The landscape of fear as a safety eco-field: Experimental evidence

doi:10.21203/rs.3.rs-2275271/v1

Download PDF

Research Article

The landscape of fear as a safety eco-field: Experimental evidence

https://doi.org/10.21203/rs.3.rs-2275271/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 02 Mar, 2023

Read the published version in Biosemiotics →

You are reading this latest preprint version

In a development of the ecosemiotic vivo-scape concept, a ‘safety eco-field’ is proposed as a model of a species response to the safety of its environment. The safety eco-field is based on the ecosemiotic approach which considers environmental safety as a resource sought and chosen by individuals to counter predatory pressure.

To test the relative safety of different locations within a landscape, 66 bird feeders (BF) were deployed in a regular 15 x 15 m grid in a rural area, surrounded by shrubs, small trees, hedgerows, and buildings. On each of 48 days in November 2021 and February and March 2022, dried mealworms were placed on each BF and counts of larvae at each BF were made at noon and dusk. The European robin (Erithacus rubecula) and the great tit (Parus major) were the most regular visitors to the BFs. Land cover at each BF was recorded. Bird behaviour at the BFs was noted from direct video recordings of the birds at nine selected BFs, totalling 32 daily sessions in March. The different behaviours of the European robin and the great tit were observable.

The safety eco-field changed according to the month and the time of day. The distance of the BF from the woodland edges seemed to be important only in the morning. In the afternoon, BFs that were more distant from the woodland edges received the highest number of visits.

A significant relationship between land cover and the number of mealworm larvae removed from the BFs was observed. Within the grid of BF, three regions were distinguishable which were related to land cover in the safety eco-field process. The experimental framework confirms the adequacy, at least for birds that have cryptic predators, to map the landscape as a proxy of safety resource. Weather conditions were found to influence the number of mealworms removed, but this requires further investigation.

From the video recordings it was noted that the European robin visits were distributed throughout the day without apparent temporal preferences, while the great tit visits were concentrated in the central part of the day. This result has the limitation of the short period of observation (March) and should be extended to the entire period of the experiment to eventually capture seasonal variation.

The experimental evidence obtained confirms that the ecosemiotic-based models of safety eco-field are an efficient approach to explain bird feeding preferences and behaviours.

ecosemiotics

safety eco-field

cryptic predators

great tit

European robin

bird feeding

According to a recent ecosemiotic theory, the vivo-scape (Farina and James 2021) – defined as ‘a field of existence or domain of an “ecological species”’ – results from the direct semiotic relationship between organisms and their operational environment. This relationship is represented by a cybernetic loop in which energy, matter, and information are exchanged between the external and internal environments of organisms. These exchanges become an integral part of the species-specific environment. Therefore, each taxon is an essential part of a relational mode which is generally expressed as “the environment”.

The primary driver of a vivo-scape is the need of each individual to obtain the resources essential to stay alive and perpetuate life. Each taxon seeks to obtain resources from the external environment through a series of events in which they use perceptual-cognitive and semiotic mechanisms of the eco-field (Farina and Belgrano 2004, 2006; Farina 2012).

The eco-field paradigm has been proposed as a tool for understanding the perceptual assessment of an environment by birds (Alderman and Hinsley 2007), to improve the concept of the ecological trap (Gilroy and Sutherland 2007), to quantify and analyze the changing structures of landscapes (Kent 2007), and to study species adaptations (for example, in carabid beetles) (Pizzolotto et al. 2009). This paradigm was mainly used to better understand how species use food resources in a dynamic environmental mosaic. However, access to food resources is conditioned not only by their distribution and abundance but also by the need of each organism to reach a compromise between acquisition of these resources and their security. Security from predators is a resource provided by the geographic aspect and land cover in just the same way that food, water, material to build a nest, or other places in which reproduction takes place are required. Seen in this way, the eco-field paradigm is a dynamic trade-off between resources provided by the external environment driven by motivations from the internal environment (e.g., hunger, a need to reproduce, staying safe from predators, etc.). In the specific case of food acquisition, there is evidence that species are driven in accessing food resources by an internal need to integrate nutritional deficits, but access to resources often leads to overexposure to predators (Parker et al. 2009; Raubenheimer et al. 2009; Birnie-Gauvin et al. 2017).

Exposure to predation has long been considered as a discriminating element of the quality of a habitat which is formalized in ecology with the term “landscape of fear” (LOF). The "landscape of fear" is a paradigm of growing interest in ecological research (Moll et al. 2017, Gaynor et al., 2019). It is the result of the spatial and temporal variation in predation risk perception by prey (Brown and Kotler 2004, Laundré et al. 2010). In particular, this paradigm recognizes non-consumption (of the prey by the predator) as an important factor acting as a regulatory mechanism between prey density and predator interactions (Gotmark and Post 1996, Gotmark and Anderson 2005), thus contributing to shape diversity in the communities (Teckentrup et al. 2018, Lodberg-Holm et al. 2019). Two processes are active in the predator-prey non-consumptive interactions: prey avoid predators, based on their perception of temporal variations in risk (the risky times hypothesis) or based on their evaluation of spatial variation in risk (the risky places hypothesis) (Creel et al. 2008, Andrews et al. 2009, Yiu et al. 2021).

