Short Communication |
Corresponding author: Jérôme Pellet ( jerome.pellet@nplusp.ch ) Academic editor: Claudia Buser
© 2023 Jérôme Pellet.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Pellet J (2023) Planning insect surveys in alpine ecosystems. Alpine Entomology 7: 201-204. https://doi.org/10.3897/alpento.7.110958
|
Most biological survey programs rely on multi-species inventories (e.g. birds, amphibians, butterflies, dragonflies). These programs usually rely on multiple visits during pre-defined time windows. The implicit goal of this popular approach is to maximize the observed species richness. Here, we present a novel method to optimize the timing of survey windows using a framework maximizing the detectable species pool. We present a proof of concept using 20 years of entomological records in Switzerland using butterflies, dragonflies, and grasshoppers. The general framework presented can potentially be applied to a wide range of biological survey schemes. It offers a new practical tool for adaptive entomological monitoring under climate change.
Lepidoptera, Odonata, Orthoptera, altitudinal levels, phenology, adaptive monitoring
Standing at the core of complex ecological food webs, insects provide insights into the health and stability of ecosystems. They are thus widely used as bioindicators at local, regional, and international scales (
Many of the ongoing entomological survey programs aim at estimating species richness among taxa. Even though recent technologies (e.g. computer vision, acoustic monitoring, radar, and molecular methods) offer new perspectives (van Klink et al. 2022), visual encounters remain the most widespread approach. This is especially true for several popular taxa that are widely surveyed in alpine ecosystems, such as butterflies/day-flying moths, dragonflies/damselflies, and crickets/grasshoppers, all of which can be readily identified or photographed in the field. Even though these taxa do not contain an overwhelming number of species compared to other taxa, surveying them remains a costly endeavor.
Entomological visual surveys are usually based on repeated visits across the activity period of the focal taxon. This is necessary because individual species fluctuate in abundance asynchronously during a year (the adult activity or flight periods of various species of insect typically only partly overlap within a focal taxon, see
Here, we present a novel approach to identify the best time windows for surveying alpine entomological communities by optimizing the encounter probabilities of every species with as few visits as possible. Using 20 years of observations for three popular taxa, we provide evidence-based, data-driven, guidance for alpine insect survey planning.
We first extracted all observations of Lepidoptera (limited to butterflies and day-flying moths), Odonata (dragonflies and damselflies), and Orthoptera (crickets and grasshoppers) from info fauna, the Swiss biological records center (www.infofauna.ch) for the period spanning 2003–2022. The data was then organized into three matrices (one for each taxon) containing (i) the species name, (ii) the year the observation was made, (iii) the altitudinal levels of the observation, (iv) 52 columns corresponding to the weeks of the calendar year. These weekly columns were then filled with the total number of adult individuals of a given species that had been observed each year at a given altitudinal level.
Species detectability in a given week at a given altitudinal level was first assumed to follow P (Xs,t) ≅ 1 − e−Ns,t, where P (Xs,t) is the probability of detecting species s during week t and Ns,t is the number of observations of species s during week t. That is, the more abundant a species is, the more likely it is that a single individual of that species will be observed. In short, we ended up with an expected number of species being potentially observed at every altitudinal level, week, and year.
Our optimization algorithm then worked through the following steps, iterating years and altitude Xs,t) (i.e. the number of species likely to be detected). For convenience, we tested 5 scenarios representing an increasing number of annual surveys (from 1 to 5). We then used this data to plot the best time windows - from a single week to a combination of 5 different weeks - that maximize the species richness likely observed by an observer.
The first draft of the introduction, discussion, and abstract of this paper has been adapted with
The optimized survey windows for 3 taxa and 3 altitudinal levels are described in Fig.
Optimal time windows to maximize potential species richness in entomological surveys for 3 taxa at 3 altitudinal levels assuming between 1 and 5 surveys each. The mean of the 2003–2022 period is represented with a white dot, the colored bars represent the standard deviation. A single survey aiming at maximizing the potential species richness of butterflies in the lowland (lower left sub-figure) would have to take place between weeks 26 and 30 of the year (first half of July). If two surveys are planned, then they should ideally take place on week 23 (early June ±1 week) and on week 28 (mid-July ±2 weeks).
For Odonata at the subalpine level (top middle sub-figure), a single visit should be made on the last week of July (the white dot representing the median best week). Depending on yearly variability, this best week can span anywhere between mid-July and the end of August. If two surveys are envisioned, then the first one should occur in mid-July and the second one in early August.
As expected, higher elevations translate into later survey windows, the amplitude of the shift being about 2 weeks between the lowland and the subalpine levels. Fig.
Running the algorithm for the 1983–2022 period (data not represented in Fig.
Insect surveys represent technically and logistically challenging operations that can prove costly (
By using a large 20-year-long dataset across multiple altitudinal levels, we closed the loop of active adaptive monitoring, where data collected in the past is used to improve future efforts (
Gerard Maze and Marine Bugnon provided help with the construction of the model and its implementation in both Python (parsing and optimization algorithms) and R (summary statistics and figure production). DO! L’agence prepared the pictograms of Fig.