Activity 1: Composition and phenology of the understorey plant community

Wooden quadrats (50 cm × 50 cm) were placed at fixed locations in the understorey of forest stands alongside the Lammi Biological Station Science Trail (central-southern Finland, 61°3.24′N 25°2.23′E, 155 m asl), in places where the plant community is composed of several key understorey species that are widespread, seasonally dynamic, and archetypal of shaded forest habitats. A grid within the quadrats was formed by string at 10-cm intervals, meaning that each quadrat contained 25 10 cm × 10 cm squares. A notebook/data-sheet with a table to fill out and an identification panel were provided at each location (Supplemental File 1: Appendix A: Supplemental Figure 1), illustrating and describing the designated plant species, to allow the participants to correctly identify the species in the quadrats. The worksheet listed the type of observations for each visitor to make, which were based on simply counting the number of plants, leaves, and flowers inside each quadrat on a given day. The worksheet also contained a box for the date, time, and weather (e.g., sunny, overcast, raining, snowing), as well as a place for the observer’s name, plus questions related to biodiversity of the plot in general, and the abundance and phenology stage of individual species. Photos of the focal species, Oxalis acetosella L., Fragaria vesca L., and Hepatica nobilis L., were provided (Supplemental File 1: Appendix A: methods, sample form, and a description of the species’ natural history).

Activity 2: How much light reaches the ground?

A set of identical wooden disks with holes drilled through them were kept in a waterproof box at each station on the trail, where participants could pick them up, together with activity sheets containing a short explanation of the activity (Supplemental File 1: Appendix B: Supplemental Figure 3). Disks had about the diameter of a saucer and the design was inspired by Brown et al. 2000 who suggested the disk be made of solid, rigid, not bendy/breakable material. We used wood because it has the advantage of being more tactile and was consequently preferred by the participants, and it is cheaper than metal. We placed at least one station in each forest stand type to allow a comparison of canopy closure of different tree species.

The participants look vertically up through the disk, with their eye at a set distance from it. To solve the problem of always maintaining the same distance, we fixed a string of 20 cm on one corner of the disk, at the base. To inspect the forest canopy, participants were instructed to hold the end of the string in one hand next to their eye, while keeping the string taught by holding the corner of the disk next to the other end of the string with their other hand (Figure 1b). By standardising the distance between the eye and the disk, any error due to the amount of the canopy seen through the disk was minimised.

The principle behind Activity 2 is to facilitate an easy way to estimate the amount of light penetrating the canopy. This is achieved by dividing the disk into sectors (12, like a clock face), each of which are split transversely in half, making 24 sectors of equal size. The participant counts, and notes down, how many of these sectors are predominantly light and dark, keeping the disk still as they do so (Figure 2). The instructions also are available to the participants as a video (https://youtu.be/M78RrsoIK8o). In practice, it is easier, cheaper, and more effective to have a disk that is solid, apart from 25 small holes of equal size (i.e., 10 mm diameter drilled through the disk), rather than one that is entirely transparent. One hole was drilled in the middle of each of the 24 sectors, so that the holes are equally spaced, plus one through the centre of the disk.

To perform the activity, participants were instructed to look up at the canopy and count the number of holes through which they could see only sky, versus the number of holes with only forest canopy made up of leaves, and the number of holes with only tree trunks and large branches. Those holes containing a roughly even mixture of these elements required a degree of subjectivity to allocate into categories. To reduce this subjectivity, the participants were asked to rotate the disk to obtain the most holes with only sky that they could see from one place. Likewise, it was necessary to control for the angle to the sky at which the disk was held, because looking vertically up gives a much shorter path-length of light through the canopy than looking towards the horizon, so would affect the values obtained. To avoid this type of error, each participant to was instructed to take one measurement looking vertically through the canopy (this will give the maximum sky and minimum trunks), and then four measurements: at 45°, at halfway between the horizon, and the vertical measurement at right angles from each other in each cardinal direction (the equivalent of facing towards each compass point).

To complete the activity, the bud development stage of the canopy trees should be estimated, allowing a connection to be made between LAI and tree phenology. At each station, sketches or a scale using photographs were given for the phenology of the principal canopy trees: birch, poplar, oak, and cherry, and the participants wrote down which developmental stage most of the tree buds had attained (Figure 3).

Data analysis

The datasheets from the two activities were categorised into groups based on their intelligibility as either: incomplete/entirely-nonsensical data, partially complete/intelligible data, or complete/intelligible data. Assessment of the quality of data was based on entries falling within the range of feasible results, and on, for example, proportional data needed to sum 1.0. The fraction of entirely nonsensical data was noted and omitted from the subsequent analyses.