The predation risk effect can be evaluated in term of energy cost for prey staying vigilant and/or reduced access to resources (Creel 2011, Matassa and Trussell 2011). In particular, to avoid predators requires the prey to pay a cost in term of use of food resources (Orrock et al. 2013).

Predators are shaping agents of the species-specific vivoscape (Farina and James 2021) and avoidance behaviour by prey reduces their exploration and use of the resources, limiting de facto the spatial dimension of the species-specific ecological niche. The niche compression can potentially increase interspecific competition and also lead to physiological effects on the prey (Danhart and Lindström 2001). For instance, an impact of predatory risk has been observed in the behaviour of ungulates with cascade consequences on the vegetation (Creel et al. 2005, Gude et al. 2006, Forester et al. 2007, Beyer et al. 2007, Fischoff et al. 2007). In this group of animals, the availability and quality of foraging areas may be a factor in the level of physiological stress experienced as a result of differences in diet that may impact the glucocorticoid concentration (Creel et al. 2009).

The landscape of fear is also a concept relevant in human life (Martinez-Abrain et al. 2020, Evens et al. 2021, Lis and Iwankowski 2021, Read et al. 2021) as well as in wildlife ecology (Stoen et al. 2015), and extensively used to understand how species balance the reduction of predatory risk while maintaining access to resources (Mendes et al. 2020, Laundré et al. 2001, Proffitt et al. 2009, Buchholtz et al. 2021). The paradigm is also becoming central in applied ecology considering some interaction between people and the environment, especially to reduce carnivore-livestock conflict (Miller and Schimtz 2019). For example, acoustic playbacks of predators like wolves, dogs, and humans have been used to increase the perception of risk in wild ungulates and, thereby, reduce crop damage produced (Widen et al. 2022). Moreover, restoring the fear landscape becomes useful for the conservation of species and to preserve the ecosystem functionality. This is the case described by Manning et al. (2009). The introduction of the wolf into the Scottish Highlands has produced non-lethal effects on deer (Cervus elaphus) numbers and changes in foraging areas, with benefits to the entire ecosystem.

Despite the growing interest in the ecology of fear landscape, species-specific risk evaluation remains a difficult factor to approach quantitatively, especially when predators are not easily detectable by the prey. To get around this obstacle, we believe it is important to consider the relation of predator-prey from an ecosemiotic perspective (Maran and Kull 2014), i.e., shifting from the predatory risk assessment to the selection of safety areas. Converted into an ecosemiotic narrative, the fear landscape becomes a “safety eco-field” (Farina and Belgrano 2006) where species recognize the structure and spatial arrangement of vegetation as carriers of meaning to assure safety against predators. Safety is an immaterial resource that can be searched/achieved when the landscape assumes specific configurations. While the ecological paradigm has predatory risk as a recurring point, the ecosemiotic paradigm has safety as a resource that plays a co-primary role with food resources. The model is not new to science and harks back in part to J. Appleton's very popular human-related model of “prospect and refuge” in environmental psychology (Appleton 1996). According to this model, this affects the choice of a site from where the environment can be observed all around and, therefore, increasing the probability of not being surprised, and at the same time having a refuge in which to hide. This model is well suited to the behaviour of many bird species, which occupy marginal areas between enclosed and open spaces. Environmental safety is, therefore, considered a real resource that each species seeks by choosing areas as its habitat that meet the safety requirements. This resource is then intercepted through semiotic and cognitive mechanisms already described by the eco-field model (Farina and Belgrano 2006) and by the vivo-scape theory (Farina and James 2021).

There is evidence that, in response to a perceived lack of safety, birds, like every other creature, flee from dangerous environments, reduce visits to areas where the landscape configuration is not a favourable habitat (few places of shelter or no shelter) and where predatory risk is potentially high (van den Hout et al. 2010). In essence, this means that a species uses cognitive templates to “scan” the landscape and produce a cognitive map that they employ when the need for “safety [resource]” emerges (Farina 2012). For example, many species of birds use the woodland border as a refuge area, restricting exploration and utilization of open spaces that are the preferred hunting areas of birds of prey (Gotmark and Post 1996). However, currently, the contribution of the spatial and temporal arrangement of food resources and their competitive effects on safety landscape evaluation is not well defined.

While food resources are physical entities that can be measured in terms of spatial and temporal distribution, safety, being an intangible resource, can only be evaluated indirectly using the spatial arrangements of the landscape mosaic as a semiotic proxy. Moreover, food resources are not evenly distributed across a landscape (Shrader et al. 2008, Farina 2012). This fact causes problems in distinguishing prey responses to potential predators when selecting a habitat, as the predators are also driven by the availability of food resources there. However, if food resources are provided (experimentally) in a uniform way, the effect of the heterogeneous distribution of food is reduced and the perception of safety prevails. Consequentially, to reduce any interference of other factors such as food and to observe the animal’s strategy to avoid predators (Emerson et al. 2011), we created, experimentally, a landscape of evenly distributed food resources (mealworms). This experimental model has been used to investigate the safety eco-field of species of birds visiting bird feeders (BF). In this experiment, the quantity of food removed at BF through the day was used as an indicator of the level of safety of the land cover configuration surrounding each BF.

