Ben P. Miller

A comparison of methods for the statistical analysis of spatial point patterns in plant ecology

We describe a range of methods for the description and analysis of spatial point patterns in plant ecology. The conceptual basis of the methods is presented, and specific tests are compared, with the goal of providing guidelines concerning their appropriate selection and use. Simulated and real data sets are used to explore the ability of these methods to identify different components of spatial pattern (e.g. departure from randomness, regularity vs. aggregation, scale and strength of pattern). First-order tests suffer from their inability to characterise pattern at distances beyond those at which local interactions (i.e. nearest neighbours) occur. Nevertheless, the tests explored (first-order nearest neighbour, Diggle’s G and F) are useful first steps in analysing spatial point patterns, and all seem capable of accurately describing patterns at these (shorter) distances. Among second-order tests, a density-corrected form of the neighbourhood density function (NDF), a non-cumulative analogue of the commonly used Ripley’s K-function, most informatively characterised spatial patterns at a range of distances for both univariate and bivariate analyses. Although Ripley’s K is more commonly used, it can give very different results to the NDF because of its cumulative nature. A modified form of the K-function suitable for inhomogeneous point patterns is discussed. We also explore the use of local and spatially-explicit methods for point pattern analysis. Local methods are powerful in that they allow variations from global averages to be detected and potentially provide a link to recent spatial ecological theory by taking the ‚plant’s-eye view’. We conclude by discussing the problems of linking spatial pattern with ecological process using three case studies, and consider some ways that this issue might be addressed.

Anthropogenic disturbance promotes hybridization between Banksia species by altering their biology

Abstract Putative hybrids between Banksia hookeriana and B. prionotes were identified among 12 of 106 populations of B. hookeriana located at or near anthropogenically disturbed sites, mainly roadways, but none in 156 undisturbed populations. Morphometrics and AFLP markers confirmed that a hybrid swarm existed in a selected disturbed habitat, whereas no intermediates were present where the two species co‐occurred in undisturbed vegetation. Individuals of both species in disturbed habitats at 12 sites were more vigorous, with greater size and more flower heads than their counterparts in undisturbed vegetation. These more fecund plants also showed a shift in season and duration of flowering. By promoting earlier flowering of B. hookeriana plants and prolonging flowering of B. prionotes, anthropogenic disturbance broke the phenological barrier between these two species. We conclude that anthropogenic disturbance promotes hybridization through increasing opportunities for gene flow by reducing interpopulation separation, increasing gamete production and, especially, promoting coflowering.

Long‐distance seed dispersal in a metapopulation of Banksia hookeriana inferred from a population allocation analysis of amplified fragment length polymorphism data

There is currently a poor understanding of the nature and extent of long‐distance seed dispersal, largely due to the inherent difficulty of detection. New statistical approaches and molecular markers offer the potential to accurately address this issue. A log‐likelihood population allocation test (aflpop) was applied to a plant metapopulation to characterize interpopulation seed dispersal. Banksia hookeriana is a fire‐killed shrub, restricted to sandy dune crests in fire‐prone shrublands of the Eneabba sandplain, southwest Australia. Population genetic variation was assessed for 221 individuals sampled from 21 adjacent dune‐crest populations of B. hookeriana using amplified fragment length polymorphism. Genetic diversity was high, with 175 of 183 (96%) amplified fragment length polymorphism markers polymorphic. Of the total genetic diversity, 8% was partitioned among populations by amova and FST. There was no relationship between genetic diversity within populations and population demographic parameters such as population size and sample size. A population allocation test on these data unambiguously assigned 177 of 221 (80.1%) individuals to a single population. Of these, 171 (77.4% of total) were assigned to the population from which they were sampled and 6 (2.7% of total) were assigned to a known population other than the one from which they were sampled. A further 9 (4.1% of total) were assigned to outside the sampled metapopulation area, and 35 individuals (15.8%) could not be assigned unambiguously to any particular population. These results suggest that both the extent [15 of 221 (6.8%) individuals originating from a population other than the one in which they occur] and distance (1.6 to > 2.5 km), of seed dispersal between dune‐crest populations is greater than expected from previous studies. The extent of long‐distance interpopulation seed dispersal observed provides a basis for explaining the survival of populations of the fire‐killed B. hookeriana in a landscape experiencing frequent fire, where local extinctions and recolonizations may be a regular occurrence.

