Potential problems with beetles as palaeoclimatic indicators

While the different methods of palaeoclimatic reconstruction using beetle fossil have been shown to produce accurate estimates of climate by reconstructing the climate of modern assemblages (e.g. Atkinson et. al., 1996; 1997; Coope & Lemdahl, 1996), and by

Figure 1.6. An example of a MLE reconstruction with five taxa. The blue bars represent the known climate envelope occupied by a taxon. The grey tails represent the MLE errors. The pale green region of the figure represents the reconstructed MST for the known ranges while the darker green represents the reconstructed MST taking the MLE errors into account. It indicates an MST of between 16.2°C and 19.9°C.

agreement with other quantitative temperature proxies (e.g. Marra et. al., 2004), some questions have been raised regarding the reliability of fossil beetles as palaeoclimatic indicators (e.g. Andersen, 1993). While these criticisms have typically been levelled at the MCR method, they apply equally to all of the methods of palaeoclimatic reconstruction using beetle fossils.

While some collection studies of modern beetles have obtained distributional data with correlated information on taxon abundance this is by no means standard (Bray et al., 2006), and one of the largest criticisms of palaeoclimatic reconstruction techniques using fossil beetles (e.g. MCR, MLE) is that they typically rely on presence/absence data to establish the climatic range occupied by a beetle taxon (Bray et al., 2006). This is a world wide problem and may result in large errors in the accuracy of a climate reconstruction even when it is based on well known data sets like those of Europe or North America (Huppert & Solow, 2004).

The only way to resolve this problem is extensive sampling of modern beetle populations with recording of species abundance at each collection site. Such work will take decades, and even then will be hampered by anthropogenic modification of the landscape as this will further restrict beetle distribution. In the meantime the development of methods which take into account the incomplete knowledge of species distribution (e.g. MLE (Marra et al., 2004)), and the modification of old methods to include error terms (e.g.

Huppert & Solow, 1993; Bray et al., 2006), are improving the accuracy of palaeoclimatic reconstructions based on fossil beetles. The MLE method of Marra et al. (2004) in particular, appears to have good potential at improving the accuracy, if not the precision, of palaeoclimatic reconstuctions, as it was specifically designed around the assumption that the modern distribution of a beetle does not represent its entire climate range.

The MLE method is not without it flaws as it assumes that the climate range occupied by a beetle species has a unimodal distribution. Unfortunately, this may not be the case in all, or even most, beetle species which often show bimodal (or even trimodal) distributions in climate space (Bray et al., 2006). Until quantitative frequency data is available, however, further development and refinement of methods of palaeoclimatic reconstruction such as the MLE method (Marra et al., 2004) or MCR ubiquity analysis (Bray et al., 2006)) are the best way to improve beetle based palaeoclimatic reconstructions.

A second criticism levelled at the reliability of beetle based palaeoclimatic reconstructions is that the macroclimatic zones to which beetles are assigned do not

accurately reflect the actual climatic tolerance of the species. This is because individual beetles actually inhabit microclimates that differ from the surrounding macroclimate (Andersen, 1993; Bray et al., 2006). For example, high-altitude ground dwelling or burrowing beetles may enjoy a warmer microclimate than indicated by the surrounding air temperature due to the fact that soil temperatures decrease more slowly with altitude than do air temperatures (Andersen, 1993). This criticism has not gone unrecognised by the palaeoentomological community, however, and is already acccounted for in the way that MCR and MLE reconstructions are created.

Both MCR and MLE methods use a ‘broad-brush’ approach to reconstructing palaeoclimate using the entire known geographic range of a species to construct its climate envelope (Atkinson et al., 1986; Coope & Lemdahl, 1996; Marra et al., 2004;

Bray et al., 2006). While it is possible that taxa may inhabit a particular of microclimate within the broad climate envelope indicated by its geographic range, this microclimate will, to a large extent, be determined by the overall macroclimate (Coope & Lemdahl, 1996; Bray et al., 2006). If this was not the case then thermophilous species would not be geographically restricted to lower, warmer latitudes, and cold-adapted taxa would not be limited to high latitudes and altitudes, as stenothermic species would be able to find suitable microclimates in a much wider variety of macroclimates (Coope & Lemdahl, 1996). Thus, while the exact microclimatic tolerance of a beetle species may be unknown, the broad macroclimatic patterns that give rise to that microclimate can be reconstructed using a broad-brush approach like MCR or MLE. This is confirmed by studies which use modern beetle assemblages from sites with known climate to test

whether those modern assemblages can accurately estimate known temperatures (e.g.

