Ecological Monographs

Valavi, Roozbeh; Guillera‐Arroita, Gurutzeta; Lahoz‐Monfort, José J.; Elith, Jane

doi: 10.1002/ecm.1486pmid: N/A

Species distribution modeling (SDM) is widely used in ecology and conservation. Currently, the most available data for SDM are species presence‐only records (available through digital databases). There have been many studies comparing the performance of alternative algorithms for modeling presence‐only data. Among these, a 2006 paper from Elith and colleagues has been particularly influential in the field, partly because they used several novel methods (at the time) on a global data set that included independent presence–absence records for model evaluation. Since its publication, some of the algorithms have been further developed and new ones have emerged. In this paper, we explore patterns in predictive performance across methods, by reanalyzing the same data set (225 species from six different regions) using updated modeling knowledge and practices. We apply well‐established methods such as generalized additive models and MaxEnt, alongside others that have received attention more recently, including regularized regressions, point‐process weighted regressions, random forests, XGBoost, support vector machines, and the ensemble modeling framework biomod. All the methods we use include background samples (a sample of environments in the landscape) for model fitting. We explore impacts of using weights on the presence and background points in model fitting. We introduce new ways of evaluating models fitted to these data, using the area under the precision‐recall gain curve, and focusing on the rank of results. We find that the way models are fitted matters. The top method was an ensemble of tuned individual models. In contrast, ensembles built using the biomod framework with default parameters performed no better than single moderate performing models. Similarly, the second top performing method was a random forest parameterized to deal with many background samples (contrasted to relatively few presence records), which substantially outperformed other random forest implementations. We find that, in general, nonparametric techniques with the capability of controlling for model complexity outperformed traditional regression methods, with MaxEnt and boosted regression trees still among the top performing models. All the data and code with working examples are provided to make this study fully reproducible.

Bozec, Yves‐Marie; Hock, Karlo; Mason, Robert A. B.; Baird, Mark E.; Castro‐Sanguino, Carolina; Condie, Scott A.; Puotinen, Marji; Thompson, Angus; Mumby, Peter J.

doi: 10.1002/ecm.1494pmid: N/A

Cumulative impacts assessments on marine ecosystems have been hindered by the difficulty of collecting environmental data and identifying drivers of community dynamics beyond local scales. On coral reefs, an additional challenge is to disentangle the relative influence of multiple drivers that operate at different stages of coral ontogeny. We integrated coral life history, population dynamics, and spatially explicit environmental drivers to assess the relative and cumulative impacts of multiple stressors across 2,300 km of the world’s largest coral reef ecosystem, Australia’s Great Barrier Reef (GBR). Using literature data, we characterized relationships between coral life history processes (reproduction, larval dispersal, recruitment, growth, and mortality) and environmental variables. We then simulated coral demographics and stressor impacts at the organism (coral colony) level on >3,800 individual reefs linked by larval connectivity and exposed to temporally and spatially realistic regimes of acute (crown‐of‐thorns starfish outbreaks, cyclones, and mass coral bleaching) and chronic (water‐quality) stressors. Model simulations produced a credible reconstruction of recent (2008–2020) coral trajectories consistent with monitoring observations, while estimating the impacts of each stressor at reef and regional scales. Overall, simulated coral populations declined by one‐third across the GBR, from an average of ~29% to ~19% hard coral cover. By 2020, <20% of the GBR had coral cover higher than 30%, a status of reef health corroborated by scarce and sparsely distributed monitoring data. Reef‐wide annual rates of coral mortality were driven by bleaching (48%) ahead of cyclones (41%) and starfish predation (11%). Beyond the reconstructed status and trends, the model enabled the emergence of complex interactions that compound the effects of multiple stressors while promoting a mechanistic understanding of coral cover dynamics. Drivers of coral cover growth were identified; notably, water quality (suspended sediments) was estimated to delay recovery for at least 25% of inshore reefs. Standardized rates of coral loss and recovery allowed the integration of all cumulative impacts to determine the equilibrium cover for each reef. This metric, combined with maps of impacts, recovery potential, water‐quality thresholds, and reef state metrics, facilitates strategic spatial planning and resilience‐based management across the GBR.

