Underground Ecosystem Conservation Through High-resolution Air Monitoring

Abstract

In cave ecosystems tourists represent moving sources of discontinuous disturbances, able to induce transient system responses whose knowledge is crucial in defining appropriate conservation measures. Here we propose an approach to evaluate the amplitude and scales of cave alterations based on high-resolution air monitoring, through the use of purposely developed low-cost monitoring stations and a consistent analytical framework for information retrieval based on time series analysis. In particular, monitoring stations adopt a modular structure based on physical computing platforms acquiring data through several sensors, with means of preventing humidity damages and guaranteeing their continuous operation. Data are then analyzed using wavelet periodograms and cross-periodograms to extract the scales of tourism-induced alterations. The approach has been exemplified in the Pertosa-Auletta Cave, one of the most important underground environments in Southern Italy, highlighting the development of monitoring stations and the information obtainable with the proposed analytical workflow. Here, 2 monitoring stations acquiring data for 1 year at 1' sampling time on temperature, relative humidity, CO₂, VOCs, and particulate matter were deployed in trails subjected to different levels of tourism. In terms of Pertosa-Auletta Cave air dynamics, the approach allowed estimating the temporal and spatial scales of tourism-induced alterations in the order of minutes and meters, respectively, with parameter-dependent variations. On more general terms, the approach proved reliable and effective, with its modularity and low-cost fostering its straightforward adoption in other underground ecosystems, where it can support the development of tailored management strategies.

Keywords: Atmospheric monitoring; Frequency spectra; Show caves; Time series; Tourist load.

Fig. 1

The Pertosa-Auletta Cave system, with indication of the sectional profiles, the natural (always open, but inaccessible to tourists) and artificial (open on-purpose for tourist transit) entrances, the main trails (in different colors), and the locations of the monitoring stations (red stars). Images of the stations with the internal electronics and installed in situ are also shown

Fig. 2

Time series of all the parameters monitored (temperature: T, relative humidity: RH, CO₂, VOC, typical particulate size: TPS, PM1.0, PM2.5, PM4.0, PM10) in the tourist (green) and fossil (red) trails, as well as the number of tourists in the cave. Shaded areas indicate, in the order, the portion of time series detailed in Figs. S2 (extract from the phase of unrestricted tourism), S1 (pandemic-related lockdown) and S3 (extract from the phase of controlled tourism)

Fig. 3

Wavelet periodograms of the time series (from top to bottom: temperature, relative humidity, CO₂, VOC, typical particulate size, PM10, number of tourists) during the first week of August 2019 (left column - from 2019-08-02 04:41:00 to 2019-08-08 04:41:00) and of August 2020 (right column - from 2020-08-02 04:41:00 to 2020-08-08 04:41:00) in the tourist trail. The x-axis indicates the minutes from the beginning of the time series, whereas the y-axis the wavelet periods (in minutes). Due to the similarity among the time series relative to the particulate matter classes, only the periodogram for PM10, as representative of the others, is shown. The wavelet power spectrum is represented on quantile scales, with white lines enclosing regions of significant (for α = 0.05) periods and black lines indicating wavelet ridges

Fig. 4

Wavelet periodograms of the time series (from top to bottom: temperature, relative humidity, CO₂, VOC, typical particulate size, PM10, number of tourists) during the first week of August 2019 (left column - from 2019-08-02 04:41:00 to 2019-08-08 04:41:00) and of August 2020 (right column—from 2020-08-02 04:41:00 to 2020-08-08 04:41:00) in the fossil trail. The x-axis indicates the minutes from the beginning of the time series, whereas the y-axis the wavelet periods (in minutes). Due to the similarity among the time series relative to the particulate matter classes, only the periodogram for PM10, as representative of the others, is shown. The wavelet power spectrum is represented on quantile scales, with white lines enclosing regions of significant (for α = 0.05) periods and black lines indicating wavelet ridges

Fig. 5

Wavelet cross-periodograms between temperature (upper panel), CO₂ (middle panel), VOC (bottom panel), and the number of tourists during the first week of August 2019 (from 2019-08-02 04:41:00 to 2019-08-08 04:41:00) in the tourist trail. The x-axis indicates the minutes from the beginning of the time series, whereas the y-axis the wavelet periods (in hours). The cross wavelet power spectrum is represented on quantile scales, with white lines enclosing regions of significant (for α = 0.05) periods and black lines indicating wavelet ridges. Arrows represent the relative phase of the tourist and the parameter wavelets: wavelets are in-phase in I and IV quadrants and out-of-phase in II and III quadrants, with the tourist leading in the I and III quadrants and lagging in the II and IV. The arrow angle indicates the phase difference between the wavelets of the two series