Modelling heterogeneity in the classification process in multi-species distribution models can improve predictive performance.

Species distribution models and maps from large-scale biodiversity data are necessary for conservation management. One current issue is that biodiversity data are prone to taxonomic misclassifications. Methods to account for these misclassifications in multi-species distribution models have assumed that the classification probabilities are constant throughout the study. In reality, classification probabilities are likely to vary with several covariates. Failure to account for such heterogeneity can lead to biased prediction of species distributions. Here, we present a general multi-species distribution model that accounts for heterogeneity in the classification process. The proposed model assumes a multinomial generalised linear model for the classification confusion matrix. We compare the performance of the heterogeneous classification model to that of the homogeneous classification model by assessing how well they estimate the parameters in the model and their predictive performance on hold-out samples. We applied the model to gull data from Norway, Denmark and Finland, obtained from the Global Biodiversity Information Facility. Our simulation study showed that accounting for heterogeneity in the classification process increased the precision of true species' identity predictions by 30% and accuracy and recall by 6%. Since all the models in this study accounted for misclassification of some sort, there was no significant effect of accounting for heterogeneity in the classification process on the inference about the ecological process. Applying the model framework to the gull dataset did not improve the predictive performance between the homogeneous and heterogeneous models (with parametric distributions) due to the smaller misclassified sample sizes. However, when machine learning predictive scores were used as weights to inform the species distribution models about the classification process, the precision increased by 70%. We recommend multiple multinomial regression to be used to model the variation in the classification process when the data contains relatively larger misclassified samples. Machine learning prediction scores should be used when the data contains relatively smaller misclassified samples.

Modelling temperature-driven changes in species associations across freshwater communities.

Due to global climate change-induced shifts in species distributions, estimating changes in community composition through the use of Species Distribution Models has become a key management tool. Being able to determine how species associations change along environmental gradients is likely to be pivotal in exploring the magnitude of future changes in species' distributions. This is particularly important in connectivity-limited ecosystems, such as freshwater ecosystems, where increased human translocation is creating species associations over previously unseen environmental gradients. Here, we use a large-scale presence-absence dataset of freshwater fish from lakes across the Fennoscandian region in a Joint Species Distribution Model, to measure the effect of temperature on species associations. We identified a trend of negative associations between species tolerant of cold waters and those tolerant of warmer waters, as well as positive associations between several more warm-tolerant species, with these associations often shifting depending on local temperatures. Our results confirm that freshwater ecosystems can expect to see a large-scale shift towards communities dominated by more warm-tolerant species. While there remains much work to be done to predict exactly where and when local extinctions may take place, the model implemented provides a starting-point for the exploration of climate-driven community trends. This approach is especially informative in regards to determining which species associations are most central in shaping future community composition, and which areas are most vulnerable to local extinctions.

Amphibian recovery after a decrease in acidic precipitation.

We here report the first sign of amphibian recovery after a strong decline due to acidic precipitation over many decades and peaking around 1980-90. In 2010, the pH level of ponds and small lakes in two heavily acidified areas in southwestern Scandinavia (Aust-Agder and Østfold in Norway) had risen significantly at an (arithmetic) average of 0.14 since 1988-89. Parallel with the general rise in pH, amphibians (Rana temporaria, R. arvalis, Bufo bufo, Lissotriton vulgaris, and Triturus cristatus) had become significantly more common: the frequency of amphibian localities rose from 33% to 49% (n = 115), and the average number of amphibian species per locality had risen from 0.51 to 0.88. In two other (reference) areas, one with better buffering capacity (Telemark, n = 21) and the other with much less input of acidic precipitation (Nord-Trøndelag, n = 106), there were no significant changes in pH or amphibians.

Relating juvenile spatial distribution to breeding patterns in anadromous salmonid populations.

1. Spatial within-population heterogeneity in density probably affects competition intensity and may have a fundamental role in shaping population dynamics and carrying capacity. This may be particularly relevant for organisms where limitations on juvenile mobility cause maternal choice of breeding locations to influence the spatial distribution of younger life stages. 2. In this study, we mapped redd locations and the resulting densities of juveniles the following year along the entire reach (9.2 km) of a river holding natural populations of anadromous salmonids (Atlantic salmon and brown trout). These data were used to quantify the spatial scale over which breeding influences juvenile densities, and hence becomes important for density-dependent processes. 3. Although the observed cumulative distributions indicated a relatively uniform distribution of breeding along the river, autocorrelation analyses identified spatial patchiness of both breeding and resulting juveniles on a local scale. Furthermore, cross-correlations suggested a close spatial relationship between distribution of redds and juveniles. 4. Using spatially explicit hockey-stick stock-recruitment functions, we found juvenile salmonid density to be mostly influenced by the amount of breeding upstream of a given location. This influence decreased rapidly within the first 75-150 m. Thus, female choice with regard to breeding location gave rise to a heterogeneous distribution of offspring on a spatial scale of almost two orders of magnitude finer than that of the whole population (9.2 km). 5. The results are consistent with smaller scale experimental studies of salmonids, and emphasize the role for maternal choice of breeding locations in causing substantial spatial heterogeneity in juvenile densities within natural populations. Due to effects of density heterogeneity on overall levels of competition, this adds another layer of complexity to the dynamics of salmonid populations even in populations where breeding appears to be relatively uniformly distributed through space, and potentially also for a range of other organisms where juvenile dispersal is constrained.