suited to facilitate conservation management at the regional scale. Instead, both the selection and the combination of environmental variables are essential to guide conservation efforts to safeguard biodiversity and ecosystem functions.

Abstract:

An understanding of the genetic diversity and structure of plant species is essential in order to comprehend the degree of biodiversity loss and to develop successful restoration programs. Handroanthus is an important genus that presents one of the most valuable timbers of South America. Handroanthus chrysanthus is an important species distributed in Central and South America. Microsatellite markers are not previously developed for this species. Ten microsatellites for Handroanthus chrysanthus developed using high-throughput sequencing are presented here. The usefulness of these microsatellite loci for the genetic analysis of subspecies H. chrysanthus subsp. chrysanthus (distributed in coastal dry forests) and subspecies H. chrysanthus subsp. meridionalis (distributed in premontane moist forests) is analyzed. At least eight polymorphic microsatellites are useful for each subspecies, seven of which can be used in both subspecies.

Abstract:

In the past two decades, the Amazon-exposed, tropical montane rain forests in south Ecuador experienced increasing deposition of reactive N mainly from Amazonian forest fires, episodic Ca and Mg inputs from Saharan dust, and a low but constant P deposition from unknown sources. To explore the response of this tropical, perhumid ecosystem to nutrient inputs, we established in 2007 a Nutrient Manipulation Experiment (NUMEX). Since 2008, we have applied 50 kg ha−1 year−1 of N as urea, 10 kg ha−1 year−1 of P as NaH2PO4·H2O, 50 kg ha−1 year−1 of N + 10 kg ha−1 year−1 of P and 10 kg ha−1 year−1 of Ca as CaCl2·H2O in a randomized block design at 2000 m a.s.l. in a natural forest of the south Ecuadorian Andes. Previous studies have shown that alkali and alkaline earth metals had beneficial effects on the functioning of N and P co-limited tropical forests occurring on acidic soils. Therefore, we determined the response of all major aqueous ecosystem fluxes of K, Ca, Mg and Na to nutrient amendments, to understand how increasing atmospheric deposition would affect their cycling in the future. Additions of N and P decreased K leaching from the organic layer and in the mineral soil, thus tightening K cycling. This suggests that increasing future N and P availability may result in K limitation in the long term. The leaching of Ca and Mg from the canopy increased in response to amendments of N and P and we observed an enhanced uptake of these nutrients also if Ca was amended alone. Although N was applied as urea, acidity of soil solutions and leaching of K, Ca, Mg and Na did not increase following separate N amendments. In spite of the acid soils and of its low cation-exchange competitivity, Na included in the P fertilizer was only partly leached from the organic layer. We suggest that it was probably required to cover an unmet Na demand of the soil fauna. Our results demonstrate the major role in the functioning of the tropical montane forests played by K, Ca and Mg as potential future growth-limiting elements and increasingly required nutrients in response to rising N and P availability, while they also support the importance of Na as a functional element in these ecosystems.

Abstract:

Summary Community trait assembly in highly diverse tropical rainforests is still poorly understood. Based on more than a decade of field measurements in a biodiversity hotspot of southern Ecuador, we implemented plant trait variation and improved soil organic matter dynamics in a widely used dynamic vegetation model (the Lund-Potsdam-Jena General Ecosystem Simulator, LPJ-GUESS) to explore the main drivers of community assembly along an elevational gradient. In the model used here (LPJ-GUESS-NTD, where NTD stands for nutrient-trait dynamics), each plant individual can possess different trait combinations, and the community trait composition emerges via ecological sorting. Further model developments include plant growth limitation by phosphorous (P) and mycorrhizal nutrient uptake. The new model version reproduced the main observed community trait shift and related vegetation processes along the elevational gradient, but only if nutrient limitations to plant growth were activated. In turn, when traits were fixed, low productivity communities emerged due to reduced nutrient-use efficiency. Mycorrhizal nutrient uptake, when deactivated, reduced net primary production (NPP) by 61–72% along the gradient. Our results strongly suggest that the elevational temperature gradient drives community assembly and ecosystem functioning indirectly through its effect on soil nutrient dynamics and vegetation traits. This illustrates the importance of considering these processes to yield realistic model predictions.

