This was intended to be the penultimate model baseline model for 2016, however no further survey work was conducted or external records imported, so in effect this is the 2016 final baseline model set. Species presence has been taken from a combination of
The final model will not be released until January 2017 once the LRC data is released (dependant on the release of several commercial bat surveys). The individual species notes give some idea of the mix of presence data used in training and testing the models. This is important because models produced from all data sources (acoustic, trapped, and rescue) have a much higher confidence than models produced from just one data source. In particular; acoustic only models with Myotis and NSL (Noctule, Serotine, Leislers') guild carry a higher probability of data errors due to incorrect classification; mist net/harp trap only data runs the risk of bias where the trapping is usually done with a lure and in a woodland context, and rescue only data would be strongly biased to urbanised areas. So, a mixed data set provides the highest confidence and least bias potential in a model of this scale.
Some further notes on interpreting these models:
Note: click on the small image to get a 1024x720 image. If you would like a high resolution image of any models, then please contact the IoW Bat Group.
Species
Model
Notes
Bbar
The Barbastelle model shows very strong preferences for older deciduous woodland areas. Barbastelles have been found to be widespread across the Island, but the peak areas of activity are Parkhurst Forest and Brighstone/Mottistone areas. The presences used in the model training/testing have been derived from all three data sources and are all high confidence.
Eser
The Serotine bat is widespread through the Island, however there are some noticeable peaks in the model and survey data. The presences used in the model training/testing have been derived from all three data sources and are all high confidence.
Malc
The model is based on a very few presence recordings, mostly in the Parkhurst Forest and Marks Corner areas, hence the prediction favours the "Parkhurst Forest like" areas. If anyone visiting the Island thinks they have recorded one of these, then please get in touch via the Isle of Wight Bat Group.
Mbec
The Bechstein's bat is extremely difficult to detect, and most of the data driving this model has come from the Isle of Wight Woodland Bat Survey conducted by Ian Davidson-Watts. We have have found Bechsteins close to Bloodstone Copse, Borthwood Copse and Parkhurst Forest using static acoustic detection with the detection threshold highly optimised to detect their calls.
The model shows a strong link to deciduous woodland, however the areas we have detected Bechsteins are all rich with older Ash trees.
The presence data is mostly derived from trapping surveys, so is high confidence. There is a risk that the model has potential to be biased since these surveys have all been conducted in woodland areas only.
Mdau
Not surprisingly, most of the Daubenton bat presence records have been close to water, and the model is showing the peaks where there is both water and woodland. There is a significant proportion of rescue data from urbanised areas, hence there is also evidence of a link with the town and village areas.
Mmys
The Whiskered bat is common across all the deciduous woodland areas we have surveyed and this has shown up the model too. This species has also shown up in significant numbers in urban areas in the rescue data.
Mnat
Although widespread, we have not encountered large numbers of Natterers bats in the surveys this year. The model shows links with urban areas, which has been driven by the relatively high percentage of bat rescue presences in the data.
Nlei
Presence records for Leislers bats are limited, and have been clustered in three sites in the north of the Island. Indications are these bats are not present all year and numbers are low. We have no grounded bat or mist net records for this species, just two independent acoustic survey data sets.
Nnoc
The Noctule can be found almost anywhere on the Island where there is food to be had. They are very common over large water areas, but will also be found foraging in open fields, or commuting to/from their roosts. The majority of records are acoustic.
Pnat
The Nathusius' Pipistelle is a common visitor to the Island, and it often shows up in our recordings. The call is distinctive from the Common Pipistrelle because it peaks at around 38kHz. We've only ever recorded the species on relatively low lying ground, hence the large areas of blue that correspond to higher ground. The majority of presence data is from acoustic survey, followed by bat rescues.
Ppip
The Common Pipistrelle is totally pervasive through the Island, but there are definite areas of peak activity that have been found to match the current model predictions. The species is so pervasive in fact, that the current model has difficulty converging, and this was the lowest AUC of all the models (0.78).
Ppyg
Most of the Soprano Pipistrelle sightings have been in the North and East of the Island, and they have never shown up on high ground, hence the habitat bias preference to the lowland areas shown in this baseline. The majority of presences are from acoustic records.
Paur
Brown Long-Eared bats are extremely difficult to detect by acoustic survey, and like the Bechstein, we have tended to pick these up from long term static surveys only. So the bulk of the data is from bat rescues and the Isle of Wight Woodland Bat Survey data. There are strong links to urban areas driven by the rescue data and sheltered woodland areas.
Paus
We've not recorded any Grey Long-Eared bats this year, and the model is informed mostly by previous bat rescue and other local record centre data. You also need extremely clear recordings in order to be able to reliably separate the Brown and Grey Long-Eared Species... As with the Brown Long-Eared bat, there are strong links to urbanised habitats where there is significant roost potential.
Rfer
We've only recorded one Greater Horseshoe bat this year (Pelham Woods), so the model has been informed by the LRC data which includes bat rescues, hibernation roost surveys, and one acoustic recording made in Parkhurst Forest last year. They have only been found where there is roost potential and/or organic cattle grazing, and particularly around woodland edges onto the grazing areas. There are not really enough data points in this model to be certain of what it is telling us, however it is peaking in the areas where there is a high probability of hibernacula and roost potential, which is the main component of the presence data.