The effects of small sample size and sample bias on threshold selection and accuracy assessment of species distribution models

Species distribution models are used for a range of ecological and evolutionary questions, but often are constructed from few and/or biased species occurrence records. Recent work has shown that the presence‐only model Maxent performs well with small sample sizes. While the apparent accuracy of such models with small samples has been studied, less emphasis has been placed on the effect of small or biased species records on the secondary modeling steps, specifically accuracy assessment and threshold selection, particularly with profile (presence‐only) modeling techniques. When testing the effects of small sample sizes on distribution models, accuracy assessment has generally been conducted with complete species occurrence data, rather than similarly limited (e.g. few or biased) test data. Likewise, selection of a probability threshold – a selection of probability that classifies a model into discrete areas of presences and absences – has also generally been conducted with complete data. In this study we subsampled distribution data for an endangered rodent across multiple years to assess the effects of different sample sizes and types of bias on threshold selection, and examine the differences between apparent and actual accuracy of the models. Although some previously recommended threshold selection techniques showed little difference in threshold selection, the most commonly used methods performed poorly. Apparent model accuracy calculated from limited data was much higher than true model accuracy, but the true model accuracy was lower than it could have been with a more optimal threshold. That is, models with thresholds and accuracy calculated from biased and limited data had inflated reported accuracy, but were less accurate than they could have been if better data on species distribution were available and an optimal threshold were used.     

The foraging ecology of two pairs of congeneric demersal fish species: importance of morphological characteristics in prey selection

The feeding habits of two sympatric species pairs of demersal fish ( Mullus barbatus-Mullus surmuletus, Serranus cabrilla-Serranus hepatus ) which occupy the shallow coastal area (25–30 m) in Iraklion Bay were investigated from samples collected on a monthly basis (August 1990 to August 1992). Stomach content analyses revealed that all of them were carnivores, feeding mainly on benthic invertebrates, and that each species consumed a narrow range of prey species with no significant dietary overlap. The morphology of their feeding apparatus was compared to examine the effect of any morphological differences on food selection and resource partitioning between the fish species. The species could be distinguished on the basis of the size of their mouth gape, the number of gill rakers and the length of their intestine. This study shows that each species pair follows a different strategy segregating along food niche dimensions. In particular, M. barbatus and M. surmuletus segregate their feeding niche consuming different prey taxa with similar sizes whereas S. cabrilla and S. hepatus differ considerably with respect to the degree to which prey species contribute to their diets coupled with differences in mean prey sizes.     

Effects of species' ecology on the accuracy of distribution models

In the face of accelerating biodiversity loss and limited data, species distribution models – which statistically capture and predict species’ occurrences based on environmental correlates – are increasingly used to inform conservation strategies. Additionally, distribution models and their fit provide insights on the broad‐scale environmental niche of species. To investigate whether the performance of such models varies with species’ ecological characteristics, we examined distribution models for 1329 bird species in southern and eastern Africa. The models were constructed at two spatial resolutions with both logistic and autologistic regression. Satellite‐derived environmental indices served as predictors, and model accuracy was assessed with three metrics: sensitivity, specificity and the area under the curve (AUC) of receiver operating characteristics plots. We then determined the relationship between each measure of accuracy and ten ecological species characteristics using generalised linear models. Among the ecological traits tested, species’ range size, migratory status, affinity for wetlands and endemism proved most influential on the performance of distribution models. The number of habitat types frequented (habitat tolerance), trophic rank, body mass, preferred habitat structure and association with sub‐resolution habitats also showed some effect. In contrast, conservation status made no significant impact. These findings did not differ from one spatial resolution to the next. Our analyses thus provide conservation scientists and resource managers with a rule of thumb that helps distinguish, on the basis of ecological traits, between species whose occurrence is reliably or less reliably predicted by distribution models. Reasonably accurate distribution models should, however, be attainable for most species, because the influence ecological traits bore on model performance was only limited. These results suggest that none of the ecological traits tested provides an obvious correlate for environmental niche breadth or intra‐specific niche differentiation.     

