On using landscape metrics for landscape similarity search

Landscape similarity search involves finding landscapes from among a large collection that are similar to a query landscape. An example of such collection is a large land cover map subdivided into a grid of smaller local landscapes, a query is a local landscape of interest, and the task is to find other local landscapes within a map which are perceptually similar to the query. Landscape search and the related task of pattern-based regionalization, requires a measure of similarity – a function which quantifies the level of likeness between two landscapes. The standard approach is to use the Euclidean distance between vectors of landscape metrics derived from the two landscapes, but no in-depth analysis of this approach has been conducted. In this paper we investigate the performance of different implementations of the standard similarity measure. Five different implementations are tested against each other and against a control similarity measure based on histograms of class co-occurrence features and the Jensen–Shannon divergence. Testing consists of a series of numerical experiments combined with visual assessments on a set of 400 3 km-scale landscapes. Based on the cases where visual assessment provides definitive answer, we have determined that the standard similarity measure is sensitive to the way landscape metrics are normalized and, additionally, to whether weights aimed at controlling the relative contribution of landscape composition vs. configuration are used. The standard measure achieves the best performance when metrics are normalized using their extreme values extracted from all possible landscapes, not just the landscapes in the given collection, and when weights are assigned so the combined influence of composition metrics on the similarity value equals the combined influence of configuration metrics. We have also determined that the control similarity measure outperforms all implementations of the standard measure.     

Self-similar land cover heterogeneity of temperate and tropical landscapes

Landscapes are complex objects often showing self-similar properties. For example, some power law exponents and fractal dimensions have been extensively used as global metrics for describing spatial arrangements of landscape land covers. This paper presents some newly found invariance properties for rural landscapes, with the hope to prepare further theory capable to link these properties mechanistically with generating processes. For this purpose, I propose a new surface-pattern analysis computing heterogeneity metrics into moving windows in a specific way. This generic method provides Multiscale Heterogeneity Maps (MHMs) and Heterogeneity Profiles (HPs) of almost any kind of landscape. Six landscapes, covering a wide range of observed mosaics have been analyzed by this way: four of them exhibit strong self-similar evenness (diversity) heterogeneity over two orders of magnitude with an apparent universality of the power law exponents (γ = ?0.19 ± 0.02). Such landscape self-organization is interpreted in terms of spatial arrangements of numerous land covers and fine scale patchy mosaics. This study suggests that most of terrestrial landscapes could exhibit power law behaviors in terms of evenness heterogeneity and could be the result of a hidden optimization of ecological and/or socio-economic processes.     

Insect pests in winter oilseed rape affected by field and landscape characteristics

Winter oilseed rape (OSR, Brassica napus) cropping is often associated with an intensive use of pesticides. The transformation of structurally rich landscapes into more monotonous landscapes may be partly responsible for this, because non-crop habitats believed to benefit natural enemies have been eliminated. We examined the influence of field (soil quality, nitrogen fertilization) and landscape characteristics (OSR area and isolation, non-crop area, landscape diversity, proportions of grassy fallows and woody areas) on three major European OSR pest groups: pollen beetles, stem weevils, and brassica pod midges. Twenty-nine landscape sectors ranging from structurally poor to complex were studied at eight spatial scales (radii 250–2000 m) centered in the studied OSR fields. Abundances of pollen beetles and stem weevils were significantly positively correlated with soil quality and negatively related to OSR area in the surroundings. Generally, abundances of all groups were positively related to woody areas, but not related to grassy fallow area. Pod midges and stem weevils tended to respond primarily to landscape variables at small (250–500 m) and medium (1000–1250 m) scales, while pollen beetles responded at medium to large (1000–2000 m) scales. The results are discussed in relation to differences in overwintering strategies and mobility of pest insects. Strategies at the field and landscape level, aiming to reduce pest pressure in OSR fields, are also discussed.     

