Skylark
Alauda arvensis is a common and charismatic species of agricultural landscapes. This animal model is one component of the nature
submodel of Persefone.jl.
The model description follows the ODD (Overview, Design concepts, Details) protocol (Grimm et al., 2006; 2010; 2020):
1. Purpose
The purpose of this animal model is to simulate the abundance and distribution of a population of Alauda arvensis in response to farm management in Central European agricultural landscapes.
2. Entities, state variables, and scales
2.1 Landscape
The simulated landscapes consist of a grid of pixels with a resolution of 10m and have an extent of 270km²-370km² (approximately; depending on the chosen input map). Each pixel is assigned a land cover class. It may also be associated with a farm plot, in which case it will contain information about the type and growth stage of the crop planted here. Farm management determines which crops are grown when, and when disturbance (e.g. mowing, harvesting, tillage) takes place.
2.2 Animals
The simulated individuals (a.k.a. agents) are mature skylarks. Each skylark is characterised by the following variables:
ID
A unique identifier for this individual, which can be used to link it to its parents and its offspring.sex
Male or female.phase
The individual's current stage in the annual/life cycle. May be one of:migration
,nonbreeding
,territorysearch
,occupation
,matesearch
,nesting
,breeding
.position
The individual's position in the simulated landscape.mate
The ID of the individual with which this individual has mated this year, if any.territory
A list of coordinates of the positions in the landscape that this individual claims as its nesting and feeding territory.nest
A coordinate giving the location of the currently active nest.clutch
The number of juvenile (i.e. not yet independent) skylarks that this individual is currently raising.
3. Process overview and scheduling
The simulation proceeds in time steps of one day. Every day, each individual executes the function associated with their current life phase:
migration
: The individual is held in a separate data structure (apart from the model landscape) and does nothing until its return date is reached. Then, it is re-introduced to the landscape and assigned the phaseterritorysearch
(for males) ormatesearch
(for females).territorysearch
: Males return first from migration. If they already have a territory from a previous year, they return to this. Otherwise, they move randomly through the landscape until they find a contiguous territory that satisfies their habitat requirements. Once a male has a territory, it changes its phase tooccupation
.matesearch
: Females return later than males from their winter migration. If they already had a partner the previous year, they have a given probability of remaining with this partner. Otherwise, they move randomly through the landscape, looking for a male with a territory and without a partner. Once the female has a partner, it changes its phase tonesting
.
If an individual fails to find a territory or a mate, it changes its phase to nonbreeding
once the breeding season is over.
occupation
: The male moves at random about its territory until the breeding season is over. Then it changes its phase tononbreeding
. (Note: Skylark males actively help with feeding their chicks. However, feeding is only modelled indirectly here, through the process of habitat selection when the male forms its territory - see section 4.1.)nesting
: The female selects a suitable location within the male's territory for the nest. Building the nest and laying eggs takes a number of days, during which she does nothing else. Then, she changes her phase tobreeding
.breeding
: The female checks for mortality. The probability of brood loss varies with the age of the clutch and the nesting habitat. If and when the chicks reach independence (30 days after hatching), they are instantiated as new individuals in thenonbreeding
phase.
If a nest fails due to predation or disturbance, or a brood leaves the nest successfully, the female resets her phase to nesting
and begins again if the breeding season is not yet over. If it is, she changes her phase to nonbreeding
.
nonbreeding
: Non-breeding mature birds move randomly around the landscape, keeping close to other individuals (flocking behaviour). Once their individual migration date is reached, they are removed from the landscape until the following year (see above). Mature birds have a mortality probability for their first summer, and others thereafter for each winter.
4. Design concepts
4.1 Basic principles
This model assumes that the two most important drivers of skylark distribution and abundance are habitat availability and juvenile mortality (see literature below). The factors and processes affecting these are therefore given the most attention in the model, while other factors and processes are only included superficially, indirectly, or not at all. Specifically, this means that the phases territorysearch
, nesting
, and breeding
are the most relevant and detailed parts of the model, as these determine the selection of habitat and the survival of offspring.
Furthermore, the model concentrates on predation and anthropogenic disturbance (through management actions such as mowing) as the main causes of juvenile mortality. Other causes, such as hunger or bad weather, are currently ignored as they are usually not significant.
The focus on habitat availability and juvenile mortality opens up two avenues by which agricultural management influences skylark populations. First, the farmers' choice of crops and date of sowing determines the quality of the habitat when skylarks select a territory. (For example, unlike summer grain, winter grain is already so high and dense in spring that it is generally avoided for nesting.) Secondly, the frequency and timing of management actions (especially mowing) is a major cause of brood loss. This means that there are direct causal links between agriculture and population trends.
Concentrating on these two drivers allows the rest of the model to be kept simple, reducing both the scientific complexity and computational costs. Thus, foraging movement (both during and after the breeding season) can be ignored or represented as random movement, as it does not directly impact either of the drivers. Likewise, chick growth and winter migration are represented very simply.
