A low‐cost <scp>IoT</scp> network to monitor microclimate variables in ecosystems

Microclimate affects all organisms. This includes both plant and animal communities, among animals, ectotherms endotherms (Kearney & Porter, 2020; Porter Gates, 1969). Monitoring the climatic conditions experienced by arthropods, plants pathogens is central to contribute a better understanding of life cycle dynamics these organisms (Rebaudo et al., 2016). Arthropods are ectothermic organisms, with little no active thermoregulation. Thus, ambient temperature main driver their body (Angilletta, 2009; Cossins Bowler, 1987). Arthropod important as almost biological processes driven temperature, from behaviour population through mortality, fecundity, development 1987; Sporleder 2017; Uvarov, 1931). Like an environmental factors involved in controlling growth (Chuine Régnière, Zaka 2017). Together relative humidity, also pathogen application early warning models (Garrett 2021). In cases, heterogeneity variables time space has been highlighted implications (Caillon 2014; Faye 2016; Saudreau Despite numerous studies describing mismatch between macroclimate data such WorldClim (Hijmans 2005) most species ecosystems (Bennie Buckley 2013; Haesen 2021; Hannah Kearney Lembrechts 2019), modelling efforts downscale temperatures (Haesen 2016), lack microclimatic empirical impeded that can be transposed finer scale If spatial challenge, same goes for temporal resolution acquisition which generally lower than desired study response variables. A applied butterfly Pieris napi L. (Lepidoptera: Pieridae) demonstrated microclimate led smaller prediction error observed compared only if sampling intervals were very small, is, acquired every 15 30 min (Schmalensee account small changes would affect history traits cannot represented mean because nonlinear form relationship traits, known Jensen's inequality (Colinet 2015; Schmalensee valid including responses applies just much arthropods (Ruel Ayres, 1999) or vertebrates (Pickett 2015). at higher frequency reduce measured values practical limitations. To capture within plot, thermal imagery employed ground unmanned aerial vehicle (UAV) success highlight importance crop canopy agricultural pests (Faye 2016, substantial decreases price UAVs associated cameras (Bossoukpe Giménez-Gallego 2021), using obtain images sufficient might too expensive cases. Also, downside UAV desirable knowing change significantly over short period (15 required example insects, see Even flights automatized, they reach phenology while minimizing errors An alternative use classical loggers, have widely used agrosystems other well, but cost remains prohibitive wide adoption, especially needs centralized real on server. promote open science, here we suggest existing technologies build low-cost Internet Things (IoT) stations macro- living It complete system, client visualization extraction analysis model building. Using monitor nothing new, several solutions recent years based either microcontrollers (Mickley 2019) single card computers developed specific objective (Sethi 2018) broader applications ecology (see review Raspberry Pi biologists Jolles, However, our knowledge, covered whole life, visualization, taking into local constraints diverse situations power devices transfer information While focus key features, extended materials methods found Supporting Information section contain reproduce IoT monitoring network, together links pointing up-to-date software codes. The purpose this evaluate whether system context ecological scientific environments (France, Kenya, Bolivia). These drastically different characterized seasons contrasting averages amplitudes. therefore locations interest testing applicability system. Our arthropod agroecosystems. network allowing objects connect exchange standard protocols (Hittinger Jaramillo, 2019). advantage Global System Mobile communications (GSM) able access cost. obstacles adoption remain technical complexity deployment energy dependence. Here, propose meet challenges related optimizing Data server PostgreSQL database (any object-relational alternative). interaction made possible thanks Representational State Transfer Application Programming Interface (RESTful API) easy data. Linux NGINX Open Source web (pronounced “engine X”; suitable heavy traffic its asynchronous management opposed traditional servers Apache) Node.js JavaScript