Extinction thresholds and negative responses of Afrotropical ant-following birds to forest cover loss in oil palm and agroforestry landscapes

Afrotropical ant-following birds are vulnerable to forest loss and disturbance, but critical habitat thresholds regarding their abundance and species richness in human-dominated landscapes, including industrial oil palm plantations, have never been assessed. We measured forest cover through Landsat imagery and recorded species richness and relative abundance of 20 ant-following birds in 48 plots of 1-km², covering three landscapes of Southwest Cameroon: Korup National Park, smallholder agroforestry areas (with farms embedded in forest), and an industrial oil palm plantation. We evaluated differences in encounter frequency and species richness among landscapes, and the presence of critical thresholds through enhanced adaptive regression through hinges. All species were detected in Korup National Park and the agroforestry landscape, which had similar forest cover (>85%). Only nine species were found in the oil palm plantation (forest cover = 10.3 ± 3.3%). At the 1-km² scale, the number of species and bird encounters were comparable in agroforests and the protected area: mean species richness ranged from 12.2 ± 0.6 in the park and 12.2 ± 0.6 in the agroforestry matrix to 1.0 ± 0.4 in the industrial oil palm plantation; whereas encounters decreased from 34.4 ± 3.2 to 26.1 ± 2.9 and 1.3 ± 0.4, respectively. Bird encounters decreased linearly with decreasing forest cover, down to an extinction threshold identified at 24% forest cover. Species richness declined linearly by ca. one species per 7.4% forest cover lost. We identified an extinction threshold at 52% forest cover for the most sensitive species (Criniger chloronotus, Dicrurus atripennis, and Neocossyphus poensis). Our results show that substantial proportions of forests are required to sustain complete ant-following bird assemblages in Afrotropical landscapes and confirm the high sensitivity of this bird guild to deforestation after industrial oil palm development. Securing both forest biodiversity and food production in an Afrotropical production landscape may be best attained through a combination of protected areas and wildlife-friendly agroforestry.     

Changes and interactions between forest landscape connectivity and burnt area in Spain

The spatial structure, functionality and dynamics of forest landscapes in peninsular Spain and the Balearic Islands were compared over the last five decades. Two particular features were studied in the sample sites: forest connectivity for wildlife and areas burnt by wildfires. 191 Squares, each 4 km × 4 km, were selected from the SISPARES (the monitoring framework designed to evaluate the trends in the structure of Spanish rural landscapes) environmental strata. Aerial photographs from 1956, 1984, 1998 and 2008 were interpreted and 11 land cover categories mapped and checked in the field, using a minimum mapping area of one hectare. The Equivalent Connected Area Index was used to assess forest connectivity over the sampling period. Social and economical factors were assessed using indicators of farm intensiveness. The Spanish forest connectivity has improved in the last five decades although two different trends can be identified: the first 40 years are characterized by positive rates of growth whereas the 10 last years are characterized by their stability. Nevertheless the area of burnt land was higher along the first 25 studied years and decreased significantly over the last decade.Our results show the climate is the main driver in the evolution of forest connectivity and burnt area in the forest landscapes, playing a direct role on forest biomass production and wildfire ignition and propagation, as well as an indirect role by keeping vertical and horizontal forest continuity through the landscape spatial pattern. Social and economic factors are very important drivers as well: Rural population density and farm size average have been tested as good indicators of landscape artificiality, highly correlated to wildfire hazard and forest connectivity.Finally, we have pointed out the evolutionary path followed by SISPARES framework as a tool for monitoring rural landscapes. It emphasises on the requirement of a 30 years time window for building-up reliable dynamic multifunctional model.     

Landscape context and elevation affect pollinator communities in intensive apple orchards

Although an extensive research has been done on the contribution of wild insects to apple pollination, most of these studies did not evaluate the effect of the surrounding landscape context on local pollinator communities. Our aim was to compare communities of wild bees in 31 equally managed apple orchards located in three contrasting landscape types (either dominated by apple, forest, or grasslands) and along an elevation gradient and to test a potential interaction between landscape context and elevation. The study was carried out in 2009 in Trentino (NE Italy), one of the major apple growing areas of Europe with ～12,000 ha of commercial orchards distributed between 150 and 950 m a.s.l. We found that apple-dominated landscapes drastically reduced wild bee species richness and abundance in the orchard compared to landscapes dominated by either grassland or forest. Forest-dominated landscapes benefited local species richness more than grassland-dominated landscapes, while abundance did not differ between grassland and forest. Total species richness and abundance further declined with increasing elevation, while no interactive effect was found between temperature and landscape context. The abundance of Apis mellifera in the apple-dominated landscapes was two to four times higher than in the landscapes dominated by forest and grasslands, respectively. Measures to restore natural pollinator communities by providing suitable habitats around the orchard would not only benefit conservation of general biodiversity, but would probably also contribute to reduce the dependence of apple pollination on managed honey bees.     

