Multi-trophic invasion resistance in Hawaii: bioacoustics, field surveys, and airborne remote sensing

We used airborne imaging spectroscopy and scanning light detection and ranging (LiDAR), along with bioacoustic recordings, to determine how a plant species invasion affects avian abundance and community composition across a range of Hawaiian submontane ecosystems. Total avian abundance and the ratio of native to exotic avifauna were highest in habitats with the highest canopy cover and height. Comparing biophysically equivalent sites, stands dominated by native Metrosideros polymorpha trees hosted larger native avian communities than did mixed stands of Metrosideros and the invasive tree Morella faya. A multi-trophic analysis strongly suggests that native avifauna provide biotic resistance against the invasion of Morella trees and exotic birds, thus slowing invasion "meltdowns" that disrupt the functioning of native Hawaiian ecosystems.     

Hyperspectral and LiDAR remote sensing of fire fuels in Hawaii Volcanoes National Park.

Alien invasive grasses threaten to transform Hawaiian ecosystems through the alteration of ecosystem dynamics, especially the creation or intensification of a fire cycle. Across sub-montane ecosystems of Hawaii Volcanoes National Park on Hawaii Island, we quantified fine fuels and fire spread potential of invasive grasses using a combination of airborne hyperspectral and light detection and ranging (LiDAR) measurements. Across a gradient from forest to savanna to shrubland, automated mixture analysis of hyperspectral data provided spatially explicit fractional cover estimates of photosynthetic vegetation, non-photosynthetic vegetation, and bare substrate and shade. Small-footprint LiDAR provided measurements of vegetation height along this gradient of ecosystems. Through the fusion of hyperspectral and LiDAR data, a new fire fuel index (FFI) was developed to model the three-dimensional volume of grass fuels. Regionally, savanna ecosystems had the highest volumes of fire fuels, averaging 20% across the ecosystem and frequently filling all of the three-dimensional space represented by each image pixel. The forest and shrubland ecosystems had lower FFI values, averaging 4.4% and 8.4%, respectively. The results indicate that the fusion of hyperspectral and LiDAR remote sensing can provide unique information on the three-dimensional properties of ecosystems, their flammability, and the potential for fire spread.     

Controls over foliar N:P ratios in tropical rain forests

Correlations between foliar nutrient concentrations and soil nutrient availability have been found in multiple ecosystems. These relationships have led to the use of foliar nutrients as an index of nutrient status and to the prediction of broadscale patterns in ecosystem processes. More recently, a growing interest in ecological stoichiometry has fueled multiple analyses of foliar nitrogen:phosphorus (N:P) ratios within and across ecosystems. These studies have observed that N:P values are generally elevated in tropical forests when compared to higher latitude ecosystems, adding weight to a common belief that tropical forests are generally N rich and P poor. However, while these broad generalizations may have merit, their simplicity masks the enormous environmental heterogeneity that exists within the tropics; such variation includes large ranges in soil fertility and climate, as well as the highest plant species diversity of any biome. Here we present original data on foliar N and P concentrations from 150 mature canopy tree species in Costa Rica and Brazil, and combine those data with a comprehensive new literature synthesis to explore the major sources of variation in foliar N:P values within the tropics. We found no relationship between N:P ratios and either latitude or mean annual precipitation within the tropics alone. There is, however, evidence of seasonal controls; in our Costa Rica sites, foliar N:P values differed by 25% between wet and dry seasons. The N:P ratios do vary with soil P availability and/or soil order, but there is substantial overlap across coarse divisions in soil type, and perhaps the most striking feature of the data set is variation at the species level. Taken as a whole, our results imply that the dominant influence on foliar N:P ratios in the tropics is species variability and that, unlike marine systems and perhaps many other terrestrial biomes, the N:P stoichiometry of tropical forests is not well constrained. Thus any use of N:P ratios in the tropics to infer larger-scale ecosystem processes must comprehensively account for the diversity of any given site and recognize the broad range in nutrient requirements, even at the local scale.     

