Description

Due to their small size, most oceanic island native forest areas have to be monitored by using preferentially high spatial resolution imagery (<10 m), although medium spatial resolution (<30 m) could be useful for some geophysical variables in island-scale mapping (e.g. soil, lithology). Consequently, the ESA Copernicus Sentinel 2 Mission (high-resolution multispectral imaging), which consists of a multispectral imagery mission with 13 spectral bands and resolutions of 10, 20 and 60 m and has a swath width of 290 km, may constitute a cost-effective (as data are fully free-of-charge), adequate and reliable remote sensing data source for supporting forest monitoring processes in islands. Despite the frequent high degree of cloudiness that characterizes many oceanic islands throughout most of the year, this mission will mitigate the impact of this issue by offering a high repeat cycle of six days at the equator and three days in mid-latitudes in the two-satellite configuration (Sentinel-2A and Sentinel-2B, respectively). Therefore, an island-scaled vegetation mapping program with high spatial (minimum mapping unit of 2500 m2) and temporal (quarterly update) resolutions with an adequate classification scheme (at the species or community level when possible) is needed to address forest characterization and monitoring issues. In terms of the main methodological approach to follow, the first phase, corresponding to the vegetation baseline map production, may consist of a classic supervised classification-based procedure, as follows:
(1) Pre-Processing: Atmospheric correction (Hadjimitsis et al. 2010) and geometric correction (Sertel et al. 2007);
(2) Processing: object-based classification (Blaschke 2010) or per-pixel classification (Lu &
Weng 2007); and
(3) Post-Processing: accuracy assessment (Foody 2002).
The second phase, to be repeatedly performed on a quarterly basis, will consist of
a multi-temporal change detection analysis (Singh 1989; Coppin et al. 2004) among the
successively acquired Sentinel-2 spectral data, in order to locate and identify changes in
forest areas.
Finally, in the third phase various spectral vegetation indices (Glenn et al. 2008)
shall be determined and compared by using the multi-date Sentinel-2 data, especially
within changed forest areas mapped in Phase 2, in order to identify the change causes
and/or trends (e.g. phenology, forest cut/clearance, biological invasion, soil erosion,
natural hazard, land-use change).

PROTOCOL 3 - Whole plot plant species survey (adapted from NETBIOME- ISLANDBIODIV project)
Taxa - All vascular plants.
Experimental design - Sampling the 50 × 50 m plots.
Frequency - This protocol is time consuming and adequate only for a first inventory.
However, some repetition can be performed every 5 or 10 years.

Description
Part A - Complete species survey
Floristic list of all the vascular plant species (ferns + phanerogams) within the plot.
The floristic list should be prepared alongside the basal area measurement (see below).
If necessary, additional time should be spent to complete the floristic list, with special
attention to microsites. Specimen samples (e.g., leaves, flowers, inflorescences, etc.) will
be collected for difficult taxa for their latter identification in the lab. Samples are stored in
plastic bags and properly labelled. After returning to the lab, samples are pressed with
drying paper in a ventilated drying oven at 50 °C, and finally deposited in a herbarium. If
the field expedition takes several days, specimens can be dried in the field using a metal
drying frame and a gas stove.
Because many species are annual or deciduous, in temperate islands inventories should be done in the spring or summer. Additionally, for most species, this is the flowering season. Two people should participate in the survey doing independent lists in order to ensure that every species is accounted for. Special care must be taken when checking for epiphytes or semi-parasites because that may be on higher branches or tree tops and are easy to miss.

Part B - Basal area and canopy height
Basal area [m2 wood at breast height (approx. 1.30 m)] of each individual tree is determined from the diameter at breast height (DBH). Results can be expressed as m2/ha, for each species present and for the whole forest community. Diameter at breast height (DBH) may be measured directly, using a diameter tape or a caliper, or indirectly by measuring the tree circumference (perimeter) with a cloth ruler (DBH is then estimated by dividing the perimeter by π). There are many situations, especially in montane cloud forests, where measuring DBH is not straightforward (e.g. tree forks below breast height, leaned trees, tree splits into several trunks close to ground). In such cases, there are rules to be followed (e.g. International Society for Arboriculture, http://www.isa-arbor.com/education/resources/educ_TreeOrdinanceGuidelines.pdf). DBH is measured for each tree or trunk (in multi-stemmed growth habits) with DBH > 10 cm. Each individual tree must be marked with a spot of degradable green paint at breast height (to avoid double measuring). In very wet environments, like montane cloud forests, using paint may not be possible. The alternative is to remove a slice of the usually dense epiphytic layer that covers the trunks. This “mark” may still be visible after 2 years but epiphytic species will gradually re-cover the bare trunk.
For canopy height, five measurements of maximum tree height should be taken: one at a distance of five meters from each corner and one at the center of the plot. Tree height may be estimated visually or measured using instruments (such as a Blum Leiss altimeter). Canopy height in the plot is defined as the mean height of the canopy based on the five measurements described.