Spatial metrics for detecting ecosystem degradation in the ridge-slough patterned landscape

Indicators of landscape condition should be selected based on their sensitivity to environmental changes and their capacity to provide early warning detection of those changes. We assessed the performance of a suite of spatial-pattern metrics selected to quantify the condition of the ridge-slough landscape in the Everglades (South Florida, USA). Spatial pattern metrics (n = 14) that describe landscape composition, geometry and hydrologic connectivity were enumerated from vegetation maps of twenty-five 2 × 2 km primary sampling units (PSUs) that span a gradient of hydrologic and ecological condition across the greater Everglades ecosystem. Metrics were assessed in comparison with field measurements from each PSU of landscape condition obtained from regional surveys of soil elevation, which have previously been shown to capture dramatic differences between conserved and degraded locations. Elevation-based measures of landscape condition included soil elevation bi-modality (BI_SE), a binary measure of landscape condition, and also the standard deviation of soil elevation (SD_SE), a continuous measure of condition. Metric performance was assessed based on the strength (sensitivity) and shape (leading vs. lagging) of the relationship between spatial pattern metrics and these elevation-based measures. We observed significant logistic regression slopes with BI_SE for only 4 metrics (slough width, ridge density, directional connectivity index – DCI, and least flow cost – LFC). More significant relationships (n = 8 metrics) were observed with SD_SE, with the strongest associations for slough density, mean ridge width, and the average length of straight flow, as well as for a suite of hydrologic connectivity metrics (DCI, LFC and landscape discharge competence – LDC). Leading vs. lagging performance, inferred from the curvature of the association obtained from the exponent of fitted power functions, suggest that only DCI was a leading metric of the loss of soil elevation variation; most metrics were indeterminate, though some were clearly lagging. Our findings support the contention that soil elevation changes from altered peat accretion dynamics precede changes in landscape pattern, and offer insights that will enable efficient monitoring of the ridge-slough landscape as part of the ongoing Everglades restoration effort.     

Polymer gel dosimeters for pretreatment radiotherapy verification using the three-dimensional gamma evaluation and pass rate maps

Polymer gel dosimeters (PGDs) have been widely studied for use in the pretreatment verification of clinical radiation therapy. However, the readability of PGDs in three-dimensional (3D) dosimetry remain unclear. In this study, the pretreatment verifications of clinical radiation therapy were performed using an N-isopropyl-acrylamide (NIPAM) PGD, and the results were used to evaluate the performance of the NIPAM PGD on 3D dose measurement. A gel phantom was used to measure the dose distribution of a clinical case of intensity-modulated radiation therapy. Magnetic resonance imaging scans were performed for dose readouts. The measured dose volumes were compared with the planned dose volume. The relative volume histograms showed that relative volumes with a negative percent dose difference decreased as time elapsed. Furthermore, the histograms revealed few changes after 24 h postirradiation. For the 3%/3 mm and 2%/2 mm criteria, the pass rates of the 12- and 24-h dose volumes were higher than 95%, respectively. This study thus concludes that the pass rate map can be used to evaluate the dose-temporal readability of PGDs and that the NIPAM PGD can be used for clinical pretreatment verifications.     

Loss of aboveground forest biomass and landscape biomass variability in Missouri,US

Disturbance regimes and forests have changed over time in the eastern United States. We examined effects of historical disturbance (circa 1813 to 1850) compared to current disturbance (circa 2004 to 2008) on aboveground, live tree biomass (for trees with diameters ≥13 cm) and landscape variation of biomass in forests of the Ozarks and Plains landscapes in Missouri, USA. We simulated 10,000 one-hectare plots using random diameters generated from parameters of diameter distributions limited to diameters ≥13 cm and random densities generated from density estimates. Area-weighted mean biomass density (Mg/ha) for historical forests averaged 116 Mg/ha, ranging from 54 Mg/ha to 357 Mg/ha by small scale ecological subsections within Missouri landscapes. Area-weighted mean biomass density for current forests averaged 82 Mg/ha, ranging from 66 Mg/ha to 144 Mg/ha by ecological subsection for currently forested land. Biomass density of current forest was greater than historical biomass density for only 2 of 23 ecological subsections. Current carbon sequestration of 292 TgC on 7 million ha of forested land is less than half of the estimated historical total carbon sequestration of 693 TgC on 12 million ha. Cumulative tree cutting disturbances over time have produced forests that have less aboveground tree biomass and are uniform in biomass compared to estimates of historical biomass, which varied across Missouri landscapes. With continued relatively low rates of forest disturbance, current biomass per ha will likely increase to historical levels as the most competitive trees become larger in size and mean number of trees per ha decreases due to competition and self-thinning. Restoration of large diameter structure and forested extent of upland woodlands and floodplain forests could fulfill multiple conservation objectives, including carbon sequestration.     