doi: 10.1002/ecm.1461pmid: N/A

Wang, Ming‐Qiang; Yan, Chuan; Luo, Arong; Li, Yi; Chesters, Douglas; Qiao, Hui‐Jie; Chen, Jing‐Ting; Zhou, Qing‐Song; Ma, Keping; Bruelheide, Helge; Schuldt, Andreas; Zhang, Zhibin; Zhu, Chao‐Dong

Stump, Simon Maccracken; Song, Chuliang; Saavedra, Serguei; Levine, Jonathan M.; Vasseur, David A.

doi: 10.1002/ecm.1493pmid: N/A

Intraspecific trait variation (ITV) is a widespread feature of life, but it is an open question how ITV affects between‐species coexistence. Recent theoretical studies have produced contradictory results, with ITV promoting coexistence in some models and undermining coexistence in others. Here we review recent work and propose a new conceptual framework to explain how ITV affects coexistence between two species. We propose that all traits belong to one of two categories: niche traits and hierarchical traits. Niche traits determine an individual's location on a niche axis or trade‐off axis, such that changing an individual's trait makes it perform better in some circumstances and worse in others. Hierarchical traits represent cases where conspecifics with different traits have the same niche, but one performs better under all circumstances, such that there are winners and losers. Our framework makes predictions for how intraspecific variation in each type of trait affects coexistence by altering stabilizing mechanisms and fitness differences. For example, ITV in niche traits generally weakens the stabilizing mechanism, except when it generates a generalist–specialist trade‐off. On the other hand, hierarchical traits tend to impact competitors differently, such that ITV in one species will strengthen the stabilizing mechanism while ITV in the other species will weaken the mechanism. We re‐examine 10 studies on ITV and coexistence, along with four novel models, and show that our framework can explain why ITV promotes coexistence in some models and undermines coexistence in others. Overall, our framework reconciles what were previously considered to be contrasting results and provides both theoretical and empirical directions to study the effect of ITV on species coexistence.

Maciá‐Vicente, Jose G.; Popa, Flavius

doi: 10.1002/ecm.1489pmid: N/A

Root‐colonizing fungi form species‐rich assemblages with key functions in principal ecosystem processes, making them prospectively important players in conservation and applied ecology. Harnessing the processes and services they drive requires a better understanding of their patterns of diversity and community structure, and how these link to function. Here, we search for possible adaptations to contrasting environmental and host conditions, indicative of participation in habitat‐specific processes. We surveyed heathland and grassland habitats across a latitudinal gradient in Western Europe, using a spatially explicit design to assess community variation at scales from centimeters, to thousands of kilometers. Root‐associated fungi assemble into strongly site‐specific communities irrespective of habitat type, shaped by environmental factors and spatial distance operating at different scales, but also by a high level of endemism, likely to be determined by local stochastic processes such as drift and dispersal limitation at short distances. Despite the high site specificity in communities, they are dominated everywhere by a core set of lineages with little preferences toward habitat conditions or host phylogeny. Our results suggest a convergent evolution across phylogenetically distant lineages toward the root‐colonizing habit, and a functional redundancy in strategies for habitat colonization and host interaction. Further efforts are needed to integrate functional trait composition in future community ecology studies of root‐colonizing fungi.

Gottschall, Felix; Cesarz, Simone; Auge, Harald; Kovach, Kyle R.; Mori, Akira S.; Nock, Charles A.; Eisenhauer, Nico

doi: 10.1002/ecm.1490pmid: N/A

There is increasing evidence that spatial and temporal dynamics of biodiversity and ecosystem functions play an essential role in biodiversity–ecosystem‐functioning (BEF) relationships. Despite the known importance of soil processes for forest ecosystems, belowground functions in response to tree diversity and spatiotemporal dynamics of ecological processes and conditions remain poorly described. We propose a novel conceptual framework integrating spatiotemporal dynamics in BEF relationships and hypothesized a positive tree species richness effect on soil ecosystem functions through the spatial and temporal stability of biotic and abiotic soil properties based on species complementarity and asynchrony. We tested this framework within a long‐term tree diversity experiment in Central Germany by assessing soil ecosystem functions (soil microbial properties and litter decomposition) and abiotic variables (soil moisture and surface temperature) for two consecutive years in high spatial and temporal resolution. Tree species richness and identity had significant effects on soil properties (e.g., soil microbial biomass). Structural equation modeling revealed that overall soil microbial biomass was partly explained by (1) enhanced temporal stability of soil surface temperature and (2) decreased spatial stability of soil microbial biomass. Overall, spatial stability of soil microbial properties was positively correlated with their temporal stability. These results suggest that spatiotemporal dynamics are indeed crucial determinants in BEF relationships and highlight the importance of vegetation‐induced microclimatic conditions for stable provisioning of soil ecosystem functions and services.