Abstract:

The amount of carbon stored in deadwood is equivalent to about 8per cent of the global forest carbon stocks1. The decomposition of deadwood is largely governed by climate2–5 with decomposer groups—such as microorganisms and insects—contributing to variations in the decomposition rates2,6,7. At the global scale, the contribution of insects to the decomposition of deadwood and carbon release remains poorly understood7. Here we present a field experiment of wood decomposition across 55 forest sites and 6 continents. We find that the deadwood decomposition rates increase with temperature, and the strongest temperature effect is found at high precipitation levels. Precipitation affects the decomposition rates negatively at low temperatures and positively at high temperatures. As a net effect—including the direct consumption by insects and indirect effects through interactions with microorganisms—insects accelerate the decomposition in tropical forests (3.9% median mass loss per year). In temperate and boreal forests, we find weak positive and negative effects with a median mass loss of 0.9per cent and −0.1per cent per year, respectively. Furthermore, we apply the experimentally derived decomposition function to a global map of deadwood carbon synthesized from empirical and remote-sensing data, obtaining an estimate of 10.93.2petagram of carbon per year released from deadwood globally, with 93per cent originating from tropical forests. Globally, the net effect of insects may account for 29per cent of the carbon flux from deadwood, which suggests a functional importance of insects in the decomposition of deadwood and the carbon cycle.

Abstract:

Tropical dry forests are an intricate ecosystem with special adaptations to periods of drought. Arbuscular mycorrhizal fungi (AMF) are essential for plant survival in all terrestrial ecosystems but might be of even greater importance in dry forests as plant growth is limited due to nutrient and water deficiency during the dry season. Tropical dry forests in Ecuador are highly endangered, but studies about AMF communities are scarce. We investigated the AMF community of a premontane semi-deciduous dry forest in South Ecuador during the dry season. We estimated AMF diversity, distribution, and composition of the study site based on operational taxonomic units (OTUs) and compared the results to those from the tropical montane rainforest and páramo in South Ecuador. OTU delimitation was based on part of the small ribosomal subunit obtained by cloning and Sanger sequencing. Nearly all OTUs were Glomeraceae. The four frequent OTUs were Glomus, and comparison with the MaarjAM database revealed these to be globally distributed with a wide range of ecological adaptations. Several OTUs are shared with virtual taxa from dry forests in Africa. Ordination analysis of AMF communities from the tropical dry and montane rainforests in South Ecuador revealed a unique AMF community in the dry forest with only few overlapping OTUs. Most OTUs that were found in both dry and rainforests and on the two continents were globally distributed Glomus.

Abstract:

Screening species-rich communities for the variation in functional raits along environmental gradients may help understanding the abiotic drivers of plant performance in a mechanistic way. We investigated tree leaf trait variation along an elevation gradient (1000–3000 m) in highly diverse neotropical montane forests to test the hypothesis that elevational trait change reflects a trend toward more conservative resource use strategies at higher elevations, with interspecific trait variation decreasing and trait integration increasing due to environmental filtering. Analysis of trait variance partitioning across the 52 tree species revealed for most traits a dominant influence of phylogeny, except for SLA, leaf thickness and foliar Ca, where elevation was most influential. The community-level means of SLA, foliar N and Ca, and foliar N/P ratio decreased with elevation, while leaf thickness and toughness increased. The contribution of intraspecific variation was substantial at the community level in most traits, yet smaller than the interspecific component. Both within-species and between-species trait variation did not change systematically with elevation. High phylogenetic diversity, together with small-scale edaphic heterogeneity, cause large interspecific leaf trait variation in these hyper-diverse Andean forests. Trait network analysis revealed increasing leaf trait integration with elevation, suggesting stronger environmental filtering at colder and nutrient-poorer sites.