Interactive effects of habitat selection, food supply and predation on recruitment of an estuarine fish

 Seagrass meadows are often important habitats for newly recruited juvenile fishes. Although substantial effort has gone into documenting patterns of association of fishes with attributes of seagrass beds, experimental investigations of why fish use seagrass habitats are rare. We performed two short-term manipulative field experiments to test (1) the effects of food supply on growth and densities of fish, and (2) effects of predation on the density and size distribution of fish recruits, and how this varies among habitat types. Experiments were conducted in Galveston Bay, Texas, and we focused on the common estuarine fish, pinfish Lagodon rhomboides. In the first experiment, replicate artifical seagrass and sand plots were either supplemented with food or left as controls. Recruitment of pinfish was significantly greater to seagrass than sand habitats; however, we detected no effect of food supplementation on the abundance of recruits in either habitat. Pinfish recruits in artifical seagrass grew at a significantly faster rate than those in sand habitats, and fish supplemented with food exhibited a greater growth rate than controls in both sand and artifical grass habitats. In our second experiment, we provided artificial seagrass and sand habitats with and without predator access. Predator access was manipulated with cages, and two-sided cages served as controls. Recruitment was significantly greater to the cage versus cage-control treatment, and this effect did not vary between habitats. In addition, the standard length of pinfish recruits was significantly larger in the predator access than in the predator exclusion treatment, suggesting size-selective predation on smaller settlers or density-dependent growth. Our results indicate that the impact of predation on pinfish recruits is equivalent in both sand and vegetated habitats, and thus differential predation does not explain the higher recruitment of pinfish to vegetated than to nonvegetated habitats. Since predators may disproportionately affect smaller fish, and a limited food resource appears to be more effectively utilized by fish in vegetated than in unvegetated habitats, we hypothesize that pinfish recruits may select vegetated habitats because high growth rates allow them to achieve a size that is relatively safe from predation more quickly. Received: 10 October 1996 / Accepted: 5 April 1997     

The effects of species manipulation on growth and survival of an assemblage of juvenile estuarine fish

Growth and survival of Cyprinodon variegatus, Fundulus heteroclitus, Menidia beryllina , and Lucania parva , held in combination with an omnivorous grass shrimp– Palaemonetes pugio , and a predatory diving beetle Tropisternus lateralis , were measured at 12–60% in outdoor static microcosms that simulate salt marsh ponds. We predicted that the low species richness of this abiotically harsh but highly productive habitat would lead to a high degree of interaction among species. The presence of the beetle had no effect on the fish or the shrimp. Removal of the shrimp similarly had no detectable effect on the fish. There was a trend in proportions of survival among three fish: C. variegatus>F. heteroclitus>M. beryllina . Survival of M. beryllina was greater when it was alone at low densities of its own species (333 v. 667 fish m ⁻³), and less when it was paired with F. heteroclitus . Interactions with other species diminished the growth of M. beryllina . Survival of F. heteroclitus was greater when it was alone at low density (333 m ⁻³), than when the same number were placed together with Menidia and Cyprinodon at a total fish density of 999 m ⁻³. There were no effects of removal of other species on survival or growth of C. variegatus at either 12'5 or 24%. This study showed that a complex array of interactions occurred among six common members of a salt marsh food web, but the degree of interaction Was less than We predicted.     