An experimental landscape was established, comprising 66 bird feeders (BF) deployed in a regular 15 m x 15 m grid across 2 ha of a rural landscape (Fig. 1) in a river terrace in Northern Italy. The BFs comprised a plastic plant saucer (12 cm in diameter), drilled to allow drainage of rainwater, and were fixed by a steel screw on the tip of an 80 cm bamboo pole (Fig. 2).

The study area is located in a rural river terrace called “Ortolano” (Fivizzano Commune (Northern Italy) (44°14'14.84" N, 10°07'08.94" E), 250 m a.s.l.). A detailed description of the study area and information on its natural history are reported in Farina (2018).

The area around the study site is composed of a mosaic of small fields cultivated with vegetables, vineyards, and grass (Fig. 1). It is surrounded by shrubs and abandoned fields that create an important habitat for ecotonal birds. Two rural houses and a wooden stable occupy approximately a fifth of the study area (Farina 2018). A small river (Rosaro) borders the western side of the study area. Despite the apparent presence of humans, this rural landscape attracts a great number of breeding and visiting birds favoured by the active cultivation, the presence of the river, and by an extended woodland. The active cultivation of this area and the great variety of spontaneous plants supply abundant food all year long to the bird community.

In order to encourage birds to visit BF, generous rations of mealworms were offered, prior to the start of the experimental period, for 15 days, during the last week of October and the first week of November 2021. For the first seven days dried mealworms were placed on the BFs and in the following 8 days defrosted mealworms were provided. It was noted that, when defrosting, the mealworms changed colour in a few hours from creamy-white to dark brown and lost the appearance of an intact larva. This probably makes them less attractive to the birds. This observation discouraged the use of frozen mealworm and, hence, only dried larvae were used during the experimental period.

At the end of the training period all remaining mealworms were removed. During the experimental period, eight dried mealworms (Tenebrio molitor) (Dretox organica™) were placed in each feeder at dusk (when feeding activity was suspended for the incoming night), providing the contemporary feeding resource at each feeder for the following day.

The experiment was run, initially, for eight days, November 11 to 19, 2021. Due to the imposition of COVID-19-related restrictions, the experiment had then to be suspended until the end of January 2022. The experiment resumed on January 31, 2022 and ran for 18 days until February 27, 2022 and resumed again for a further 22 days between March 1 to March 27, 2022, totalling 48 days of field experiments. To better investigate the effects of the season on the mealworm removal, data were aggregated according to three sessions (November, February, and March). The count of 31 January was incorporated for convenience in the February period.

The level of use of each feeder was ascertained twice a day at approximately 12:00 noon (labelled in the data processing as “Morning”) and, to coincide with dusk, at around 18:00 hours (labelled as “Afternoon”) by counting the number of mealworms removed from each BF. The sum of the mealworms removed during the Morning and the Afternoon was indicated as “Daily”.

It took 30 minutes to count the mealworms at the feeders at noon and one hour at dusk to count the residual number of mealworms and to replenish the BFs. During bad weather conditions (heavy rains and strong winds), the counts were suspended, losing 16 days in all. On three days (21 and 27 February, 17 March), counts were limited to the morning due to bad weather in the afternoon.

A nominal code based on the distance from edges was assigned to each BF: 1 = close to the edge, 2 = intermediate, and 3 = furthest from the edge (Fig. 3). With the category of edge, we included shrubs, trees, hedgerows, and gardens close to the buildings. These three categories of distances were used as the independent variable to map the safety eco-field (Fig. 3). The shape of the safety eco-field was estimated using the number of mealworms removed from the BFs by birds: the higher the number of mealworms removed at a BF, the higher the perceived safety of the location from predation.

In order to confirm the species visiting the BFs and their temporal distribution during the day, nine BFs of those most used by the birds were selected for video recording using SQ11 Full HD 1080P ™ video cameras (Fig. 2; Table 6s). Due to the low autonomy (one hour) of the internal battery, video cameras were replaced every hour. BFs were supplied with ad libitum mealworms every hour at the same time when the video cameras were being replaced.

The cameras operated from 07:00 to 18:00, recording a file every 5 minutes. VideoPAD (NCH Software™) was used to collate every 5 minutes shots in files for each of the one-hour sessions. The 3DF Zephyr lite™ was used to extract a frame every second from the video to identify and count the number of birds at the BFs. A total of 715,320 frames was visually inspected (AF) and bird species were identified. Due to the restricted temporal window during which the video cameras were deployed (March 2022), it was not possible to make a comparison across seasons.

The climatic parameters (temperature, rain, wind speed and direction) were collected from the “Valle” forecast station (Davis Vantage Pro2™) situated 800 m from the study area in Fivizzano, 328 m a.s.l., [44°13'47" N-10°07'12" E] (Table 1s).