Remote analysis of biological invasion and the impact of enemy release

Escape from natural enemies is a widely held generalization for the success of exotic plants. We conducted a large-scale experiment in Hawaii (USA) to quantify impacts of ungulate removal on plant growth and performance, and to test whether elimination of an exotic generalist herbivore facilitated exotic success. Assessment of impacted and control sites before and after ungulate exclusion using airborne imaging spectroscopy and LiDAR, time series satellite observations, and ground-based field studies over nine years indicated that removal of generalist herbivores facilitated exotic success, but the abundance of native species was unchanged. Vegetation cover <1 m in height increased in ungulate-free areas from 48.7% +/- 1.5% to 74.3% +/- 1.8% over 8.4 years, corresponding to an annualized growth rate of lambda = 1.05 +/- 0.01 yr(-1) (median +/- SD). Most of the change was attributable to exotic plant species, which increased from 24.4% +/- 1.4% to 49.1% +/- 2.0%, (lambda = 1.08 +/- 0.01 yr(-1)). Native plants experienced no significant change in cover (23.0% +/- 1.3% to 24.2% +/- 1.8%, lambda = 1.01 +/- 0.01 yr(-1)). Time series of satellite phenology were indistinguishable between the treatment and a 3.0-km2 control site for four years prior to ungulate removal, but they diverged immediately following exclusion of ungulates. Comparison of monthly EVI means before and after ungulate exclusion and between the managed and control areas indicates that EVI strongly increased in the managed area after ungulate exclusion. Field studies and airborne analyses show that the dominant invader was Senecio madagascariensis, an invasive annual forb that increased from < 0.01% to 14.7% fractional cover in ungulate-free areas (lambda = 1.89 +/- 0.34 yr(-1)), but which was nearly absent from the control site. A combination of canopy LAI, water, and fractional cover were expressed in satellite EVI time series and indicate that the invaded region maintained greenness during drought conditions. These findings demonstrate that enemy release from generalist herbivores can facilitate exotic success and suggest a plausible mechanism by which invasion occurred. They also show how novel remote-sensing technology can be integrated with conservation and management to help address exotic plant invasions.     

Temporal variability of forest fires in eastern Amazonia

Widespread occurrence of fires in Amazonian forests is known to be associated with extreme droughts, but historical data on the location and extent of forest fires are fundamental to determining the degree to which climate conditions and droughts have affected fire occurrence in the region. We used remote sensing to derive a 23-year time series of annual landscape-level burn scars in a fragmented forest of the eastern Amazon. Our burn scar data set is based on a new routine developed for the Carnegie Landsat Analysis System (CLAS), called CLAS-BURN, to calculate a physically based burn scar index (BSI) with an overall accuracy of 93% (Kappa coefficient 0.84). This index uses sub-pixel cover fractions of photosynthetic vegetation, non-photosynthetic vegetation, and shade/burn scar spectral end members. From 23 consecutive Landsat images processed with the CLAS-BURN algorithm, we quantified fire frequencies, the variation in fire return intervals, and rates of conversion of burned forest to other land uses in a 32 400 km2 area. From 1983 to 2007, 15% of the forest burned; 38% of these burned forests were subsequently deforested, representing 19% of the area cleared during the period of observation. While 72% of the fire-affected forest burned only once during the 23-year study period, 20% burned twice, 6% burned three times, and 2% burned four or more times, with the maximum of seven times. These frequencies suggest that the current fire return interval is 5-11 times more frequent than the estimated natural fire regime. Our results also quantify the substantial influence of climate and extreme droughts caused by a strong El Ni?o Southern Oscillation (ENSO) on the extent and likelihood of returning forest fires mainly in fragmented landscapes. These results are an important indication of the role of future warmer climate and deforestation in enhancing emissions from more frequently burned forests in the Amazon.     