For Activity 1, observations from consecutive days were averaged together when there was only one survey per day. These data were compared with wider surveys by researchers of understorey plant development in the two stands. The results of Activity 2 were compared to calculations of LAI made from hemispherical photos of the site before, during, and after the period of canopy closure, and senescence in the autumn, to assess the consistency of the two approaches. These LAI data were collected by researchers in the CanSEE group as part of a broader research project (Hartikainen et al. 2020).

Activity 1: Composition and phenology of the understorey plant community

Patterns in understorey growth and phenology

Differences between performance of the three species in the spruce and oak stands were visible in the results of the activity summarised in Table 1 and provided in full in the supplemental material. Seasonal patterns in folding and unfolding of Oxalis leaves were recorded, with more unfolded leaves in the spruce than in the oak stand, and leaves were folded more often in early spring prior to spring canpy flush and on sunny days when sunflecks reached the forest floor (Table 1). Fragaria was the only species to favour the oak stand, producing stolons throughout the summer there, but only during autumn in the spruce stand (Table 1). Hepatica plants produced a second annual cohort of leaves only in the spruce stand, which contributed to their greater cover under spruce (4–8 leaves) than in the oak stand (2–6 leaves) (Table 1).

Considering trends in density of the three most common understorey species in quadrats in the oak and spruce stands (Figure 4), Oxalis produced new leaves in the spring of 2017 in both stands (Figure 4a) but it was present at far greater density in the spruce stand. It was possible to identify differences in the phenology of winter leaves in Fragaria, which started to grow about 5 days earlier in the spruce (year-month-day: 2016-09-26) than in the oak stand (2016-10-01) (Figure 4b). It was not possible to distinguish summer 2016 trends in Hepatica growth, but in spring 2017 (2017-05-27), their over-wintered leaves started to senesce (decline in number) in the spruce stand (Figure 4c). Hepatica both produces new leaves and loses leaves to senescence in spring, and together with its relatively low cover, this likely confounded any relationship between the two datasets.

Figure 4 

Plot showing a time series of changes in understorey plant cover of three common species, (a)Oxalis acetosella, (b)Hepatica nobilis, and (c)Fragaria vesca), over one year in spruce and oak stand. Data are median ± 1 SE for each set of recordings. Non-equispaced dates of recordings are given on the x-axis, and a grey vertical dotted line marks the start of the winter period under snow when no recordings were possible. Note that Oxalis in the oak stand are plotted on a secondary axis at a far smaller scale for ease of comparison of the trends. Raw data given in the Supplemental Files 2 and 3.

Activity 2: How much light reaches the ground?

Participation in the activity

In total, the activity was completed 143 times over 50 different days between 2016-06-07 and 2017-06-24. Since multiple school and university groups passed along the trail during this period, Figure 5 includes several days when many individuals completed the activities at the same time. Of the 143 datasheets completed, 18 were adjudged incomplete/nonsensical datasheets (omitted from analysis), 41 were partially complete/intelligible datasheets, and 85 were complete/intelligible datasheets (the latter two sets were included in the analysis). From these data, we calculated daily averages, or averages over a period of observations on 2 to 3 consecutive or nearby dates (15 cases); where datasheets were only partially intelligible, these data were averaged over a period of several days. In total, this gave 24 time-points with measurements of canopy closure, each time-point being a mean average of between 2 and 17 datasheets (the median number was 4 datasheets). Our feedback suggested that some participants struggled to exercise the subjectivity required to choose which category to put the canopy into (i.e., to determine when a mixture of branches and sky was visible through the holes versus only leaves, or only branches, or only sky. The instruction to rotate the disk to obtain the maximum number of holes with only sky was added to help overcome this problem.

Figure 5 

The percentage composition of the canopy viewed through the canopy disk. One year of measurements, twenty-four occasions, at the same location in a Betula sp. stand on Lammi Biological Station Science Trail. (a) Vertical observations looking directly up through the canopy, and (b) “horizontal” cardinal observations, looking at a 45° angle above the horizon in each of the four cardinal directions.