Table 1

Level of significance (Kruskal-Wallis non-parametric ANOVA test) obtained by comparing the difference in the average number of mealworms removed according to BF categories of distance from borders during the Morning count (circa 12:00 noon) and the Afternoon count (circa 18:00 hrs).
K-W ANOVA	p
Nov-21	Morning	6.749	0.034
	Afternoon	6.16	0.046
Feb-22	Morning	18.15	0.0001
	Afternoon	5.28	0.071
Mar-22	Morning	17.28	0.0002
	Afternoon	18.15	0.0001

A DJI™ mini 2 drone was used for the detailed survey of the landscape. DJI Fly ™ application release 1.5.10 for IOS system (iPhone 11™) was used to drive the drone and to take pictures of the study area from a height of 60 m. To create a georeferenced orthomosaic of the study area, 166 pictures were taken during a single flight and combined using the open source WebODM® (OpenDroneMap™) drone mapping software (https://www.opendronemap.org/webodm/).

The same drone was used to survey the land cover centred on each BF, taking images from an altitude of 25 m. A circular area of 225 m² with a radius of 7.5 m around each BF was analyzed and the land cover of each BF was quantified by superimposing each BF image with a gridded square mask composed of 49 sub-squares of 2 x 2 m. Examples of such BF-centred images are presented in Fig. 4. The total land cover of the study area was obtained by combining the data from all 3,234 survey squares (49 x 66 BF). When multiple land covers were present in a square, the dominant one (> 50%) was entered into the first of three columns, the second column was used for a percentage of cover < 50% and > 5%, and the third column was used for land cover < = 5% of the entire surface of each square.

In order to evaluate the importance of the land cover in the choice of different BFs by bird species, the land cover classes observed within the study area were aggregated using a cluster analysis and by imposing three levels of aggregation (high: 3 clusters; medium: 5 clusters; and low: 10 clusters). The correspondence between the number of larvae removed in each BF differentiated by month and by count (Morning, Afternoon) and the three levels of land cover aggregation as the result from the cluster analysis was verified. Statistical analysis was performed using JMP™ (Sall et al. 2017). The Kruskal-Wallis ANOVA, a non-parametric statistical test, was used to test the significance between data. Cluster analysis was performed using the clustering function of SPSS™ [amalgamation (linkage) rule: Ward’s method; distance measure: square Euclidean distance].

In addition to the systematic methods set out, the study area was observed casually throughout the study. This was a continuation of such observations over many years by one of the authors (AF), on whose land the site is situated. These observations allowed identification of the bird species present at the site and salient aspects of their behaviour.

A non-consumptive human apex-predator, a Eurasian sparrow hawk (Accipiter nisus), and feral cats are the three major agents of fear landscape affecting the bird community living in the study area. We reduced human intrusion during the experiment by limiting the presence at the BF only during the noon and dusk inspections. Cats have been observed in the experimental area, but we have not observed direct interactions with birds at the BFs.

Several field observations of the study site conducted during the experimental sessions confirmed the sparrow hawk as the only diurnal predator. During the February session, the sparrow hawk was observed chasing resident breeding species, including blackbird (Turdus merula), cirl bunting (Emberiza cirlus), and chaffinch (Fringilla coelebs).

Based on the casual observations of the site, it was ascertained that only the robin and the great tit were regular visitors of the BF, with the sole exception of an individual black redstart, who for a single day went to BF 3,12, repeatedly removing some larvae. Only once has a wren been seen to take to the perch 2,6 but without eating. This was despite the presence, in the study areas, of several species of birds which would potentially feed on mealworms, for example, wren (Troglodytes troglodytes), house sparrow (Passer domesticus), blue tit (Cyanistes caeruleus), blackcap (Sylvia atricapilla), and jay (Garrulus glandarius).

As it was not possible to differentiate the number of mealworms taken by the European robin from those taken by the great tit, a safety eco-field for the two species combined was established (see Sections 3–5 below).

1 Number of mealworms removed during the day

There are significant differences in the numbers of mealworms removed between November, February, and March in both the mornings (ANOVA Chi sqr [n = 65, df = 2] = 18.179, p = 0.00011) and afternoons (ANOVA Chi sqr [n = 65, df = 2] = 19.75, p = 0.00005), with a maximum in March in the morning and in the afternoon in November (Fig. 5).

2 Number of mealworms removed in relation to the distance from the border

There were significant differences between the number of mealworms removed in the three distance categories of BF in November, February, and March and in the morning and afternoon counts with one exception – the afternoons in February (Table 1 and Fig. 6).

In the mornings, the highest number of mealworms removed was in distance category 1 (the closest to the edge) and lowest distance category 3 (the furthermost from the edge). In the afternoons, the highest number of mealworms removed was in distance category 3 (Fig. 6).

3 Relationship between sub-areas and removal of mealworms

We identified three sub-areas (A, B, and C) inside the study area (with the following coordinates- A:2,2 to 5,6; B:1,7 to 7,11; C: 3,12 to 6,18), Sub-area A (the south of the study area – the left-hand portion of Fig. 1) where the number of mealworms removed was the highest, sub-area B (central part), where the number is intermediate, and the third (sub-area C) to the north (the right-hand side of Fig. 1), where the fewest mealworms were removed. (Fig. 1). These three sub-areas show a significant difference in number of mealworms removed in all three seasons and for the mornings and afternoons (Table 2).

