4
Electronic supplementary material - Appendix
Detailed methodology
The 18 islands that we utilized include nine that had been invaded by rats and nine that had never been invaded. These islands are off the north-eastern coast of New Zealand [1]. Those who either own the islands or serve as kaitiaki (guardians) of them include various local iwi and organizations (Ngāti Hako, Ngāti Hei, Ngāti Manuhiri, Ngāti Paoa, Ngāti Puu, Ngāti Rehua and Ngātiwai, and the Aldermen Islands Trust and the Ngāmotuaroha Trust) and individuals (John McCallum, Bryce Rope and the Neureuter family); others are administered by New Zealand’s Department of Conservation. They vary in size between 3 and 163 hectares.
The rats colonized the nine invaded islands between 150 and 50 years ago [2]. Densities of burrows created by seabirds during nesting were on average 24 times greater on the nine rat-free than the nine rat-invaded islands [1]. We used islands that are as similar as possible in their geographic features; it has previously been shown that the nine invaded islands and the nine uninvaded islands do not differ significantly in latitude, longitude, area, elevation, distance to mainland, or the distance to the nearest island [1]. Previous work on these islands, in which each of the 9 invaded and uninvaded islands serves as a statistically independent replicate, have provided evidence free of confounding factors that rat invasion in this system causes reduction of soil fertility [1], populations or biomasses of many soil invertebrate groups [1, 4], losses of soil organic matter [5], and foliar nutrition and litter decomposability of most common plant species [3]. These effects override any positive effects that rats, or the process of seabird consumption of seabirds by rats, may exert on soil fertility. Further details of these islands are given in [1].
On each island we sampled within a 20m × 20m area in well-developed secondary forest on each island, in the vicinity of plots used for previous studies [1, 3-5]. In this area we took approximately 30 soil samples each about 1L to 10 cm depth; previous work has shown that over 80% of the soil biota on these islands occurs in the 0-10 cm depth layer [1]. These samples were bulked to provide one sample per island. For the purposes of this study, the 18 islands were arranged into 9 pairs of two islands (one invaded and one non-invaded island per pair) with pairing performed on the basis of latitude and island size; this pairing is identical to that used in earlier work on these islands [3].
Each soil sample was returned to the laboratory, homogenized and loosely crumbled, and divided into four subsamples each subsample being subjected to one of four sterilization treatments for determination of the effect of soil biota (i.e., bacteria, fungi, microfauna, mesofauna and small macrofauna) on plant growth through established methodology [6, 7]. As such, the four subsamples should have the same abiotic properties, and should differ only in the direct and indirect effects of death of soil biota caused by sterilization or addition of soil biota caused by reinoculation. These four sterilization treatments are:
(i) Soil not sterilized, i.e., resident soil biota left intact.
(ii) Soil sterilized by g-irradiation (25 kGy), an approach that is widely used in soil feedback experiments for killing all soil biota while causing minimal disruption to the soil environment [7].
(iii) Soil sterilized as for (ii) but then re-inoculated with non-sterilized (live) soil from the same island containing resident soil biota, by mixing sterilized and non-sterilized soil at a ratio of 19:1 [7].
(iv) Soil sterilized as for (ii) but then re-inoculated with non-sterilized soil from the other island of the same pair (so that sterilized soil from each invaded island was re-inoculated with live soil from a non-invaded island and vice versa), by mixing sterilized and non-sterilized soil at a ratio of 19:1.
If soil biota is important in regulating plant growth then plants will grow least in treatment (ii). If soil biota is more important in promoting plant growth on non-invaded than on invaded islands then plants will grow more when sterilized soil is inoculated with live soil from a non-invaded island than with soil from an invaded island (‘biotic pathway’ in Fig. 1). This will be equally true for reinoculation of sterilized soil from invaded islands (where plants will grow more in treatment (iv) than treatment (iii), and for reinoculation of sterilized soil from uninvaded islands (where plants will grow more in treatment (iii than treatment (iv)). If in contrast soil biota is unimportant in determining plant growth and indirect effects of rat invasion on plant growth occur solely through alteration of soil abiotic characteristics (‘abiotic’ pathway in Fig. 1), then we would expect plants to grow better in soil from non-invaded than invaded islands islands for each of the four treatments (i), (ii), (iii) and (iv), but no difference in plant growth among these four treatments for either non-invaded or invaded islands.
