Thresholds to Restoration

Science Center Objects

Mesic forests of Hawai‘i island provide an ideal system for the study of forest restoration because they have a similar history to other tropical and subtropical forests globally, while maintaining a relatively simple species assemblage.  Many of these forests were cleared for grazing, and then later abandoned, to become dominated by pasture grasses that form competitive layers such that native woody species have difficulty reestablishing after the initial disturbance is removed. Our research goals are to explore where ecosystem thresholds to Hawaiian forest restoration lie, and the feedback mechanisms that create these thresholds. In addition, we hope to expand our understanding of ASE (Alternative Stable Equilibrium) theory by evaluating how single versus multiple interacting factors influence the lack of recovery of degraded tropical forests.

Aerial seed trap in an ohia tree

Seedrain trap in the canopy of an ‘ōhi‘a tree being used to measure seed dispersal by forest birds. Photo: S. Yelenik


Ecologists are increasingly recognizing that disturbed landscapes often do not undergo natural succession back to a native-dominated state simply because the original stressors (e.g., deforestation, grazing) have been removed. This may be because habitats are now exhibiting alternative stable equilibrium (ASE). In contrast to succession, ASE theory suggests that more than one state of species composition (e.g., native-dominated versus exotic-dominated) is possible in a single environment depending on the order that species arrive, and that states do not move easily from one to the other.

Mesic forests of Hawaii Island are dominated by two canopy trees, nitrogen-fixing koa (Acacia koa), and non-fixing ʻōhiʻa (Metrosideros polymorpha), with a dense understory of subcanopy trees, shrubs and ferns. Like many forests worldwide, Hawaiian forests were cleared and planted with exotic pasture grasses for livestock grazing. While large forest areas were cleared of all woody species, in other areas, some canopy trees were left as shelter for cattle. Exotic pasture grasses remain persistent after grazer removal, suggesting ASE.

Theoretically, phase shifts can be forced by altering state variables, such as mass outplantings of native plant species, but empirical data to support these ideas are lacking for most terrestrial systems. In addition, we are not clear on where these thresholds reside, which is important information for managers trying to best utilize scarce resources for activities such as native plant restoration. Our research goals are to explore where ecosystem thresholds to Hawaiian forest restoration lie, and the feedback mechanisms that create these thresholds. In addition, we hope to expand our understanding of ASE theory by evaluating how single versus multiple interacting factors influence the lack of recovery of degraded tropical forests.

Project Objectives:

Technician surveys for plants in the understory

Contractor Eli Rose surveys plant communities under remnant koa and ‘ōhi‘a trees at Hakalau Forest National Wildlife Refuge. Photo: M. Stewart

We propose that prior to forest degredation three positive feedbacks previously existed in this system and that in their absence, exotic grasses remain dominant and succession does not occur towards native understory woody species. These feedbacks were between 1) understory shrubs and frugivorous birds, 2) dominant native plants and litter layers and 3) understory species and mutualistic mycorrhizal fungi. We will use surveys and field experiments to tease apart the relative importance of these factors and to identify thresholds needed for native ecosystem recovery using sites at the Hakalau Forest National Wildlife Refuge where long-term recovery is evident under some remnant ʻōhiʻa trees. Understanding the natural recruitment patterns, or lack thereof, will help guide restoration of these forests, which are critical habitat for several threatened and endangered plant and bird species.

Using degraded habitat patches that vary in overstory tree species and understory states, we will evaluate how the contributions of avian seed dispersal, exotic grass competition, canopy-litter-understory feedbacks and the composition of mycorrhizal fungi influence the recovery of degraded Hawaiian mesic forests. Our goal is to understand whether these factors act as feedbacks and if so, if manipulation of individual feedbacks can set succession in motion, whether threshold dynamics exist for certain feedbacks, and whether all factors need to be manipulated simultaneously for succession to occur.

In habitat patches with varying understory density, we will monitor aerial seedrain from birds, seedling emergence and survival, litter and soil nutrient dynamics, and mycorrhizal identity and abundance. In addition, we will use experimental manipulations of grass biomass and seedrain to explore where thresholds to seedling emergence lie. We will use detailed demographic data to explore models of thresholds on seedling emergence and survival, and thus ecosystem succession or lack thereof.