Optimal sampling design and the accuracy of occupancy models

We present general theoretical limits on the possible accuracy (mean squared error or MSE) of occupancy estimates for a large range of occupancy study designs with imperfect detection and confirm our theoretical results via a simulation study. In particular, we show that for a given total survey effort, the best possible MSE is driven by two design-related factors: the fraction of visits made at occupied sites (regardless of whether that occupancy status is known or not) and the number of visits made to each site with unknown occupancy status (ie, sites with no detections). The limits reveal that there is very little room for improvement over optimal implementations of the three existing occupancy design paradigms: standard design (visit S sites K times each), removal design (visit S sites up to K times each, halting visits to each site following a positive detection), and conditional design (visit S sites once, then resurvey sites with a positive detection an additional $K-1$ times). For the small portion of the occupancy-detection parameter space where improvement can be achieved, we introduce a new hybrid survey design with accuracy closer to the theoretical limit, which we illustrate by reanalyzing an existing coyote (Canis latrans) camera trap dataset. Our results provide new clarity and intuition regarding key factors of occupancy study design.     

Length-weight relationships of six tropical estuarine fish species from Pulicat lagoon,India

Length–weight relationships (LWRs) of six finfish species under two families of the order Perciformes collected from Pulicat lagoon (India) are presented in this study. The fish specimens were collected weekly from the catches of shore seine net (with a mesh size of 50mm at the main webbing and 10 mm at the cod end) operated in the lagoon at an average depth of 2.5 m by the local fishermen during the period from July 2018 to April 2019. This study reports the LWRs for five gobiid and one siganid fish species which have no earlier reports on LWRs. The calculated ‘b’ value ranged from 3.538 (Favonigobius reichei) to 2.677 (Oxyurichthys microlepis) and ‘a’ value varied from 0.0041 (Favonigobius reichei) to 0.0199 (Acentrogobius cyanomos). This study also reports the new maximum total length (TL_max) for the gobiid fish species Oxyurichthys microlepis.     

Past and present fish species recorded in the estuarine Lake Ichkeul,northern Tunisia

Lake Ichkeul in northern Tunisia is a Ramsar site, a MAB Biosphere Reserve and a UNESCO World Heritage Site. The system is one of the most important coastal wetlands in North Africa, especially as an over-wintering area for migratory birds, particularly Palaearctic waterfowl. The present study was aimed at diagnosing the status of fish species in Lake Ichkeul and documenting their annual and seasonal occurrence within the system. Fish samples were collected monthly at 22 sites from March 2011 to April 2012 using a variety of gears, including a dragnet, gillnet, trammelnet, frynet and beach-seine. Temperature and salinity measurements were taken at each site, while fishes were obtained from only 11 sites, representing mainly lacustrine stations. A total of 18 fish species belonging to 14 genera and 12 families were identified, with Mugilidae dominating with four species, followed by Syngnathidae with three species. This study represents a relative recovery in terms of fish species richness, since only 13 species were recorded here between 2003 and 2007. However, the numbers of fish caught, especially of the commercial taxa, seem to have declined in recent years, and the number of species currently present is much lower than that of several decades ago. Increasing human pressures, particularly reductions in catchment freshwater inputs, a breakdown in ecosystem connectivity, pollution and over-fishing appear to be the principal causes for the decline in fish abundance and diversity.     

Distribution of diatom species on an estuarine mud flat and experimental analysis of the selective effect of stress

The distribution patterns of seven dominant diatom species on an estuarine mud flat transect were related to salinity gradients and organic pollution. The temporal distribution of species can be explained partly by the seasonal variation in irradiance and temperature, and partly by the large discharges of organically polluted fresh water during the late autumn. The selective effects of stress factors, such as low or high salinity, high concentrations of ammonia and free sulphide, and high temperatures were studied by measuring: (1) the tolerance of natural diatom populations isolated from the sediment; (2) the tolerance and capacities of unialgal cultures: and (3) the effects of the stress factors on the species composition of the populations in sediment samples incubated in the laboratory. A low salinity (1‰), but not a high salinity (20‰), markedly altered the species composition of natural diatom populations kept in the laboratory. Navicula salinarum Grunow and N. cf. cryptocephala Kützing outcompete other species only at a salinity of 1‰, which is below the optimum salinity for growth of these species. High concentrations (2–4 mM) of ammonia inhibited the photosynthesis of N. phyllepta Kützing and N. flanatica Grunow and decreased the cell numbers of these species in mixed populations kept in the laboratory. N. salinarum and Gyrosigma fasciola (Ehr) Cleve were relatively ammonia-tolerant and reached their highest abundance during periods of large discharges of organically polluted water. The populations on the mud-flat stations that had black, sulphide-containing layers 1 mm below the surface of the sediment, were dominated by the relatively sulphide-tolerant Navicula salinarum and N. pygmaea Kützing.High values of irradiance and temperature were a selective factor explaining the absence of N. flanatica in summer. Uptake of organic substrates and the inhibitory effect of high population densities are discussed in regard to their possible modifying influence on the distribution of diatom species on the mud flat.     