Following sterilization treatment, approximately 1.8L of each of the four soil subsamples was placed in each of two pots of 14.5 cm top diameter and 11.0 cm bottom diameter, and of 11.0 cm depth (yielding 8 pots per island) and left to equilibrate in abiotic conditions for 12 weeks before planting. At planting one of the two pots from each sterilization treatment was planted with a seedling of the tree Melicytus ramiflorus, and the other with a seedling of the tree Kunzea ericoides; seedlings at the time of planting were approximately 24 weeks old and 2 cm tall. These two species were selected because they occur commonly on the islands but differ vastly in their physiological traits, with Melicytus having traits associated with rapid resource acquisition (i.e., large, less defended and more nutritious leaves) and Kunzea having traits associated with greater resource conservation (electronic supplementary material, table S2). The experiment was then left in a glass house for 240 days before it was destructively harvested. The glasshouse conditions were exactly as described in [8]; the air temperature was within 5 ° C of the ambient temperature (range: – 4 ° C to + 33 ° C), pots were watered as needed (normally every 3 – 10 days), and incident light reduction by the shade house was 30%.
For each seedling at harvest, the root system was thoroughly washed, and the plant material then separated into root, stem, and leaf tissue. The total leaf area for the seedling, and the oven dry weight for each of the three tissue types (60oC, 72h), was then determined. Specific leaf area was determined as the ratio of total leaf area to total leaf weight.
Each response variable was analyzed by split plot ANOVA as described in [3] with island invasion status as the main plot factor, tree seedling species as a subplot factor, and sterilization treatments as a sub-subplot factor; interaction effects were tested among all combinations of factors and pairs of islands served as the units of replication. When ANOVA results were significant at P = 0.05, Tukey’s honestly significant difference test was used to assess significance of differences among means. If soil biota was to have some role in influencing plant growth then we would expect ANOVA to yield a significant effect for the sterilization factor, and could then explore the nature of this effect by using Tukey’s test to compare the means of the four sterilization treatments ((i), (ii), (iii) and (iv)); if soil biota was to have no effect on plant growth (meaning that only abiotic factors were important) then the ANOVA would yield a non-significant effect for the sterilization treatment and Tukey’s text would not be required for comparisons of means . For all data analyses, data was transformed as needed to satisfy assumptions of ANOVA.
4
References cited in detailed methodology
1. Fukami, T., Wardle, D. A., Bellingham, P. J., Mulder, C. P. H., Towns, D. R., Yeates, G. W., Bonner, K. I., Durrett, M. S., Grant-Hoffman, M. N. & Williamson, W. M. 2006 Above- and belowground impacts of introduced predators in seabird-dominated island systems. Ecol. Lett. 9, 1299–1307.
2. King, C. M. Editor. 2005. The handbook of New Zealand mammals. Oxford: Oxford University Press.
3. Wardle, D. A., Bellingham, P. J., Bonner, K. I. & Mulder, C. P. H. 2009 Indirect effects of invasive predators on plant litter quality, decomposition and nutrient resorption on seabird-dominated islands. Ecology 90, 452–464.
4. Towns, D.R., Wardle, D.A., Mulder, C.P.H., Yeates, G.W., Fitzgerald, B.M., Parrish, G.R., Bellingham, P.J. Bonner, K.I. 2009. Predation of seabirds by invasive rats: multiple indirect consequences for invertebrate communities. Oikos 118, 420–430.
5. Wardle, D. A., Bellingham, P. J., Mulder, C. P. H. Fukami, T. 2007 Promotion of ecosystem carbon sequestration by invasive predators. Biol. Lett. 3, 479–482.
6. Kulmatiski, A. & Kardol, P. 2008. Getting plant-soil feedback out of the greenhouse: experimental and conceptual approaches. Progr. Botany 69, 449–472.
7. Kardol, P., Bezemer, T. M. & van der Putten, W. H. 2006.. Temporal variation in plant-soil feedback controls succession. Ecol. Lett. 9, 1080–1088.
8. Bellingham, P. J., Walker, L. R. & Wardle, D. A. 2001. Differential facilitation by a nitrogen fixing shrub during primary succession influences relative performance of canopy tree species. J. Ecol. 89, 861–875.