Highlights and Key Findings:

Preliminary results suggest that high levels of understory are needed before thresholds to restoration are achieved. For example, it is not until the highest understory densities that exotic grasses drop out of the ecosystem naturally. In addition, these patterns are only ever found under ʻōhiʻa, never koa, suggesting that the high nitrogen environment under koa trees exacerbates exotic grass biomass and vigor. Finally, a student project (REU) over the summer suggests that tree architecture of ohia can influence whether understory seedlings are probable to germinate and survive. Specifically, trees with complex architecture that offer somewhat horizontal surfaces for seed to land and germinate on the tree itself tend to have more understory regeneration. These patterns are also highly influenced by neighborhood tree density.

Understory plant communities and soil properties:

Remnant and/or afforestation koa on its own is not facilitating passive regeneration of native woody understory in Hakalau Forest NWR. In contrast, remnant ‘ōhi‘a trees once cleared of understory show some spontaneous native understory recruitment, although this was highly variable among and within sites. It is possible that sites closer to intact forest have greater seed input, and thus greater understory development. Seedtrap data will help resolve this in the future. Grass cover remains high in all sites, ranging from 40 to 100% under ‘ōhi‘a, and 70 to 100% under koa. There was a significant negative relationship between grass and native understory woody cover, suggesting that grass is limiting native woody understory regeneration. Grass removal experiments will help resolve this question in the future.

Soil properties, grass biomass, and litter biomass also differed as a function of canopy species. ‘Ōhi‘a trees had higher pools of standing litter, less grass biomass with lower foliar N content, higher soil organic matter content, and lower inorganic soil N than koa trees.  Woody species’ litter and grass biomass were negatively correlated, suggesting that a build-up of litter under ‘ōhi‘a trees may facilitate understory regeneration, either by negatively affecting grasses and/or creating substrate that is beneficial for native seedling germination and survival. In contrast, soil nitrogen was greatest under koa, which also had greater grass biomass. This suggests that koa increases soil N over time, facilitating grass growth, which is hampering native understory regeneration.

Bird-mediated seed dispersal:

Seedrain samples collected from 18th May to 23rd December 2015, contained 3,717 bird dispersed seeds from six native species (Cheirodendron trigynumCoprosma spp., Ilex anomalaStyphelia tameiameiaeRubus hawaiensisVaccinium spp.) and one non-native species (Rubus argutus). Monthly averages showed clear signals of reflecting of fruit phenology patterns. While seed of some species were abundant only briefly during summer months when they were also seen fruiting (Vaccinum spp.), others occurred in low numbers in seedrain traps for the entire sampling period, reflecting year-round fruiting (ie. Cheirodendron trigynum). Seedrain collections from traps continued through May 2016 to complete a full year of monitoring.

USGS personnel establish plots to study grass competition

USGS Botanist, Stephanie Yelenik (left), and volunteer, Meredith Stewart (right), establishing an experiment to investigate the effects of grass competition on seedling emergence and survival. Photo: E. Rose


We have completed two different outplanting experiments to look at the role of canopy tree identity, exotic grass biomass, native understory presence, and litter layers in the survival and growth of native trees. These experiments will inform future restoration efforts and help elucidate ecological processes that facilitate or stall the regeneration of native forest species in Acacia koa reforestation areas. A portion of the study has been published in the manuscript "Linking dominant Hawaiian tree species to understory development in recovering pastures via impacts on soils and litter" in the journal Restoration Ecology.

We have also completed one full year of sampling bird-dispersed seed rain using aerial seed-rain traps. Methodology and the sampling design using aerial seedrain trapsare are described in the manuscript "Methods for measuring bird-mediated seed-rin: insights from a Hawaiian mesic forest" in the journal Pacific Science

Planned Work:

We will take final measurements on various outplanting experiments. We will end the biweekly census of seedling and grass biomass in our large scale seed addition experiment. We will take the final basal stem diameter measurements (after 1 full year) to quantify stem diameter growth of various native woody understory species. We will begin analysis of data and manuscript preparation. We will start our demographic modeling effort to predict growth of native understory populations in different habitat types. Finally, we will add management "levers" to these models to test how different management actions will affect forest succession and restoration.