4.2 Emergence
Multiple patterns emerge from the basic principles outlined above. The most important are listed here:
- Territory size and population density: The model assumes that skylarks occupy only as much area as they need to satisfy their nesting and foraging requirements, and that population size is limited by the amount of available habitat. This means that territories in high-quality habitat are smaller than in low-quality habitat. Scaling up, this leads to a pattern whereby population densities are highest in open landscapes with a diversity of crops, grassland, semi-natural habitat, and lower in landscapes with low habitat diversity or many woody features.
- Ecological traps: Jenny (1990) describes a strong ecological trap effect whereby skylarks avoid winter grain crops, preferentially nesting in more open grassland sites. However, the mowing frequency associated with modern agriculture means that nest loss in grassland is almost assured, since there is insufficient time between two mowing dates to raise a brood. This means that landscape composition leads skylarks to breed in habitats that have a high mortality, resulting in population declines.
4.3 Adaptation
In the model, skylarks primarily adapt to their surroundings by choosing suitable territories. These are chosen by evaluating the quality of surrounding habitats for breeding and foraging, and occupying as much area as needed to satisfy requirements (see section 7.1).
4.4 Objectives
Skylarks' main objective in the model is to have sufficient habitat available to raise a brood. Habitat quality is calculated as a function of habitat type, vegetation height, vegetation cover, and distance to vertical structures (see section 7.1).
4.5 Learning
The model includes no learning by individuals.
4.6 Prediction
The model includes no predictions by individuals.
4.7 Sensing
Skylarks can perceive the landscape structure in a given radius around them (habitat type, vegetation height and cover). They can also see nearby conspecifics and are aware of the territories claimed by other individuals. When mating, they recognise whether another individual already has a mate, and mated individuals share information about their territory and brood status.
4.8 Interaction
The model includes two direct forms of interaction. First, during mating, females move around the landscape looking for males who have a territory but no mate yet. Once they have found one, the two individuals set each other as their mate. Secondly, after the breeding seasons, individuals move around the landscape, keeping close to other individuals in their vicinity (flocking behaviour).
There are also indirect interactions, in that there is a competition for habitat (territory that has been claimed by one male cannot be occupied by another) and males (males that have mated with one female will not mate with another in the same season).
4.9 Stochasticity
Stochasticity is used when modelling mortality and movement. Predation mortality is modelled as an age- and habitat-dependent probability, while migration mortality is a simple probability. Dispersal movement (when searching for a territory or a mate) is modelled as a random walk, as it is assumed that skylarks are not significantly impeded in their long-range movement by habitats that are unsuitable for foraging or nesting. Foraging movement by the male and by non-breeding individuals is also random, as it is desirable to show movement (to help model analysis) but unimportant to model this exactly.
Persefone.jl includes a seed
parameter, which is used to initialise the random number generator. This can be used to ensure reproducibility (simulation runs of the same model version with the same parameter values will be identical in outcome). The model saves all input files used for a simulation run alongside the run's output data, so that all run's can be repeated if necessary.
4.10 Collectives
After the breeding season, skylarks move around in loose agglomerations (flocking behaviour). However, this has no relevant ecological effect.
4.11 Observation
The model collects three sets of data. The first set gathers daily abundance data, listing the number of individuals currently alive in each life phase. The second gathers information about each nesting attempt, with the date, habitat choice, and territory size. The third records the cause of death of all animals that die during the simulation. All data are saved as CSV files, with several figures automatically created from these at the end of the run.
5. Initialisation
At the beginning of a model run, pairs of skylarks are created on grassland and agricultural land, keeping a distance of 60m to vertical structures and allowing each pair approximately 3ha of suitable habitat (an average territory size in agricultural landscapes).
For details, see the source code and the associated documentation.
6. Input data
The general input to Persefone (i.e. land use maps and weather data) is described here.
The following extract from the source code lists the species parameters and values used by the Skylark model, based on the literature cited below:
@species Skylark begin
# juveniles
eggtime::Int64 = 11 # days from laying to hatching
nestlingtime::Int64 = 9 # days from hatching to leaving nest
fledglingtime::Int64 = 21 # days from leaving the nest to independence
eggpredationmortality::Float64 = 0.03 # per-day egg mortality from predation
nestlingpredationmortality::Float64 = 0.03 # per-day nestling mortality from predation
fledglingpredationmortality::Float64 = 0.01 # per-day fledgling mortality from predation
firstyearmortality::Float64 = 0.38 # total mortality in the first year after independence
# migration
migrationdeparture::Tuple{AnnualDate,AnnualDate} = ((September, 15), (November, 1))
migrationarrival::Tuple{AnnualDate,AnnualDate} = ((February, 15), (March, 1))
migrationdelayfemales::Day = Day(15)
migrationmortality::Float64 = 0.33 # chance of dying during the winter
# habitat requirements
minimumterritory = 5000m² # size of territory under ideal conditions
limitterritory = false # limit the territory sizes to a diameter of `movementrange`?