Environmental filtering and land-use history drive patterns in biomass accumulation in a mediterranean-type landscape

Aboveground biomass (AGB) reflects multiple and often undetermined ecological and land-use processes, yet detailed landscape-level studies of AGB are uncommon due to the difficulty in making consistent measurements at ecologically relevant scales. Working in a protected mediterranean-type landscape (Jasper Ridge Biological Preserve, California, USA), we combined field measurements with remotely sensed data from the Carnegie Airborne Observatory's light detection and ranging (lidar) system to create a detailed AGB map. We then developed a predictive model using a maximum of 56 explanatory variables derived from geologic and historic-ownership maps, a digital elevation model, and geographic coordinates to evaluate possible controls over currently observed AGB patterns. We tested both ordinary least-squares regression (OLS) and autoregressive approaches. OLS explained 44% of the variation in AGB, and simultaneous autoregression with a 100-m neighborhood improved the fit to an r2 = 0.72, while reducing the number of significant predictor variables from 27 variables in the OLS model to 11 variables in the autoregressive model. We also compared the results from these approaches to a more typical field-derived data set; we randomly sampled 5% of the data 1000 times and used the same OLS approach each time. Environmental filters including incident solar radiation, substrate type, and topographic position were significant predictors of AGB in all models. Past ownership was a minor but significant predictor, despite the long history of conservation at the site. The weak predictive power of these environmental variables, and the significant improvement when spatial autocorrelation was incorporated, highlight the importance of land-use history, disturbance regime, and population dynamics as controllers of AGB.     

Climatic/edaphic controls on soil carbon/nitrogen response to shrub encroachment in desert grassland.

The proliferation of woody plants in grasslands over the past 100+ years can alter carbon, nitrogen, and water cycles and influence land surface-atmosphere interactions. Although the majority of organic carbon in these ecosystems resides belowground, there is no consensus on how this change in land cover has affected soil organic carbon (SOC) and total nitrogen (TN) pools. The degree to which duration of woody plant occupation, climate, and edaphic conditions have mediated SOC and TN responses to changes in life-form composition are poorly understood. We addressed these issues at a desert grassland site in Arizona, USA, where the leguminous shrub velvet mesquite (Prosopis velutina) has proliferated along an elevation/precipitation/temperature gradient and on contrasting soil morphologic surfaces. On sandy loam complexes of mid-Holocene origin, mean SOC and TN of soils in the grassland matrix increased approximately 68% and approximately 45%, respectively, with increasing elevation. Soil organic carbon pools were comparable and TN pools were approximately 23% higher in Pleistocene-aged clay loam complexes co-occurring with Holocene-aged soils at the upper elevation/climatic zone. Across the site, belowground resources associated with large Prosopis plants were 21-154% (SOC) and 18-127% (TN) higher than those in the grassy matrix. The variance in SOC and TN pools accounted for by Prosopis stem size (a rough surrogate for time of site occupation) was highest at the low- and mid-elevation sites (69-74%) and lowest at the upper elevation site (32-38%). Soil delta15N values ranged from 5.5 per thousand to 6.7 per thousand across the soil/elevation zones but were comparable in herbaceous and shrub-impacted soils and exhibited a weak relationship with Prosopis basal stem diameter (r2 < 0.1) and TN (r2 < 0.08). The SOC delta13C values decreased linearly with increasing Prosopis basal diameter, suggesting that size and isotopic composition of the SOC pool is a function of time of Prosopis site occupation. Isotopic mixture models indicate that encroachment of C3 woody plants has also promoted SOC additions from C4 plant sources, indicative of long-term herbaceous facilitation. Grassy sites in contrasting soil/elevation combinations, initially highly distinctive in their SOC pool size and delta13C, appear to be converging on similar values following approximately 100 years of woody plant proliferation.     

Incorporating connectivity into conservation planning for the optimal representation of multiple species and ecosystem services

Conservation planning tends to focus on protecting species’ ranges or landscape connectivity but seldom both—particularly in the case of diverse taxonomic assemblages and multiple planning goals. Therefore, information on potential trade-offs between maintaining landscape connectivity and achieving other conservation objectives is lacking. We developed an optimization approach to prioritize the maximal protection of species’ ranges, ecosystem types, and forest carbon stocks, while also including habitat connectivity for range-shifting species and dispersal corridors to link protected area. We applied our approach to Sabah, Malaysia, where the state government mandated an increase in protected-area coverage of approximately 305,000 ha but did not specify where new protected areas should be. Compared with a conservation planning approach that did not incorporate the 2 connectivity features, our approach increased the protection of dispersal corridors and elevational connectivity by 13% and 21%, respectively. Coverage of vertebrate and plant species’ ranges and forest types were the same whether connectivity was included or excluded. Our approach protected 2% less forest carbon and 3% less butterfly range than when connectivity features were not included. Hence, the inclusion of connectivity into conservation planning can generate large increases in the protection of landscape connectivity with minimal loss of representation of other conservation targets.     