Table 2

Mean number of mealworms removed during morning and afternoon periods and aggregated per sub-areas (A, B, and C).
			Mean	SD	Chi square	p
Morning	November	A	6.923611	1.451507	14.9403	0.0006
		B	6.305556	1.7453
		C	4.189909	2.40846
	February	A	7.135802	0.898886	15.0872	0.0005
		B	4.786008	2.608323
		C	3.444618	3.091055
	March	A	7.183586	1.018417	9.7349	0.0077
		B	5.527978	2.022727
		C	5.820686	2.147956
Afternoon	November	A	0.652778	1.00051	16.0695	0.0003
		B	1.236111	1.141089
		C	2.288503	1.346589
	February	A	0.934444	0.491232	11.0594	0.004
		B	1.557407	0.731419
		C	0.954286	0.855141
	March	A	0.387778	0.542582	6.5375	0.0381
		B	0.994074	0.916449
		C	0.643684	0.674934

4 Relationship between land cover and removed mealworms

In total, 42 typologies of land cover were found in the study areas, of which the most frequent were: grassland “lawn” (47% of the entire dataset), forbs were located at the edges of the study site and on slopes (19%), crop fields 16%, black locust (Robinia pseudoacacia) (3%), olive (Olea europaea) (2%) and bramble (Rubus spp.) (2%). The frequency of each land cover is summarized in Table 2s. The three frequency categories (1: >50%; 2: <50 > 5%; 3: <5%) were used to create BF clusters with 3, 5 and 10 levels of aggregations (Table 3s, 4s and 5s).

Table 3

Relationship between mealworm removed and BF aggregated in 3, 5 and 10 clusters according to the land cover.
		Morning						Afternoon
		Nov		Feb		Mar		Nov		Feb		Mar
Cluster	#	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD
1	16	5.8717	2.5304	5.8264	2.8303	7.1458	0.9473	1.4202	0.3514	1.0456	0.6580	0.2938	0.4031
2	13	5.7981	1.6198	3.6282	2.2762	4.2742	1.7832	1.5673	0.3775	1.7646	0.6875	1.7169	0.7678
3	37	5.4831	2.4136	5.1426	2.8913	6.2203	1.9188	1.4291	0.2237	1.0900	0.7418	0.5222	0.6028
		Chi square	p	Chi square	p	Chi square	p	Chi square	p	Chi square	p	Chi square	p
		0.66042	0.7392	5.0036	0.0819	16.2528	0.0003	0.374	0.895	9.4986	0.0087	23.1091	0.001
		Morning						Afternoon
		Nov		Feb		Mar		Nov		Feb		Mar
Cluster	#	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD
1	1	7.8750	.	6.7222	.	6.9091	.	0.1250		1.5600		0.4000
2	8	5.8839	2.3650	5.0764	3.2404	6.9069	1.2622	1.6317	1.6203	1.3288	0.7480	0.4038	0.5299
3	7	5.6429	1.8434	3.3333	2.3393	3.9647	2.1584	1.6964	1.3087	1.7671	0.5216	1.7657	0.8072
4	8	7.0625	1.0938	6.6458	1.9000	6.9345	1.1303	0.6875	0.7196	1.1725	0.9720	0.5425	0.8587
5	42	5.2679	2.4393	4.9246	2.8943	6.0632	1.9385	1.5610	1.3708	1.0967	0.7295	0.6193	0.6914
		Chi square	p	Chi square	p	Chi square	p	Chi square	p	Chi square	p	Chi square	p
		4.9477	0.2927	6.4102	0.1705	9.475	0.0503	3.6446	0.4562	6.2891	0.1786	12.7315	0.0127
		Morning						Afternoon
		Nov		Feb		Mar		Nov		Feb		Mar
Cluster	#	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD	mean	SD
1	1	7.8750		6.7222		6.9091		0.1250		1.5600		0.4000
2	6	5.8750	2.0218	6.3796	1.7038	7.5729	0.3616	1.7292	1.5581	0.9817	0.6505	0.1833	0.2058
3	3	8.0000	0.0000	8.0000	0.0000	7.7576	0.2099	0.0000	0.0000	0.5100	0.0173	0.0000	0.0000
4	4	4.7054	2.1257	2.0694	2.7787	6.0411	1.3247	2.7009	1.1177	1.7025	0.5282	0.7125	0.6033
5	2	4.0000	5.6569	7.9722	0.0393	7.2745	0.5168	0.0000		0.4700	0.0424	0.1750	0.2475
6	8	6.6719	1.4046	4.3542	2.4792	4.7486	2.0027	0.9844	0.9220	1.5700	0.6838	1.6950	0.8207
7	5	4.4000	0.6697	2.4667	1.4308	3.5152	1.1617	2.5000	0.8197	2.0760	0.6345	1.7520	0.7666
8	1	7.0000		5.0000		5.8636		1.0000		1.1300		1.1400
9	16	6.8516	1.5970	6.9826	1.2473	7.1752	0.9748	0.8438	1.1924	0.9931	0.5953	0.2913	0.4397
10	19	4.2829	2.4885	3.7164	3.1617	5.4130	2.2487	1.9474	1.3796	1.1742	0.8881	0.6742	0.6860
		Chi square	p	Chi square	p	Chi square	p	Chi square	p	Chi square	p	Chi square	p
		22.5677	0.0072	29.9528	0.0004	28.5577	0.0008	21.4718	0.0107	19.3475	0.0224	32.5087	0.0002

