Phenology and abundance in relation to climatic variation in a sub-arctic insect herbivore–mountain birch system

Abstract

The two forest-defoliating geometrid moth species Operophtera brumata and Epirrita autumnata are known to exhibit different altitudinal distribution patterns in northern birch forests. One possible explanation for this is that altitudinal climatic variation differentially affects the performance of two species through mismatching larval and host plant phenology. We explored this hypothesis by investigating the relationship between larval phenology and leaf phenology of Betula pubescens, which is the main host plant of both moth species, along ten replicate altitudinal transects during two springs with contrasting climate in northern Norway. There was a distinct monotonous cline in host plant phenology with increasing altitude in both years of the study, but the development of the leaves were generally 14 days later in the first of the 2 years due to cold spring weather. We found that larval development of both species closely tracked host plant leaf phenology independent of altitude and year. However, at the time of sampling, E. autumnata was approximately one instar ahead of O. brumata at all altitudes, probably reflecting that E. autumnata has faster early instar growth than O. brumata. The abundance of O. brumata was lowest at the altitudinal forest-line, while E. autumnata was lowest near sea level. Our results do not indicate that the altitudinal distribution patterns of the two moth species is due to any phenological mismatch between larval and host plant phenology. We suggest rather that natural enemies at low altitudes limit larval survival and thus abundance of E. autumnata, while an early onset of winter at the forest limit reduces survival of late eclosing adults of O. brumata.

Appendices for electronic archive

1.1 Appendix 1

At all transects stations a 50 m long sampling line was drawn horizontally on each side of the station. Vegetation cover variables for dominating species of the tree-, shrub- and field layers, and occurrence of streams and boulders was recorded at each meter along the sampling lines (see Table 2 for a description of variables). Additionally, at each station and at the two ends of the sampling line stand density, canopy height, snow depth and slope were recorded. Long-term average snow depth was measured by the mean height position of the lichen Parmelia olivacea on birch stems (Sonesson et al. 1994).

Table 2 Codes and description for variables used to characterize the forest along the altitudinal transects

In the PCA of forest characteristics, two principal components, PC1 and PC2, accounted for 44.3% of the variance between sampling stations (Fig. 8). PC1 was mainly associated with tree height and accounted for 29.3%, while PC2 was associated with snow level and accounted for 15.0%. Tree height was positively correlated with tree density, presence of deadwood, high perennials and litter, and negatively correlated with presence of bushes and moss in mats. Snow level was positively correlated with slope and presence of blueberry, and negatively correlated with peat moss, moss in tufts, heather and grass.

Fig. 8

The association between the original forest characteristics variables and the two first principal components derived from the PCA. For description of variables see Table 2. The unit circle indicates the maximum correlation that could be observed between each of the PC and the original variables. The variables with arrows close to the circle are very well explained by the first two PC axes

1.2 Appendix 2

Results of the multinomial modelling of instar structure, where the different models are ranked according to the two model selection criteria Akaike’s Information Criterion (AIC) and Bayesian Information Criterion (BIC). The models with the strongest support according to AIC/BIC weights are considered to be the most appropriate (i.e. the models in bold for both species). Terms included in different models are indicated with a X. Alt_q: altitude as a continuous variable; Alt_P2: second order polynomial for Alt_q; Alt_c: Altitude as a categorical variable. Np Number of parameters. The model Date: (Alt_q+Year) fitted both O.brumata (Pearson’s chi square=19.39, df=23, P=0.68) and E. autumnata (Person’s chi square=28.95, df=23, P=0.18; excluding one observation with a very large residual, but not affecting model parameters estimates)

Species	Alt_q	Alt_P2	Alt_c	Date	Year	AltP2:Year	AltP2:Date	Alt_c:Year	Alt_c:Date	Date:Year	Date:Year:Alt_c	Np	AICc	BIC	AIC weigths	BIC weights
O. brumata	X	X		X	X		X					26	0	0	0.32	0.84
	X	X		X	X							14	21.78	4.2	0.00	0.10
	X	X		X	X	X	X					30	1.48	7.3	0.15	0.02
	X	X		X	X		X			X		32	-0.82	8.0	0.48	0.02
			X	X	X					X		22	14.05	8.2	0.00	0.01
			X	X	X							16	24.99	10.3	0.00	0.00
			X	X	X			X		X		28	12.02	15.0	0.00	0.00
			X	X	X				X			34	6.99	18.7	0.01	0.00
	X			X	X							12	39.42	18.9	0.00	0.00
			X	X	X			X				22	26.74	20.9	0.00	0.00
			X	X	X				X	X		40	6.6	27.1	0.01	0.00
			X	X	X			X	X			40	10.78	31.3	0.00	0.00
			X	X	X			X	X	X		46	4.39	33.7	0.04	0.00
				X	X			X	X	X	X	64	19.67	75.4	0.00	0.00
E. autumnata
			X	X	X			X		X		28	0	0	0.99	0.53
			X	X	X					X		22	9	0.2	0.01	0.47
	X	X		X	X		X			X		32	19.54	25.4	0.00	0.00
			X	X	X			X				22	36.99	28.2	0.00	0.00
				X	X			X	X	X		46	21.53	47.9	0.00	0.00
				X	X				X	X		40	33.16	50.7	0.00	0.00
	X			X	X							12	89.89	66.4	0.00	0.00
	X	X		X	X							14	92.15	71.6	0.00	0.00
			X	X	X							16	92.52	74.9	0.00	0.00
	X	X		X	X	X	X					30	76.34	79.3	0.00	0.00
			X	X	X			X	X	X	X	64	36.99	89.8	0.00	0.00
	X	X		X	X		X					26	106.58	103.6	0.00	0.00
			X	X	X			X	X			40	94.18	111.8	0.00	0.00
				X	X				X			34	119.13	127.9	0.00	0.00