Spatial patterns in species‐rich sclerophyll shrublands of southwestern Australia

Abstract Question: The drivers of spatial patterning among plant species and the implications of those patterns for the structure and function of plant communities are of ongoing interest and debate. Here we explore the spatial patterning shown by individual species in species-rich plant communities. We (1) compare the levels of aggregation in these communities to those observed in other species-rich communities, in particular tropical rain forests, and (2) consider how abiotic conditions might influence the levels of aggregation observed. Location: We describe the spatial structure of four species-rich Mediterranean-type shrubland communities near Eneabba, Western Australia. The four sites each contain > 10000 plants and up to 113 species, and differ in substrate-type, species richness and composition. Methods: We analysed the spatial patterning of all species with more than 20 individuals (233 species patterns), and used point process models for aggregated patterns to separate first-order gradient effects from second-order clustering. Results: Aggregated distributions were most common at all sites, but especially at the site with the highest resource availability and heterogeneity and lowest species richness. A Poisson cluster process best described the majority of aggregated species, suggesting that local interactions drive fine-scale patterns in these communities. Conclusions: As with many previous studies, we found that most species showed strong local aggregation. The proportion of species showing aggregation was less than has been described in species-rich tropical rainforests but was higher than observed in many temperate plant communities. The highest proportion of aggregated species was seen at the most resource-abundant site; this is in direct contrast to conceptual models that suggest that competition should be weakest, and aggregation most prevalent, in the most resource-limited sites.

SOIL VS. CANOPY SEED STORAGE AND PLANT SPECIES COEXISTENCE IN SPECIES-RICH AUSTRALIAN SHRUBLANDS

The fire-prone shrublands of southwestern Australia are renowned for their high plant species diversity and prominence of canopy seed storage (serotiny). We compared species richness, abundance, and life history attributes for soil and canopy seed banks in relation to extant vegetation among four sites with different substrate conditions and high species turnover (50-80%) to identify whether this unusual community-level organization of seed storage might contribute to maintenance of high species richness. Soil seed bank (SSB) densities were low to moderate (233-1435 seeds/m2) compared with densities for other Mediterranean-type vegetation and were lowest for sites with highest canopy seed bank (CSB) species richness and lowest nutrient availability, but not richness or abundance of resprouters. Annuals were infrequent in the lowest nutrient sites, but there was no evidence that small SSB size was due to low seed inputs or a trade-off between seed production/storage and seed size in response to low nutrient availability. Sorensens similarity between SSB and extant vegetation was 26-43% but increased to 54-57% when the CSB was included, representing levels higher than reported for most other ecosystems. Resprouting species were well represented in both the SSB and CSB, and there was no evidence for lower seed production in resprouters than in non-sprouters overall. The SSB and CSB held no species in common and were characterized by markedly different seed dispersal attributes, with winged or small seeds in the CSB and seeds dispersed by ants, birds, and wind (though none with wings) in the SSB. There was no evidence of spatial differentiation in the distribution of seeds of SSB species between vegetated and open microsites that might facilitate species coexistence, but most woody non-sprouters showed aggregation at scales of 1-2 m, implying limited seed dispersal. High similarity between overall seed bank (SSB + CSB) and extant species composition, high number of resprouting species, and seed dispersal processes before (SSB) and after fire (CSB) leading to differential spatial aggregation of post-fire recruits from the two seed bank types may buffer species composition against rapid change and provide a mechanism for maintaining species coexistence at the local scale.