Validation of the approach through controlled tests in multiple sites

Vertical, horizontal, and combined canopy disk data all gave strong positive correlations against summer LAI, but the closest relationship was obtained by weighting vertical and average horizontal readings equally at each measurement point (R² = 0.54; Figure 6a). Comparing the non-leaf canopy disk readings with LAI calculated from photos taken during the winter did not yield a meaningful correlation among deciduous stands. The inclusion of data from the evergreen Picea abies stand did produce a highly significant overall R² but this was only because of the very large difference in LAI between evergreen and deciduous canopies (R² = 0.92; Figure 6b). The slope of this relationship (5.48) was at least similar between the summer and winter (Figure 6), allowing the relationship to be used as a conversion factor from canopy disks to estimate the LAI using our nature trail data.

Figure 6 

Calibration of the canopy disk through comparison of canopy cover calculated from the mean of the vertical and cardinal measurements against the leaf area index (LAI) calculated from hemispherical photographs (according to Hartikainen et al. 2018). Parallel measurements with the two approaches taken from four measurement points in five different forest stands in Viikki Arboretum, Helsinki, in (a) summer and (b) winter.

Patterns in canopy closure compared with hemispherical photo data

The results cover only one full year in the forest stand from June 2016, when spring flush was already almost complete (phenology stage 7), up to the completion of spring canopy flush in June 2017. This allowed us to follow a trend in canopy openness and closure through the autumn and spring seasons (Figure 7). As expected, the vertical observation allowed more sky to be seen than the horizontal observations, so by employing both orientations, participants could capture more of the variability in canopy openness at different times of year.

Figure 7 

Leaf Area Index (LAI) in the Betula sp. stand at Lammi Biological Station over the one-year period, calculated from data collected through the canopy disk activity compared with LAI calculated from hemispherical photographs.

When the annual cycles in LAI in the understorey of the Betula stand at Lammi Biological Station are compared using hemispherical photos and the canopy disk method, with LAI calculated using the calibration from Figure 6a, the relative effectiveness of the methods can be assessed (Figure 7). Some errors or inconsistencies in the canopy disk method are apparent during the transitions through bud burst and autumn leaf fall, but in general the timing of these changes are in agreement between the two techniques. The canopy disk approach is able to capture some of the fine-scale variability in the time course that might not otherwise be visible (Figure 7). In winter, the canopy disk method gave a higher estimate of LAI than our hemispherical photos. This may be because of the weight given to the “mixed” category in winter or because of viewing some ground through the disks. This weighting could be lowered to improve the match to hemispherical photos data, but given that there is also subjectivity in selecting the exposure of the photos, improving the match doesn’t necessarily give a truer result. This most important consideration is to use consistent criteria in the assessment among stands and measurements to enable them to be compared.

Participation in the activities, and their educational value

The Lammi Biological Station Science Trail was originally designed to activate visitors and to promote learning through a participatory approach. In this way, the traditional passive learning approach common to nature trails, where subject matter is presented on interpretive signs, is augmented by tasks that activate learners. There is a great deal of evidence to suggest that active environments promote more effective learning (e.g., Freeman et al. 2014; Deslauriers et al. 2019) and help to narrow achievement gaps among learners (Theobald et al. 2020). By implementing citizen science–based activities on the trail, our aim was to engage learners in scientific fieldwork, helping them to better understand this aspect of the scientific process, while gaining specific knowledge about ecological processes, phenology, and taxonomy.

Since most participants identified themselves on the datasheets, we could determine that school children tended to collect data of better quality than older university or college students, probably due to a higher motivation or to supervision of the activity. Additionally, data collected from volunteers seemed more complete and accurate than data collected by those university students who had been asked to do the activity as part of their duties. This reenforces the suggestion that the level of enthusiam for the activity, and its perceived relevance, as explained directly to the participants, plays an important role in the reliability of the data. A similar conclusion was reached from a study of marine debris that found primary school students to be more effective citizen scientists than secondary school students, who lacked focus and motivation (van der Velde et al. 2017). There is also evidence from studies of citizen scientists feeding birds that young adults find ecological studies rewarding when their participation is time-effective (Martin and Greig 2019). Participants are more engaged when they feel that the results they produce will be of scientific values, as here in climate-change ecological research, and if there is a social or companiable aspect to the activity (Martin and Greig 2019). These factors may all have contributed to the better results that we obtained from younger groups of participants who received more training and interaction than older participants.

The quadrat vegetation survey (Activity 1) allowed the participants to learn to identify plants at different stages of their life cycle. Participants were encouraged to think about how the composition of the plant community depends on the canopy trees and to question which factors are most important for determining plant community assembly (e.g., effects of tree cover on light, on temperature, and on the leaf litter that is recycled into the soil). Activity 1 illustrates to participants how the time of year affects the diversity and condition of understorey vegetation, and how much this changes among sites. In comparison, the canopy observation (Activity 2) draws participants’ attention to differences in the timing and effect of tree leaf-flush phenology over the spring, while also stimulating ideas about how and why light conditions vary among different types of forest. The two activities together provoke contemplation of the consequences of differences in sunlight and canopy cover for biodiversity under each canopy.