Abstract:

Theory predicts positive effects of species richness on the productivity of plant communities through complementary resource use and facilitative interactions between species. Results from manipulative experiments with tropical tree species indicate a positive diversity–productivity relationship (DPR), but the existing evidence from natural forests is scarce and contradictory. We studied forest aboveground productivity in more than 80 humid tropical montane oldgrowth forests in two highly diverse Andean regions with large geological and topographic heterogeneity and related productivity to tree diversity and climatic, edaphic and stand structural factors with a likely influence on productivity. Main determinants of wood production in the perhumid study regions were elevation (as a proxy for temperature), soil nutrient (N, P and base cation) availability and forest structural parameters (wood specific gravity, aboveground biomass). Tree diversity had only a small positive influence on productivity, even though tree species numbers varied largely (6–27 species per 0.04 ha). We conclude that the productivity of highly diverse Neotropical montane forests is primarily controlled by thermal and edaphic factors and stand structural properties, while tree diversity is of minor importance.

Abstract:

A new satellite-based algorithm for rainfall retrieval in high spatio-temporal resolution fo Ecuador is presented. The algorithm relies on the precipitation information from the Integrated Multi-SatEllite Retrieval for the Global Precipitation Measurement (GPM) (IMERG) and infrared (IR) data from the Geostationary Operational Environmental Satellite-16 (GOES-16). It wa developed to (i) classify the rainfall area (ii) assign the rainfall rate. In each step, we selected the most important predictors and hyperparameter tuning parameters monthly. Between 19 April 2017 and 30 November 2017, brightness temperature derived from the GOES-16 IR channels and ancillary geo-information were trained with microwave-only IMERG-V06 using random forest (RF). Validation was done against independent microwave-only IMERG-V06 information not used for training. The validation results showed the new rainfall retrieva technique (multispectral) outperforms the IR-only IMERG rainfall product. This offers using the multispectral IR data can improve the retrieval performance compared to single-spectrum IR approaches. The standard verification scored a median Heidke skill score of ~0.6 for the rain area delineation and R between ~0.5 and ~0.62 for the rainfall rate assignment, indicating uncertainties for Andes’s high elevation. Comparison of RF rainfall rates in 2 km2 resolution with daily rain gauge measurements reveals the correlation of R = ~0.33.

Abstract:

Background: Since the 1990’s, afforestation programs in the páramo have been implemented to offset carbon emissions through carbon sequestration, mainly using pine plantations. However, several studies have indicated that after the establishment of pine plantations in grasslands, there is an alteration of carbon pools including a decrease of the soil organic carbon (SOC) pool. The aim of this study is to investigate the impact of the establishment of pine plantations on the carbon stocks in different altitudes of the páramo ecosystem of South Ecuador. Results: At seven locations within an elevational gradient from 2780 to 3760 m a.s.l., we measured and compared carbon stocks of three types of land use: natural grassland, grazed páramo, and Pinus patula Schlltdl. & Cham. plantation sites. For a more accurate estimation of pine tree carbon, we developed our own allometric equations. There were significant (p < 0.05) differences between the amounts of carbon stored in the carbon pools aboveground and belowground for the three types of land use. In most of the locations, pine plantations revealed the highest amounts of aboveground and belowground carbon (55.4 and 6.9 tC/ha) followed by natural grassland (23.1 and 2.7 tC/ha) and grazed páramo sites (9.1 and 1.5 tC/ha). Concerning the SOC pools, most of the locations revealed significant lower values of plantations’ SOC in comparison to natural grassland and grazed páramo sites. Higher elevation was associated with lower amounts of pines’ biomass. Conclusions: Even though plantations store high amounts of carbon, natural páramo grassland can also store substantial amounts above and belowground, without negatively affecting the soils and putting other páramo ecosystem services at risk. Consequently, plans for afforestation in the páramo should be assessed case by case, considering not only the limiting factor of elevation, but also the site quality especially affected by the type of previous land use.