Atkinson et al., 1987; Coope & Lemdahl, 1996). Typically the results of these studies are close approximations of the known temperature (e.g. Atkinson et al., 1987; Coope &

Lemdahl, 1996), clearly validating these ‘broad-brush’ methods of palaeoclimatic reconstruction.

While the preferred microclimate of a beetle taxon does not appear to impact broad scale reconstructions of palaeo-macroclimate (Coope & Lemdahl, 1996), the point that beetles tend to occupy a microclimate (Andersen, 1993; Bray et al., 2006) highlights the fact that beetles are constrained in their distribution by more than macroclimatic effects. This means that any palaeoclimatic reconstruction is therefore subject to some degree of error.

It is for this reason, as mentioned above, that phytophagus taxa are typically excluded from palaeoclimatic reconstructions (Atkinson et al., 1987; Elias, 1991; Bray et al., 2006). The exception to the exclusion of photophagous taxa has been the MLE reconstructions undertaken in New Zealand which use as many taxa in the assemblage as possible.

There are two main reasons for the inclusion of phytophagus taxa in New Zealand studies. Firstly, New Zealand fossil assemblages are routinely dominated by herbivourous weevils (Curculionidae) and include few predatory ground beetles (Carabidae) (Marra, 2003; 2007). The lack of ground beetles differs dramatically from fossil assemblages obtained from other parts of the world. These fossils assemblages typically include many species of ground beetle (e.g. Cong & Ashworth, 1996) and they comprise a substantial

proportion of the taxa used in palaeoclimatic reconstructions. The reason for the lack of carabid beetles in New Zealand’s fossil assemblages has been related to two factors.

Firstly, as previously mentioned, carabid beetles occur in extremely low densities in the New Zealand fauna (Marra, 2003), and secondly New Zealand lacks carabid beetles associated with swamps and peats where most fossil beetle assemblages form (Marra, 2003; 2007).

In addition to the lack of carabids and the dominance of weevils, the New Zealand beetle fauna is also particularly poorly understood and many taxa lack good distributional data (pers. observation). Excluding phytophagus taxa, therefore, leaves few species left in an assemblage to reconstruct the climate. It has therefore been necessary to include all taxa with available distributional data, regardless of which tropic level they occupy, in order to produce palaeoclimatic reconstructions from New Zealand fossil assemblages.

The inclusion of phytophagus species in palaeoclimatic reconstructions from New Zealand does, however, raise the question of whether the climatic reconstructions produced reflect the climate at the time of assemblage formation, or the modern day distribution of New Zealand’s native vegetation. While there is no way to be completely sure of the answer to this question, the fact that every climate envelope produced using the MLE method adds a statistical error to the climate range, in order to account for the fact that the full distribution of the beetle is unknown (Marra et al., 2004), may help to alleviate any problems associated with the restricted distribution of New Zealand’s native vegetation.

Furthermore, the MLE method has been shown to produce estimates of temperature (e.g.

Marra et al., 2004) that agree with temperature estimates from other quantitative proxies such as sea-surface temperatures (e.g. Barrows & Juggins, 2005), equilibrium line estimates (ELAs) for glaciers (e.g. Porter, 1975) and from transfer function results from chironomids (Woodward & Shulmeister, 2007) suggesting that it is a robust method of measuring the New Zealand palaeoclimate. The reconstructions presented in this thesis therefore continue to use the MLE method of Marra et al. (2004).

1.5 Vegetation history of the West Coast since the penultimate (Waimea) glaciation