The influence of freshwater inflows on spawning success and early growth of an estuarine resident fish species, Acanthopagrus butcheri

The influence of freshwater inflows and salinity on spawning success of black bream Acanthopagrus butcheri (Sparidae) was investigated over 2 years in a small estuary on the east coast of Tasmania, Australia. The individual spawning seasons experienced quite different freshwater inflows; 2004-2005 was characterized by low flows throughout the season whereas during 2005-2006 there were three relatively large discharge events in the first part of the season. Macroscopic gonad staging of adults was used to define the spawning season and daily increment analysis of otoliths from recently settled recruits was used to backcalculate spawning dates. Gonad staging indicated that adults were in spawning condition over a 3 to 4 month period during spring and summer. The timing and duration of successful spawning, however, differed markedly between years and was linked to the timing of freshwater inflows and salinity conditions, with successful spawning occurring during periods of low freshwater discharge and when salinities in the upper estuary were above c. 15. Growth rates of the recently settled recruits did not differ between years, nor did the timing of spawning within the season influence growth rates. While the latter finding was unexpected, especially given within season temperature variability, these results imply that by the onset of winter earlier spawned fish would be larger than later spawned individuals, potentially conferring advantages for survival and competition for food. Climate change predictions for eastern Tasmania indicate a decrease in river flows in spring and an increase during summer, potentially increasing environmental variability between and within years, with implications for spawning success and subsequent recruitment.     

Distribution and ecology of anthozoa in the estuarine region of the rivers Rhine,Meuse and Scheldt

Data on the ecology and distribution in the estuarine region of the rivers Rhine, Meuse and Scheldt are given for ten species of Anthozoa. These data have been correlated with several environmental variables, but especially salinity. Communication nr. 142 of the Delta Institute for Hydrobiological Research.     

Evaluating the sampling bias in pattern of subterranean species richness: combining approaches

We investigated the pattern of species richness of obligate subterranean (troglobiotic) beetles in caves in the northwestern Balkans, given unequal and biased sampling. On the regional scale, we modeled the relationship between species numbers and sampling intensity using an asymptotic Clench (Michaelis–Menten) function. On the local scale, we calculated Chao 2 species richness estimates for 20 × 20 km grid cells, and investigated the distribution of uniques, species found in only one cave within the grid cell. Cells having high positive residuals, those with above average species richness than expected according to the Clench function, can be considered true hotspots. They were nearly identical to the observed areas of highest species richness. As sampling intensity in a grid cell increases the expected number of uniques decreases for any fixed number of species in the grid cell. High positive residuals show above average species richness for a certain level of sampling intensity within a cell, so further sampling has the most potential for additional species. In some cells this was supported by high numbers of uniques, also indicating insufficient sampling. Cells with low negative residuals have fewer species than would be expected, and some of them also had a low number of uniques, both indicating sufficient sampling. By combining different analyses in a novel way we were able to evaluate observed species richness pattern as well as identify, where further sampling would be most beneficial. Approach we demonstrate is of broad interest to study of biota with high levels of endemism, small distribution ranges and low catchability.     