Landscape heterogeneity metrics as indicators of bird diversity: Determining the optimal spatial scales in different landscapes

Species distribution models are often used to study the biodiversity of ecosystems. The modelling process uses a number of parameters to predict others, such as the occurrence of determinate species, population size, habitat suitability or biodiversity. It is well known that the heterogeneity of landscapes can lead to changes in species’ abundance and biodiversity. However, landscape metrics depend on maps and spatial scales when it comes to undertaking a GIS analysis.We explored the goodness of fit of several models using the metrics of landscape heterogeneity and altitude as predictors of bird diversity in different landscapes and spatial scales. Two variables were used to describe biodiversity: bird richness and trophic level diversity, both of which were obtained from a breeding bird survey by means of point counts. The relationships between biodiversity and landscape metrics were compared using multiple linear regressions. All of the analyses were repeated for 14 different spatial scales and for cultivated, forest and grassland environments to determine the optimal spatial scale for each landscape typology.Our results revealed that the relationships between species’ richness and landscape heterogeneity using 1:10,000 land cover maps were strongest when working on a spatial scale up to a radius of 125–250 m around the sampled point (circa 4.9–19.6 ha). Furthermore, the correlation between measures of landscape heterogeneity and bird diversity was greater in grasslands than in cultivated or forested areas. The multi-spatial scale approach is useful for (a) assessing the accuracy of surrogates of bird diversity in different landscapes and (b) optimizing spatial model procedures for biodiversity mapping, mainly over extensive areas.     

Landscapes as gradients: The spatial structure of terrestrial ecosystem components in southern Ontario,Canada

The conventional view of the environment – consisting of discrete patches that repeat themselves across the landscape – has seldom been tested. Across a wide spatial scope in southern Ontario, Canada, we investigated the spatial structure of physical and biological features of the environment: vegetation communities, moisture, pH, and organic content of soil at local scales (10–1250 m), and mammalian communities, mean annual temperature and precipitation at regional scales (10–650 km). Spatial structure was quantified using log–log regression of variance (V) with distance (D), according to the power formula, V = aD^z. All these ecosystem components exhibited gradients. Slopes (z) of log–log regressions were positive (0.065 < z < 0.703) and were significantly steeper at the regional scale than the local scale. Variance appeared to increase without bound as distance between sampling locations increased. The results support the view of landscapes as continua and gradients. These patterns represent a challenge to the conventional view of how the natural environment is organised.     

Determinants of smooth-coated otter occupancy in a rapidly urbanizing coastal landscape in Southeast Asia

Urbanization often has negative impacts on wildlife, nevertheless many species can persist in heavily modified habitats. Understanding factors that promote species persistence in urbanizing landscapes is therefore important for maintaining biodiversity in changing landscapes and may inform more biodiversity-friendly development. We investigated effects of landscape-scale variables on habitat occupancy of Smooth-coated otter (Lutrogale perspicillata) in the Inner Gulf of Thailand. In this internationally important wetland complex, 86% of natural habitats have been altered and are now highly urbanized. We conducted track and sign surveys in 60, 25 km² grid cells encompassing 1,474 km². Within each cell, we quantified the landscape cover types including urban areas, natural habitat (predominately thin ribbons of mangrove along waterways), agriculture, aquaculture, and potential prey availability using a GIS and field surveys. We used occupancy models to identify habitat variables that affected probability of detection and occupancy. Estimated otter occupancy, based on the top model, was 0.33 ± 0.07 (95% CI 0.18–0.48) and detection probability was 0.50 ± 0.05 (95% CI 0.41–0.60). Otter occupancy was positively associated with the proportion of natural habitat and the cover of traditional aquaculture ponds, but negatively associated with agriculture and urban cover. The remaining natural patches appear to act as critical refuges for otter, allowing them to persist in an otherwise heavily transformed landscape. Because aquaculture ponds are likely important prey sources for otters, it may lead to conflict with aquaculture farmers. Further studies of feeding and movement patterns in cooperation with aquaculture farmers would be beneficial for developing detailed management plans for the species in this human dominated landscape.     

