4 DISCUSSION
Here we provide one of the first empirical measures of animal occlusion
from different predator viewing angles and the first use of observer
height as a factor for acuity modelling. Measurements of camouflage from
arrays of different distances are increasingly being used in
publications on the functions of animal colour patterns (Caves et al.,
2018; Nokelainen et al., 2021). However, studies frequently fail to
account for occlusion in determining whether or not the viewing
distances used for visual models are biologically relevant. Our results
show that ‘openness’ at a human scale does not reflect openness at
scales relative to the nests (Allen et al., 2011), with nest occlusion
being more likely to limit detection distance than visual acuity.
Especially when viewed at the height of terrestrial predators, where the
scales of the clutches and observers render the 3D scene more akin to a
closed habitat, the bowl shape of the nest occluding the clutches at low
angles. The ability to obtain a broader array of unobscured viewing
angles, independent of physical height and topography, is a likely
driver of the increased acuity of aerial predators. Short terrestrial
predators should not be under selection for visual acuities capable of
segmenting objects from further than they are capable of observing.
Previous work investigating the search behaviour of foxes and domestic
dogs trained to find nests have found them to possess a short
localisation distance, less than 2m, for nests. Previous work
investigating the search behaviour of foxes and domestic dogs trained to
find nests has found them to have a short localisation distance of
<2 metres (Seymour et al., 2003; Storaas et al., 1999). Both
our ΔS measurements and occlusion measures support this observation.
Discrimination of the clutch outline at short distances is likely to be
the mechanism of egg detection for all, barring the few poorly
background-matching background matching clutches. Nests with greater
visibility (less vegetated) were also found to have better colour match
in the corvid visual model. Whether the increased differences were due
to higher selection intensity when less occluded or limitations in the
avian egg colour palette’s ability to match live vegetation is difficult
to disentangle with our current dataset (Hanley et al., 2015).
Previous research on landscape effects on lapwing nest success has shown
that increased proximity to taller ground vegetation, being at a greater
distance from the tree line and having surrounding bodies of water
decrease the risk of nest predations. The lapwings within our study
system were shown to nest preferentially in local habitats with higher
3D variation at scales above the size of the clutch. Habitats that
feature depressions and topography (plough, cattle and horse grazing)
with similar scales to their nests should decrease lapwing predation by
increasing the amount of noise at the scales relative to nests
(Swaisgood et al., 2018). Existing guidelines for creating suitable
lapwing nesting sites, promoted by the UK conservation organisations
(e.g., RSPB, BTO and GWCT), recommend fields with short patchy
vegetation in pastoral sites (Ausden and Hirons, 2002; Smart et al.,
2013). Analysis of lapwing habitat structure with our 3D scans supports
this preference for patchy local sites with 3D variation above the scale
of their nests. These results also emphasise previous work advising the
avoidance of grazing species that create homogenous and flat vegetation,
such as sheep (Winter et al., 2005). The null scans of the arable sites
were more similar to those of the nests than the pastoral sites. Chalk
arable sites offer both better colour match and local 3D variation match
to the lapwings’ nests. While not significant, these sites had the
lowest proportion of predations, 0, but were also under intense predator
control. Northern lapwing populations have long been associated with
spring cropland throughout Eurasia (Galbraith, 1988; Salek and Cepáková,
2006). Selection of these habitats has been thought to be and is likely,
driven by the large-scale match to the locally preferred background 3D
and colour features found naturally within wet grassland. Nesting
preference at these sites may be sub-optimal for survival at later
stages of their life history, acting as a sensory/ecological trap, with
higher chick predation and lower food availability present within these
sites (Baines, 1990; Schekkerman et al., 2009).
Modelling occlusion with handheld 3D scanners can be a useful tool for
estimating an object’s visibility; however, it does not account for
taller features at greater distances. The nests of the sampled lapwing
were found in fields without much obstruction except at the boundaries
(hedgerows & forests) (MacDonald & Bolton, 2008). Other UK
ground-nesting waders, e.g., Eurasian curlew Numenius arquata and
redshank Tringa totanus , and populations of lapwing in more
forested areas are more likely to have visibility influenced by
structures further from the nest than in our 3D scans. Using large-scale
LIDAR scans in conjunction with fine scale scans could provide a broader
map of visibility and cover of nests (Lone et al., 2014). It is also
worth reminding that observing from lower visual angles will in of
itself influence the match to the surrounding background. Partial and
self-occlusion will reduce the visible area of the clutch and nesting
material and mask recognisable features such as the clutch’s shadow and
edge (Lovell et al., 2013; Webster, 2015). Future work should consider
measuring camouflage in the presence of obstruction and/or from
different visual angles. In particular, experiments measuring the
survival of sedentary objects, such as eggs or model animal targets,
where object motion and changes in the local 3D environment are less
prevalent an issue. Using of 3D multispectral models or
colour-calibrated video cameras may also provide potential alternate
technological solutions to the challenges of measuring visibility from
multiple viewing angles (Miller et al., 2022; Vasas et al., 2022).
However, these methods are slower and more computationally expensive
than our 3D phone scans. Finally, our study serves as a reminder of how
occlusion is integral to understanding the distances with which visual
systems can interact with natural objects and the adaptations required
to break camouflage from biologically relevant distances.