Part C - Species density
On each corner, and in the center, of the main (2500 m2) plot a 5 m × 5 m subplot is delimited. Every shoot with a DBH > 1 cm is counted in order to determine the density of shoots per species (expressed as the mean number of shoots per square meter and per hectare). For ferns, when possible, the number of adult individuals of each species may be counted (density is expressed as the number of individuals per square meter and per hectare). If individual determination is not possible (i.e. when a mat of a particular species covers an area with individual fronds emerging from the ground) a possible solution is to count the number of mats of each species.

Part D - Species regeneration
At each corner, and in the center, of the main (2500 m2) plot a 2 m × 2 m subplot is delimited. In these subplots the number of basal sprouts, seedlings and saplings of woody plants (trees, shrubs and woody lianes) are counted. For every seedling and sapling, identify the species and record maximum height, number of leaves and vitality (on an ordinal scale).

SAMPLING EFFORT: One day per plot with two researchers.

PROTOCOL 4 - Population monitoring of rare plant species
Taxa - All.
Experimental design - Subsampling within the 50 × 50 m plots.
Frequency – This survey protocol can be implemented within both short-term (seasonal variation) and long-term studies. This protocol is adequate for yearly sampling.
This count-based approach should be used for species of special interest, usually those with very small number of populations and few individuals per population (i.e., close to extinction), which cannot be monitored adequately using other standardized protocols (see Protocol 3).

Description
Monitoring plots should be set up that include all existing individuals. Every single individual should be marked using GPS and followed up on a yearly basis. For rare trees and shrubs (such as Euphorbia stygiana or Taxus baccata in the Azores), this can be done
by rangers or amateur botanists. For short-lived herbs or annual species, this is more challenging and time-consuming and even setting up the plots correctly will often take several years because the number of individuals can vary greatly (e.g., Myosotis azorica in the Azores). The best time of the year for the fieldwork is species-specific (usually corresponds to the flowering or fruiting periods but sometimes the autumn leaf colour change period may be appropriate) and needs to be adapted accordingly. It is important to recognize that not all island systems to which this protocol may be relevant have an autumn. They may, however, have seasonal variation and so there may be optimal times for sampling. So, I think some modification even if just in parenthesis to acknowledge that other seasonal variation modes exist to the four temperate zone seasons. For each plant individual, performance parameters should be recorded and for each population the threats should be assessed (e.g. herbivore grazing, invasive plants, herbicide usage or cutting of trees).
It is important to mention fresh water/riparian communities, which are usually of high biodiversity on islands, where they occur. Their components are good indicators of water quality (Luis et al. 2015).
In regular intervals (5–10 years, depending on generation time), leaf samples should be collected in zip-lock plastic bags with silica gel for assessment of genetic diversity within and between populations.

References
Blaschke T (2010) Object based image analysis for remote sensing. ISPRS-J Photogramm Remote Sens 65:2-16. doi:10.1016/j.isprsjprs.2009.06.004
Coppin P, Jonckheere I, Nackaerts K, Muys B, Lambin, E (2004) Digital change detection methods in ecosystem monitoring: a review. Int J Remote Sens 25:1565-1596. doi:10.1080/0143116031000101675
Foody G (2002) Status of land cover classification accuracy assessment. Remote Sens Environ 80:185-201. doi:10.1016/S0034-4257(01)00295-4
Hadjimitsis D, Agapiou A, Papadavid G, Themistocleous K, Retalis A, Michaelides S, Chrysoulakis N, Toulios L, Clayton C (2010) Atmospheric correction for satellite remotely sensed data intended for agricultural applications: impact on vegetation indices. Nat Hazards Earth Syst Sci 10:89-95. doi:10.5194/nhess-10-89-2010
Gil A, Fonseca C, Lobo A, Calado H (2012) Linking GMES Space Component to the development of land policies in Outermost Regions - the Azores (Portugal) case-study. Eur J Remote Sens 45:263-281. doi:10.5721/EuJRS20124524
Glenn EP, Huete AR, Nagler PL, Nelson SG (2008) Relationship between remotely-sensed vegetation indices, canopy attributes and plant physiological processes: What vegetation indices can and cannot tell us about the landscape. Sensors 8:2136-2160. doi:10.3390/s8042136
Luis L, Bergamini A, Sim-Sim M (2015) Which environmental factors best explain variation of species richness and composition of stream bryophytes? A case study from mountainous streams in Madeira Island. Aquat Bot 123:37–46. doi: 10.1016/j.aquabot.2015.01.010
Singh A (1989) Digital change detection techniques using remotely-sensed data. Int J Remote Sens 10:989-1003. doi:10.1080/01431168908903939