The effect of semi-natural habitats on aphids and their natural enemies across spatial and temporal scales

Semi-natural habitats in agricultural landscapes are generally assumed to enhance the biological control of insect pests based on native beneficial insects, by providing alternative prey and hosts, resources and refuges for overwintering. We hypothesized that natural enemies of winter wheat aphids should arrive sooner in fields near semi-natural habitats. We compared aphid, hoverfly (larvae and eggs) and parasitized aphid (mummies) abundances in 54 winter wheat fields located in southern France from 2003 to 2007. Six surveys were recorded each spring and were split into the early period (defined as the period before the peak of aphid growth) and the late period (after the peak). The wheat fields differed by their surrounding landscape composition measured as the proportion of semi-natural habitats (woods, hedges and grasslands), at three different spatial scales: 200 m, 500 m, and 1200 m. Despite great variability in abundance data between years, the abundance of hoverflies appeared more sensitive to landscape composition than aphid abundance was. Early abundance for both aphids and hoverflies was positively related to wood cover, but not late abundance in spring. The abundance of hoverflies was positively related to hedge and grassland cover at all spatial scales and both periods considered. Aphid parasitism was higher near hedges at the small spatial scale late in the spring. Our results confirmed that higher proportions of semi-natural habitats in agricultural landscapes enhance the biological control of pests, but this effect depends on the spatial scale, the time period in the spring and the natural enemies considered.     

2D and 3D CT Radiomics Features Prognostic Performance Comparison in Non-Small Cell Lung Cancer

OBJECTIVE: To compare 2D and 3D radiomics features prognostic performance differences in CT images of non-small cell lung cancer (NSCLC). METHOD: We enrolled 588 NSCLC patients from three independent cohorts. Two sets of 463 patients from two different institutes were used as the training cohort. The remaining cohort with 125 patients was set as the validation cohort. A total of 1014 radiomics features (507 2D features and 507 3D features correspondingly) were assessed. Based on the dichotomized survival data, 2D and 3D radiomics indicators were calculated for each patient by trained classifiers. We used the area under the receiver operating characteristic curve (AUC) to assess the prediction performance of trained classifiers (the support vector machine and logistic regression). Kaplan–Meier and Cox hazard survival analyses were also employed. Harrell's concordance index (C-Index) and Akaike's information criteria (AIC) were applied to assess the trained models. RESULTS: Radiomics indicators were built and compared by AUCs. In the training cohort, 2D_AUC = 0.653, 3D_AUC = 0.671. In the validation cohort, 2D_AUC = 0.755, 3D_AUC = 0.663. Both 2D and 3D trained indicators achieved significant results (P < .05) in the Kaplan-Meier analysis and Cox regression. In the validation cohort, 2D Cox model had a C-Index = 0.683 and AIC = 789.047; 3D Cox model obtained a C-Index = 0.632 and AIC = 799.409. CONCLUSION: Both 2D and 3D CT radiomics features have a certain prognostic ability in NSCLC, but 2D features showed better performance in our tests. Considering the cost of the radiomics features calculation, 2D features are more recommended for use in the current study.     

Modeling landscape condition for biodiversity assessment—Application in temperate North America

Conservation decisions are well supported by predictive spatial models that indicate the relative ecological condition of a given place. The intent of this 90 m pixel landscape condition model is to use nationally available spatial data from the USA, Mexico, and Canada to express assumptions regarding the relative ecological effects of land uses on terrestrial natural communities and species. This approach emphasizes and updateable and transparent design which takes advantage of empirical biodiversity data from the USA to both calibrate and validate the model. Map layers depicting infrastructure, land use, and modified vegetation were each scored for site impact and distance decay, and then combined into one map surface. Field observations from Natural Heritage Programs, each scored for relative ecological condition (in categories A = excellent to D = poor), were used to calibrate distance decay parameters. Some 90,000 observations for at-risk species, invasive plant species, and natural communities were used for model validation. Statistically significant distinctions in ecological condition among validation samples were predicted by the resultant spatial model. Variation in landscape condition was then summarized by regional U.S. Landscape Conservation Cooperatives (LCCs) in terms of areas approximating A–D condition. Montane and desert LCCs scored on average much higher in area approximating “A” and “B” landscape condition, while LCCs with more substantial agricultural and urban footprints scored overwhelmingly within the “D” range of condition. Similar analyses illustrated range-wide scoring of landscape condition for major vegetation types across temperate North America.     