Table 4

Relationship between temperature, rain and wind and number of mealworms removed at Morning, Afternoon and at Daily count.
	Pearson r	Spearman’s rho	p	Chi square	p
Morning
C°	0.0133	-0.005	0.9
rain mm	-0.2839	-0.16	0.273
wind (km/hour)	-0.3036	-0.2324	0.112
wind direction				20.1262	0.0053	NE
Afternoon
C°	0.2947	0.3063	0.0407
rain mm	0.6821	0.4952	0.0005
wind (km/hour)	0.0816	0.0542	0.7237
wind direction				8.0765	0.3259
Daily
C°	0.16785	0.2775	0.1012
rain mm	0.1449	0.1921	0.2063
wind (km/hour)	-0.1763	-0.1231	0.4203
wind direction				17.6818	0.0135	SW

The level of aggregation that showed a significant constant relationship with the number of larvae removed in the three months and in both counts was the 10-clusters model. The 5-clusters model showed significant relationship only in March for both morning and evening counts. The 3-clusters model showed a significant relationship in the morning in March and in February and March in the afternoon.

5 Relationship between weather conditions and mealworms removed

As reported in Table 4 the weather seems to play a relevant role in the removal of mealworms. A rise in temperature and an increase in rain events significantly increase the number of mealworms removed in the afternoons. Wind direction is a discriminant factor in the mornings but not in the afternoons. In the morning, wind from the north-east and, at a daily scale, from the south-west are the dominant wind conditions that contribute to differentiate the effects of wind direction on the rate of mealworms removed.

6 Species and their spatio/temporal distribution at BF, from video recordings

Within the selected BFs, there were significant differences in the number of visits by the European robin (Chi square 33.0376, Df 6, p < .0001); and the great tit (Chi square 18.2712, Df 5, p < .0026). The great tit made the highest number of visits to the BFs (Chi square 24.4093, Df:1, p < .0001). The number of visits at every hourly interval (from 7.00 a.m. to 6 p.m.) differs significantly for the great tit (Chi square 25.6454, Df 11, p < .0073), with a peak in the middle of the day (Fig. 7a, b) but this does not occur for the robin (Chi square 14.6573, Df 11, p < .1981) although a trend with a maximum in the mornings and in a minimum in the afternoons was observed.

In this study, we propose a paradigm shift from prey behaviour regulated by risk assessment mechanisms for a species based on fear, to one in which prey behaviour is based on the dynamic assessment of the level of safety due to species-specific perception of the spatial configuration of the landscape (a safety eco-field). The safety eco-field is defined as a spatial configuration that carries meaning in interpreting resources. That is to say that it is an ecosemiotic interpretation of landscape. Seen in this way, safety becomes a resource in the landscape and can be viewed in the same way as any other resource.

The safety eco-field is based on a cognitive template formed by an individual scanning their surroundings, searching for a spatial configuration of vegetation that satisfies their specific safety criterion. Safety is an intangible resource. Being one of many resources (e.g., food, mating, territoriality, etc.) in a landscape and an individual has to choose which resources to access at what time. All resources are revised/constrained by environmental conditions (e.g., climate, human disturbance) or by the physiological rhythms of species (e.g., seasonal, daily species-specific dynamics).

From our experiment it emerges that the distance from the edge, where the safety resource is high, is an important factor structuring the safety eco-field. In fact, it is evident that the European robin and the great tit discriminate between BFs (i.e., a food resource) according to the distance from edges. As the BFs are placed in a regular grid (15 m x 15 m) and as the species show a clear preference among the individual BFs according to their distance from the edge, it is reasonable to assume that this distance is sufficient for these birds to discriminate between resources at this scale. The assessment of the safety eco-field appears to be dynamic. In the morning it was smaller than in the afternoon. As resources nearer the place of high safety resource were depleted, these two species travelled further from the edge. This indicated a shift in the safety threshold probably affected by the necessity to reach the optimum body weight before the long and cold nights, which were a characteristic of the study site from November 2021 to March 2022 and to reduce the risk of starvation (Bonter et al. 2013). This observation illustrates the dynamic interaction of the internal and external environments. In this case, the internal environment (the need to build body weight) dominated the external environment (the species explored areas in which the safety resource was lower than what they had explored earlier in the day).

The perception of edge in birds seems more important than suspected and this has been confirmed by Rodriguez et al. (2001) who observed great tits avoiding edges in February with low light. Our results are not in accordance with this, and a further investigation is required to better understand this apparent discrepancy. Despite the fact that during dusk or dawn we never observed nocturnal predators in the study area, and occasional observations during the experiments confirm the presence of a sparrow hawk in the middle of the morning. Finally, sparrow hawk and feral cat predation may be affected by the presence of humans, thus reducing the predatory pressure on birds.

