1.3 Landscape complexity

Purpose of indicator

Relevant for the post-2020 Global Biodiversity Framework Goal B on conserving services provided by ecosystems and Target 10 on ‘Proportion of agricultural area under productive and sustainable agriculture’ (CBD/COP/15/L.26). Aligns with SDG indicators 2.4.1 and 15.3.1 on proportion of agricultural area under productive and sustainable agriculture. Maintaining at least 20% natural and semi-natural habitat in agricultural landscapes is important for safeguarding food security, biodiversity, and ecosystem services (Garibaldi et al. 2020).

Key Metadata

Metadata Item	Description
Indicator Name	Landscape complexity
Theme	Biodiversity
SDGs Targeted	SDG 2.4.1, SDG 15.3.1
Data Source	Satellite data, or direct measurement by walking around natural or semi-natural features on-farm with a GPS. In both cases, the field or farm boundary needs delineating.
Measurement	Percentage of farmland area that is covered with natural or semi-natural vegetation.
Measurement Units	Percentage cover, disaggregated by type of feature

Guidance on Measurement

We discuss three options for measurement, i) farmer estimates validated through field surveys (default option in HOLPA), ii) using satellite imagery, iii) direct measurement.

Farmer estimates

A simplistic and rapid way to assess this indicator is to ask respondents to estimate the proportion of their farmland covered by natural or semi-natural vegetation (e.g. bushland, fallow land, hedgerows, natural grassland, ponds, lakes, forest patches). This can be validated by assessing the proportion of natural vegetation contained within three 10x10m plots sampled on the farm, taking a GPS point to allow for satellite-based analysis of the proportion of natural vegetation at larger spatial scales.

Satellite imagery

Natural or semi-natural features on farmland (see Definitions) are often visible on high resolution (<50m and ideally <10m) satellite imagery, such as Google Earth imagery. In a GIS, import the boundary of the field or farm of interest (see Definitions). If no boundary is available, generate a 1km diameter circular plot around each fieldwork plot, using a GPS point collected at the centre of the plot on the ground. In Google Earth Engine, digitise all visible natural or semi-natural features within the area of interest. Calculate the area of land with natural or semi-natural features to obtain an estimate of the share of land covered with natural or semi-natural features. Where possible this should be verified on the ground.

Known problems:

Distinguishing between tree canopies and shade can be difficult.
Determining what is farmland and what is natural land can be difficult.
Distinguishing between grass and shrubs is not always easy.
The condition of the vegetation is an unknown but important variable.

Direct measurement

Natural or semi-natural features coverage can be determined by walking around the field or farm of interest including all natural or semi-natural features, recording the boundaries with a handheld GPS. These areas can be imported into a GIS and used to obtain an estimate of the percentage of farmland covered with natural or semi-natural features.

Definitions

Farmland includes areas delineated with generally linear boundaries and inside which there are one of the following:

Regular rows of woody or herbaceous crops
Visible tractor tracks
Visible rows of tilled soil
Evidence of irrigation, e.g. area is much greener than surrounding land, or visible sprinkler system
Hay bales in field

Natural or semi-natural features on farmland includes linear or areal features on farmland comprising:

Single or small groups of trees or shrubs not regularly arranged within the field. These include woodlots, forest remnants.
Narrow strips of trees, shrubs or other vegetation along field edge that have no visible management, e.g., are not arranged in rows, no signs of tractor or tillage activity, irregular shape. These include hedgerows, flower strips, grass borders.
Other natural or semi-natural features, including long-term (>6 month) fallow land, rivers, ponds.

This list of natural or semi-natural features should be discussed and adjusted with local communities and/or based on expert knowledge of the context.

Guidance on Data Entry and Reporting

Record which natural or semi-natural features were detected, e.g. hedgerows, flower strip. For fallow land, record how long the land has been in fallow.

Calculation Method

Farmer estimates (using HOLPA Survey Data)

  tmp <- main_surveys %>%

    ### ensure required variables are numeric
    mutate_at(
      vars(
        natural_vegetation,
        bushland,
        fallow_land,
        hedgerows,
        grassland,
        ponds,
        forest_patches,
        wetlands,
        woodlots,
        other_land_covering
      ),
      as.numeric
    ) %>%

    ### map diversity variables from survey optiosn to scale of 1 - 5
    mutate_at(
      vars(
        bushland_diversity,
        fallow_land_diversity,
        hedgerows_diversity,
        grassland_diversity,
        forest_patches_diversity,
        wetlands_diversity,
        woodlots_diversity
      ),
      function(x) {
        case_when(
          x == "high" ~ 5,
          x == "medium" ~ 3.66,
          x == "low" ~ 2.33,
          x == "none" ~ 1 #check as HOLPA script says NA rather than 1
        )
      }
    ) %>%
    rowwise() %>%
    mutate(
      ### take the median of each landscape type score
      lc1 = median(
        c_across(c(
          natural_vegetation,
          bushland,
          fallow_land,
          hedgerows,
          grassland,
          ponds,
          forest_patches,
          wetlands,
          woodlots,
          other_land_covering
        )),
        na.rm = TRUE
      ),

      ### take the median of the calculated diversity scores

      lc2 = median(
        c_across(c(
          bushland_diversity,
          fallow_land_diversity,
          hedgerows_diversity,
          grassland_diversity,
          forest_patches_diversity,
          wetlands_diversity,
          woodlots_diversity
        )),
        na.rm = TRUE
      ),

      ### overall landscape complexity is the median of the 2 medians

      kpi7_landscape_complexity = median(c_across(lc1:lc2), na.rm = TRUE)
    )

  ### include the new variable in the performance_indicators data frame
  performance_indicators <- performance_indicators %>%
    left_join(
      tmp %>%
        select(farm_id, owner_id, submission_id, kpi7_landscape_complexity)
    )