Turner, Monica G.; Braziunas, Kristin H.; Hansen, Winslow D.; Hoecker, Tyler J.; Rammer, Werner; Ratajczak, Zak; Westerling, A. Leroy; Seidl, Rupert

doi: 10.1002/ecm.1485pmid: N/A

As temperatures continue rising, the direction, magnitude, and tempo of change in disturbance‐prone forests remain unresolved. Even forests long resilient to stand‐replacing fire face uncertain futures, and efforts to project changes in forest structure and composition are sorely needed to anticipate future forest trajectories. We simulated fire (incorporating fuels feedbacks) and forest dynamics on five landscapes spanning the Greater Yellowstone Ecosystem (GYE) to ask the following questions: (1) How and where are forest landscapes likely to change with 21st‐century warming and fire activity? (2) Are future forest changes gradual or abrupt, and do forest attributes change synchronously or sequentially? (3) Can forest declines be averted by mid‐21st‐century stabilization of atmospheric greenhouse gas (GHG) concentrations? We used the spatially explicit individual‐based forest model iLand to track multiple attributes (forest extent, stand age, tree density, basal area, aboveground carbon stocks, dominant forest types, species occupancy) through 2100 for six climate scenarios. Hot‐dry climate scenarios led to more fire, but stand‐replacing fire peaked in mid‐century and then declined even as annual area burned continued to rise. Where forest cover persisted, previously dense forests were converted to sparse young woodlands. Increased aridity and fire drove a ratchet of successive abrupt declines (i.e., multiple annual landscape‐level changes ≥20%) in tree density, basal area, and extent of older (>150 yr) forests, whereas declines in carbon stocks and mean stand age were always gradual. Forest changes were asynchronous across landscapes, but declines in stand structure always preceded reductions in forest extent and carbon stocks. Forest decline was most likely in less topographically complex landscapes dominated by fire‐sensitive tree species (Picea engelmannii, Abies lasiocarpa, Pinus contorta var. latifolia) and where fire resisters (Pseudotsuga menziesii var. glauca) were not already prevalent. If current GHG emissions continue unabated (RCP 8.5) and aridity increases, a suite of forest changes would transform the GYE, with cascading effects on biodiversity and myriad ecosystem services. However, stabilizing GHG concentrations by mid‐century (RCP 4.5) would slow the ratchet, moderating fire activity and dampening the magnitude and rate of forest change. Monitoring changes in forest structure may serve as an operational early warning indicator of impending forest decline.

doi: 10.1002/ecm.1460pmid: N/A

Firkowski, Carina R.; Thompson, Patrick L.; Gonzalez, Andrew; Cadotte, Marc W.; Fortin, Marie‐Josée

doi: 10.1002/ecm.1484pmid: N/A

Environmental fluctuations influence patterns of synchrony and stability in species abundances. Most of our understanding of synchrony and stability stems from competitive community and metacommunity ecology, when in reality species interact in more complex ways. Therefore, there is a mounting need for the integration of multi‐trophic interactions into metacommunity ecology. In particular, knowledge is lacking on: (1) whether synchrony and stability respond to environmental fluctuations similarly under competitive and multi‐trophic metacommunities; (2) how synchrony and stability change across the hierarchical levels of a metacommunity; and (3) whether trophic groups differ in their contributions to synchrony and stability. Here, we address these questions through a complementary approach, using model simulations to derive theoretical expectations for the effects of environmental fluctuations on synchrony and stability, and a microcosm experiment to test observations against these expectations. We created spatially heterogeneous metacommunities populated by eight protist and one rotifer species organized in a multi‐trophic food‐web. We controlled environmental fluctuations so that they were spatially uncorrelated and species were assumed to respond differently to environmental conditions. We contrasted the control of constant environmental conditions to the effects of periodic environmental fluctuations. We show that environmental fluctuations can reduce synchrony between patches and increase stability, but can also decouple asynchrony between species and increase population and metapopulation variability. We discuss how some of these findings apply to both competitive and multi‐trophic metacommunities but changes are stronger in multi‐trophic metacommunities, and how trophic groups differ in their contributions to synchrony and variability.