The measurement sites are located in north-east Germany in the federal state of Brandenburg. Two sites were selected according to 1) the homogeneity of tree species composition within the forest stands, 2) the size of the ecosystem patches (fragment), 3) the cardinal direction of their zero line (edge), 4) the age and management of the forest, 5) the homogeneity of the surrounding landscape, 6) the management of the agricultural land, 7) the distance to Müncheberg, Germany and 8) the willingness of farmers, forest managers and land owners to cooperate. According to this framework, the following specific criteria were set and evaluated using geographic information systems (GIS) to prepare the selection:

1. continental, temperate forest with coniferous trees (comprise ca. 75% in the region) resulting in pine (Pinus sylvestris L.) and larch (Larix decidua) (it was not possible to find sites with the same main species composition that fit all other criteria at the same time),
2. >10 ha total area of the forest patch and more than 500 m of forested area perpendicular to the zero line (edge) of the forest,
3. one east-facing and one west-facing edge to avoid too strong effects of the cardinal directions north and south,
4. managed, planted forest with an age of trees between 40 and 60 years,
5. a minimum of additional wind-breaking landscape elements and flat terrain within 500 m of the measurement sites,
6. agricultural land cropped for more than 10 years prior to the study,
7. less than 50 km distance to Müncheberg to remain operable,
8. approval to conduct all measurements from June 2016 to September 2017.

We identified an east-facing (the arable land is east of the forest) site in Ihlow, Germany (52°37'23.8"N, 14°03'40.4"E) and a west-facing site in Elisenhof, Germany (52°29'37.1"N, 14°11'03.4"E). Both sites are located in the cold temperate area with summers colder than 22°C on average and 4 months where the temperature is above 10°C (Dfb; Peel et al. 2007). See Fig. 3 for an examplary picture of a forest edge. Besides the main tree species composition, the edge was characterised by grassland as well as scattered shrubs (bird cherry (Prunus padus) at the east facing site) and deciduous trees (birch (Betula pendula) at the west-facing site). The zero line is the abrupt change in ploughed arable land to the unmanaged forest soil.

Figure 1.  

Cross-sectional scheme of the measurement design for both sites. The height for litterfall and biomass sampling is not scaled.

Figure 2.  

Topview on to the measurement design for east-facing site. The west-facing site is mirrored. The width is only scaled to the forest inventory plots. The symbols for weather stations, litterfall and biomass sampling are offset.

Figure 3.  

Examplary picture of the forest edge at the west-facing site. View towards north.

The microclimate was measured in 105 m long transects with five weather stations – one at the zero line (edge), two in the arable land (see Fig. 4 for the setting) and two in the forest (Fig. 1, Fig. 2). The measured variables were air and soil temperature as well as moisture, wind speed and direction, air pressure, precipitation and solar radiation. The microclimatic data are given separately for the west-facing site (Suppl. material 1) and the east-facing site (Suppl. material 2).

Figure 4.  

Setting of weather stations in the arable land at the east-facing site. View towards south. Picture taken in autumn.

The measurements started on 15 June 2016 and ended on 17 July 2017. The distances were chosen based on a previous literature analysis (Schmidt et al. 2017).

The following sensors were used (see Table 1 for specifications):

Biomass of crops

In 2016, at four plots of 1 m² each and at 0 m, 7.5 m, 15 m, 30 m distance from the zero line, the aboveground biomass was harvested manually (Figs 1, 2, 5) approximately one week before the harvest by the farmers (12 July). In 2016, at the east-facing site, peas were grown and at west-facing site oil-seed rape. In 2017, another biomass harvest was conducted with the same design. On 3 July, approximately 10 days before the official date, winter barley was harvested at the east-facing site and winter wheat at the west-facing site. After chopping the plants, they were oven-dried for 48 h at 65°C and the dry mass was weighed thereafter (Suppl. material 3).

Figure 5.  

Delayed flowering of rape in the transition zone to the forest at the west-facing site. View towards south. Picture taken in summer.

Figure 6.  

Setting of weather stations in the forest (zero line, -35 m and -70 m) with rain gauges.

Figure 7.  

Setting of litter fall traps at the zero line (edge).

Stem diameter and height of trees

At the end of April 2016 within the east-facing site and at the beginning of May within the west-facing site, we measured the stem perimeter at 1.3 m height above the ground and the height of the trees (Suppl. material 4; hypsometer: Haglöf Vertex IV). Per site, we defined three plots with a width of 80 m parallel to the zero line and a depth (length) of 20 m perpendicular to the zero line. The plots were 0-20 m, 50-70 m and 130-150 m away from the zero line (edge). As tree density was different at both sites, we measured every fourth tree in a range of 5 m within the east-facing (A, Fig. 8) and every third tree in a range of 7 m within the west-facing site (B, Fig. 8) resulting in n=30 per plot.

Figure 8.  

Design for tree inventory with the dots as examplary trees, the arrows as walking direction during measurements and the circles as the range of counting the trees with 5 m range in A and 7 m range in B.

Litterfall

Litterfall was measured with buckets (see Fig. 7) of 52 cm inner diameter resulting in an area of ca. 0.21 m² per bucket trap fall. Per distance (0 m, 35 m and 70 m; Figs 1, 2), n=10 litter fall traps were used. Sampling was conducted from September 2016 to May 2017 at five dates (13th Sept., 10th Oct., 7th Nov., 10th Jan., 22nd May). After each sampling, the needles of larch (Larix decidua) at the west-facing site and pine (Pinus sylvestris L.) at the east-facing site were oven-dried for 48 h at 65°C and, subsequently, dry mass was determined by weighing (Suppl. material 5). The samples also comprised small amounts of other biomass.

The soil was sampled at two depths – 20±3 cm and 40±3 cm – along the transects (Figs 1, 2). The samples were sieved at 2 mm, air dried and analysed (n=3) for total soil carbon content and total soil nitrogen content (see Suppl. material 6). All samples were carbonate-free as tested in the field using hypochloric acid (see also pH in Suppl. material 6). Soil types were determined in the field according to WRB (FAO 2014; Table 2) using a soil drill (1 m length).

The microclimatic data from all ten weather stations were merged (see Suppl. material 7 for the R script). Moreover, the original header contained non-unicode symbols which had to be transformed. Data had to be rearranged as two weather stations were mixed up after ploughing and had to be replaced to their original positions. Further, data was cut according to the official starting and end points of the experiment and pauses due to management activities of the farmers (west-facing: 12/07/2016 till 25/10/2016; east-facing: 13/07/2016 till 19/08/2016 and 13/09/2016 till 24/10/2016).

We deleted two outliers for precipitation (error in measurements shortly after installation and the measurement date was wrong) and one for relative humidity (relative humidity of 4.8 not possible) from the whole data set.

The data logger (EM-50G) used the time zone of the plugged device. Due to that and the switch of standard time during the year, we had to unify the time to Central European Time (CET) manually.