Recovery of forest structure and spectral properties after selective logging in lowland Bolivia.

Effective monitoring of selective logging from remotely sensed data requires an understanding of the spatial and temporal thresholds that constrain the utility of those data, as well as the structural and ecological characteristics of forest disturbances that are responsible for those constraints. Here we assess those thresholds and characteristics within the context of selective logging in the Bolivian Amazon. Our study combined field measurements of the spatial and temporal dynamics of felling gaps and skid trails ranging from <1 to 19 months following reduced-impact logging in a forest in lowland Bolivia with remote-sensing measurements from simultaneous monthly ASTER satellite overpasses. A probabilistic spectral mixture model (AutoMCU) was used to derive per-pixel fractional cover estimates of photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV), and soil. Results were compared with the normalized difference in vegetation index (NDVI). The forest studied had considerably lower basal area and harvest volumes than logged sites in the Brazilian Amazon where similar remote-sensing analyses have been performed. Nonetheless, individual felling-gap area was positively correlated with canopy openness, percentage liana coverage, rates of vegetation regrowth, and height of remnant NPV. Both liana growth and NPV occurred primarily in the crown zone of the felling gap, whereas exposed soil was limited to the trunk zone of the gap. In felling gaps >400 m2, NDVI, and the PV and NPV fractions, were distinguishable from unlogged forest values for up to six months after logging; felling gaps <400 m2 were distinguishable for up to three months after harvest, but we were entirely unable to distinguish skid trails from our analysis of the spectral data.     

Woody plants in grasslands: post-encroachment stand dynamics

Woody plant abundance is widely recognized to have increased in savannas and grasslands worldwide. The lack of information on the rates, dynamics, and extent of increases in shrub abundance is a major source of uncertainty in assessing how this vegetation change has influenced biogeochemical cycles. Projecting future consequences of woody cover change on ecosystem function will require knowledge of where shrub cover in present-day stands lies relative to the realizable maximum for a given soil type within a bioclimatic region. We used time-series aerial photography (1936, 1966, and 1996) and field studies to quantify cover and biomass of velvet mesquite (Prosopis velutina Woot.) following its proliferation in a semidesert grassland of Arizona. Mapping of individual shrubs indicated an encroachment phase characterized by high rates of bare patch colonization. Upon entering a stabilization phase, shrub cover increases associated with recruitment and canopy expansion were largely offset by contractions in canopy area of other shrub patches. Instances of shrub disappearance coincided with a period of below-average rainfall (1936-1966). Overall, shrub cover (mean +/- SE) on sandy uplands with few and widely scattered shrubs in 1902 was dynamically stable over the 1936-1996 period averaging approximately 35% +/- 5%. Shrub cover on clayey uplands in 1936 was 17% +/- 2% but subsequently increased twofold to levels comparable to those on sandy uplands by 1966 (36% +/- 7%). Cover on both soils then decreased slightly between 1966 and 1996 to 28% +/- 3%. Thus, soil properties influenced the rate at which landscapes reached a dynamic equilibrium, but not the apparent endpoint. Although sandy and clayey landscapes appear to have stabilized at comparable levels of cover, shrub biomass was 1.4 times greater on clayey soils. Declines in shrub cover between 1966 and 1996 were accompanied by a shift to smaller patch sizes on both sandy and clayey landscapes. Dynamics observed during the stabilization phase suggest that density-dependent regulation may be in play. If woody cover has transitioned from directional increases to a dynamic equilibrium, biomass projections will require monitoring and modeling patch dynamics and stand structure rather than simply changes in total cover.