Because temperatures are higher or more continuously favourable towards the tropics, it has often been suggested that herbivory (and predation) might be more sustained threats there (Pianka, 1966; Janzen, 1970; Connell, 1971). Belief that high levels of herbivory are an important determinant of plant strategy in the tropics has become widespread amongst ecologists, despite the relative scarcity of evidence for such a pattern. The best data we have on this relationship at present comes from collation of data from previous studies, the biggest of which have been performed by Coley and co-workers. In 1996, Coley and Barone brought together data from 42 studies, finding that average annual leaf damage was 7.1% in temperate sites, compared with 11.1% in shade-tolerant species from tropical wet forests, 48% in gap specialists in tropical wet forests, and 14.2% in tropical dry forests. Similarly, collation of data from seventeen studies showed significantly higher rates of herbivory towards the tropics (Coley & Aide 1991), as did a meta-analysis of the relationship between mammalian winter browsing intensity and latitude across six species of woody plants (Swihart & Bryant 2001). However, even in the most comprehensive of these reviews, the authors urged readers to accept the results only as a “working hypothesis”, and to bear in mind the problems associated with pooling data from studies in which different methods have been employed (Coley & Barone 1996).

Our main aim in this research is to provide a fair comparison of the magnitude of the latitudinal gradient in herbivory. We will do this by using consistent methods to study herbivory at a large number of sites continuously spread along a latitudinal gradient. We will gain information about mechanisms underlying any latitudinal gradient in herbivory through the following questions:

a) How much of the variation in herbivory can be explained by latitude?
We will begin by quantifying the proportion of variation in herbivory that is explained by latitude. However, it is clearly not latitude per se that generates gradients in biological processes or species traits. We will therefore investigate which of the environmental variables associated with latitude (e.g. net primary productivity, rainfall, temperature and soil fertility) are most closely correlated with levels of herbivory.

b) Are there more herbivores per unit of available leaf near the equator?
 Most theory regarding the relationship between herbivory and latitude suggests that greater abundance of herbivores near the equator will lead to higher levels of herbivory. However, present evidence regarding the relationship between latitude and herbivore (or seed predator) abundance is both sparse and inconsistent. One study showed hemipterans to be more abundant closer to the equator, but found no particular relationship between the abundance of Coleoptera and latitude (N. Andrew, unpublished data), and there was no relationship between latitude and the abundance of small mammals trapped over a five-year period in Sweden (Hansson 1992). In this project, we aim to quantify the relationship between invertebrate herbivore biomass and latitude, across a wide range of sites. We will also measure the amount of leaf area that is available to the herbivores at a given latitude, in order to quantify the relationship between the biomass of herbivores supported per unit leaf mass and latitude. Possible outcomes are:

1)   There is more herbivore biomass per unit leaf mass in the tropics. This might come about as a result of the more continuously favourable climate in the tropics.

2) The ratio of herbivore biomass per unit leaf mass remains constant across the latitudinal gradient. This might occur if herbivore biomass was determined primarily by the mass of available food – that is, if increases in plant biomass were matched by increases in herbivore biomass. This pattern would lead us to predict no relationship between herbivory and latitude. In support of this idea, Moran and Southwood (1982) found high consistency in the proportion of species of different guilds on several tree species in two ecosystems.

3)   There is less herbivore biomass per unit leaf mass in the tropics. This might result from greater potency of anti-herbivore defences in tropical environments (discussed below), or from tropical herbivore populations being more limited by the abundance of their predators than by food availability (Hairston et al. 1960).

In order to determine whether increases in physical or chemical defences are correlated with decreases in the value of a given food source to a herbivore, it is important to consider both leaf nitrogen (herbivore reward) and leaf defences (costs to a herbivore of eating a given resource). We will quantify the relationships between leaf defences, leaf nitrogen and latitude, and investigate whether plants receive a greater level of herbivory for a given level of leaf defence per unit nitrogen nearer the equator.

Our working hypothesis is that vertebrate herbivory will become relatively more important at greater distances from the equator, as the ectothermic invertebrates become increasingly restricted by cool temperatures.

We will quantify the proportion of leaf area produced that is lost to herbivores across a large number of sites. In order to maximise the number of sites included (to increase power to detect geographic patterns in herbivory and to determine what environmental variables are most closely correlated with the observed patterns), we will study only the five species most abundant (by foliage cover) at each site. For each species, we will monitor herbivory on expanding and mature leaves by tagging leaves that have recently completed expansion, and leaves that have just begun expansion (methods as for Moles & Westoby, 2000). Tagged leaves will be monitored four times during the year, in order to capture seasonal variation in levels of herbivory, and to ensure that herbivory is not underestimated as a result of unobserved total leaf removal (this is the method recommended by Coley & Barone in their 1996 review of herbivory in tropical forests).

We will quantify invertebrate biomass using pyrethrum fogging (methods as in Kitching et al. 1993), carried out seasonally on standardised volumes of foliage at each site. Samples will be sorted to broad feeding guilds, by a part-time research assistant. Only invertebrate biomass (not number of individuals or morphospecies) will be measured, in order to reduce processing time and thus allow an assay of invertebrate biomass to be made at the full suite of sites. The amount of leaf available at each site will be estimated using leaf area index values (measured with a Licor LAI 2000), combined with monitoring the growth of previously tagged branches to quantify the total biomass of new leaf produced during a year for each of the study species. The relative importance of vertebrate herbivores will be quantified by comparing the amount of herbivory inside vs outside wire mesh cages excluding vertebrates but allowing free access to invertebrates.

We will quantify physical (toughness, water content, specific leaf area etc) and chemical (yet to be determined) leaf defenses. Leaf toughness will be quantified using a leaf fracture toughness tester (methods as in Wright & Cannon 2001). We are still deciding exactly what chemical traits to measure, but we will definitely quantify leaf nitrogen content using a Leco C:N analyser. Nitrogen content gives some indication of the nutritional benefit gained by a herbivore from eating a given amount of leaf tissue, and relates to other aspects of leaf strategy (Cunningham et al. 1999, Reich et al. 1999).