The variability observed in the number of mealworms removed from the BFs in November, February, and March probably reflects the dynamics of physiological requirements (Ebling and Barrett 2008), combined with the availability of natural resources. Again, this illustrates the interaction between internal and external environments and the dynamic assessment of eco-fields. This characteristic has been indirectly demonstrated in experiments by Fransson and Weber (1997) on the blackcap (Sylvia atricapilla). When this species was exposed to an increase in predatory risk, they adjusted their stopover behaviour accordingly. Some large mammals, like black bears (Ursus americanus), alter their movements according to the time of day and season. This species uses developed areas when human activity is low, demonstrating that they are able to navigate across a landscape by adapting their safety eco-field (Zeller et al. 2019).

The land cover surrounding each BF seems to contribute an explanation for the number of visits by the birds but the strong correlation between distance categories from edges and the nature of edges reduces the importance of land cover as a distinct independent factor. For example, the presence of bramble (Rubus agg) and acacia (Acacia pseudoacacia) in the edge seems to be a discriminant favouring visits to BFs, but this cannot be verified in this experiment because this vegetation was only present in some parts of edge and the edge was highly heterogeneous. It is also important to recognize that land cover was classified using human criteria and that this will not be fully coincident with how other animals like birds perceive the landscape, due to differences in the olfactory, acoustic, and visual senses.

There is evidence that during migration birds select the best stopover areas: birds navigate more frequently and longer in a safety area, avoiding, if possible, unsafe areas (Schekler et al. 2022). This ability is probably also extended to permanent residents and wintering species when navigating across habitats. However, the decision to remain inside the perceived safe area or to move outside it also depends on other functional resources, such as food, mating success, territorial patrolling, etc.

The difference found between the number of European robins and great tits visiting BFs probably depends on their life history (Palmer et al. 2017). A lower density of robins in the study area is probably due to winter territoriality (Lack 1939), and the flocking behaviour of great tits outside the breeding period (Morse 1978) probably explains the constant presence of this species at BFs during the period of video recording (March). However, the visual inspection of the videos confirmed that the great tit is dominant over the European robin (great tits consistently chased European robins away from the BFs) and hence had a more prolonged foraging behaviour. The significantly higher number of visits in the middle of the day by great tits could be explained by habits linked with a major abundance of their food resources and their ability to detect these resources (in natural conditions) when light and temperature are at their optimal seasonal level. It is well known that the European robin has crepuscular habits outside the breeding season, however, the trend that emerges from the results is contradictory to this as they show a uniform distribution throughout the day (Fig. 6a) (Lack 1939). This discrepancy is probably due to the different habits that emerge from populations located in different parts of their home range (Cabej 2012). The different number of visits to the BF of the two species could also be related to a difference in body size. Dierscke (2003) has found that stopover light birds were more exposed to predation than heavy ones. And this could be an additional hypothesis of the different behaviour between the European robin and the great tit at a BF.

The offering of food may influence the behaviour of birds because it represents a strong encouragement, especially when individuals are under predatory pressure elsewhere. In Spain free-ranging rodents were unexpectedly observed to use the live traps as refuges to escape predatory pressure (Hernandez et al. 2018). Although bird behaviour is different from micromammal behaviour, it is reasonable to hypothesize that birds can also be encouraged by the presence of food to inspect feeders (in this experimental case represented by the plastic dish) when they are under a predatory pressure. Unfortunately, we have no quantitative data on the direct predatory pressure of the sparrow hawk to corroborate this hypothesis: we have only sporadic observations of individuals’ hunting successes by some plucking posts.

Birds living in the study area probably use at least three typologies of safety eco-field to neutralize the presence of direct and indirect predation from sparrow hawks, humans, and feral cats. It is reasonable to suppose that the safety eco-field to prevent cat predation is more related to the height of vegetation close to a BF where this predator can be concealed. The safety eco-field from human presence is more related to a combination of visual and acoustic cues produced by human field activity. The safety eco-field to prevent sparrow hawk attacks is more linked to the spatial arrangement of trees, shrubs, and hedgerows surrounding the BFs. Although our model considers only the safety eco-field for a bird of prey (sparrow hawk), a trade-off between these three emerging cognitive templates certainly influences the behaviour of the European robin and the great tit. For instance, a similar combined effect has been described by Van Donselaar et al. (2018) in black-capped chickadees (Poecile atricapillus). The trophic control of ecosystems is the result of several drivers that influence the distribution and density of organisms (Schmitz et al. 1997, 2008, Creel and Christianson 2009, Iribarren and Kotler 2012). The presence of a predator produces a cognitive anticipatory stress that may have a great influence on the health status of species (Boonstra 2013) but it remains a difficult process to investigate (Seress et al. 2011). The European robin and the great tit have no possibility of adjusting their position in the landscape according to the continued presence of predators, as commonly happens, for instance, for ungulates, for this species, the safety eco-field that is based on the perception of landscape configuration as a carrier of meaning to track the “safety resource” seems more adapted to investigate the prey-predator dynamics than a risk assessment approach.