Abstract:

Morphological traits provide the interface between species and their environment. For example, body size affects the fitness of individuals in various ways. Yet especially for ectotherms, the applicability of general rules of interspecific clines of body size and even more so of other morphological traits is still under debate. Here we tested relationships between elevation (as a proxy for temperature) and productivity with four ecologically relevant morphological traits of orthopteran assemblages that are related to fecundity (body size), dispersal (wing length), jumping ability (hind femur length), and predator detection (eye size). We measured traits of 160 orthopteran species that were sampled along an extensive environmental gradient at Mt. Kilimanjaro (Tanzania), spanning elevations from 790 to 4,410 m above sea level (a.s.l.) with different levels of plant productivity. For traits other than body size, we calculated the residuals from a regression on body length to estimate the variation of traits irrespective of body size. Bayesian analyses revealed that mean body size of assemblages, as well as the means of relative wing length, hind femur length, and eye size, decreased with increasing elevation. Body size and relative eye size also decreased with increasing productivity. Both phylogenetic relationships, as well as species‐specific adaptations, contributed to these patterns. Our results suggest that orthopteran assemblages had higher fecundity and better dispersal and escape abilities, as well as better predator detection at higher temperatures (low elevations) than at low temperatures (high elevations). Large body sizes might be advantageous in habitats with low productivity because of a reduced risk of starvation. Likewise, large eye size might be advantageous because of the ability to detect predators in habitats with low vegetation cover, where hiding possibilities are scarce. Our study highlights that changes in temperature and productivity not only lead to interspecific changes in body size but are also related to independent changes of other morphological traits that influence the ecological fit of organisms in their environment.

Abstract:

Habitat loss and landscape fragmentation are important drivers of changes in biodiversity. In fragmented landscapes, bird species are able to use multiple forest patches that may therefore share an important portion of the regional biodiversity. In turn, these patches are linked through their shared bird diversity (i.e. species-habitat networks). Identifying the importance of nodes (e.g. forest patches) in species-habitat networks is increasingly important to improve conservation planning. Within this network approach, the relative importance of patches for birds can be identified via centrality indices-measures of the magnitude of shared bird diversity of patches across the entire network (i.e. patch centrality). Here, we tested for changes in patch centrality within bird species-patch networks of two habitat guilds, forest specialists and generalists, in relation to patch area, patch shape irregularity and within-patch habitat characteristics across 15 Polylepis woodland patches in a páramo landscape of southern Ecuador. Patch centrality for forest specialists decreased with greater influence of surrounding páramo plants, i.e. an increasing proportion of bunch-grasses and small shrubs, in the within-patch habitat and was unaffected by either patch area or patch shape irregularity. On the other hand, patch centrality for generalists was positively influenced by patch shape irregularity but was unaffected by patch area or the influence of surrounding páramo plants in the within-patch habitat. Patch centrality reveals that the relative importance of Polylepis woodlands lies in their habitat quality. Forest specialists are dependent on mature Polylepis woodland patches, while generalists benefit from the natural irregular shape of the woodlands. Finally, a species-habitat network approach facilitates the recognition of important Polylepis patches and their characteristics for conservation of the Andean bird community at a landscape scale.

Abstract:

The Random Forest (RF) algorithm, a decision-tree-based technique, has become a promising approach for applications addressing runoff forecasting in remote areas. This machine learning approach can overcome the limitations of scarce spatio-temporal data and physical parameters needed for process-based hydrological models. However, the influence of RF hyperparameters is still uncertain and needs to be explored. Therefore, the aim of this study is to analyze the sensitivity of RF runoff forecasting models of varying lead time to the hyperparameters of the algorithm. For this, models were trained by using (a) default and (b) extensive hyperparameter combinations through a grid-search approach that allow reaching the optimal set. Model performances were assessed based on the R2, %Bias, and RMSE metrics. We found that: (i) The most influencing hyperparameter is the number of trees in the forest, however the combination of the depth of the tree and the number of features hyperparameters produced the highest variability-instability on the models. (ii) Hyperparameter optimization significantly improved model performance for higher lead times (12- and 24-h). For instance, the performance of the 12-h forecasting model under default RF hyperparameters improved to R2 = 0.41 after optimization (gain of 0.17). However, for short lead times (4-h) there was no significant model improvement (0.69 &lt; R2 &lt; 0.70). (iii) There is a range of values for each hyperparameter in which the performance of the model is not significantly affected but remains close to the optimal. Thus, a compromise between hyperparameter interactions (i.e., their values) can produce similar high model performances. Model improvements after optimization can be explained from a hydrological point of view, the generalization ability for lead times larger than the concentration time of the catchment tend to rely more on hyperparameterization than in what they can learn from the input data. This insight can help in the development of operational early warning systems.