Hay crop of boreal rich fen communities traditionally used for haymaking

Abstract

The traditional use of fen areas for hay production had been extensive in boreal Europe, but few studies have investigated the hay crop of different fen plant communities. We studied the hay crop from upper boreal (sub-alpine) rich fens using data from 81 permanent plots over more than 30 years in one coastal (oceanic) area and one inland (continental) area in central Norway. Permanent 12.5 m² plots were mown with a scythe every year, every 2nd year or every 4th year. A large majority of the plots under study were lawn and open margin communities classified within or related to the phytogeographical order Caricetalia davallianae. There was no difference in the hay crop between the study areas in these communities, indicating that biomass production is about the same in ecologically similar rich fens that share the same dominant species. The first hay crop (including litter) from lawn communities was about 160 g/m² after 20–30 years of abandonment. Regular mowing every second year reduced the hay crop by more than 30 %, and stabilized it after three mowings. The hay crop decreased with increasing mowing frequency; in lawn and open margin communities, mowing every 4th, every 2nd and every year yielded on average 140, 113 and 65 g/m², respectively. In earlier times, it used to take farmers about 10 days’ work to harvest one hectare. Thus, the traditional practice of mowing every 2nd year was efficient in terms of the hay crop and labour input, and the quality of the hay was improved due to a lower litter fraction.

Appendix

Characteristics of three minimal adequate models (GLMM, normal distribution) developed and used in the analyses of the hay crop (ln [g/m²]). The minimal adequate models were identified through stepwise exclusion of factors and interactions between factors (starting with full models) based on AIC-values and deviance tests. The full models contain the random factors ‘locality’, ‘plot’ and ‘year’ (all three modelling approaches), and the fixed factors ‘study area’ (model 1 and 3), ‘mowing regime’ (model 1 and 2) and ‘community group’ (model 1 and 2) or ‘community’ (model 3). Model 1 has two levels of the mowing regime (mown every 4th or 2nd year) and applies to both study areas, whereas model 2 includes a third mowing regime (mown every year) and applies to the inland study area (Sølendet) alone. Model 3 deals with the hay crop on the community level, the dataset is restricted to plots mown every 2nd year, and the model applies to both study areas. Contrasts between levels of fixed factors (all minimal adequate models) and interactions between fixed factors (model 1 only) are summarized with estimates and accompanying t-values. The estimate of the intercept in model 1 is the scenario fen carpet mown every 4th year at Tågdalen, and the intercept in model 2 is fen carpet mown every 4th year. The intercept in model 3 is community I (see text and Table  2 for details). T – Tågdalen, S – Sølendet, 0.25 – mown every 4th year, 0.5 – mown every 2nd year, 1 – mown every year, C – fen carpet, L – fen lawn, M – fen margin, I–VI – six plant communities (Moen et al. 2012).

Model No.	Study area	n	Random factors	Fixed factors	Contrast	Estimate	t
1	Tågdalen and Sølendet	702	Locality, plot and year	Intercept	–	4.68	22.65
				Study area	T – S	−0.04	−0.27
				Mowing regime	0.25 – 0.5	−0.68	−5.38
				Community group	C – L	0.15	0.74
					C – M	0.55	1.99
				Study area : mowing regime	T:0.5 – S:0.5	0.23	1.85
				Community group : mowing regime	C:0.5 – L:0.5	0.34	2.59
					C:0.5 – M:0.5	−0.02	−0.09
2	Sølendet	654	Locality and year	Intercept	–	4.59	33.01
				Mowing regime	0.25 – 0.5	−0.27	−3.79
					0.25 – 1	−0.71	−9.13
				Community group	C – L	0.37	2.78
					C – M	0.34	2.46
3	Tågdalen and Sølendet (mown every 2nd year)	627	Locality, plot and year	Intercept	–	4.25	24.52
				Community	I–II	−0.53	−1.40
					I–III	0.22	1.08
					I–IV	0.40	2.11
					I–V	0.42	2.14
					I–VI	0.63	2.13