**Research methods**

In the years up to 2010, the team investigated the floral and vegetative components of the test plots using methods derived from the Braun-Blanquet methods [25]. On each sub-plot of the roof, a complete species list was produced by noting the percentage of the area covered by each vascular higher plant genus. Additional information included the total coverage value of the indicator group of sedum and grasses.

The green roof turf mats on building 3 were a pre-grown professional turf layer with a mixture of grasses and sedum species. Such industrially produced mats have been used in many comparable projects throughout Germany. The plant species development is a role model for many typical extensive green roofs. On building 3, one question examined how these well-suited vegetation layers performed compared to green roofs started with sedum cuttings only, as an alternative approach to provide a cheaper vegetation layer on flat roofs. In the following text, the west, north, and south sides with turf mats are compared to the east side with the sedum cuttings (see Figure 2a,b).

(**a**) (**b**)

**Figure 2.** (**a**) Building 2, May 2001: The turf mats were placed on the north plots, as was done on the west and south plots. The growing media can be seen in the background: light red Op, dark in the front: Ulo. (**b**) Sedum cutting area on the east roof in August 2002: Right: Ulo, sedum developed slowly, left: Sedum and several spontaneously sown grasses covered the Op successfully. Maintenance work on the weather station caused some open areas in the vegetation cover.

> In contrast to this, a vegetation layer completely based on sedum cuttings and grass seeds was constructed on building 2 in 1999.

> The test plots on building 3 were divided in half, with one half using the professional Op growing medium (Optima-extensive media) and the other half Ulo (Ulopor—expanded slate 2–11 mm sizes) growing medium. Op, similar to the Zi (Zinco growing media on building 2), is a professional growing medium that satisfies all the growing media requirements of FLL 2018 – Standard [24] in regard to the water retention capacity, grain size distribution, and basic fertilizer. On building 2, the Blä substrate was used as a test. This is simply crushed expanded clay as a recycled product with no additional nutrition, merely being a supplement for extensive green roofs. The same applies to Ul (Ulopor). This expanded slate has good performance in terms of its water retention capacity, but its dark color means that it heats up considerably in summer, much more than the typical

gravel layer of 16/32-mm stones. On the east side the vegetation reached the maximum coverage of 80% on Op and 75% on Ulo in 2002, the second year. This was ultimately a comparison between the different layers to learn more about the long-term performance of the materials for improving the growing media mixtures and to use the results to contribute to the updates of the FLL guideline [24].

#### **Harvest on 10-cm media**

In 2017, from the growing media test plot Zi, Blä, Op, Ulo, and Op-2, all above-ground and root mass was harvested, with three replicates for each test plot. All phytomass material was divided into the following groups: above-ground vascular plants shoots, roots, mosses, and lichens. The material was dried to a constant weight. The total carbon was 50% of the dry mass. The CO2 concentration was calculated by multiplying with 3.65 to provide a guide value [26–29].

#### **Harvest on 30-cm media**

On building 2, the 26 planter boxes on the roof terraces were used as phytomass test plots. Using scissors, the above-ground phytomass was harvested in 2011, 2012, 2013, 2014, 2015, and 2017. The similar 13 boxes on building 3 were treated in the same way in 2013, 2014, 2015, and 2017. The oven-dried material allowed an initial estimate of the annual growth rate on these materials.