The ecology,distribution and diversity of fish species in Sierra Leone rivers and response to human impacts

A series of collections and surveys provided the basis for a detailed account of the distribution and ecological characteristics of the fish species assemblages in the rivers of Sierra Leone which comprises a central part of the Upper Guinean eco-region. Within and between river differences were analysed and a high degree of statistical similarity was found between the species composition of the major rivers. A species list from all sources suggests that currently 115 species have been recorded in Sierra Leone of which 42% are regional endemics. Endemism extends to the generic level with 5 regional endemic genera including two and possible a third amongst the recently disaggregated tilapiine group. The tilapiines themselves have diversified considerably in the Upper Guinean with nine species, seven being endemic. The drivers of this considerable diversification generally in the Upper Guinean are discussed.Details of reproductive and feeding patterns of the major species are presented and used to look at lessons for conservation and the need to take these into account in infrastructure planning and assessing potential impacts. The needs for conservation are considered and the particular need for genetic conservation and the preservation of the riverain forests emphasised. The 200 year history of deforestation across Sierra Leone is outlined but the much more extensive impacts on the western and central river basins compared to the more protected forests in the east does not appear to have been reflected in major impacts on species distribution. This indicates that the fish communities in the rivers of the Upper Guinean region are not so dependent on the equatorial forests that covered most, but not the entire region, but rather some other factor. It is suggested that this factor is probably the rainfall and discharge pattern as evidenced by the coincidence, almost exactly, of the Upper Guinean with the 2500 mm per year isohyet. Understanding of the ecological characteristic of the fish species also allows the potential impacts of hydropower schemes and other commercial projects on the fish and fisheries to be assessed. Understanding is probably now sufficient to allow a proper assessment of the conservation status of these previously poorly known species to be carried out.     

The relationship between biogeography and ecology: envelopes,models, predictions

This paper reviews ideas on the relationship between the ecology of clades and their distribution. Ecological biogeography represents a tradition that dates back to ancient times. It assumes that the distribution of organisms is explained by factors of present environment, especially climate. In contrast, modern systematics, following its origins in the Renaissance, concluded with Darwin that ‘neither the similarity nor the dissimilarity of the inhabitants of various regions can be accounted for by their climatal and other physical conditions’. In many cases, species distribution models – ecological niche models – based on the current environment of a species (its environmental envelope) fail to predict the actual distribution of the species. In particular, they often over‐predict distributions. In addition, a group's niche often varies in space and time. These results provide valuable evidence that Darwin was correct, and many ecologists now recognise that there is a problem with the niche theory of distribution. Current ecological processes explain distribution at smaller scales than do biogeographical and evolutionary processes, but the latter can lead to patterns that are much more local than many ecologists have assumed. Biogeographical phenomena often occur at a much smaller scale than that of the Wallacean regions. In areas that have been subjected to marine inundation or intense tectonism, many centres of endemism are only tens of kilometres across. © 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 115 , 456–468.     

The iterative ensemble modelling approach increases the accuracy of fish distribution models

Methodological absences, i.e. when a species is not detected although it is actually present, are known to reduce the prediction accuracy of species distribution models (SDMs). To deal with this problem, we assessed whether a new iterative ensemble modelling (IEM) approach better predicts the spatial distribution of a set of 31 freshwater fish species, exhibiting a wide range of prevalence and methodological absences. Model efficiency was compared using one threshold‐independent (AUC) and three threshold‐dependent indicators of model predictive performance: the percentage of misclassified sites; the Kappa index; and the True Skill Statistic. We then reconstructed species assemblages from individual species predictions and compared observed assemblages to those predicted using EM and IEM using the Jaccard index. Compared to an EM approach, IEM improved model predictive performance for most difficult‐to‐detect species. The iterative approach outperformed EM at modelling the distribution of difficult‐to‐detect species, provided that presence data are representative of the niche of the species. At the assemblage level, the discrepancy between observed and IEM predicted assemblages was significantly lower than that between observed and EM predicted assemblages, showing that IEM can be used to predict the distribution of entire species assemblages. The IEM approach provides a way to consider difficult‐to‐detect species in species distribution models.