There are other factors that may produce modifications in the safety eco-field: the increase in fragmentation of forest habitats is responsible for a change in predatory pressure and consequently on the safety eco-field configuration (Whytock et al. 2020). Confidence at lack of predator is facilitated by the inter-individual interaction and can reduce the requirement of a safety eco-field (Wooster et al. 2021). The number of predators may alter the daily activity of the great tit. For instance, this species has been found to be more sensitive to two predators than to one (Krams 2000). Other factors can impact on the efficiency of a safety eco-field. For instance, the capacity of the great tit to react to the alarm calls of neighbouring individuals in the presence of a predator (Tilgar et al. 2022) can be reduced by anthropogenic noise that impacts on the safety eco-field. The decrease in risk evaluation by the great tit in urban areas compared to forests may be connected with human disturbance that distract birds (Vincze et al. 2019).

Viewing safety as a resource in a landscape allows safety to be considered alongside other resources. This in turn allows landscape resources to be managed and, hopefully, will result in increased biodiversity. Viewing landscape in an ecosemiotic way helps us understand how animals interact with their external environment, how those interactions are modified by changes in their internal environment, and how landscapes function for the benefit of species.

Aknowledgments

We would like to express our very great appreciation to Flora MacDonald-MacTavish for her valuable and constructive suggestions for the final version of the paper.

Alderman, J., & Hinsley, S. A. (2007). Modelling the third dimension: incorporating topography into the movement rules of an individual-based spatially explicit population model. Ecological Complexity, 4(4), 169–181.
Andrews, K. S., Williams, G. D., Farrer, D., Tolimieri, N., Harvey, C. J., Bargmann, G., & Levin, P. S. (2009). Diel activity patterns of sixgill sharks, Hexanchus griseus: the ups and downs of an apex predator. Animal Behaviour, 78(2), 525–536.
Appleton, J. (1996). The experience of landscape (p. 282). Chichester: Wiley.
Beyer, H. L., Merrill, E. H., Varley, N., & Boyce, M. S. (2007). Willow on Yellowstone's northern range: evidence for a trophic cascade? Ecological Applications, 17(6), 1563–1571.
Birnie-Gauvin, K., Peiman, K. S., Raubenheimer, D., & Cooke, S. J. (2017). Nutritional physiology and ecology of wildlife in a changing world.Conservation Physiology, 5(1).
Bonter, D. N., Zuckerberg, B., Sedgwick, C. W., & Hochachka, W. M. (2013). Daily foraging patterns in free-living birds: exploring the predation–starvation trade-off. Proceedings of the Royal Society B: Biological Sciences, 280(1760), 20123087.
Boonstra, R. (2013). Reality as the leading cause of stress: rethinking the impact of chronic stress in nature. Functional Ecology, 27(1), 11–23.
Brown, J. S., & Kotler, B. P. (2004). Hazardous duty pay and the foraging cost of predation. Ecology Letters, 7(10), 999–1014.
Buchholtz, E. K., Spragg, S., Songhurst, A., Stronza, A., McCulloch, G., & Fitzgerald, L. A. (2021). Anthropogenic impact on wildlife resource use: Spatial and temporal shifts in elephants’ access to water. African Journal of Ecology, 59(3), 614–623.
Creel, S. (2011). Toward a predictive theory of risk effects: hypotheses for prey attributes and compensatory mortality. Ecology, 92(12), 2190–2195.
Creel, S., Winnie Jr, J., Maxwell, B., Hamlin, K., & Creel, M. (2005). Elk alter habitat selection as an antipredator response to wolves. Ecology, 86(12), 3387–3397.
Creel, S., Winnie Jr, J. A., Christianson, D., & Liley, S. (2008). Time and space in general models of antipredator response: tests with wolves and elk. Animal Behaviour, 76(4), 1139–1146.
Creel, S., & Christianson, D. (2009). Wolf presence and increased willow consumption by Yellowstone elk: implications for trophic cascades. Ecology, 90(9), 2454–2466.
Creel, S., Winnie Jr, J. A., & Christianson, D. (2009). Glucocorticoid stress hormones and the effect of predation risk on elk reproduction. Proceedings of the National Academy of Sciences, 106(30), 12388–12393.
Dänhardt, J., & Lindström, Å. (2001). Optimal departure decisions of songbirds from an experimental stopover site and the significance of weather. Animal Behaviour, 62(2), 235–243.
Dierschke, V. (2003). Predation hazard during migratory stopover: are light or heavy birds under risk? Journal of Avian Biology, 34(1), 24–29.

No competing interests reported.

Download PDF

Journal Publication

published 02 Mar, 2023

Read the published version in Biosemiotics →

Editorial decision: Major revision
13 Jan, 2023
Reviews received at journal
23 Nov, 2022
Reviewers agreed at journal
18 Nov, 2022
Reviewers invited by journal
18 Nov, 2022
Editor assigned by journal
17 Nov, 2022
Submission checks completed at journal
16 Nov, 2022
First submitted to journal
15 Nov, 2022

You are reading this latest preprint version

The landscape of fear as a safety eco-field: Experimental evidence

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Experimental Design And Materials

Results

1 Number of mealworms removed during the day

2 Number of mealworms removed in relation to the distance from the border

3 Relationship between sub-areas and removal of mealworms

4 Relationship between land cover and removed mealworms

5 Relationship between weather conditions and mealworms removed

6 Species and their spatio/temporal distribution at BF, from video recordings

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1