runtime Express framework provide features (Mardan, 2018). Thanks API, delivered Javascript Object Notation (JSON) format gateways Internet. gateway defined device two systems formats. Sensors directly connected gateways. sensors microcontroller units (MCUs) equipped Long Range (LoRa) Radio Transceiver communication radio frequencies (Figure 1a). LoRa telecommunications protocol low speed (as high offered GSM/cellular networks), long distance distances Bluetooth, Wi-Fi, Zigbee, Z-Wave). With IoT, makes it take locally free low-speed global level high-speed via 2). solution not depend subscription plans transmission (apart Internet), commercial networks Sigfox LoRaWan. RFM95W modules HOPE Microelectronics CO. designed so place Rapsberry Pi-based near any building outlets connection, ESP32-based surrounding fields Wi-Fi range powered batteries solar panels, Arduino-based MCUs (called PI2Puino rest text) distant rechargeable panels 1b). Transceiver. resulting topology star intermediate relays. Environmental depending level. For MCU communicating (PI2Puino, Figure 1b), sensor (DS18B20; accuracy ±0.5°C) humidity (DHT22; ±0.5°C ±2%RH), ESP32 outlet panel, pressure (Adafruit module BME280/680; ±3%). Gateways light SI1145 visible, Ultra Violet, Infra Red light). Although chosen accuracy, genericity community users, added substitute those presented here. sent embedded antenna lightweight Secure encryption algorithm (SIT) keep consumption minimum (Usman reason does get after sending data, stored Electrically Erasable Programmable Read-Only Memory (EEPROM) duration configurable user again next reading loop. fixed sleep_enter function Arduino code, 100 cycles corresponding approximately S1). Between readings, enters sleep mode limit consumption. code was written sketch IDE 1.8. (a program). AA short-term usage panel 3.7 V 2600 mAh Lithium-Ion battery (together lithium-ion charger voltage regulator: S1 detailed description). Once received module, checks integrity (numerical value) forwards JSON-encoded message. forwarded Python 3.9. case unhandled exception, python script relaunched automatically systemd unit. RF95W works similar way, forwarding key. electric outlet, EP32-based battery. order minimum, enabled reads. defined. version Examples provided S1. Sequelize Relational Mapping (ORM) tool, allows abstract software. retrieved RESTful API. Any accessed identification number, results returned JSON format. documentation Swagger available https://pi2p.ird.fr/documentation/api/. Here number 57 request last records curl (https://curl.se/) UNIX terminal, response: 1 https://pi2p.ird.fr/sensors/57/datas?page_size=2 2 3 {"metadata":{ 4 "link":{"first":"https://pi2p.ird.fr/sensors/57/datas?page=0&page_size =2","next":"https://pi2p.ird.fr/sensors/57/datas?page=1&page_size =2","last":"https://pi2p.ird.fr/sensors/57/datas?page=12242931& page_size=2"} 5 }, 6 "data":[ 7 {"id":24521090,"sensorId":57,"value":27.413593,"createdAt":"2022‐07‐15 T14:05:44.707Z","updatedAt":"2022‐07‐15T14:05:45.035Z"}, 8 {"id":24521068,"sensorId":57,"value":27.576876,"createdAt":"2022‐07‐15 T14:04:39.833Z","updatedAt":"2022‐07‐15T14:04:40.163Z"} 9 ]} equivalent R programming language httr package (Wickham, 2022): library("httr") # 1.4.3 r <‐ httr::GET("https://pi2p.ird.fr/sensors/57/datas?page_size=2") resp httr::content(r) print(data.frame(resp$data)) id sensorId value createdAt updatedAt 24521143 27.33977 2022‐07‐15T14:07:56.121Z &#x2026; 24521119 27.34836 2022‐07‐15T14:06:49.580Z production quick overview preferred way should done however convenient check trends series dynamic rich tool browser. We vis.js edition browser library (https://visjs.org/; vis-network) Plotly Graphing Library (https://plotly.com/javascript/; 3a,b). displayed dynamically each sensor, sorted institution gateway. deployed France, Kenya Bolivia (temperate climate Paris tropical Andean mountain El Alto Bolivia, Nairobi Kenya). gathered more (from January 2020 November 2022) selection three then WorlClim (Fick Hijmans, 2017), addressing term (monthly) differences macro microclimate. day, calculated maximum average boxplot monthly distributions, well extracted reference. Regarding devices, estimate 93€ gateway, 75€ 44€ basis unit purchases (Table 1). Collected sites aggregated perform comparison 4). aspects may subject future developments, note configured send back event connectivity failure side. missing