Contrasting impacts of pollen and seed dispersal on spatial genetic structure in the bird-pollinated Banksia hookeriana

In plants, pollen- and seed-dispersal distributions are characteristically leptokurtic, with significant consequences for spatial genetic structure and nearest-neighbour mating. However, most studies to date have been on wind- or insect-pollinated species. Here, we assigned paternity to quantify effective pollen dispersal over 9 years of mating, contrasted this to seed dispersal and examined their effects on fine-scale spatial genetic structure, within the bird-pollinated shrub Banksia hookeriana (Proteaceae). We used 163 polymorphic amplified fragment length polymorphism markers to assess genetic structure and pollen dispersal in a spatially discrete population of 112 plants covering 0.56 ha. Spatial autocorrelation analysis detected spatial genetic structure in the smallest distance class of 0–5 m (r=0.025), with no significant structure beyond 8 m. Experimentally quantified seed-dispersal distances for 337 seedlings showed a leptokurtic distribution around a median of 5 m, reaching a distance of 36 m. In marked contrast, patterns of pollen dispersal for 274 seeds departed strikingly from typical near-neighbour pollination, with a distribution largely corresponding to the spatial distribution of plants. We found very high multiple paternity, very low correlated paternity and an equal probability of siring for the 50 closest potential mates. Extensive pollen carryover was demonstrated by multiple siring in 83 of 86 (96.5%) two-seeded fruits. Highly mobile nectar-feeding birds facilitate this promiscuity through observed movements that were effectively random. As the incidence of bird-pollination is markedly greater in the Southwest Australian Floristic Region than elsewhere, our results have broad and novel significance for the evolution and conservation for many species in Gondwanan lineages.

Resistance and resilience to changing climate and fire regime depend on plant functional traits

Changing disturbance-climate interactions will drive shifts in plant communities: these effects are not adequately quantified by environmental niche models used to predict future species distributions. We quantified the effects of more frequent fire and lower rainfall - as projected to occur under a warming and drying climate - on population responses of shrub species in biodiverse Mediterranean-climate type shrublands near Eneabba, southwestern Australia. Using experimental fires, we measured the density of all shrub species for four dominant plant functional groups (resprouter/non-sprouter × serotinous/soil seed bank) before and after fire in 33 shrubland sites, covering four post-fire rainfall years and fire intervals from 3-24 years. Generalized linear mixed effects models were used to test our a priori hypotheses of rainfall, fire interval and plant functional type effects on post-fire survival and recruitment. At shortened fire intervals, species solely dependent on seedling recruitment for persistence were more vulnerable to local extinction than were species with both seedling recruitment and vegetative regrowth. Nevertheless, seedling recruitment was essential for population maintenance of resprouting species. Serotinous species were less resilient than soil seed storage species regardless of regeneration mode. Critically, in relation to changing climate, a 20% reduction in post-fire winter rainfall (essential for seedling recruitment) is predicted to increase the minimum inter-fire interval required for self-replacement by 50%, placing many species at risk of decline. Synthesis. Our results highlight the potentially deleterious biodiversity impacts of climate and fire regime change, and underscore weaknesses inherent in studies considering single impact factors in isolation. In fire-prone ecosystems characterized by a projected warming and drying climate, and increasing fire hazard, adaptive approaches to fire management may need to include heightened wildfire suppression and lengthened intervals for prescribed fire to best support the in situ persistence of perennial plant species and of plant biodiversity. This conclusion is at odds with the view that more managed fire may be needed to mitigate wildfire risk as climate warms.

Fire behaviour in south-western Australian shrublands: evaluating the influence of fuel age and fire weather

Fuel age (time since last fire) is often used to approximate fire hazard and informs decisions on placement of shrubland management burns worldwide. However, uncertainty remains concerning the relative importance of fuel age and weather conditions as predictors of fire hazard and behaviour. Using data from 35 experimental burns across three types of shrublands in Western Australia, we evaluated importance of fuel age and fire weather on probability of fire propagation (hazard) and four metrics of fire behaviour (rate of spread, fireline intensity, residence time, surface temperature) under moderate to high fire danger weather conditions. We found significant support for a threshold effect of fuel age for fire propagation but limited evidence for an effect of fuel age or fire weather on rates of spread or fireline intensity, although surface heating and heating duration were significantly related to fuel age and shrubland type. Further analysis suggested that dead fuel mass and accumulation rate rather than live fuels were responsible for this relationship. Using BEHAVE, predicted spread rates and intensities were consistently lower than observed values, suggesting further refinement is needed in modelling shrubland fire behaviour. These data provide important insight into fire behaviour in globally significant, fire-adapted shrublands, informing fire management and relationships between fire frequency and fire intensity.