mindistancetoedge = 60m # minimum distance of habitat to vertical structures
maxforageheight = 50cm # maximum preferred vegetation height for foraging
maxforagecover = 70 # maximum preferred vegetation cover for foraging (%)
nestingheight = (15cm, 25cm) # min and max preferred vegetation height for nesting
nestingcover = (20, 50) # min and max preferred vegetation cover for nesting (%)
offfieldnesting = true # allow nesting on unmanaged grassland
# breeding
matefaithfulness = 0.5 # chance of a female retaining her previous partner
nestingbegin::Tuple{AnnualDate,AnnualDate} = ((April, 10), (April, 20)) # begin nesting in the middle of April
nestbuildingtime::UnitRange{Int64} = 4:5 # 4-5 days needed to build a nest (doubled for first nest)
eggsperclutch::UnitRange{Int64} = 2:5 # eggs laid per clutch
nestingend::AnnualDate = (August, 15) # end of the nesting period
# mating / non-breeding
movementrange::Length = 500m
visionrange::Length = 200m
flockingdistance::Length = 30m
end
7. Submodels
7.1 Territory formation
In the territorysearch
phase, male skylarks look for a breeding territory upon returning from migration. If they still have a designated territory from the previous year, they occupy this again. Otherwise, they take random steps through the landscape until they find a territory (one step per day of movementrange
length). Each step, they inspect the landscape in concentric circles around their location, adding landscape pixels to their putative territory until they have accumulated a sufficiently large effective habitat area. The area that each pixel contributes to the effective area is weighted by its forage quality, which is calculated as a function of its openness (bare ground is best, quality decreases as vegetation height and cover increase). If the required effective habitat area cannot be reached because of other territories in the surrounding, or an insufficiently large contiguous block of habitat, the search breaks off and is repeated in a different location the next day.
7.2 Juvenile mortality
Juvenile mortality comes from two different sources: predation and disturbance (i.e. management). Predation is an age-dependent constant probability applied daily before chicks become independent, then one-off for the first summer upon independence. (Skylarks are only instantiated as code objects at independence, therefore applying first-year mortality as a one-off probability means that only those need to be instantiated that will survive the year, this saves processing time.) Apart from the constant background mortality of predation, the management actions of tillage and harvest also lead to juvenile death (100 % probability). Except for the first-year mortality, which is individual-specific, all other causes of juvenile mortality affect the whole brood, i.e. the whole nest is lost at once.
8. Testing & Validation
The only critical parameters we found were nestingheight
and nestingcover
, as these determine where skylarks can build their nests. The original values tested here (15–25 cm and 20–50 %) resulted in rapid population loss, as the crop models predict values outside this range for most of the breeding season. After another look at the literature, we therefore raised the maximum nesting height to 45 cm, as different studies report different values here. We also raised the maximum nesting cover to 100 %, as AquaCrop and ALMaSS both predict very high canopy cover proportions very quickly, which would have precluded any nesting otherwise. (In addition, there were concerns about methodological mismatches between the values generated by our crop models and those measured in empirical studies.)
Two other parameters we changed from the initial settings where offfieldnesting
and limitterritory
. We disabled the former because we were observing a lot of skylarks nesting in areas that were not under agricultural management in the model; something that we judged to be a modelling artefact rather than a real ecological effect. We enabled the latter to curtail the creation of very large territories (>40 ha) in a few cases; however, a closer inspection of the literature showed that these sizes can in fact be reached, so we turned this setting off again.
With these described parameter adjustments, and good calibration of the AquaCrop model, the skylark model reproduced a range of complex ecological patterns without further changes needed.
9. References
Bauer, H.-G., Bezzel, E., & Fiedler, W. (Eds.). (2012). Das Kompendium der Vögel Mitteleuropas: Ein umfassendes Handbuch zu Biologie, Gefährdung und Schutz (Einbändige Sonderausg. der 2., vollständig überarb. und erw. Aufl. 2005). AULA-Verlag
Delius, J. D. (1965). A Population Study of Skylarks Alauda Arvensis. Ibis, 107(4), 466–492.
Donald et al. (2002). Survival rates, causes of failure and productivity of Skylark Alauda arvensis nests on lowland farmland. Ibis, 144(4), 652–664.
Glutz von Blotzheim, Urs N. (Ed.). (1985). Handbuch der Vögel Mitteleuropas. Bd. 10. Passeriformes (Teil 1) 1. Alaudidae - Hirundidae. AULA-Verlag, Wiesbaden. ISBN 3-89104-019-9
Jenny, M. (1990). Territorialität und Brutbiologie der Feldlerche Alauda arvensis in einer intensiv genutzten Agrarlandschaft. Journal für Ornithologie, 131(3), 241–265.
Jeromin, K. (2002). Zur Ernährungsökologie der Feldlerche (Alauda arvensis L. 1758) in der Reproduktionsphase [Doctoral thesis]. Christian-Albrechts-Universität zu Kiel.
Püttmanns et al. (2022). Habitat use and foraging parameters of breeding Skylarks indicate no seasonal decrease in food availability in heterogeneous farmland. Ecology and Evolution, 12(1), e8461.