February 2021 Kenyan site confirm problem failure, August 2022 exemplified side 4b). cope problems, collected manually. contrast, save few days (depending acquisition), store term. addition, ESP32-type charged continue operate absence sunlight holds. At one item min, autonomy experimentally darkness. follows due location (microclimate), where daily expected transferred successfully period, minor loss outages 5; 55 out 942, i.e. 94% 1,649,460 readings). API language. could greatly reduced larger orders. Due strong fluctuations component prices, amounts estimates. total solutions. estimation comparable Onset Computer Corporation configurator (https://www.onsetcomp.com/hobonet-configurator; Micro RX Station AC (1264€), without radiation shield attached (208€), remote HOBOnet (299€) costs 1771€, against less 200€ necessary integrate labour construction, maintenance (e.g. changing batteries, updating server), warranty, after-sales service product follow-up. ±0.25°C solution. ±2% ±2.5% condition specifically ecosystems, adapted requiring (temperature, atmospheric pressure, light), extra wind sensors, soil moisture). attempt quality 1; Maclean discussion measurements), did subscription-based Internet) consequence electricity require alternatives. operating field kilometres line sight, physical hills dense vegetation will (about 25 meters forests). internet impossible, storage envisaged, USB SD card, constraint longer having time. maintain feature locations, GSM 2:5G (high speed, traffic), LoRaWan Roe NarrowBand-IoT (reduced rate, uses others, fully compatible implement volume transmitted, networks, coverage area operators. when areas Wi-Fi. size 1) whose cover sunshine reduced, latitudes winter under partial cover. As mentioned introduction, variable vast literature covers agriculture (Guo Zhang specificity contribution source covering (Tzounis offer choice options, strategies availability. Compare offering services acquisition, availability Corporation, Campbell Scientific), offers users flexibility adapt components needs. control transparency storage, formats facilitate reuse many people possible. integration type new full management, providing lowest cost, disadvantage deployment, appendix implementation result, proposed measurements stages collection processing, projects budgetary resources represent constraint. capacity academic use, confirms possibility agreement (Jolles, limited functionality solutions, flexibility. Measurements minute carried Bolivia. problems encountered 5) hardware, rather adjustment stable reliable latter overcome manually unloading device. target insects high-level sensitivity association crops damage feeding yield losses. allow us understand insect pest populations essential agroecosystems temperate inter-tropical zones. frequency, conditions. reference, highlights quantifies discrepancy François Rebaudo conceived ideas methodology; Rebaudo, Bruno Condori, Reinaldo Quispe-Tarqui, Paul-André Calatayud, Soledad Chavez Vino Henri Tonnang data; Titouan Soulard Lucas Bessière methods; writing manuscript. All authors contributed critically drafts gave final approval publication. work ANR-19-CE32-0001 PI2P project supported French National Research Agency (ANR). part PRIMA program European Union. thank help Kenya: Bonoukpoé Sokame, Julius Obonyo Glenn Sequiera ICIPE-Nairobi, Béatrice Elesani ICIPE-Muhaka, Antony Odiwuor Owuor ICIPE-Mbita, Jefferson Joel Mwamburi Dembwa, Jenipher Nyambura Mwadime Mjomba Maktau, Baptiste Régnier IRD. Bolivia: Franz Callizaya, Beatriz Ortiz Marco Antonio Pari Fundación PROINPA, INNOVA-AF funded IICA FIDA installing Cintis (https://innova-af.iica.int/). farmers individuals who plots connection. They Editor, Associate Editor anonymous reviewers constructive comments, helped improve conflict declare. peer article https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/2041-210X.14062. codes paper supporting S1, GitHub repository (https://github.com/frareb/pi2p/) Zenodo al. (2022); https://doi.org/10.5281/zenodo.7479887). Codes distributed GNU AFFERO GENERAL PUBLIC LICENCE Version file repository. https://pi2p.ird.fr/ https://frareb.github.io/pi2p_docu/. https://pi2p.ird.fr/documentation/api/ result dataverse https://dataverse.ird.fr/dataverse/pi2p updates once year. S1: Detailed methods. Please note: publisher responsible content supplied authors. queries (other content) directed author article.