Late Quaternary climate change and spatial genetic structure in the shrub Banksia hookeriana

Spatial genetic variation within species is influenced by both contemporary and historical factors. We attempted to assess the impact of increased aridity and lower temperatures associated with the last glacial maximum on possible refugia and the structuring of genetic variation in Banksia hookeriana, a shrub species restricted to deep sands on the Eneabba sandplain and adjacent Gingin Scarp/Dandaragan Plateau, centred 300 km north of Perth, Western Australia. We used optically stimulated luminescence (OSL) dating to estimate the last time these sands were mobile, as well as amplified fragment length polymorphisms (AFLP) to infer spatial patterns and the phylogeographical history of genetic variation among 15 populations of B. hookeriana. While genetic variation at the species level was high, with 96.6% of 238 AFLP markers polymorphic, average within population gene diversity was low (Hpop = 0.16). Of the total genetic variation, an analysis of molecular variance (amova) partitioned 70% within populations, 24% among populations within substrate and 6% between substrates. There was an isolation‐by‐distance effect among populations within the same substrate, but not across substrates, and ordination highlighted genetic differentiation between the sandplain and scarp/plateau populations. A neighbour‐joining tree identified the sandplain populations as a distinct clade, with the exception of the most northern sandplain population, which clustered with two northern and eastern plateau populations. The most southern plateau populations formed a clade sister to the sandplain clade. OSL dating of sand at six extant populations suggested that dunes were last mobile 15 000–35 000 years ago, with no clear difference in the ages of sandplain and plateau dunes. These data are consistent with a historical scenario of (re)colonization from isolated refugia of smaller populations either within the patchily vegetated sandplain and/or refugia at the northern, eastern and southern sandplain/scarp margins following postglacial climate amelioration and dune stabilization. Historic interpretations were confounded by the possible effects of long‐distance dispersal, natural selection by substrate, and weak and/or ancient introgression with the sister species, Banksia prionotes.

Low-dimensional trade-offs fail to explain richness and structure in species-rich plant communities

Mathematical models and ecological theory suggest that low-dimensional life history trade-offs (i.e. negative correlation between two life history traits such as competition vs. colonisation) may potentially explain the maintenance of species diversity and community structure. In the absence of trade-offs, we would expect communities to be dominated by ‘super-types’ characterised by mainly positive trait expressions. However, it has proven difficult to find strong empirical evidence for such trade-offs in species-rich communities. We developed a spatially explicit, rule-based and individual-based stochastic model to explore the importance of low-dimensional trade-offs. This model simulates the community dynamics of 288 virtual plant functional types (PFTs), each of which is described by seven life history traits. We consider trait combinations that fit into the trade-off concept, as well as super-types with little or no energy constraints or resource limitations, and weak PFTs, which do not exploit resources efficiently. The model is parameterised using data from a fire-prone, species-rich Mediterranean-type shrubland in southwestern Australia. We performed an exclusion experiment, where we sequentially removed the strongest PFT in the simulation and studied the remaining communities. We analysed the impact of traits on performance of PFTs in the exclusion experiment with standard and boosted regression trees. Regression tree analysis of the simulation results showed that the trade-off concept is necessary for PFT viability in the case of weak trait expression combinations such as low seed production or small seeds. However, species richness and diversity can be high despite the presence of super-types. Furthermore, the exclusion of super-types does not necessarily lead to a large increase in PFT richness and diversity. We conclude that low-dimensional trade-offs do not provide explanations for multi-species co-existence contrary to the prediction of many conceptual models.