Reliability of the data

The quadrat vegetation survey used in Activity 1 is a widely tested method in plant science for the estimation of vegetation cover and species composition (Floyd and Anderson 1987). According to our results, the estimates of growth rates and phenological stages, and their differences among stands by citizen scientists, matched our stand-wide assessments. These data were more consistent with researcher-collected data when using the median of observations rather than the mean, which was biased by outliers. Nevertheless, the trends were quite noisy, and it could be expected that with more participants their reliability would be improved. In contrast, we do not suggest the use of data collected in this way to estimate species composition and abundance as it is difficult to control for monitoring of an area of forest representative of the whole stand.

Participants sometimes had difficulty recognising the target species, especially in absence of their flowers. Additionally, a higher replication (i.e., more quadrats at different points along the trail) would be necessary to obtain more representative results in terms of species composition across a stand. In our activity, we decided to use only two quadrats (one per stand) to avoid making the activity too time-consuming or tedious for the participants. The issue of replication alternatively could be addressed by using larger quadrats, as suggested in previous studies (Archaux et al. 2007), instead of more quadrats. When we tested the activity, we identified locations to place the quadrats, ensuring that they contained all of the target species present inside the quadrat area. While this allowed the volunteers to observe all the species, it might also have biased the results, since quadrats should be placed randomly in the stand in a well-designed study to determine the species composition (Crocker and Tiver 1948; Evans and Love 1957). Nevertheless we favour fixed quadrats, as allowing quadrats to be placed at random by participants would add noise to the results and would require that participants wonder through the understorey, potentially trampling some species.

The canopy disk used in Activity 2 proved to be a valid method to estimate canopy cover and LAI, presenting a good correlation with the estimates calculated by the research group through hemispherical pictures. A similar approach was tested before, and successfully proved to be a simple and reliable method for the estimate of canopy cover (Brown et al. 2000). While several alternative techniques are available, the canopy disk is considerably cheaper, more easily replicable, and more straightforward to use than most of these (Černý et al. 2019; Russavage et al. 2021). While a certain subjectivity is required and could produce a bias, the high number of replications obtained through the involvement of citizen scientists would compensate for individual user subjectivity when all are pooled. Additionally, the simplicity of the method allows untrained volunteers, like families and passers-by, to participate in the activity.

Potential refinements

The discrepancies discussed between the results of these citizen sciences activities and researchers’ findings suggest that there is room for improvement in the instructions and protocols of both activities. Additionally, we might expect that trained volunteers should be able to collect better data as was previously shown in several studies (Feldman et al. 2018). Below, we discuss specific improvements for both activities.

A compromise is required to retain the participants’ interest while achieving high replication. Ideally, to obtain well-replicated data in the quadrat vegetation survey, one station with two to four quadrats should be located at multiple points representing each different canopy type along a trail. However, individuals shouldn’t feel under pressure to survey quadrats at all points. This would allow for a more varied sample that should be more representative of the composition of understorey species in a forest stand. Using this approach, the quadrats at different locations could accommodate some species in common and some that differ to be more representative of the stand as a whole; for example, Oxalis and Hepatica dominate the understorey of the evergreen stands, while deciduous/broadleaved stands, though typically more diverse, may contain Fragaria or Oxalis in the understory.

Recently, various devices for monitoring LAI, and mobile phone apps for canopy photography, have been developed, providing everyday alternatives to the canopy disk method (Černý et al. 2019; Arietta 2022). Using a mobile phone app designed to create a hemispherical image from fixed permanent locations under several forest types has been shown to give good reproducibility (Arietta 2022). This is a functional approach, but lacks the pedagogical aspect of the canopy disk and is subject to the same difficulties in standardisation as traditional hemispherical photography (Ribas Costa et al. 2022; Russavage et al. 2021; Torresan et al. 2021). In comparison, our canopy observation activity provides a cheap, simple, and novel way to estimate the penetration of light to the understorey. A densitoimeter could be provided to use as an alternative approach (e.g., YouTube video, how to use a densitoimeter), although obtaining accurate estimates using densitoimeters is known to be challenging (Applegate 2000; Russavage et al. 2021).