Urban permeation of landscapes and sprawl per capita: New measures of urban sprawl

Urban sprawl (dispersed urban development) has increased at alarming rates in Europe and North America over the last 50 years. Quantitative data are urgently needed in monitoring systems for sustainable development. However, there is a lack of reliable measures of urban sprawl that take into account the spatial configuration of the urban areas (not just total amount). This paper introduces four new measures of urban sprawl: degree of urban dispersion (DIS), total sprawl (TS), degree of urban permeation of the landscape (UP), and sprawl per capita (SPC). They characterize urban sprawl from a geometric point of view. The measures are related through TS = DIS × urban area, UP = TS/size of the landscape studied, and SPC = TS/number of inhabitants.The paper investigates the properties of the new measures systematically using 13 suitability criteria which were derived from a clear definition of urban sprawl as discussed in a previous paper. The scale of analysis is specified by the so-called horizon of perception. Second, the new measures are applied to three examples from Switzerland. Subsequently, the measures are briefly compared to other measures of urban sprawl from the literature. We demonstrate that UP is an intensive and area-proportionately additive measure and is suitable for comparing urban sprawl among regions of differing size, while SPC is most appropriate when comparing sprawl in relation to human population density. The paper also provides practical advice for calculating the new measures. We conclude that the new method is more suitable than previous methods to quantify the indicator “urban sprawl” in monitoring systems as this method distinguishes the phenomenon of urban sprawl from its various causes and consequences. This article is part II of a set of two papers.     

Potential use of green macroalgae Ulva lactuca as a feed supplement in diets on growth performance,feed utilization and body composition of the African catfish,Clarias gariepinus

This study aimed to evaluate the effects of diet containing the green macroalgae, Ulva lactuca, on the growth performance, feed utilization and body composition of African catfish Clarias gariepinus. Four experimental diets were formulated: D1 as a control group and D2, D3 and D4 which included 10%, 20% and 30% U. lactuca meal, respectively. 180 African catfish, weighing 9.59 ± 0.43 g, and with an average length of 11.26 ± 0.21, (mean ± SE) were divided into four groups corresponding to the different feeding regimes. The final body weight of the fish showed insignificant differences (P > 0.05) between the control and fish fed D2, whereas, there was a significant difference (P < 0.05) between these two diets compared with D3 and D4, with weights of 70.52, 60.92, 40.57 and 35.66 g recorded for D1, D2, D3 and D4, respectively. In the same trend significant differences were also evident in weight gain, specific growth rate and feed utilization. Fish fed with a diet containing 20% or 30% U. lactuca meal had poorer growth performance and feed utilization. Protein productive value, protein efficiency ratio, daily dry feed intake and total feed intake were also significantly lower in fish fed with D3 and D4 than in the control D1 and D2.Overall, the results of the experiment revealed that African catfish fed a diet with U. lactuca included at 20% and 30% levels showed poorer growth and feed utilization than the control group and fish fed diets containing 10% of U. lactuca.     

Rarity in large data sets: Singletons,modal values and the location of the species abundance distribution

Species abundance data in 12 large data sets, holding 10 × 10³ to 125 × 10⁶ individuals in 350 to 10 × 10³ samples, were studied. Samples and subsets, for instance the summarized data of samples over years, and whole sets were analysed. Two methods of the binning of data, assigning abundance values to classes for histograms, have been applied in the past: bins of equal size and bins of exponentially increasing size (‘octaves’). A hump in a histogram with exponential bins does not represent a mode of primary, non-transformed abundance values, but of log transformed abundance values. A proper interpretation of the hump is given. Moreover, the extrapolation to the left of a histogram with exponential bins, below an abundance of unity, lifting a ‘veil’, hiding species present in the community but absent from the sample, is rejected. The literature is confusing at these points and, as a result, prevents a proper view on the species abundance distribution. Applying bins of equal size, modal values equalled or approached unity. The number of singletons increased with sample size in some data sets but decreased in others. However, singletons remain present in large samples, subsets or sets, in agreement with the results on modal values. The relatively high number of singletons in small samples is no artefact of undersampling. The mode at unity, that is at the left end of the species abundance distribution, independent of scale (sample, subset or set), is an important statistical property of the species abundance distribution. Our results may have implications for theory development in community ecology: the selection and/or development of an accurate species abundance model, and, connected to this, the formulation of improved assembly rules, and the selection and/or development of more precise species richness estimators.